Research in Progress
in the Howard Hughes Medical Institute, 1992-1993
Founded in 1953 and incorporated in Delaware, the Howard Hughes Medical Institute is a nonprofit scientific
and philanthropic organization. It is dedicated to the promotion of human knowledge within the basic sciences,
principally medical research and medical education, and the use of such knowledge for the benefit of humanity.
The Institute is directly engaged in medical research and is qualified as a medical research organization under the
U.S. tax code. It is an equal opportunity employer.
Howatxl Hughes Medical Institute
6701 Rockledge Drive
Bethesda, Maryland 20817
(301) 571-0200
After February 8, 1993:
Howard Hughes Medical Institute
4000 Jones Bridge Road
Chevy Chase, Maryland 20815-6789
(301) 215-5000
Developing eye of the common fruit fly Drosophila meianogaster, as imaged by a computer-based
confocal microscope. Red and yellow represent the highest concentrations of Notch, a cell-surface
protein that is necessary for the cell-to-cell interactions that decide the developmental fate of each
cell within the eye.
From Pehon, R.G., Johansen, K., Rebay, I., and Artavanis-Tsakonas, S. 1991. J Cell Biol
1 13:657-669. Copyright permission by the Rockefeller University Press.
Copyright © 1992 by the Howard Hughes Medical Institute
Ail rights reserved. Address correspondence to
Dr. W. Maxwell Cowan
Vice President and Chief Scientific Officer
Howard Hughes Medical Institute
Printed in the United States of America
Howard Hughes Medical Institute
Research
in Progress
1992-1993
Contents
xix Foreword
Purnell W. Choppin, M.D.
xxi Trustees, Officers, and Principal Staff Members
xxiii Medical Advisory Board
xxiv Scientific Review Board
xxvi Locations of Howard Hughes Medical Institute Laboratories
xxix Introduction
xxix Cell Biology and Regulation Program
XXXix Genetics Program
xlvi Immunology Program
liv Neuroscience Program
Ixi Structural Biology Program
W. Maxwell Cowan, M.D., Ph.D.
1 Electrical Activity of Nerve Cells
Paul R. Adams, Ph.D.
3 Three-Dimensional Macromolecular and Cellular Structure
David A. Agard, Ph.D.
5 Molecular Mechanisms of Ion Channel Function
Richard W. Aldrich, Ph.D.
7 Divergent Members of the SRY Family of Transcriptional Regulators
Bind an Insulin-Responsive Element, IRE-A
Maria C. Alexander-Bridges, M.D., Ph.D.
9 Genetic Mechanisms Involved in the Generation of the
Antibody Repertoire
Frederick W. Alt, Ph.D.
11 Control of Cell Fate During Vertebrate Neuronal Development
David J. Anderson, Ph.D.
13 Cell Fate Choices in the Nervous System and Elsewhere
Spyridon Artavanis-Tsakonas, Ph.D.
15 The Complement System
John P. Atkinson, M.D.
1 7 The Molecular Biology of Smell
Richard Axel, M.D.
19 Mammalian Developmental Genetics
Gregory S. Barsh, M.D., Ph.D.
21 Cell Cycle Control
David H. Beach, Ph.D.
Hi
Contents
23 Genetic Control of Morphogenesis in Drosophila
Philip A. Beachy, Ph.D.
25 Molecular Studies of Human Genetic Disease
Arthur L. Beaudet, M.D.
27 Molecular Genetics of Diabetes Mellitus
Graeme I. Bell, Ph.D.
29 Development of the Drosophila Peripheral Nervous System
Hugo J. Bellen, D.V.M., Ph.D.
31 Genetic Manipulation of Hematopoietic Stem Cells
John W. Belmont, M.D., Ph.D.
33 Proteins of the Spectrin- based Membrane Skeleton
G. Vann Bennett, M.D., Ph.D.
35 TNF and the Molecular Pathogenesis of Shock
Bruce A. Beutler, M.D.
37 Cytotoxic T Lymphocyte Recognition
Michael J. Bevan, Ph.D.
39 Vascular Endothelium in Inflammation and Metastasis
Michael P. Bevilacqua, M.D., Ph.D.
41 Structural Studies of Molecules Involved in the Immune Recognition
of Infected Cells
Pamela J. Bjorkman, Ph.D.
43 Mechanisms of Insulin Action
Perry J. Blackshear, M.D., D.Phil.
45 Intracellular Protein Traffic and Nuclear Organelles
Giinter Blobel, M.D., Ph.D.
47 Immunity and Pathogenesis of Third World Diseases: Leprosy
and Tuberculosis
Barry R. Bloom, Ph.D.
49 Molecular Biology of the Extracellular Matrix
Jeffrey F. Bonadio, M.D.
51 Functional Heterogeneity in CD4-bearing T Lymphocytes
H. Kim Bottomly, Ph.D.
53 Retroviral Replication
Patrick O. Brown, M.D., Ph.D.
55 Regulation of Cellular Processes by Protein- Tyrosine Phosphorylation
Joan S. Brugge, Ph.D.
iv
Contents
57 Computational Structural Biology
Axel T. Briinger, Ph.D.
59 Biophysical Studies of Eukaryotic Gene Regulation and
Molecular Recognition
Stephen K. Burley, M.D., D.Phil.
61 Molecular Studies of Calcium Channels and the Dystrophin-
Glycoprotein Complex
Kevin P. Campbell, Ph.D.
63 Gene Targeting
Mario R. Capecchi, Ph.D.
65 Genetic Control of Pattern Formation in Drosophila
Sean B. Carroll, Ph.D.
69 Human Disease Gene Identification and Correction
C. Thomas Caskey, M.D.
71 Enzymatic RNA Molecules and the Structure of Chromosome Ends
Thomas R. Cech, Ph.D.
73 Molecular and Cellular Physiology of Acute Inflammatory Cytokines
David D. Chaplin, M.D., Ph.D.
75 Hormonal Regulation of Gene Expression
William W. Chin, M.D.
77 Technologies for Genome-sequencing Projects
George M. Church, Ph.D.
79 Molecular Genetics of Limb Development in Drosophila
Stephen M. Cohen, Ph.D.
81 Tracking Genes That Cause Human Disease
Francis S. Collins, M.D., Ph.D.
83 Development of the Immune System
Max D. Cooper, M.D.
85 Structure and Function of RNA Polymerase II
Jeffry L. Corden, Ph.D.
87 Mechanically Activated Ion Channels
David P. Corey, Ph.D.
89 Genetic Regulatory Mechanisms in Cellular Differentiation
Gerald R. Crabtree, M.D.
91 The Mechanism of a Bacterial Transposition Reaction
Nancy L. Craig, Ph.D.
V
Contents
93 Mechanisms of Antigen Processing
Peter Cresswell, Ph.D.
95 Regulation of Human Retroviral Gene Expression
Bryan R. Cullen, Ph.D.
"97 Mechanism of Retrovirus Infection
James M. Cunningham, M.D.
99 The Nuclear Pore Complex
Laura I. Davis, Ph.D.
1 01 Molecular Approaches to T Lymphocyte Recognition and Differentiation
Mark M. Davis, Ph.D.
103 Signal Transduction by the Epidermal Growth Factor Receptor
Roger J. Davis, Ph.D.
105 Traffic of Synaptic Vesicle Proteins in Neurons and Endocrine Cells
Pietro De Camilli, M.D.
107 Three-Dimensional Structures of Biological Macromolecules
Johann Deisenhofer, Ph.D.
109 Molecular Mechanisms of Lymphocyte Differentiation
Stephen V. Desiderio, M.D., Ph.D.
113 Transcription Control During Early Drosophila Development
Claude Desplan, Ph.D.
115 Immune Evasion by Parasites Causing Tropical Diseases
John E. Donelson, Ph.D.
11 7 Post-transcriptional Regulation of Gene Expression, RNA-Protein
Complexes, and Nuclear Structures
Gideon Dreyfuss, Ph.D.
119 Genetic Basis of Hearing Loss
Geoffrey M. Duyk, M.D., Ph.D.
121 Molecular Genetics of Intracellular Protein Sorting
Scott D. Emr, Ph.D.
123 Mechanisms Involved in Preventing Unwanted Blood Clots
Charles T. Esmon, Ph.D.
125 Molecular Genetics of Steroid and Thyroid Hormone Receptors
Ronald M. Evans, Ph.D.
129 Molecular Mechanisms Involved in the Actions of Calcium-mediated
Hormones
John H. Exton, M.D., Ph.D.
Contents
131 Tumor-Suppressor Genes
Andrew P. Feinberg, M.D., M.P.H.
133 Genetics, Structure, and Function of Histocompatibility Antigens
Kirsten Fischer Lindahl, Ph.D.
137 Genetic Approaches to Immune Function and Tolerance
Richard A. Flavell, Ph.D.
139 Biophysical Genetics of Protein Structure and Folding
Robert O. Fox, Ph.D.
141 Molecular Basis of Genetic Diseases and Chromosome Mapping
Uta Francke, M.D.
143 Molecular Biology of Obesity and Diabetes
Jeffrey M. Friedman, M.D., Ph.D.
147 Regulation of Keratin Expression During Differentiation and
Development in Human Skin
Elaine Fuchs, Ph.D.
149 The Molecular Basis of Viral Replication and Pathogenesis
Donald E. Ganem, M.D.
151 Second Messengers and Cell Regulation
David L. Garbers, Ph.D.
153 Molecular Genetics of the Major Histocompatibility Complex
Jan Geliebter, Ph.D.
155 The Decoding Code in mRNA
Raymond F. Gesteland, Ph.D.
157 Molecular Analysis of Proteins Involved in Human Disease
Mary-Jane H. Gething, Ph.D.
159 Signal Transduction Pathways in B lymphocytes
Sankar Ghosh, Ph.D.
161 Molecular Genetics of Blood Coagulation
David Ginsburg, M.D.
163 Uncovering the Molecular Basis of X-linked Disorders
Jane M. Gitschier, Ph.D.
165 Membrane Lipids and Cell Regulation
John A. Glomset, M.D.
167 Determination and Maintenance of Cell Type
Richard H. Gomer, Ph.D.
169 Growth Cone Guidance and Neuronal Recognition in Drosophila
Corey S. Goodman, Ph.D.
vii
Contents
1 71 Mechanisms of Immunological Self-Tolerance and Autoimmunity
Christopher C. Goodnow, B.V.Sc, Ph.D.
1 73 Developmental Control of Gene Expression
Rudolf Grosschedl, Ph.D.
175 Polypeptide Hormone Gene Regulation
Joel F. Habener, M.D.
1 77 Structural Studies of Macromolecular Assemblies
Stephen C. Harrison, Ph.D.
1 79 Control of Gene Expression During the Cell Cycle and Development
of the Mammalian Cerebellum
Nathaniel Heintz, Ph.D.
181 Structural Biology of CD4 and CDS Involvement in the Cellular
Immune Response
Wayne A. Hendrickson, Ph.D.
183 Variegated Position Effects in Drosophila
Steven Henikoff, Ph.D.
185 Biological Roles and Expression of Complement Receptors
V. Michael Holers, M.D.
187 Genetic Control of Nematode Development
H. Robert Horvitz, Ph.D.
189 Protein Folding in Vivo
Arthur L. Horwich, M.D.
191 Molecular Mechanisms in the Regulation of Synaptic Transmission
Richard L. Huganir, Ph.D.
193 Molecular Aspects of Signal Transduction in the Visual System
James B. Hurley, Ph.D.
195 The Molecular Basis of Cell Adhesion in Normal and
Pathological Situations
Richard O. Hynes, Ph.D.
197 Molecular Genetics of Intracellular Microorganisms
Ralph R. Isberg, Ph.D.
201 Genetic Approaches to the Control of Mycobacterial Diseases
William R. Jacobs, Jr., Ph.D.
203 Mechanisms of Neurotransmitter Storage and Release
Reinhard Jahn, Ph.D.
205 Molecular Studies of Voltage-Sensitive Potassium Channels
Lily Y.Jan, Ph.D.
via
Contents
207 Neural Development in Drosophila
Yuh Nungjan, Ph.D.
205^ Activation ofCD4 T Cells
Charles A. Janeway, Jr., M.D.
211 Control of Cell Pattern in the Developing Nervous System
Thomas M. Jessell, Ph.D.
213 Energy-transducing Membrane Proteins
H. Ronald Kaback, M.D.
215 Control of the Immunoglobulin Heavy-Chain Gene
Thomas R. Kadesch, Ph.D.
21 7 Genetic Control of Hemoglobin Synthesis
Yuet Wai Kan, M.D., D.Sc.
219 Cell Biological Studies of Memory
Eric R. Kandel, M.D.
221 The T Cell Repertoire
John W. Kappler, Ph.D.
223 The Genetic Control of Morphogenesis
Thomas C. Kaufman, Ph.D.
225 Protein Folding and Macromolecular Recognition
Peter S. Kim, Ph.D.
227 RNA Viral Genetics
Karla A. Kirkegaard, Ph.D.
229 Adrenergic Receptor Structure and Function
Brian K. Kobilka, M.D.
231 Molecular Genetics of Lymphocyte Development and Neoplasia
Stanley J. Korsmeyer, M.D.
233 Molecular Genetics of Neuromuscular Disease
Louis M. Kunkel, Ph.D.
235 Structural Studies on DNA-Replication Enzymes, src -related Oncogene
Products, and Oxidoreductases
John Kuriyan, Ph.D.
237 Molecular Analysis of Down Syndrome
David M. Kurnit, M.D., Ph.D.
239 Replication and Pathogenesis of RNA Viruses
Michael M.-C. Lai, M.D., Ph.D.
241 Molecular Biology of Human Papillomaviruses
Laimonis A. Laimins, Ph.D.
ix
Contents
243 Genetic Studies in Cardiovascular Disease
Jean-Marc Lalouel, M.D., D.Sc.
245 Structure and Replication of Influenza Virus and Paramyxoviruses
Robert A. Lamb, Ph.D., Sc.D.
2?7 Cancer and Genetic Modification
Philip Leder, M.D.
249 From Molecular Biology to Therapy of Human Disease
Fred D. Ledley, M.D.
251 Molecular Biology of Hormone and Drug Receptors in Health
and Disease
Robert J. Lefkowitz, M.D.
253 Axis Formation and Germline Determination in Drosophila
Ruth Lehmann, Ph.D.
255 Regulation of Gene Expression During Cellular Differentiation
and Activation
Jeffrey M. Leiden, M.D., Ph.D.
257 Chemical Communication
Michael R. Lerner, M.D., Ph.D.
259 Structural Determinants of Human a-Globin Gene Expression
Stephen A. Liebhaber, M.D.
261 The Heat-Shock Response
Susan L. Lindquist, Ph.D.
263 T Cell Surface Glycoproteins in Development and Viral Infections
Dan R. Littman, M.D., Ph.D.
265 The Biology of T Lymphocyte Development
Dennis Y.-D. Loh, M.D.
267 Molecular Genetics of Mammalian Glycosyltransferases
John B. Lowe, M.D.
269 Mechanisms of Embryonic Induction in Vertebrates
Richard L. Maas, M.D., Ph.D.
271 Cell Cycle Control
James L. Mailer, Ph.D.
273 The Role of T Cells in Health and Sickness
Philippa Marrack, Ph.D.
275 Cell Regulation by Transforming Growth Factors
Joan Massague, Ph.D.
Contents
277 Structural Basis of Interactions Within and Between Macromolecules
Brian W. Matthews, Ph.D., D.Sc.
279 What Viruses Are Telling Us About Gene Regulation in Mammalian Cells
Steven Lanier McKnight, Ph.D.
281 Fundamental Mechanisms of Ion Channel Proteins
Christopher Miller, Ph.D.
283 Neural Foundations of Vision
J. Anthony Movshon, Ph.D.
285 Human Retroviral Gene Expression and Cellular Transcription
Gary J. Nabel, M.D., Ph.D.
287 Molecular Analysis of Muscle Development and Function
Bernardo Nadal-Ginard, M.D., Ph.D.
289 The Genomic Response to Growth Factors
Daniel Nathans, M.D.
291 Molecular Biology of Visual Pigments
Jeremy Nathans, M.D., Ph.D.
293 Gene Regulation in Animal Cells
Joseph R. Nevins, Ph.D.
295 Molecular Genetics of X-linked Disease
Robert L. Nussbaum, M.D.
297 Function of Oncogenes in Early Embryogenesis
Roel Nusse, Ph.D.
25^5^ Molecular Mechanisms That Regulate B Cell Development
Michel C. Nussenzweig, M.D., Ph.D.
301 Mechanism of DMA Replication
Michael E. O'Donnell, Ph.D.
303 Large-Scale Analysis of Yeast and Human DNA
Maynard V. Olson, Ph.D.
305 Molecular Genetic Studies of Hematopoietic Cells
Stuart H. Orkin, M.D.
307 Albinism and Tyrosinase
Paul A. Overbeek, Ph.D.
309 Structural Studies of DNA-binding Proteins
Carl O. Pabo, Ph.D.
313 The X and Y Chromosomes in Mammalian Development
David C. Page, M.D.
xi
Contents
315 Mammalian Development and Disease
Richard D. Palmiter, Ph.D.
31 7 Regulation of Gene Expression in Steroid Hormone Biosynthesis
Keith L. Parker, M.D., Ph.D.
319 Molecular Neuroimmunology
Donald G. Payan, M.D.
321 Molecular Basis of Lymphocyte Signaling
Roger M. Perlmutter, M.D., Ph.D.
323 Genetic Dissection of a Signal Transduction Pathway in
Drosophila melanogaster
Norbert Perrimon, Ph.D.
325 Gene Regulation and Immunodeficiency
B. Matija Peterlin, M.D.
327 Mechanism of Action of Polypeptide Growth Factors
Linda J. Pike, Ph.D.
329 Protein Structures, Molecular Recognitions, and Functions
Florante A. Quiocho, Ph.D.
331 Molecular Approaches to Olfaction
Randall R. Reed, Ph.D.
333 The Molecular Basis of Hereditary Diseases of the Kidney
Stephen T. Reeders, M.D.
335 Extracellular Factors Affecting Neuron Development
Louis F. Reichardt, Ph.D.
33 7 Molecular Genetics of RNA Processing and Behavior
Michael Rosbash, Ph.D.
339 Molecular Mechanisms of Transcription, Regulation, and Development
of the Neuroendocrine System
Michael G. Rosenfeld, M.D.
341 Development of the Drosophila Visual System
Gerald M. Rubin, Ph.D.
345 The Regulation of Blood Coagulation
J. Evan Sadler, M.D., Ph.D.
347 Molecular Mechanism of Transmembrane Signal Transduction by
G Protein-coupled Receptors
Thomas P. Sakmar, M.D.
349 Molecular Genetics of Development in Drosophila
Shigeru Sakonju, Ph.D.
Contents
351 Generating a Repertoire of Antigen-Specific Receptors
David G. Schatz, Ph.D.
353 Intracellular Protein Transport
Randy W. Schekman, Ph.D.
355 Development and Function of the Synapse
Richard H. Scheller, Ph.D.
357 Molecular Pathogenicity Studies of Enteric Bacteria
Gary K. Schoolnik, M.D.
361 Three-Dimensional Structure of Eukaryotic Chromosomes
John W. Sedat, Ph.D.
363 A Molecular Basis of Familial Hypertrophic Cardiomyopathy
Jonathan G. Seidman, Ph.D.
367 Computational Neurobiology of Sensory Representations
TerrenceJ. Sejnowski, Ph.D.
369 Adenovirus as a Model for Control of Gene Expression
Thomas E. Shenk, Ph.D.
371 Growth Control of Myeloid Cells
Charles J. Sherr, M.D., Ph.D.
373 The Role of Second Messengers in Ion Channel Regulation
Steven A. Siegelbaum, Ph.D.
375 Chemistry of Cellular Regulation
Paul B. Sigler, M.D., Ph.D.
377 The Mitochondrial Genome of Trypanosomes
Larry Simpson, Ph.D.
379 Regulation of Gene Activity During B Cell Development
Harinder Singh, Ph.D.
381 Regulation of Gene Expression in Developing Lymphocytes
Stephen T. Smale, Ph.D.
383 Developmental Genetics
Philippe M. Soriano, Ph.D., D.Sc.
385 Understanding How Eggs Work
Allan C. Spradling, Ph.D.
387 Structural Studies of Regulatory Proteins
Stephen R. Sprang, Ph.D.
389 Insulin and the Islets of Langerhans
Donald F. Steiner, M.D.
xiii
Contents
391 Autoantibody Probes for Mammalian Gene Expression
Joan A. Steitz, Ph.D.
393 Structural Studies of Protein-Nucleic Acid Interactions
Thomas A. Steitz, Ph.D.
397 Pattern Formation and Neuronal Cell Recognition in the Drosophila
Visual System
Hermann Steller, Ph.D.
399 Molecular Genetics of Nematode Development and Behavior
Paul W. Sternberg, Ph.D.
401 Why Do We Drink Coffee and Tea?
Charles F. Stevens, M.D., Ph.D.
403 Morphogen Gradients and the Control of Body Pattern in Drosophila
Gary Struhl, Ph.D.
405 Secretory Pathways in Neurons
Thomas C. Siidhof, M.D.
407 Transcription Factors in Cell Growth and Kidney Differentiation
Vikas P. Sukhatme, M.D., Ph.D.
409 Structure and Function of Voltage-Dependent Calcium Channels
Tsutomu Tanabe, Ph.D.
411 The Molecular Biology of Liver Regeneration
Rebecca A. Taub, M.D.
413 Protein- Tyrosine Phosphatases and the Control of Lymphocyte
Activation
Matthew L. Thomas, Ph.D.
415 Molecular Regulation of Lymphoid Cell Growth and Development
Craig B. Thompson, M.D.
41 7 The Molecular Basis of Metamorphosis
Carl S. Thummel, Ph.D.
419 The Regulation of Mammalian Development
Shirley M. Tilghman, Ph.D.
421 Mechanisms of Gene Regulation in Animal Cells
Robert Tjian, Ph.D.
423 Studies on T Lymphocytes and Mammalian Memory
Susumu Tonegawa, Ph.D.
427 Molecular Engineering Applied to Cell Biology and Neurobiology
Roger Y. Tsien, Ph.D.
xiv
Contents
429 Genetic Defects in the Metabolic Pathways Interconnecting the Urea
and Tricarboxylic Acid Cycles
David L. Valle, M.D.
431 Human Molecular Genetics in Two X-linked Diseases
Stephen T. Warren, Ph.D.
433 The MyoD Gene Family: A Nodal Point During Specification of Muscle
Cell Lineage
Harold M. Weintraub, M.D., Ph.D.
435 Structural and Functional Studies of the T Cell Antigen Receptor
Arthur Weiss, M.D., Ph.D.
437 Following the Life History of Lymphocytes
Irving L. Weissman, M.D.
439 Function and Regulation of the Cystic Fibrosis Transmembrane
Conductance Regulator
Michael J. Welsh, M.D.
441 Identification of the Gene Responsible for Adenomatous Polyposis
Raymond L. White, Ph.D.
443 Mechanisms of the Biological Activities of Membrane Glycoproteins
Don C. Wiley, Ph.D.
445 Studies of Blood Cell Formation
David A. Williams, M.D.
447 Growth Factor-stimulated Cell Proliferation
Lewis T. Williams, M.D., Ph.D.
449 Somatic Cell Gene Transfer
James M. Wilson, M.D., Ph.D.
453 Normal and Abnormal Lymphocyte Growth Regulation
Owen N. Witte, M.D.
455 Translational Regulation
Sandra L. Wolin, M.D., Ph.D.
457 Molecular Genetics and Studies Toward Gene Therapy
for Metabolic Disorders
Savio L. C. Woo, Ph.D.
455^ Paracrine Control of Blood Vessel Function: Role of the Endothelins
Masashi Yanagisawa, M.D., Ph.D.
461 Mechanism of Phototransduction in Retinal Rods and Cones
King-Wai Yau, Ph.D.
463 Molecular Mechanisms of Ion Channel Function
Gary Yellen, Ph.D.
XV
Contents
465 Drosophila Behavior and Neuromuscular Development
Michael W. Young, Ph.D.
469 Control of Transcription by Transmembrane Signals
Edward B. Ziff, Ph.D.
471 Cell-Cell Interactions Determine Cell Fate in the Drosophila Retina
S. Lawrence Zipursky, Ph.D.
4 73 Molecular Genetics of Sensory Transduction
Charles S. Zuker, Ph.D.
475 Investigators by Location
481 International Research Scholars
483 Molecular Biology and Epidemiology for Control of Rotavirus Diarrhea
Carlos F. Arias, Ph.D.
485 Molecular Genetics of Normal and Leukemic Hematopoiesis
Alan Bernstein, Ph.D.
487 Molecular Biology of Two Enteropathogenic Bacteria
Edmundo Calva, Ph.D.
489 Functional Heterogeneity in Prolactin- secreting Cells
Gabriel Cota, Ph.D.
491 Ionic Channels in Sea Urchin Sperm Physiology
Alberto Darszon, Ph.D.
493 Host-Pathogen Interactions in Microbial Pathogenesis
B. Brett Finlay, Ph.D.
45^5 Mechanisms of Transcriptional Regulation
Jack Greenblatt, Ph.D.
497 Ionic Homeostasis in White Blood Cells
Sergio Grinstein, Ph.D.
499 Genetic Basis of Multidrug Resistance
Philippe Gros, Ph.D.
501 Control of Bacterial Protein Synthesis During Viral Infection
Gabriel Guarneros Pena, Ph.D.
503 Molecular Genetics of Photosynthesis and Carbon Assimilation in Plants
Luis R. Herrera-Estrella, Ph.D.
507 Gene Pattern Expression in Early Embryogenesis
Alexandra L. Joyner, Ph.D.
509 Diagnostic Use of UNA Replication in Infectious Diseases
Paul M. Lizardi, Ph.D.
xvi
Contents
511 Lineage-Specific Gene Expression in Caenorhabditis elegans
James D. McGhee, Ph.D.
513 Cytokine Regulation of Effector Functions in Immune Responses
Tim R. Mosmann, Ph.D.
515 Cellular and Molecular Basis of Variability in Entamoeba histolytica
Esther Orozco, Ph.D.
51 7 Phosphorylation and Protein-Protein Interactions in
Signal Transduction
Tony Pawson, Ph.D.
519 Chemical and Functional Characterization of Scorpion Toxins
Lourival Domingos Possani, Ph.D.
521 Protein Crystallography in the Study of Infectious Diseases
Randy J. Read, Ph.D.
523 Representations and Transformations of Tactile Signals in Somatic
and Frontal Motor Cortices
Ranulfo Romo, M.D., Ph.D.
525 Anterior-Posterior Patterning in the Early Mammalian Embryo
Janet Rossant, Ph.D.
527 Response of the Cerebral Cortex to Spatial Information
Jean-Pierre Roy, M.D.
529 Molecular Studies on Neuronal Calcium Channels
Terry P. Snutch, Ph.D.
531 Cystic Fibrosis, Gene Expression in the Mammalian Lens, and Mapping
of Chromosome 7
Lap-Chee Tsui, Ph.D.
533 Index
xvii
THE Howard Hughes Medical Institute (HHMI) presents the seventh volume of Research
in Progress, a series begun in 1986 to describe the research activities of the Institute.
Early in 1992 the scientists associated with HHMI at that time were asked to provide
statements in nontechnical terms of their past accomplishments, current work, and plans for
the future, in order to provide a concise but informative picture of the Institute's research
in progress. You will read their personal essays in the pages that follow. We are pleased that
this annual publication has come to be a useful source of information for scientists and
nonscientists interested in biomedical research, as well as for the members of the HHMI family.
The Institute also publishes as a companion volume the formal Annual Scientific Report.
This is the official archival record of the research of each HHMI laboratory and includes
yearly bibliographies as well as descriptions of other HHMI scientific activities. A general
Annual Report of the Howard Hughes Medical Institute describes the various programs for
a lay audience and gives a summary of the Institute's financial data. With the recent
publication of From Egg to Adult, which describes the exciting process of biological
development, the Institute has expanded its series of reports on subjects of current scientific
interest written for a general audience. This booklet joins the widely acclaimed Finding the
Critical Shapes and Blazing a Genetic Trail that have been in great demand by a broad
spectrum of readers, especially teachers who have used these reports in classrooms across
this country and abroad.
In recognition of the fact that the boundaries of biomedical science are not constrained by
national borders, the Institute launched its International Research Scholars Program in the
spring of 1991- This is a limited and experimental program with the purpose of providing
research support in selected countries for promising scientists whose careers are in a
developing phase but who have already made significant contributions to biomedical
research. The international program is separate from the Institute's ongoing medical research
program, described in the first section of this volume, in which scientists join HHMI as full-
time employee-investigators and are supported through the Institute's operation as a Medical
Research Organization (MRO). The Scholars do not join the HHMI staff, but instead each
receives research support through a grant.
For the initial awards in this program, selected scientists in the countries that are our
immediate neighbors, Canada and Mexico, were invited to compete. Twenty-four outstanding
investigators received five-year awards (14 from Canada and 10 from Mexico). You will read
their descriptions of the research in their laboratories in the second section of this volume.
For awards in 1993, the Institute has turned to scientists in the United Kingdom, Australia,
and New Zealand. Those selected for grant support will be announced in late 1992, and their
work will be included in the next volume of Research in Progress.
The Institute has a large grants program that complements its MRO activities. In addition to
the International Research Scholars initiative, the grants program supports education in the
biomedical and related sciences at the precollege, college undergraduate, graduate, and
postdoctoral levels. New in 1992 were grants awarded to certain science museums, science
and technology centers, children's museums, and natural history museums to assist in
education and outreach programs for elementary school children and their families. The
Institute's annual publication Grants for Science Education details these initiatives.
A very important event for HHMI occurred in the spring of 1991, when construction began
on the buildings that will become our new headquarters and conference center complex in
Chevy Chase, Maryland. At this writing, the construction is entering its final phases, and the
attractiveness of the buildings of this large facility and how they fit into the gentle hills and
valleys of the lovely site are exceeding our original expectations. We expect to occupy the
buildings in early 1993- The facility provides the administrative focus for the Institute's 222
investigators (as of July 1, 1992) whose laboratories are located at 53 institutions across
this country.
We invite you to share the excitement of the research of the HHMI investigators whose
work forms the core of the Institute's activities and of the studies of our International
Foreword
Research Scholars that have further enriched our scientific program. We are especially
pleased that the Institute's research on problems of medical concern to developing countries
is extended through the basic studies of a number of these Scholars. The advances toward
understanding biological processes and disease mechanisms that are described herein move
us further tovv'ard our goal of the betterment of the human condition.
Purnell W. Choppin, M.D.
President
XX
Trustees, Officers, and Principal Staff Members
Trustees
Alexander G. Beam, M.D.
Adjunct Professor
The Rockefeller University
Professor Emeritus of Medicine
Cornell University Medical College
Former Senior Vice President
Merck Sharp & Dohme, International
Helen K. Copley
Chairman of the Corporation and
Chief Executive Officer
The Copley Press, Inc.
Frank William Gay
Former President and Chief
Executive Officer
SUMMA Corporation
James H. Gilliam, Jr., Esq.
Executive Vice President
Beneficial Corporation
Hanna H. Gray, Ph.D.
President
The University of Chicago
William R. Lummis, Esq.
Chairman of the Board of Directors
SUMMA Corporation
Irving S. Shapiro, Esq., Chairman
Of Counsel
Skadden, Arps, Slate, Meagher & Flom
Former Chairman and Chief
Executive Officer
E.I. du Pont de Nemours and Company
George W. Thorn, M.D.,
Chairman Emeritus
Professor Emeritus
Harvard Medical School
James D. Wolfensohn
President
James D. Wolfensohn Incorporated
Officers
Purnell W. Choppin, M.D.
President
W. Maxwell Cowan, M.D., Ph.D.
Vice President and Chief Scientific Officer
Graham O. Harrison
Vice President and Chief Investment Officer
Joseph G. Perpich, M.D , J.D.
Vice President for Grants and
Special Programs
Jose E. Trias, Esq.
Vice President and General Counsel
Robert C. White
Vice President and Chief Financial Officer
Trustees, Officers, and Principal Staff Members
Principal Staff Members
Craig A. Alexander, Esq.
Associate General Counsel
Stephen A. Barkanic
Grants Program Officer
W. Emmett Barkley, Ph.D.
Director of Laboratory Safety
Lillian H. Blucher
Managing Director — Investments
WinfredJ. Clingenpeel
Director of Purchasing
David Davis-Van Atta
Grants Program Officer
Barbara Filner, Ph.D.
Grants Program Officer
Patricia S. Gage, Esq.
Associate General Counsel
James R. Gavin III, M.D., Ph.D.
Senior Scientific Officer
Donald H. Harter, M.D.
Senior Scientific Officer and
Director, HHMI-NIH Research
Scholars Program
Patricia J. Hoben, Ph.D.
Grants Program Officer
(through September 4, 1992)
John A. Jones
Director of Computer Services
David W. Kingsbury, M.D.
Senior Scientific Officer
Laura A. Kumin, Esq.
Associate General Counsel
Joan S. Leonard, Esq.
Associate General Counsel
Robert H. McGhee
Director of Research Facilities Planning
Alan E. Mowbray
Director of Management Services
Robert C. Mullins
Director of Internal Audit
Edward J. Palmerino
Assistant Controller
Robert A. Potter
Director of Communications
Donald C. Powell
Director of Human Resources
Ellen B. Safir
Managing Director — Investments
Mark W. Smith
Controller
Claire H. Winestock, Ph.D.
Senior Scientific Officer
Medical Advisory Board
Michael S. Brown, M.D.
Paul J. Thomas Professor of Medicine and
Genetics
Director, Center for Genetic Disease
University of Texas Southwestern Medical
Center at Dallas
William A. Catterall, Ph.D.
Professor and Chair
Department of Pharmacology
University of Washington School of Medicine
John E. Dowling, Ph.D.
(through December 31, 1992)
Maria Moors Cabot Professor of Natural
Science
Department of Cellular and Developmental
Biology
Harvard University
Thomas J. Kelly, Jr., M.D., Ph.D.
( effective January 1, 1993 )
Chairman
Department of Molecular Biology and
Genetics
The Johns Hopkins University School of
Medicine
Hugh O. McDevitt, M.D.
( through December 3 1, 1992 )
Burt and Marion Avery Professor of
Immunology
Professor of Microbiology, Immunology,
and Medicine
Stanford University School of Medicine
Dinshaw Patel, Ph.D.
(effective January 1, 1993)
Abby Rockefeller Mauze Chair in
Experimental Therapeutics
Member, Cellular Biochemistry and
Biophysics Program
Memorial Sloan-Kettering Cancer Center
William E. Paul, M.D.
Chief, Laboratory of Immunology
National Institute of Allergy and Infectious
Diseases
National Institutes of Health
Frederic M. Richards, Ph.D.
( through December 31, 1992 )
Sterling Professor Emeritus of Molecular
Biophysics and Biochemistry
Yale University
Janet D. Rowley, M.D.
Blum-Riese Distinguished Service Professor
Departments of Medicine and Molecular
Genetics and Cell Biology
The University of Chicago
David D. Sabatini, M.D., Ph.D.
Frederick L. Ehrman Professor and Chairman
Department of Cell Biology
New York University School of Medicine
Phillip A. Sharp, Ph.D.
John D. MacArthur Professor of Biology and
Head, Department of Biology
Massachusetts Institute of Technology
Melvin I. Simon, Ph.D.
(effective January 1, 1993)
Biaggini Professor of Biology
California Institute of Technology
William S. Sly, M.D.
(through December 31, 1992)
Alice A. Doisy Professor and Chairman
Edward A. Doisy Department of
Biochemistry and Molecular Biology
St. Louis University School of Medicine
Lloyd H. Smith, Jr., M.D., Chairman
Professor of Medicine and Associate Dean
University of California, San Francisco,
School of Medicine
Jonathan W. Uhr, M.D.
(effective January 1, 1993)
Professor and Chairman
Department of Microbiology
University of Texas Southwestern Medical
Center at Dallas
xxiii
Scientific Review Board
Cell Biology and Regulation
Alfred G. Oilman, M.D., Ph.D.
Professor and Chairman
Department of Pharmacology
University of Texas Southwestern Medical
Center at Dallas
Tony Hunter, Ph.D.
Professor of Molecular Biology and Virology
The Salk Institute
Stuan Kornfeld, M.D.
Professor of Medicine
Washington University School of Medicine
David J. L. Luck, M.D., Ph.D.
Professor of Cell Biology
The Rockefeller University
Thomas D. Pollard, M.D.
Professor and Director
Department of Cell Biology and Anatomy
The Johns Hopkins University School
of Medicine
Lucille Shapiro, Ph.D.
Joseph D. Grant Professor in the School of
Medicine and Chairman, Department of
Developmental Biology
Stanford University School of Medicine
Genetics
David Botstein, Ph.D.
Professor and Chairman
Department of Genetics
Stanford University School of Medicine
Ira Herskowitz, Ph.D.
Professor and Chairman
Department of Biochemistry and Biophysics
Head, Division of Genetics
University of California, San Francisco,
School of Medicine
Anthony P. Mahowald, Ph.D.
Louis Block Professor and Chairman
Department of Molecular Genetics
and Cell Biology
The University of Chicago
Thomas P. Maniatis, Ph.D.
Professor of Biochemistry and Molecular
Biology
Harvard University
Malcolm A. Martin, M.D.
Chief, Laboratory of Molecular Microbiology
National Institute of Allergy
and Infectious Diseases
National Institutes of Health
Carolyn W. Slayman, Ph.D.
Professor and Chairman
Department of Genetics
Professor, Department of Cellular
and Molecular Physiology
Yale University School of Medicine
Harold E. Varmus, M.D.
American Cancer Society Professor
of Molecular Virology
Department of Microbiology and
Immunology
University of California, San Francisco,
School of Medicine
Immunology
Joseph M. Davie, M.D.
(through June 30, 1992)
President
G.D. Searle and Company
Matthew D. Scharff, M.D.
Professor of Cell Biology
Albert Einstein College of Medicine
Ursula Storb, M.D.
( effective January 1, 1993 )
Professor
Department of Molecular Genetics
and Cell Biology
The University of Chicago
Emil R. Unanue, M.D.
Professor and Chairman
Department of Pathology
Washington University School of Medicine
xxiv
Scientific Review Board
Ellen S. Vitetta, Ph.D.
Professor of Microbiology and Director,
Cancer Immunobiology Center
University of Texas Southwestern Medical
Center at Dallas
Neuroscience
Floyd E. Bloom, M.D.
Chairman, Department of
Neuropharmacology
Member, The Scripps Research Institute
Arthur M. Brown, M.D., Ph.D.
Professor and Chairman
Department of Molecular Physiology
and Biophysics
Baylor College of Medicine
Zach W. Hall, Ph.D.
Professor and Chairman
Department of Physiology
University of California, San Francisco,
School of Medicine
Stephen F. Heinemann, Ph.D.
( effective January 1, 1995 )
Faculty Chairman
Director of Molecular Neurobiology
Laboratory
The Salk Institute
CarlaJ. Schatz, Ph.D.
{effective July 1, 1993)
Professor of Neurobiology
Department of Molecular and Cell Biology
University of California, Berkeley
Melvin I. Simon, Ph.D.
(through December 31, 1992)
Biaggini Professor of Biology
California Institute of Technology
Structural Biology
David R. Davies, Ph.D.
Chief, Section on Molecular Structure
Laboratory of Molecular Biology
National Institute of Diabetes and Digestive
and Kidney Diseases
National Institutes of Health
Michael N. G. James, D.Phil.
Professor of Biochemistry
University of Alberta
Dinshaw Patel, Ph.D.
(through December 31, 1992)
Abby Rockefeller Mauze Chair in
Experimental Therapeutics
Member, Cellular Biochemistry
and Biophysics Program
Memorial Sloan-Kettering Cancer Center
Michael G. Rossmann, Ph.D.
Hanley Professor of Biological Sciences
Purdue University
Peter E. Wright, Ph.D.
(effective January 1, 1993)
Member and Chairman
Department of Molecular Biology
The Scripps Research Institute
XXV
Locations of Howard Hughes Medical Institute
Laboratories
Alabama
California
Colorado
Connecticut
Georgia
Illinois
Indiana
Iowa
Maryland
Massachusetts
Michigan
Missouri
New Jersey
New York
University of Alabama at Birmingham and associated hospitals
California Institute of Technology and associated hospitals, Pasadena
The Salk Institute for Biological Studies, San Diego
Stanford University and the Stanford University Hospital, Palo Alto
University of California, Berkeley, and associated hospitals
University of California, Los Angeles, and associated hospitals
University of California, San Diego, and the UCSD Medical Center
University of California, San Francisco, and associated hospitals
University of Southern California, Los Angeles, and associated hospitals
National Jewish Center for Immunology and Respiratory Medicine, Denver
University of Colorado at Boulder and the University's Health Sciences Center
University of Colorado Health Sciences Center, Denver, and associated hospitals
Yale University and associated hospitals, New Haven
Emory University School of Medicine, Atlanta, and associated hospitals
Northwestern University and associated hospitals, Evanston
The University of Chicago and The University of Chicago Hospitals
Indiana University, Bloomington, and associated hospitals
Indiana University School of Medicine, Indianapolis, and associated hospitals
The University of Iowa and associated hospitals, Iowa City
The Carnegie Institution of Washington, Baltimore, and The Johns
Hopkins Hospital
The Johns Hopkins University and Hospital, Baltimore
Brandeis University, Waltham, and associated hospitals
Brigham and Women's Hospital, Boston
The Children's Hospital, Boston
Harvard College, Arts and Sciences, Cambridge
Harvard Medical School, Boston
Massachusetts General Hospital, Boston
Massachusetts Institute of Technology and associated hospitals, Cambridge
Tufts University School of Medicine and associated hospitals, Boston
University of Massachusetts, Worcester, and associated hospitals
University of Michigan and associated hospitals, Ann Arbor
Washington University and associated hospitals, St. Louis
Princeton University and associated medical centers, Princeton
Albert Einstein College of Medicine of Yeshiva University, Bronx, and
associated hospitals
Cold Spring Harbor Laboratory and associated hospitals. Cold Spring Harbor
Columbia University and associated hospitals. New York City
Cornell University Medical College, New York City
Memorial Sloan-Kettering Cancer Center, New York City
New York University (Medical Center and Washington Square) and associated
hospitals. New York City
The Rockefeller University and Rockefeller University Hospital, New York City
State University of New York at Stony Brook and University Hospital at
Stony Brook
North Carolina Duke University, including Duke University Medical Center, Durham
xxvi
Locations of Howard Hughes Medical Institute Laboratories
Oklahoma
Oregon
Pennsylvania
Tennessee
Texas
Utah
Washington
Oklahoma Medical Research Foundation and associated hospitals,
Oklahoma City
University of Oregon and associated hospitals, Eugene
University of Pennsylvania and associated hospitals, Philadelphia
St. Jude Children's Research Hospital, Memphis
Vanderbilt University, including Vanderbilt University Hospital, Nashville
Baylor College of Medicine and associated hospitals, Houston
Rice University and associated hospitals, Houston
University of Texas Southwestern Medical Center at Dallas and
associated hospitals
University of Utah, including University of Utah Medical Center, Salt Lake City
Fred Hutchinson Cancer Research Center, Seattle
University of Washington and associated hospitals, Seattle
Wisconsin
University of Wisconsin-Madison, and associated hospitals
Other Institute Facilities
Maryland HHMI-NIH Cloister at the Mary Woodard Lasker Center for Science and
Education on the NIH campus, Bethesda
New York Synchrotron Beam Lines, Brookhaven National Laboratory (under construction)
xxvii
I
I
(
I
(
I
I
I
I
I
1
1
!
I
Introduction
THE annual publication of Research
in Progress has become one of the
highlights of the Howard Hughes Medi-
cal Institute's research activities. By pro-
viding a synoptic overview of the re-
search being conducted by the Institute's
investigators, in essentially nontechnical
language, it has come to be greatly ap-
preciated not only by those interested in
the Institute's activities as a Medical Re-
search Organization, but by many others
who view it as an accessible introduction
to current biomedical research.
The present volume in this series follows fairly
closely the pattern that has evolved over the past
three or four years: it provides a snapshot of the
status of the Institute's research in early 1992. As
in the previous volume, this issue includes a se-
ries of introductory essays that are intended to
assist the reader who has had little or no back-
ground in biology or medicine. These essays are
not intended to be exhaustive nor simply to repli-
cate material readily available in most standard
biology or medical texts; rather they are intended
to serve as an expanded glossary of terms, de-
fining in a general way many of the terms used in
the individual reports. To emphasize this feature
of the essays, these technical terms are shown
in bold type. The essays are based on an initial
set of drafts prepared for each of the five pro-
grammatic areas in which the Institute conducts
research: cell biology and regulation; genetics;
immunology; neuroscience; and structural biol-
ogy. The original drafts were kindly prepared by
Drs. Richard O. Hynes, Philip Leder, Charles A.
Janeway, Eric R. Kandel, and Stephen C. Harrison,
respectively. We are grateful to them and to a
number of their colleagues within the Institute
who commented on the essays or provided addi-
tional information and, in some cases, clarified
ambiguities or corrected errors in the original
text. As in last year's volume the text of the essays
has been supplemented by a number of illustra-
tions, derived for the most part from general texts
or scientific papers that are identified in the ac-
companying legends. We hope that these dia-
grams and photographs will make the essays even
more useful.
For readers who wish to know about specific
research being conducted by HHMI investigators,
we have provided once again a detailed index.
The index lists almost every topic or term, from
abd-A (a gene that directs the segmental develop-
ment of the abdomen — first identified in the fruit
fly Drosophila melanogaster) to the zif268
gene and, in between, the less arcane subjects of
AIDS, cancer, cystic fibrosis, diabetes, hemo-
philia, muscular dystrophy, and obesity. Several
readers have remarked how useful the addition of
an index has been, and we are grateful once more
to Diana Witt for preparation of the index.
The greater part of the volume consists of a
series of short reports by the investigators asso-
ciated with the Institute during 1992. This por-
tion of the volume is again larger than its prede-
cessor, reflecting the increased number of
investigators whose work is represented. In addi-
tion, reports prepared by the Institute's Interna-
tional Research Scholars are included in a section
following the investigators' reports. These 24
scientists are the first recipients of research
grants awarded under the Institute's new interna-
tional initiative. The reports submitted by each
investigator and International Research Scholar
have been collated and edited by Dr. Claire H.
Winestock, Senior Scientific Officer; Elizabeth
Cowley, Copy Editor; William T. Carrigan, Edi-
tor/Writer; Gail Markley, Manager of Publica-
tions; and Kimberly A. Cornejo, Permissions Edi-
tor. Laura North also assisted with the editing of
manuscripts. We are grateful to them and to the
many investigators who have provided us with
suitable illustrations taken from their ongoing re-
search. Not only do the illustrations help to clar-
ify the written text, but they also greatly enhance
the aesthetic appeal of the volume.
Cell Biology and Regulation Program
The Cell Biology and Regulation Program is the
oldest of the Institute's research programs. Origi-
nally referred to as Metabolic Regulation, which
reflected the program's roots in clinical studies
of metabolic and endocrine disorders, the title of
this program was changed six years ago to reflect
more accurately its principal theme — the biol-
ogy of cells, the factors that regulate their normal
growth and distinctive functions, and the ways in
which cells interact with each other. In this sense
the program has come to occupy a critical posi-
tion in the Institute's research endeavors, and it
lies firmly within the mainstream of contempo-
rary biological research.
Introduction
The human body contains several trilHon (i.e.,
milHon million) cells of a thousand or more dis-
tinct types. Research in the field of cell biology
seeks to understand how these various cells are
constructed and organized, how they differ from
one another, how they sense and respond to out-
side influences, how they interact with their
neighbors to form more complex tissues and or-
gans, and, in general, how the cells of the body
are integrated to produce an appropriately func-
tioning organism. Equally important, research in
this area is aimed at understanding how these cel-
lular functions are perturbed by disease. To this
extent the problems addressed by investigators in
the Cell Biology and Regulation Program inevita-
bly impinge on related work in genetics, develop-
ment, neuroscience, and immunology. For a
Medical Research Organization it is especially
gratifying to see how many of the insights gleaned
from these studies are already beginning to throw
light on such medically important problems as
diabetes, heart disease, cancer, muscular dys-
trophy, cystic fibrosis, and a number of other ge-
netic disorders.
The strikingly rapid advances that have oc-
curred in cell biology in recent years have been
due, in large part, to earlier progress in biochem-
istry and cellular physiology, but especially to
the dramatic developments that have occurred in
molecular biology since the early 1960s. The
techniques developed in these fields have been
invaluable to cell biology, which has always been
quick to apply different and newly emerging ap-
proaches to the solution of the many problems of
cell structure and function. To understand the
types of research being conducted in contempo-
rary cell biology, it may be helpful to begin with a
general account of a "typical" animal cell (Fig-
ure 1).
Near the center of each cell is the nucleus,
which contains the genes (Figure 2) that are
made of DNA and encode the information neces-
sary to construct an entire organism and maintain
its day-to-day activities. The entire complement
of genes is the genome, which comprises a set of
instructions encoded in the sequences of the DNA
molecules (as described more fully in the section
on genetics). The human genome consists of 46
chromosomes — 22 pairs of autosomes and 2 sex
chromosomes. The two copies of each autosome
are inherited from the mother and the father, re-
spectively; in females there are two X chromo-
somes (one from each parent) , while in males the
Y chromosome is always inherited from the father
and the X chromosome (Figure 3) is always in-
herited from the mother. The 46 chromosomes
comprise a total of about 3 billion pairs of nu-
cleotides. Estimates vary, but it is thought that
there may be as many as 100,000 genes in the
human genome. These genes vary in length from
around 1,000 to about 2 million nucleotides.
Each gene encodes the information for a particu-
lar cellular structure or function. This informa-
tion, which can be likened to a computer lan-
guage, is first read (transcribed) into RNA, and
the message contained within the nucleotide se-
quence of the RNA molecule is then decoded
(translated) by the machinery of the cell into a
different language, or chemical structure. While
the RNA transcript mirrors exactly the DNA se-
quence of the gene, the messenger RNA (mRNA)
that is translated into the amino acid sequence of
the encoded protein is a highly edited message
(Figure 4). Generally several intervening (or
noncoding) sequences called introns are selec-
tively removed from the transcript. Introns ac-
count for a considerable proportion of the DNA in
all higher organisms and, together with several
other noncoding stretches of DNA (spacer DNA,
satellite DNA, and other repetitive DNA se-
quences) , are sometimes referred to as selfish or
junk DNA. In humans it is estimated that more
than 90 percent of the DNA in the genome is of
this kind. The possible functions (if any) of this
noncoding DNA are not known, but the mecha-
nisms whereby the coding sequences or exons
are spliced out and joined together (sometimes in
different order — a process known as alternative
splicing) is a subject of considerable interest at
present. The chemical language of the cell has 20
different characters or units, known as amino
acids, which are linked together, again in linear
arrays, to make proteins.
Whereas DNA is the blueprint directing the
cell's development and function, proteins are the
molecules from which cells are built and which
carry out most cellular functions. Most genes en-
code proteins, and each cell contains about
10,000 different types of protein. That is, each
cell uses only about 10 percent of the total set of
genes at any one time. This raises two of the cen-
tral questions in cell biology today: 1) How are
genes turned on and off so that each cell type
expresses only its appropriate set of genes and
contains only its correct complement of proteins?
2) How are the genes in a given cell regulated, so
that the cell can respond appropriately to outside
influences by changing either the pattern of
Introduction
Figure 1. A small section of a typical mammalian cell as seen in the electron microscope.
Part of the nucleus (N) with its surrounding membrane and nucleolus (Nu ) are shown in
the upper right; the cytoplasm that occupies the rest of the electron micrograph contains
several different organelles, including several mitochondria (M), parts of the Golgi com-
plex (GC), and the rough endoplasmic reticulum (RER). Magnification approximately
50,000 X.
Micrograph provided by David D. Sabatini.
Introduction
Figure 2. From genes and chromosomes to cells, organs, and entire bodies.
Adapted from an illustration by Warren Isensee for The Chronicle of Higher Educa-
tion, September 3, 1986, with permission from the publisher.
Introduction
X
Figure 3- A schematic representation of the X chro-
mosome showing the location of the characteristic
banding patterns and the distribution of several of
the genes that have been localized to this
chromosome.
Excerpt from Stephens, J.C., Mador, M.L., Cava-
naugh, M.L., Gradie, M.I., andKidd, K.K. The Human
Genome Map. Science 250: October 12, 1990. Copy-
right 1990 by the AAAS.
INTRON
EXON INTKON
(CODING (NONCODING
REGION) REGION)
, 1 ,
PNA-
1
■ 2B 3 ■ 4 ■ 5
rUANSC:RII'TION
CAP
PRIMARY
niRNAl
TRANSt:HIIT
I'OLY A
INTRONS ARE CUT OUT AND
COOING REGIONS ARE
SI'LIGED HX;ETHtR
MATURE inRNA TRANSCRIH 1 |fl TT
Figure 4. The primary RNA transcript of a gene is very much larger than the message that is
actually translated to form a protein. A variable number of noncoding sequences called introns
are edited out, and the remaining exons are spliced together by a complex mechanism shown in
the drawings to the left.
From Dugaiczyk, A., Woo, S.L.C., Colbert, D.A., Lai, B.C., Mace, M.L., Jr., and O'Malley, B.
1979. Proc Natl Acad Sci USA 76:2253-2257.
OCOCOCl I %
Introduction
DNA
mRNA
RNA POLYMERASE
FORMING
PROTEIN CHAIN
FORMING
PROTEIN
Figure 5. Transcription of a gene by the enzyme RNA polymerase and the subsequent translation of this messenger RNA ( mRNA )
by ribosomes to form a protein.
From Raven, P.H., and Johnson, G.B. 1988. Understanding Biology. St. Louis, MO: Times Mirror /Mosby College Publishing,
p. 267.
EXTRACELLULAR
SPACE
apical plasma
membrane
transport
vesicle
Golgi
apparatus
transport
vesicle
membrane-bound
polyribosomes
in ER
Figure 6. The synthesis and intracellular
trafficking of proteins from the endoplas-
mic reticulum through the Golgi apparatus
to the cell membrane.
From Alberts, B., Bray, D., Lewis, J., Raff,
M., Roberts, K., and Watson, J. D. 1989. Mo-
lecular Biology of the Cell, 2nd edition. New
York: Garland, p. 455.
basal plasma membrane
Introduction
genes it uses (and thus the kinds of proteins it
produces) or the amounts of each protein it
makes (Figure 5)? In the next section we describe
the considerable progress that has been made
recently in deciphering the DNA sequence ele-
ments that determine whether a gene is tran-
scribed. Proteins, commonly known as tran-
scriptional factors, bind to these DNA
sequence elements and determine when, and
how frequently, a gene is transcribed. Since this
often influences how much of the encoded pro-
tein is produced by a cell, such information pro-
vides one level of insight about how cells differ
from one another.
Another large issue concerns the way in which
proteins are deployed in the cell. A cell is not just
a bag of randomly distributed molecules; it has a
highly organized internal structure. The DNA that
makes up the genes and the machinery for gene
transcription are packaged in the nucleus, which
is surrounded by a membrane with distinctive
pores that separates it from the cytoplasm,
which comprises the rest of the cell. Nearly all
cellular functions are compartmentalized in
other cellular structures collectively referred to
as organelles. Among the more prominent or-
ganelles in animal cells are the mitochondria,
which are the cells' principal energy source; ly-
sosomes, which are concerned largely with the
degradation of foreign materials and of cellular
proteins whose functions have been fulfilled; the
rough endoplasmic reticulum, a complex, ri-
bosome-studded system of membranes responsi-
ble for the synthesis of proteins secreted by the
cells; and the Golgi apparatus, which both mod-
ifies proteins (by adding other chemical groups
such as sugars) and also packages them for trans-
port to their appropriate locations, such as the
cell membrane. Most of these organelles are
themselves surrounded by membranes that sepa-
rate their functions from those of the rest of the
cell. It is easy to see how such a compartmental-
ized structure allows the cell to organize its dif-
ferent processes efficiently, but it poses an organi-
zational problem that is of considerable current
interest in cell biology: How are particular pro-
teins routed to the correct organelles? As we be-
gan to learn about the structures of individual
proteins, it was discovered that there are specific
"signals" built into proteins that target them to
particular organelles or particular locations
within the cell and that there are distinctive cel-
lular machineries that "detect" these signals and
steer the proteins in particular directions (Figure
6). Thus certain proteins are directed to the nu-
cleus, while others are targeted for insertion into
the surface membrane of the cell, and yet others
are destined for export out of the cell as secretory
products.
At another level of organization, it has become
evident that the organelles within a cell are also
not distributed randomly. In many cells one can
identify a distinct "top" and "bottom." Other
cells, while not polarized in this manner, have
asymmetric structures arranged in such a way that
given organelles are distributed in different, but
quite reproducible, patterns. Still other cells, es-
pecially nerve cells, have unusual extensions or
processes that may be many hundred times as
long as the body of the cell. In each case organ-
elles have to be transported to particular loca-
tions and maintained there; they do not drift
about haphazardly inside the cell.
The asymmetric shapes of cells and the loca-
tions of their organelles both rely on cellular
structures known collectively as the cytoskele-
ton. The cytoskeleton consists of several types of
elongated threads or filaments (microfilaments,
microtubules, intermediate filaments), each
made of specific proteins that are so designed as
to assemble spontaneously into filaments. These
cytoskeletal elements serve as a form of internal
scaffolding to maintain the shape of the cell, and
as a system of tracks along which organelles can
be transported. Recent research has disclosed a
variety of proteins that function as molecular mo-
tors that can move proteins and organelles that
attach to them along particular cytoskeletal fila-
ments to various locations within the cell. We
also know that the appropriate organization of
cytoskeletal filaments and motor proteins can, in
some cases, contribute to cell motility, that is,
the movement of the entire cell from one loca-
tion to another. Such cellular movements are es-
pecially important in development but continue
to play an integral role in the life of certain cells
even in adult life.
All these processes — gene transcription, pro-
tein targeting, organelle movement, and cell mo-
tility— must be carefully regulated so that cells
respond appropriately to different situations. The
same is true of many other cellular processes. For
example, the proliferation of cells that takes
place by cell division involves copying or repli-
cating the genes, the breakdown of the nuclear
membrane, the separation of the duplicated
chromosomes into two equivalent sets, divi-
sion of the cell into two daughter cells, the re-
Introduction
formation of a nucleus in each of the daughter
cells, and finally the resumption of normal func-
tion in both cells. This whole process, which is
known as the cell cycle, takes place whenever
cells divide and remains an important part of the
life of all but a few cell types. Many types of cells
^for example, the cells in the blood and skin —
are continually being formed and replaced.
Other cells proliferate rarely, and some divide
only during early development; the nerve cells
that make up the brain, for example, proliferate
rapidly during development, but no further cell
division occurs before death of the individual
some 70 or more years later. Obviously, cell pro-
liferation must be tightly controlled. Recent ad-
vances have shown that the cell cycle is con-
trolled by a set of proteins whose role is to modify
other proteins selectively and thus regulate their
functions. One common way in which proteins
are modified (but certainly not the only one) is to
attach a small chemical group, such as a phos-
phate group, to a protein. Proteins that attach
phosphate groups to other proteins are called
protein kinases. Control of many aspects of the
cell cycle and, indeed, of many other cellular
functions relies on complex control networks of
protein kinases acting on key proteins at pivotal
stages in the life of the cell.
Thus far we have considered only processes oc-
curring within a cell. An important related set of
issues concerns how cells interact with each
other and how they respond to the external envi-
ronment. Each cell is surrounded by a surface
(or plasma) membrane, which serves as a se-
lective barrier separating the inside of the cell
from the world outside itself. Embedded in this
membrane are several types of proteins. One es-
sential class are transporters, specialized for
the ordered movement in and out of the cell of
nutrients, ions and other small molecules that are
essential for normal cell function.
A second group of cell surface proteins are re-
ceptors, which bind other types of molecules
that interact with the cell. As the name suggests,
receptors serve to receive input from the cell's
external environment. They are of many different
types. The largest group binds peptide hormones
or diffusible factors produced locally or at a dis-
tance by other cells, but another important group
serves to transport materials like cholesterol from
outside to the interior of the cell. Typically these
receptors have three parts: an external part or
ligand-binding domain that can bind the hor-
mone or diffusible factor, a transmembrane do-
main that spans the cell membrane, and an intra-
cellular part that can interact with internal
components of the cell. The binding of a hor-
mone or other diffusible factor to such a receptor
triggers in some way, as yet undetermined, a sig-
nal inside the cell. These signals are of many
types. Some receptors are protein kinases that are
selectively activated by binding the appropriate
external factor; others, when activated, lead to
the release of diffusible, small molecules, such as
calcium ions or cyclic nucleotides. These diffusi-
ble second messenger molecules in turn acti-
vate other control mechanisms inside the cell,
including protein kinases and other regulatory
molecules. In this way the triggering of a recep-
tor from outside cells can result in a cascade of
events that ultimately controls the various intra-
cellular processes discussed earlier. One of the
current "hot topics" in cell biology research
concerns the nature and mechanisms of cell sur-
face receptor signaling and the control circuits
inside cells that link receptor activity to other
control mechanisms, including those that regu-
late gene function and the control of the cell
cycle.
A special subset of this class of receptors are
those that respond to the release of chemical sig-
nals (transmitters) at the specialized endings of
nerve cell processes (Figure 7) . The released neu-
rotransmitters bind to the external part of the cell
surface receptor and in doing so may open an ion
channel or trigger the activation of a second in-
tracellular message. Since the majority of nerve
signals are transmitted from cell to cell in this
way, the analysis of this class of receptors is (as
we shall see in the section on neuroscience)
one of the central issues in contemporary
neuroscience.
Another class of cell surface receptors is in-
volved in the adhesion of cells, either to their
neighbors or to the extracellular matrix, a
complex group of secreted proteins and polysac-
charides that assemble into an organized mesh-
work on the cell surface. Depending upon the
cell type and environment, the extracellular ma-
trix performs various functions (Figure 8). In a
petri dish, for example, the extracellular matrix
provides a cushion on which the cell sits. In the
epidermis, the extracellular matrix helps to form
the basement membrane, which anchors the epi-
dermis to the rest of the skin. In connective
tissues, the extracellular matrix completely
surrounds most cells and is often more extensive
in its distribution than the cells themselves. In
OCOCOC l^l
Introduction
postsynapik: ce)l
Action potential
triggers entry of
calcium ions (Ca^*)
into the presynaptic
terminal.
Synaptic vesicles
fuse withi tfie presynaptic
membrane, releasing
transmitter.
Tfie transmitter binds to proteins
in tfie postsynaptic membrane,
changing their conformation and
either opening an ion channel or
activating a second messenger
system v^/ifhin the postsynaptic cell.
The transmitter is removed
from the synaptic cleft, and
the postsynaptic receptors
revert to their original
conformation.
Figure 7. A summary of the essential events at a chemical synapse following the arrival of an action potential in the axon
terminal.
Adapted from Alberts, B., Bray, D., Lewis, J., Raff, M., Roberts, K., and Watson, J. D. 1989- Molecular Biology of the Cell, 2nd
edition. New York: Garland, p. 1077.
Figure 8. A: A section of human skin stained with an anti
body against the protein keratin. Note how the cells are
packed together to form the distinctive structure of skin. The
dark structures within each cell are the nuclei that do not
contain keratin and so are not labeled.
B: A fluorescence micrograph showing the presence of kera-
tin filaments in cultured skin cells. The keratin filaments
have been revealed by a fluorescently labeled antibody
against the keratin protein.
Courtesy of Elaine V. Fuchs.
xxxvii
Introduction
this case, the extracellular matrix helps to pro-
vide the body's architectural framework.
Cellular adhesion, which plays a crucial role
in cell, tissue, and organ structure and in cell
movements, depends on specialized cell adhe-
sion receptors that are connected to the intracel-
lular cytoskeleton. It is also likely that cells can
signal to one another via cell adhesion receptors.
Decisions as to whether a cell remains stationary,
or where and when it should move, and what
shape it should take up, are all affected by cell
adhesion events. And these, in turn, are largely
dependent on the pattern and functions of cell
adhesion receptors deployed on the surfaces of
the cells.
Elucidating the normal functions of cells is an
important first step in understanding how these
processes go awry in a number of human diseases .
For example, it is now known that alterations in
certain normal genes (called oncogenes) can
contribute to cancer. It is also clear that many
oncogenes encode proteins that are involved in
the regulation of specific cellular functions:
some oncogene-encoded proteins are growth fac-
tors; others are cell surface receptors; yet others
are signaling molecules, protein kinases, and
transcription factors. Many other human diseases
are known to be due to defects of one or another
of the cellular processes reviewed above. Thus
disturbances in insulin production lead to dia-
betes mellitus, defects in the extracellular matrix
can cause osteogenesis imperfecta, and abnormal-
ities in cell adhesion receptors may result in
various bleeding disorders. Indeed, one of the
major insights in pathology and medicine is that
all disease processes are ultimately attributable
to the breakdown of one or more cellular func-
tions. In the twenty-first century, we can be fairly
certain that medicine will be concerned largely
with the identification and treatment of specific
disordered cell functions. We already know of
many disorders that can be attributed to distur-
bances in particular organelles. What is particu-
larly encouraging to researchers in cell biology
today is that new avenues are beginning to be
perceived for therapy, as the molecular bases of
various disordered functions become known. In-
deed, one of the especially appealing aspects of
modern cell biological research is the imme-
diacy with which fundamental research advances
are having an impact on medically important
problems.
Investigators in the Cell Biology and Regulation Program
Alexander-Bridges, Maria C, M.D., Ph.D.
Beach, David H., Ph.D.
Bennett, G. Vann, M.D., Ph.D.
Beutler, Bruce A., M.D.
Bevilacqua, Michael P., M.D., Ph.D.
Blackshear, Perry J., M.D., D.Phil.
Blobel, Gunter, M.D., Ph.D.
Bonadio, Jeffrey F., M.D.
Brugge, Joan S., Ph.D.
Campbell, Kevin P., Ph.D.
Caron, Marc G., Ph.D.'
Carroll, Sean B., Ph.D.
Chin, William W., M.D.
Crab tree, Gerald R., M.D.
Craig, Nancy L., Ph.D.
Cunningham, James M., M.D.
Davis, Laura I., Ph.D.
Davis, Roger J., M.D.
Dreyfuss, Gideon, Ph.D.
Emr, Scott D., Ph.D.
Esmon, Charles T., Ph.D.
Exton, John H., M.D., Ph.D.
Fuchs, Elaine, Ph.D.
Ganem, Donald E., M.D.
Garbers, David L., Ph.D.
Gething, Mary-Jane H., Ph.D.
Glomset, John A., M.D.
Gomer, Richard H., Ph.D.
Grosschedl, Rudolf, Ph.D.
Habener, Joel F., M.D.
Heintz, Nathaniel, Ph.D.
Hynes, Richard O., Ph.D.
Isberg, Ralph R., Ph.D.
Kaback, H. Ronald, M.D.
Kim, Peter S., Ph.D.
Kirkegaard, Karla A., Ph.D.
Kobilka, Brian K, M.D.
Lai, Michael M.-C, M.D., Ph.D.
Lamb, Robert A., Ph.D., Sc.D.
Lefkowitz, Robert /., M.D.
Lehmann, Ruth, Ph.D.
Maas, Richard L., M.D., Ph.D.
Mailer, James L., Ph.D.
Massague, Joan, Ph.D.
McKnigbt, Steven Lanier, Ph.D.
Nusse, Roel, Ph.D.
O'Donnell, Michael E., Ph.D.
Parker, Keith L., M.D., Ph.D.
Pike, Linda J., Ph.D.
Sadler, J. Evan, M.D., Ph.D.
Schekman, Randy W., Ph.D.
Schoolnik, Gary K., M.D.
Shenk, Thomas E., Ph.D.
Sherr, Charles J., M.D., Ph.D.
xxxviii
Introduction
Simpson, Larry, Ph.D.
Spradling, Allan C, Ph.D.
Steiner, Donald P., M.D.
Welsh, Michael J., M.D.
Williams, Lewis T., M.D., Ph.D.
Wilson, James M., M.D., Ph.D.
Genetics Program
The emergence of what is sometimes referred
to as the "new genetics" has contributed more to
our fundamental understanding of biology and
medicine in the past two decades than any other
advance in biomedical science. Not surprisingly,
this development has come to assume a central
place in all biological research, making it possi-
ble to examine biological processes at a level of
resolution that was considered quite impossible
just 20 years ago. "Gene cloning," "recombinant
DNA technology," "genetic engineering," and
the "Human Genome Initiative" are phrases that
have entered everyday language, but the possibili-
ties they offer for major advances in biology and
medicine have yet to be fully appreciated. Given
the central role of genetics in modern biology
and the current sense of excitement that the new
genetics has generated, it is appropriate that the
Genetics Program should be the largest research
program within HHMI .
Historically, one of the first applications of
Mendel's classic laws of inheritance was to the
analysis of certain human diseases. In the early
years of this century, Archibald Garrod, an En-
glish physician, noted that a number of relatively
rare diseases tended to occur in families, often in
families with consanguineous marriages. The pat-
tern of occurrence of these rare diseases followed
Mendel's laws, discovered almost 50 years ear-
lier. Mendel's work on inherited characteristics
in plants was neglected for many years, but after it
was rediscovered, around the turn of the century,
it was quickly established that the laws of genet-
ics are universal and govern inheritance in organ-
isms as disparate as peas and worms, mice and
fruit flies, bacteria and human beings. Beginning
with Garrod, these genetic laws were applied to a
host of inherited diseases, as it was realized that
the genetic makeup of an individual can have a
profound effect on his or her health and well-
being. However, an understanding of what genes
are and how they function had to await the discov-
ery, in 1944, that the genetic material is DNA,
and, in 1953 and 1961, of the double helix and
the genetic code, respectively.
Much of our early understanding of the action
Wolin, Sandra L., M.D., Ph.D.
Yanagisawa, Masashi, M.D., Ph.D.
* This investigator was appointed after manu-
scripts were submitted for publication. His re-
search will be described in the next volume.
of genes came from experiments that took advan-
tage of the universality of gene action by using
simple organisms, especially bacteria and their
viruses, as model systems. The cardinal discovery
was the identification of deoxyribonucleic acid
(DNA) as the fundamental chemical in which ge-
netic information is encoded. But the finding that
DNA has a double-stranded, mirror image struc-
ture provided the first clear insight as to how this
information could be replicated and passed on
from one generation of organisms to the next (Fig-
ures 9 and 10). Understanding how the chemical
language of DNA could be used to direct the syn-
thesis of other cellular constituents, especially
proteins, came with the discovery of the nature of
the genetic code. These great advances will al-
ways be viewed as the high watermark of the early
molecular stage of modern genetics.
Notwithstanding these dramatic developments
— arguably the most important in biology since
the publication of Darwin's essay On the Origin
of Species in 1859 — the molecular details of the
genes of higher organisms remained hidden from
view by the enormous complexity of their ge-
nomes (Figure 11). Fortunately, the genes of sim-
ple organisms were accessible, because they are
relatively few in number (involving in many
cases an assemblage of as few as 3,000 base
pairs, as the building blocks of DNA are called)
compared to the human genome, which probably
contains about 3 billion base pairs.
The problem of genetic complexity has been
finally overcome in the past 1 0 years by the devel-
opment of the powerful new genetic methods
known collectively as recombinant DNA tech-
nology. This technology allows researchers to
isolate specific genes from complex mixtures,
to prepare them in sufficiently large amounts
that their entire molecular structure can be de-
termined, and to move them from one group of
cells or from one organism to another, so that
their functional properties can be identified and
their products produced in abundance. In a num-
ber of cases, medically valuable products such
as human insulin, growth hormone, antihemo-
philic factor, erythropoietin, and TPA (tissue
rV^ 4 fV*
»Ar fcA' »vV
Introduction
sugar-phosphate
backbone
Figure 9- The famous DNA dou-
ble helix.
Prom Alberts, B., Bray, D.,
Lewis, J., Raff, M., Roberts, K.,
and Watson, J.D. 1989. Molecu-
lar Biology of the Cell, 2nd edi-
tion. New York: Garland, p. 99.
Figure 10. An illustration of how DNA is repli-
cated so as to form two new strands that are
exactly complementary to the original DNA
sequences from which they derive their
sequences.
From Raven, P.H., and Johnson, G.B.
1988. Understanding Biology. St. Louis, MO:
Times Mirror /Mosby College Publishing,
p. 277.
xl
chromosome of 1 50,000,000 nucleotide pairs, containing about 3000 genes
0.5% of chromosome, containing 15 genes
one gene of 100,000 nucleotide pairs
1^
regulatory region
intron
exon
DNA TRANSCRIPTION
primary RNA transcript
RNA SPLICING
3'
mRNA
Figure 11. The organization of genes on a typical vertebrate chromosome. Proteins
that bind to the DNA in regulatory regions determine whether a gene is transcribed;
although often located on the 5' side of a gene, as shown here, regulatory regions can
also be located in introns, in exons, or on the 3' side of a gene. The intron sequences
are removed from the primary RNA transcripts that encode protein molecules to
produce a messenger RNA (mRNA ) molecule. The figure given here for the number of
genes per chromosome is only a minimal estimate.
From Alberts, B., Bray, D., Lewis, J., Raff, M., Roberts, K., and Watson, J.D. 1989.
Molecular Biology of the Cell, 2nd edition. New York: Garland, p. 487.
Introduction
Normal hemoglobin beta chain
Vol
ine
Histidi
me
Leucine
Th
reonme
Prol
ine
Sickl e-cell anemia hemoglobin beta chain
Vali
ne
Histidi
ne
Leucine
Th
reonme
Proli
ne
Glutamic
acid
Vol
ine
Glutamic
acid
Glutamic
acid
Figure 12. A drawing of the hemoglo-
bin molecule and the single-amino
acid change in the P-chain of the mole-
cule that causes sickle cell anemia.
From Raven, P.H., and Johnson,
G.B. 1988. Understanding Biology. St.
Louis, MO: Times Mirror/Mosby Col-
lege Publishing, p. 25 5.
Figure 13. We can determine where particular genes
are being expressed by using a radioactively labeled
probe that selectively binds to the UNA transcript of
the gene and examining the distribution of the radio-
activity in a tissue autograph. This technique, known
as in vitro hybridization, was used in this case to show
the expression of one of the keratin genes in a hair
follicle. The micrograph on the left is viewed under
darkfield optics ( the exposed silver grains in the auto-
radiograph appear as white dots ); that on the right is
seen with conventional optics, and the radioactivity is
marked by black dots.
Courtesy of Elaine V. Fuchs.
xlii
Introduction
plasminogen activator) have been commercially
manufactured in this way and are now being used
therapeutically.
Most of these accomplishments depend on the
technique commonly referred to as gene clon-
ing. In this process a gene (for example, the hu-
man insulin gene, which is a few thousand base
pairs in length) is isolated from a complex mix-
ture of human DNA fragments and transferred
into the genetic apparatus of a much simpler or-
ganism such as a bacterium. Since bacteria multi-
ply very rapidly, the inserted human insulin gene
is amplified (cloned) along with the genes of its
simple host. It is then a relatively straightforward
process to purify the cloned gene and to deter-
mine the sequence of nucleotides that encodes
the relevant genetic instructions.
One of the most important developments to
emerge from gene cloning and sequencing is
that they permit geneticists to determine exactly
how a gene may have been altered (genetic alter-
ations of this kind are called mutations) to pro-
duce an inherited disease. While a majority of the
investigators in the HHMI Genetics Program are
directing their efforts toward understanding the
principles that govern the action of genes, a large
and growing number are trying to identify the
mutations responsible for some of the estimated
4,000 human genetic defects. Among the genes
being studied are those responsible for mus-
cular dystrophy, cystic fibrosis, several forms
of hemoglobinopathy, chronic granulomatous
disease, phenylketonuria, polyposis coli, neuro-
fibromatosis, osteogenesis imperfecta, and hemo-
philia. Most of the genes responsible for these
diseases have already been cloned and their
structures determined; in this way the precise mo-
lecular effects of the mutations that cause the dis-
orders are now being determined. Information
that has been derived in this way is being used in
several ways. For example, in some cases it is be-
ing used to counsel affected families about the
risks they face in having additional children; in
others it is being used to develop tests that are
critical for prenatal diagnosis, and, in the case of
hemophilia and chronic granulomatous disease,
it is already being used to provide appropriate
therapy. Information about some of these genes
and their protein products has also told us a great
deal about the role of the relevant proteins in
development (e.g., the protein dystrophin,
which is normally found in muscle but is missing
in most cases of muscular dystrophy) or about the
mechanisms required for normal function (e.g..
in the lungs and pancreas in cystic fibrosis, or for
the mechanisms of blood clotting in hemo-
philia). It is the continuous interplay between
the development of new basic knowledge and its
application to the understanding of human dis-
ease that is so dramatically informing modern
medicine.
The execution of a genetic program in a cell
obviously cannot be left to chance. To bring
about an ordered series of changes in a develop-
mental or metabolic process, each genetic in-
struction must be activated at a specific time, and
the product of the gene must be produced in an
amount commensurate with the needs of the or-
ganism. As was pointed out in the section on cell
biology, different sets of genes are expressed in
each cell type so, for example, many of those ex-
pressed in muscle cells are likely to be quite dif-
ferent from those expressed in skin cells or in
bone. Since with few exceptions (one of which
will be discussed in the section on immunology)
each cell contains the same genetic information,
there must be some mechanism or mechanisms
that insure that the appropriate genes are being
expressed in each cell type. The processes that
govern these crucial steps are generally referred
to as gene regulation.
Once again, the study of inherited human dis-
eases has revealed mutations that can disrupt
those regulatory programs. For example, it is now
known that certain inherited anemias are brought
about by specific mutations that affect the regula-
tory apparatus of the genes that encode the red
blood cell protein, hemoglobin (Figure 12). An
even larger number of examples could be cited
from work on simpler organisms to show how
powerful this mutational analysis has been for
our understanding of genetic regulation. How-
ever, it will suffice to say here that the work of a
considerable number of HHMI investigators has
this as its primary theme.
As discussed in the section on cell biology and
regulation, genetic information is usually con-
veyed through a series of steps from the nucleus
of the cell to the surrounding cytoplasm, where it
is decoded or translated to form the protein prod-
ucts that execute the relevant genetic program.
This process involves making many copies of the
gene, in the closely related chemical, RNA. The
process of copying a gene from DNA into RNA is
called transcription. This is one of the key
points at which the flow of genetic information
can be regulated. Genes that are not scheduled
for expression in a particular cell type are not
xliii
Introduction
transcribed: while still present, they remain si-
lent and ineffective.
Since it plays such a central role in regulating
the expression of genetic information, it is not
surprising that the study of transcription is one of
the most active areas in modern genetic research
(Figure 13). Investigators have taken two success-
ful approaches in an effort to understand this pro-
cess. One involves the study of mutations that
disrupt regulation in a variety of organisms. The
other is biochemical and involves isolating spe-
cific proteins that are parts of the transcription
machinery and then determining their mode of
action at a molecular level. The ultimate goals of
both approaches are to provide a complete ac-
count, in chemical terms, of the processing of
genetic information and to provide a sound basis
for our understanding of the types of disturbances
in this process that can lead to various genetic
defects and many diseases in adult life.
It is perhaps worth commenting on the broad
spectrum of organisms that are currently used as
models in genetic investigations. Yeast, worms,
fruit flies, bacteria, viruses, mice, and human be-
ings all provide instructive paradigms. For exam-
ple, because the structure of the nervous system
is relatively well understood in the primitive
worm, Caenorhabditis elegans, it has proved to
be particularly useful for elucidating the genetic
program that directs the formation of the nervous
system. The fruit fly Drosophila tnelanogaster,
which has an uncommonly rich background of
genetic variation, has proved to be the most valu-
able organism for studying the genes involved in
early embryonic development. Similarly, the
mouse, with a generation time of just a few
months, is a convenient stand-in for genetic ex-
periments that cannot be contemplated in most
mammals or human beings. Thus it has become
an everyday technique to introduce new genes
into the genetic makeup of mice. Many of the
resulting animals, called transgenic mice, have
proved to be powerful models for certain human
diseases. Transgenic mice that are genetically
cancer-prone or that have sickle cell anemia have
been generated recently and are proving useful
not only for understanding the molecular bases of
these disorders but also for exploring possible
means for their treatment. Complementing this
approach has been the development of tech-
niques that allow one to knock out specific genes
selectively and to generate mice of essentially any
specified genotype. The power of this new meth-
odology (that involves homologous recombi-
nation) for understanding mammalian develop-
ment, for studying the function of the nervous
system, and for modeling known human diseases
has already captured the attention of many genet-
icists, including several in the HHMI Genetics
Program. At the other end of the spectrum, the
simple baker's yeast, Saccharomyces cerevisiae,
is proving to be especially useful for cloning
large segments of mammalian chromosomes. Spe-
cial carrier elements called YACs (yeast artificial
chromosomes) have been developed that can ac-
commodate up to 500,000 base pairs of DNA.
Since this is more than 20 times as large as the
fragments that have traditionally been cloned in
bacteria and viruses, this approach and the asso-
ciated method for separating large DNA fragments
are beginning to play an important part in the
international effort to map the human genome.
The development of a complete map of the
human genome is one of the great challenges
engaging the attention of geneticists worldwide,
including several HHMI investigators. Such a map
will permit the identification of genes that are
close to, or responsible for, a large number of as
yet uncharacterized genetic diseases. Further-
more, knowledge of the map will allow the devel-
opment of easily identifiable genetic markers for
specific diseases. These genetic markers, called
restriction fragment length polymorphisms
(RFLPs), can be used to detect carriers of many
genetic disorders, to determine paternity, and to
identify individuals for forensic purposes (a pro-
cess commonly referred to in the press as DNA
fingerprinting) .
Four or five years ago an extremely sensitive
technique for detecting these DNA marker frag-
ments was developed. This technique (the poly-
merase chain reaction [PGR]) greatly amplifies
the genetic signal and has made these studies
both simpler and faster. With this and other
emerging technologies, the structures of the ge-
nomes of several simple bacteria should be
known in the near future, and the way is already
clear to begin the systematic structural analysis of
the genomes of a number of more complex organ-
isms, such as the nematode worm C. elegans and
the fruit fly Drosophila. It is confidently pre-
dicted that with further improvements in the
technology for sequencing genes, the complete
structure of the human genome should be known
within 12 or 15 years.
While the new genetics has opened the door to
an understanding of an important range of funda-
mental problems such as the mechanism of ch«-o-
xliv
Introduction
mosome replication, the genetic basis of embry-
onic development, and the control of gene
regulation, it also holds promise for an assault on
the even more complex problems that confront
those interested in human disease. Cancer has a
clear genetic basis. Specific genes, generally those
concerned with cell growth and gene regulation,
have been shown to be involved in many different
malignancies. Other disorders, such as heart dis-
ease and hypertension, have clear genetic compo-
nents. Even distressing behavioral disorders like
manic depression and schizophrenia have a genetic
basis that may offer important clues to their nature
and perhaps, in time, to their treatment.
The real power of the new genetics is that it
Investigators in the Genetics Program
Barsh, Gregory S., M.D., Ph.D.
Beachy, Philip A., Ph.D.
Beaudet, Arthur L., M.D.
Bell, Graeme I., Ph.D.
Bellen, Hugo J., D.V.M., Ph.D.
Belmont, John W., M.D., Ph.D.
Brown, Patrick O., M.D., Ph.D.
Capecchi, Mario R., Ph.D.
Caskey, C. Thomas, M.D.
Cech, Thomas R., Ph.D.
Church, George M., Ph.D.
Cohen, Stephen M., Ph.D.
Collins, Francis S., M.D., Ph.D.
Corden, Jeffry L., Ph.D.
Cullen, Bryan R., Ph.D.
Desiderio, Stephen V., M.D., Ph.D.
Desplan, Claude, Ph.D.
Donelson, John E., Ph.D.
Duyk, Geoffrey M., M.D., Ph.D.
Feinberg, Andrew P., M.D., M.P.H.
Francke, Uta, M.D.
Friedman, Jeffrey M., M.D., Ph.D.
Geliebter, Jan, Ph.D.
Gesteland, Raymond F., Ph.D.
Ginsburg, David, M.D.
Gitschier, Jane M., Ph.D.
Henikoff, Steven, Ph.D.
Norwich, Arthur L., M.D.
Kadesch, Thomas R., Ph.D.
Kan, Yuet Wai, M.D., D.Sc.
Kaufman, Thomas C, Ph.D.
Kunkel, Louis M., Ph.D.
Kurnit, David M , M.D , Ph.D.
Laimins, Laimonis A., Ph.D.
Lalouel, Jean-Marc, M.D., D.Sc.
Leder, Philip, M.D.
Ledley, Fred D., M.D.
Liebhaber, Stephen A., M.D.
Lindquist, Susan L., Ph.D.
allows investigators to approach these complex
problems using a reductionist approach. If, for
instance, a gene contributes to the development
of a behavioral disorder, it is reasonable to as-
sume that it also plays an important role in pro-
gramming normal behavior. By identifying such
genes and understanding the chemistry of their
protein products, we should obtain important in-
sights about the biological basis of behavior.
Thus the ultimate promise of the new genetics is
that it will help us to understand diseases that are
not commonly thought of as being primarily ge-
netic and, in doing so, it will tell us much about
those biological processes that lie at the very core
of our humanity.
Lowe, John B., M.D.
Malim, Michael H., Ph.D.*
Nabel, Gary J., M.D. Ph.D.
Nadal-Ginard, Bernardo, M.D., Ph.D.
Nathans, Daniel, M.D.
Nathans, Jeremy, M.D., Ph.D.
Nevins, Joseph R., Ph.D.
Nussbaum, Robert L., M.D.
Olson, Maynard V., Ph.D.
Or kin, Stuart H., M.D.
Overbeek, Paul A., Ph.D.
Page, David C, M.D.
Palmiter, Richard D., Ph.D.
Perrimon, Norbert, Ph.D.
Reeders, Stephen T., M.D.
Rosbash, Michael, Ph.D.
Sakonju, Shigeru, Ph.D.
Seidman, Jonathan G., Ph.D.
Singh, Harinder, Ph.D.
Soriano, Philippe M., Ph.D., D.Sc.
Steitz, Joan A., Ph.D.
Sternberg, Paul W., Ph.D.
Sukhatme, Vikas P.. M.D., Ph.D.
Taub, Rebecca A., M.D.
Thummel, Carl S., Ph.D.
Tilghman, Shirley M., Ph.D.
Tjian, Robert, Ph.D.
Valle, David L., M.D.
Warren, Stephen T., Ph.D.
Weintraub, Harold M., M.D., Ph.D.
White, Raymond L., Ph.D.
Williams, David A., M.D.
Woo, Savio L. C, Ph.D.
Young, Michael W., Ph.D.
* This investigator was appointed after manu-
scripts were submitted for publication. His re-
search will be described in the next volume.
xlv
Introduction
Immunology Program
One of the most important developments that
occurred during the evolution of vertebrates was
the emergence of protective mechanisms that en-
able animals to defend themselves against inva-
sion by foreign organisms, especially viruses, bac-
teria, and a number of disease-causing parasites.
These defenses employ tw^o distinct but related
strategies: the first detects, v^^ith exquisite sensi-
tivity, the presence of foreign organisms, cells, or
proteins; the second involves a series of mecha-
nisms that act cooperatively to rid the body of the
invading organisms or, at the least, to neutralize
their potentially harmful effects. The task of ef-
fecting both strategies falls to the immune
system.
Recognizing the strategic importance of the
immune system in both health and disease, the
Institute selected Immunology to be one of its
earliest research programs. The wisdom of that
decision has been amply borne out by the truly
remarkable progress that has been made in immu-
nology in the past two decades. With the notable
exception of molecular genetics, no field of bio-
medical research has witnessed such an astonish-
ing series of successes at almost every level, from
understanding the immune system's unique rec-
ognition mechanisms to the elucidation of the
cellular and chemical means used to destroy or
neutralize invading organisms.
The body's initial line of defense against inva-
sion by foreign organisms is the continuously pa-
trolling system of macrophages and other types
of blood-borne phagocytic cells that act both as
an early warning system and as a "first-strike" de-
fense. These cells respond by ingesting and
breaking up the invading organisms and by
releasing soluble signaling molecules like
interleukin-1 that serve, among other things, to
mobilize the next line of defense, the immune
response (Figure 14). This response involves
two classes of lymphocytes, called T and B ceils,
reflecting their origin from the thymus and bone
marrow, respectively.
The first step jn the immune response is the
activation of a special subclass of T lympho-
cytes called helper T cells. Macrophages pre-
sent fragments of foreign proteins, or antigens,
on their surfaces. Recognition of these antigens
by specialized receptors found on helper T cells
then initiates the two responses: a cell-mediated
immune response and a humoral immune re-
sponse. The cell-mediated response involves
principally the stimulation of another subclass of
T lymphocytes called cytotoxic T cells that rec-
ognize and destroy infected cells. The humoral
response, on the other hand, involves the activa-
tion of the second major class of lymphocytes, the
B cells, to produce circulating antibodies. Anti-
bodies recognize and neutralize soluble antigens
and mark cells or viruses bearing antigens for de-
struction by phagocytic cells.
One of the central problems in immunology
concerns the way in which lymphocytes recog-
nize antigens. The complexity of this problem
may be gauged from the observation that humans
and other higher vertebrates are capable of form-
ing antibodies against virtually any molecule or
part of a molecule (epitope), including even
those that do not occur naturally but are chemi-
cally synthesized in a laboratory. How does this
occur? And how does the immune system distin-
guish foreign molecules from those produced by
its own cells? In a word, how do lymphocytes
distinguish self from non-self?
The key to the first issue, as we now know, is to
be found in the almost unlimited variety of re-
ceptors on the surfaces of lymphocytes. The dis-
covery of how just a few hundred genes are capa-
ble of producing such extraordinary receptor
diversity is one of the great success stories of mod-
ern immunology. The essential features of the im-
mune system's capacity for generating molecular
diversity can be summarized briefly by stating
that lymphocyte receptors are formed by pairs
of protein chains that are chemically linked to
form a complex receptor structure. Each chain of
the pair has a constant domain and a variable do-
main. The variable domain of the two chains is
responsible for antigen recognition and the dis-
crimination between self and non-self. The con-
stant (invariant) domain is physically linked to
other membrane proteins of the receptor com-
plex that activates the lymphocyte's internal sig-
naling and effector mechanisms. T and B cells
triggered via their antigen receptors respond to
auxiliary signaling molecules by proliferating
and differentiating to a mature effector stage. In
the case of B cells, the maturation process ulti-
mately results in the generation of plasma cells
that produce large amounts of antibody for secre-
tion into bodily fluids, chiefly the bloodstream.
The complex structure of the variable parts
of the receptors is due to several processes.
First, and most important, the genes responsi-
ble for this portion of the receptor are assembled
from a large number of different gene segments
xlvi
Introduction
Pathogens Humoral immunity
Lymphokines Lysis
Figure 14. When a pathogen invades the body, the immune system responds with three types of reaction. The cells of the
humoral immune system (B cells ) secrete antibodies that can bind to the pathogen. Cells of the cellular system (T cells ) carry
out two major types of functions. One type of T cell (the cytotoxic T cells, CTL ) develops the ability to kill pathogen-infected
cells. Helper T cells, on the other hand, secrete factors (lymphokines) that stimulate the body's overall response.
From Molecular Cell Biology, 2nd edition. James Darnell, Harvey Lodish, and David Baltimore, p. 1005. Copyright ® 1990
by Scientific American Books, Inc. Reprinted with permission by W.H. Freeman and Company.
xlvii
Introduction
Vh segments (hundreds)
D segments (-20)
Jh segments (4)
Germ-line
organization
iVH
First rearrangement
1 Iv.i ioi iol..!
|jh
N
Second rearrangement
Vh
Jh IJh
E
N N
Figure 15. The specificity of antibodies and the receptors on T cells is brought about by a series of complex
rearrangements of various gene segments like those shown in this figure.
From Molecular Cell Biology, 2nd edition. James Darnell, Harvey Lodish, and David Baltimore, p. 1027.
Copyright ® 1990 by Scientific American Books, Inc. Reprinted with permission by W.H. Freeman and
Company.
Figure 16. A: A molecular model of an antibody mol-
ecule. Each amino acid is represented by a small
sphere. The heavy chains are colored blue, the light
chains red. The four chains wind about one another
to form a Y shape, with two identical antigen-bind-
ing sites at the arms of the Y and a tail region that
serves to direct the antibody to a particular portion
of the immune response.
B: A schematic drawing of an antibody molecule.
Each molecule is composed of two identical light (L)
chains and two identical heavy (H) chains. Carbo-
hydrate is sometimes associated with the H chain.
While the antigen- binding sites are formed by a com-
plex of both H and L chains, the tail region is formed
by H chains alone.
From Raven, P.H., and Johnson, G.B. 1988. Under-
standing Biology. St. Louis, MO: Times Mirror/Mosby
College Publishing, p. 692.
xlviii
Introduction
THYMUS
Precursors
Hemopoietic
Stem Cell
CD3/TCR/S
o
/
Epithelial
Stromal CD3/TCRa/3
Cells
CDS/TCR/S
^(~\- ySTCell
— ^rnn +
Cytotoxic T Cell
\ CD3/TCRa/3 X^-^CD8-
(J) (^^^^^f^-=^ ^ CD3/TCR 0/3
CD4,8+ Helper T Cell
CD4'
^~.CD8±
IgM + lgD igM
a— o^(s>^
Natural Killer Cell
Antibodies
Stromal Cell
sIgM IgM+lgD+IgG, 2 30,4 '9Gi,2.3or4
cllis^ ^ \\lgm+lgD+lgA,<„2 IgA, or
Fetal Liver
Bone Marrow
igM
igG,
2.3 or 4
IgM+lgD+lgE
IgA, or 2
Mature B Cells
Plasma Cells
Figure 17. Contemporary model of immune system development in mammals. T and B cells, which form the major
recognition and effector limbs of the immune system, are derived from multipotent stem cells that also give rise to the
other types of blood cells. Definition of the T cell receptors and accessory cell surface molecules has allowed the
identification of two major sublineages of the thymus-derived T cells, each of which expresses a different type of T cell
receptor and exhibits specialized function. The B lineage cells, generated in hemopoietic tissues, initially express one
class of antibody receptors called immunoglobulin M ( IgM ), but may then switch to the expression of other antibody
classes, each of which has special biological advantages. Note the third lineage of lymphocytes, the natural killer
cells. They are the most recently recognized members of the lymphocyte family, but may well prove to be the most
phylogenetically ancient elements in this constellation of cells involved in host defense.
Courtesy of Max D. Cooper.
Introduction
Figure 18. A schematic representation of
how the thymus gland selects T cells that are
able to recognize foreign but not self
proteins.
From Molecular Cell Biology, 2nd edition.
James Darnell, Harvey Lodish, and David
Baltimore, p. 1037. Copyright ® 1990 by
Scientific American Books, Inc. Reprinted
with permission by W.H. Freeman and
Company.
Bone-marrow cells
Rearrange gene segments to make
a unique T-cell receptor (TCR)
for the cell surface
TCR
Thymus
gland
Into the
selection
system
Rejected: TCR binds to
self-MHC plus self-peptides
(negative selection)
Rejected: TCR does not bind
tightly enough to self-MHC
(positive selection)
Accepted (-5% of cells): TCR
appropriate to recognize foreign
peptides bound to self-MHC
j I To peripheral lymphoid
y J system
Introduction
(Figure 15). Each gene segment exists in several
— and in some cases hundreds — of different cop-
ies. These segments randomly recombine to form
new genes that encode the virtually limitless re-
pertoire of recognition elements. To take just one
example, T cells form their receptors by combin-
ing a number of different gene sequences: V
(variable), D (diversity), and J (joining) seg-
ments. From this array any given T cell derives 1
from about 100 possible V segments, 1 from
about 6 D segments, and 1 from about 50 J seg-
ments to form its so-called a- or heavy chain,
and about 1 in 20 V, 1 in 2 D, and 1 in 12 J
segments to form its light chain. The random
recombination of V, D, and J segments in the two
chains can thus code for literally millions of dif-
ferent possible receptor structures. The receptors
on B cells are formed in basically the same way,
although the numbers of V, D, and J segments
available for selection and recombination differ
somewhat. Antibodies (which are secreted by
plasma cells) are generated in much the same
way as their receptors, and have the same almost
unlimited capacity for diversity (Figure 1 6) .
At one time it was thought that antigens were
capable of shaping the structure of lymphocyte
receptors and antibodies so that the binding sites
of receptors and antibodies would mold them-
selves in some way to fit the shape of the antigen,
much as a rubber glove molds itself to fit one's
hand. We now know that this "instructional" hy-
pothesis is wrong. Rather, as indicated above, the
immune system produces very large numbers of
different types of receptors and antibodies, and
collectively these can "fit" essentially every pos-
sible antigen (Figure 1 7) . Each T or B cell bears
only one type of receptor on its surface (although
there are thousands of receptor molecules of that
given structure on each cell). In the same way,
each B cell secretes antibodies of only a single,
defined structure. Thus the capacity of the body
to respond to an enormous variety of different
antigens is due to the existence of an enormous
number of different T or B cells, each able to rec-
ognize a single antigen (or more commonly, a
part of a complex antigen known as an antigenic
determinant). And when the cell recognizes
and binds to an antigen, it responds by proliferat-
ing to form a large number of cells of the same
type. Such a population of cells, all derived from
a single progenitor, is known as a clone, and the
hypothesis put forward to account for the selec-
tive proliferation of lymphocytes of particular re-
ceptor type in response to a specific antigen is
known as the clonal selection hypothesis. This
theory, first advanced by Sir F. Macfarlane Burnet,
has withstood every test and is rightly viewed as
one of the cornerstones of modern immunology.
The intriguing question as to how the cells of
the immune system distinguish foreign mole-
cules from those on the surface of the cells of
their own host was also addressed by Burnet. He
suggested — and there is now a large body of evi-
dence to support this view — that lymphocytes
that recognize the body's own tissues (so-called
self antigens) are selectively eliminated during
early development — a process known as clonal
deletion. The mechanisms responsible for clonal
deletion of T and B cells are still under investiga-
tion, but at least for the T cells it appears that
during development the thymus may actively se-
lect, for export to the rest of the immune system,
only those T cells that are capable of functioning
in the host (Figure 18). These "useful" T cells
are allowed to survive and mature, while the po-
tentially harmful cells die and are removed. This
process results in the death of about 90 percent
or more of the T cells that are initially formed.
The bone marrow selection of B cells for survival
may be equally stringent. Although this seems an
astonishingly wasteful process, comparable cell
deaths are known to be a rather common feature
in the development of virtually all organs and of
all multicellular organisms.
For reasons that remain to be elucidated, in
some conditions — commonly referred to as au-
toimmune disorders — the immune system may
mistakenly mount an attack on components of the
host organism's own cells. For example, the neu-
rological condition myasthenia gravis involves
the production of circulating antibodies directed
against the receptor molecules on the surfaces of
muscle cells that normally enable them to re-
spond to the release of the neurotransmitter ace-
tylcholine from the motor nerves. When the re-
ceptor molecules are damaged or destroyed,
there is a progressive loss of neuromuscular con-
trol, and if the respiratory muscles are involved
the condition may be fatal. Similarly, type I or
juvenile-onset diabetes is now known to be due
to the combined attack of T cells and antibodies
directed against the /S-cells of the pancreas that
normally produce insulin, the hormone that regu-
lates sugar metabolism.
Another topic of considerable current interest
in immunology concerns antigen presenta-
tion. We now know that this involves a complex
set of genes called the major histocompatibil-
li
Introduction
ity complex (MHC). The membrane proteins
encoded by these genes (of which there are two
types called class I and class II) are able to selec-
tively bind short segments of partially digested
protein antigens, termed peptides. These pep-
tides, arising from protein breakdown inside the
cell, reach the surface of the cell together with
the relevant MHC molecule. Recent x-ray crystal-
lographic studies indicate that the peptide anti-
gen is lodged within a distinctive groove on the
outer surface of the MHC molecule, where it can
be detected by a lymphocyte bearing the appro-
priate receptor. The receptors on T cells are spe-
cialized to recognize antigens only in the form of
such a peptide: MHC complex. There are acces-
sory molecules on the surfaces of T cells, called
CD4 and CDS, that are selectively expressed on
cells that recognize antigens presented by MHC II
and MHC I molecules, respectively. Functionally
these accessory molecules form part of the T cell
receptor for peptide:MHC complexes by binding
to both the MHC molecule and the T cell recep-
tor. For this reason CD4 and CDS are sometimes
called co-receptors.
Like the antigen receptors on B and T cells,
MHC molecules show considerable diversity.
However, this diversity is not due to the recombi-
nation of different gene segments but rather to
genetic polymorphism. There may be as many as
100 different genetic sequences (alleles) at a
single MHC locus, and T cells are selected during
development only if they can recognize peptides
presented by self MHC molecules. How this oc-
curs is unknown, but its role in T cell selection is
obviously fundamental.
The CD4-bearing T cells (also called T4 cells,
helper lymphocytes, or CD4^ cells) have become
widely known because of their role in the devel-
opment of AIDS (acquired immune defi-
ciency syndrome) . The virus that causes AIDS
— the human immunodeficiency virus (or
HIV) — selectively invades these cells, because
the CD4^ molecule fortuitously serves also as a
specific receptor for the virus (Figure 19). On
entering the CD4^ T cells, the genetic material of
the virus (which is formed of RNA) is reverse
transcribed into DNA, and this, in turn, becomes
integrated into the T cell's own genome. In this
way the virus subverts the cell's genetic machin-
ery and, when activated, the cell produces more
and more virus, until ultimately the cell is killed.
When the cell dies it releases virus into the bodily
fluids, where it is free to invade other CD4^ T
cells, and the whole process may be repeated un-
til the entire T cell population is effectively de-
pleted. Since, as we have seen, T helper cells are
essential for mounting both cell-mediated and
humoral immune responses, patients with AIDS
become progressively more vulnerable to all
forms of infection and commonly succumb to op-
portunistic infections that would normally be eas-
ily overcome.
A second important component of the immune
system is the complement system, which con-
sists of a complex series of proteins in the serum
and in cell membranes (Figure 20). These pro-
teins perform essential roles in the immune re-
sponse to foreign organisms such as bacteria and
viruses, and in the response to tumors. Deficien-
cies of any of the complement proteins may
lead to diseases, including those that involve
infection, hemolysis of red blood cells, or
autoimmune diseases such as systemic lupus
erythematosus.
The complement system is activated by two
general mechanisms. First, antibodies (Ab) can
activate complement when they bind their anti-
gen (Ag). In addition to this so-called classical
pathway there is an alternative pathway that is
continuously active at a low level marking for-
eign organisms for which there are no preformed
antibodies available.
In addition to these roles, complement pro-
teins help to regulate the immune system by an-
other mechanism. This involves the interaction of
specific activated complement protein fragments
with receptors, or binding proteins, that are on
the surface of immune system cells. These recep-
tors allow for communication with the interior of
the cell, and their activation leads to a change in
the function or fate of the cell.
Overall, the complement system plays a funda-
mental role in normal or abnormal immune re-
sponses. Current study in this area is directed
toward understanding not only the molecular
mechanisms of complement activation and regu-
lation but also the general effects on the immune
response of experimentally altering complement
function.
The devastating consequences of AIDS, con-
genital immunodeficiency disorders, and the
frequent rejection of transplanted organs have
made the public increasingly aware of the im-
portance of the immune system in medical prac-
tice. The development of immunosuppressive
drugs has gone a long way toward overcoming
the problem of tissue rejection, and there is now
considerable interest in the possible develop-
lii
I
Introduction
Figure 19- Stereoview of a region on the macromolecule CD4, the T cell receptor for HIV. CD4
contains four tandem immunoglobulin-like domains, D1-D4. The view represents Dl's major
binding region for the virus coat glycoprotein gpl20. Atoms of residues 41-59 are shown in red.
They are enveloped by the surface in contact with a water molecule probe. This figure was produced
with the display program QUANTA.
Reprinted by permission from Ryu, S.-E., Kwong, P.D., Truneh, A., Porter, T.G., Arthos,/., Rosen-
berg, M., Dai, X., Xuong, N.-h., Axel, R., Sweet, R.W., and Hendrickson, W.A. Nature 348:419-426.
Copyright ® 1990 Macmillan Magazines Ltd.
Biologic
Activation Regulation Effects
Ag clearance
Tumor killing
Cellular activation
Hemolysis
fragments
Figure 20. The complement system. Ag, antigen; Ab, antibodies.
Courtesy of V. Michael Holers.
liii
Introduction
ment of vaccines to limit the spread of HIV.
Many of us remember how some 30 years ago
poliomyelitis was to all intents and purposes
eliminated in this country by the development
of the Salk and Sabin vaccines, and we are all
conscious of the fact that many illnesses such as
measles, rubella, whooping cough, and even
smallpox have been brought under control. But
it is still not widely appreciated that the immune
system is itself subject to a number of serious
disorders such as lymphoma and leukemia. Our
Investigators in the Immunology Program
Alt, Frederick W., Ph.D.
Atkinson, John P., M.D.
Bevan, Michael J., Ph.D.
Bjorkman, Pamela J., Ph.D.
Bloom, Barry R., Ph.D.
Bottomly, H. Kim, Ph.D.
Chaplin, David D., M.D., Ph.D.
Cooper, Max D., M.D.
Cress well, Peter, Ph.D.
Davis, Mark M., Ph.D.
Fischer Lindahl, Kirsten, Ph.D.
Flavell, Richard A., Ph.D.
Ghosh, Sankar, Ph.D.
Goodnow, Christopher C, B.V.Sc, Ph.D.
Holers, V. Michael, M.D.
Jacobs, William R., Jr., Ph.D.
Janeway, Charles A., Jr., M.D.
Kappler, John W., Ph.D.
Neuroscience Program
Among the most challenging problems in bio-
medical research are those posed by the human
brain. How do we perceive the world around us?
How do we learn from past experiences? How do
we store and recall information derived from
those experiences? How do we determine when
to act and what actions to carry out? What is
thought? And what are the neural mechanisms
that underlie language? In a word, how are ail
those aspects of our lives that most specifically
define our humanity instantiated in the function-
ing of our brains? The answers to these questions
still lie far in the future, but in the past two de-
cades considerable progress has been made in our
understanding of some of the cellular and molec-
ular mechanisms involved in brain function. Rec-
ognizing this, in 1983 the Institute initiated its
Neuroscience Program. Until recently, the Neu-
roscience Program has been largely focused on
the ways in which nerve cells conduct signals
ability to deal with these malignant conditions is
still very limited, but we are beginning to under-
stand what may cause them. While these dis-
orders present the most urgent challenges to
clinical immunology, even relatively minor al-
lergic disorders continue to pose problems both
for the practicing physician and for the patients
who suffer from them. Many of the reports in this
volume indicate how these and other problems
associated with the immune system are currently
being addressed.
Korsmeyer, Stanley J., M.D.
Leiden, Jeffrey M., M.D., Ph.D.
Littman, Dan R., M.D., Ph.D.
Loh, Dennis Y.-D., M.D.
Marrack, Philippa, Ph.D.
Nussenzweig, Michel C, M.D., Ph.D.
Pay an, Donald G., M.D.
Perlmutter, Roger M., M.D., Ph.D.
Peterlin, B. Matija, M.D.
Schatz, David G., Ph.D.
Smale, Stephen T., Ph.D.
Thomas, Matthew L., Ph.D.
Thompson, Craig B., M.D.
Tonegawa, Susumu, Ph.D.
Weiss, Arthur, M.D., Ph.D.
Weissman, Irving L., M.D.
Witte, Owen N., M.D.
and communicate with each other and with the
effector tissues of the body (such as muscle and
gland cells) and on the cellular mechanisms in-
volved in the development of the nervous system.
Modern neuroscience is founded on two funda-
mental concepts that derive from the late nine-
teenth and the early years of the twentieth cen-
tury. The first of these, commonly referred to as
the neuron doctrine, is that the fundamental
functioning units of the nervous system are nerve
cells, or neurons. Among the cells of the body,
neurons are distinguished anatomically by the
fact that they all extend processes (some of con-
siderable length) that are of two general types:
shorter tapering processes (dendrites) that
mainly serve to receive information from other
cells, and longer processes (axons), of more un-
iform diameter, that serve to transmit information
to other parts of the nervous system or to the body
at large. The second basic concept is that infor-
liv
Introduction
mation in the nervous system is principally en-
coded in a series of signals called nerve im-
pulses, or action potentials. These are brief,
usually all-or-nothing electrical changes in the
nerve cell membrane that are propagated along
the axons at rates between about 3 and 400 feet
per second. A necessary corollary of this concept
is the notion that nerve cells communicate this
encoded information to each other at specific
sites called synapses, where the axon of one cell
functionally interacts with the dendrites or the
bodies of other neurons.
The essential morphological features of neu-
rons were established in the 1870s and 1880s,
with the aid of a number of selective staining pro-
cedures, notably the metallic impregnation tech-
nique developed by the Italian microscopist Ca-
millo Golgi. And the fundamental principle of
the neuron doctrine, namely, that nerve cells are
anatomically and functionally discrete entities,
was convincingly demonstrated around the turn
of the century by the great Spanish neurohistolo-
gist, Santiago Ramon y Cajal. The biophysical
mechanisms responsible for the nerve impulse
and for synaptic transmission were established in
the early 1950s, principally through the work of
Hodgkin, Huxley, Katz, and Eccles (Figures 21,
22, and 23). In brief, activation of a nerve cell
results in the successive opening of pores or ion
channels along the length of the axon that result
in the temporary reversal of the voltage between
the inside and the outside of the axon (this tran-
sient change in potential is the action potential;
see Figure 7) . When the action potential reaches
the ends of the axon it causes the release of a
neurotransmitter that diffuses across the micro-
scopic gap between the axon terminal and the
postsynaptic cell. The binding of the neurotrans-
mitter to specialized receptors in the membrane
of the postsynaptic cell in turn triggers a response
in that cell which may either be the opening of an
ion channel or the activation of a second intracel-
lular messenger in the cell. In either case the
binding of the transmitter to the receptor is re-
flected in the generation of a graded voltage
change across the membrane of the postsynaptic
cell, called a synaptic potential. Depending on
the nature of the transmitter receptor, the re-
sponse may be either excitatory or inhibitory;
i.e., the postsynaptic cell may either be activated
or rendered less likely to discharge an impulse.
Finally, the released neurotransmitter is either
broken down by a specific enzyme within the
synaptic cleft, or taken up by selective transport
mechanisms into the axon terminal (where it can
be reutilized) or into the surrounding nonneural
(glial) cells.
In the past 10 years we have learned a good
deal about the molecular mechanisms involved
in both impulse conduction and synaptic trans-
mission, largely as the result of the successful
cloning of the genes for a number of the ion chan-
nels involved (e.g., for Na"^, K"*", and Ca'^"'") and for
many neurotransmitter receptors like those for
acetylcholine, glutamate, 7-aminobutyric acid
(GABA), serotonin, norepinephrine, dopamine,
and various neuropeptides. From the nucleotide
sequence of these genes it has been possible not
only to deduce the primary amino acid sequence
of the channel or transmitter proteins (and from
this to infer the probable arrangement of the rele-
vant protein in the membrane) but also to gener-
ate hybridization probes to identify other related
channels or receptors. And using some of the es-
tablished techniques of genetic engineering, like
site-directed mutagenesis, it has been possible in
some cases to establish the regions within the
channel molecules that are sensitive to changes
in voltage (Figure 24), or the ligand-binding and
second messenger-activating domains of recep-
tors. One example will suffice to demonstrate the
importance of this approach to our understand-
ing of these fundamental processes.
It has been known for almost 50 years that the
relatively simple molecule acetylcholine is the
transmitter at the junctional region between mo-
tor nerve fibers and muscle cells and also at cer-
tain synapses in the brain and spinal cord. With
the discovery in the 1970s that the clinical con-
dition myasthenia gravis (previously discussed
in the immunology section) is caused by circulat-
ing antibodies directed against the receptor for
acetylcholine in the muscle membrane, a major
effort was mounted to purify and biochemically
characterize the acetylcholine receptor (AChR).
This work served to establish that the AChR con-
sists of five subunits: two designated a, and one
each called /3, 7, and b.
In the early 1980s Heinemann, Patrick, and
their colleagues succeeded in cloning the genes
for the a-subunit, and in 1983 Numa and his co-
workers presented the complete nucleotide se-
quences encoding all four kinds of subunits.
From these sequences we gained several impor-
tant insights. First, the four subunits showed a
high degree of homology, which suggested that
their genes were probably derived — by duplica-
tion and divergence — from a single ancestral
Iv
Introduction
cell body
dendrites
axon (less than 1/20 inch to
more than three feet in length)
terminal branches of axon
Figure 21. A schematic diagram of a typi-
cal neuron. The arrows indicate the di-
rection in which nerve signals are con-
veyed. The largest axons in the human
brain and spinal cord extend for about
three feet and have a diameter of less
than 1/1, 000th of an inch. Many axons
are covered by an insulating layer
known as the myelin sheath. The myelin
sheath is interrupted at intervals known
as nodes of Ranvier.
From Alberts, B., Bray, D., Lewis, J.,
Raff, M., Roberts, K., and Watson, J.D.
1989. Molecular Biology of the Cell, 2nd
edition. New York: Garland, p. 1061.
1
Figure 22. A few of the many types of neu-
rons in the vertebrate nervous system.
(S. Ramon y Cajal, Histologic du Systeme
Nerveux de I'Homme et des Vertebres.
Paris: Maloine, 1901- 191 1; reprinted,
Madrid: C.S.I.C., 1972.)
From Alberts, B., Bray, D., Lewis, J.,
Raff, M., Roberts, K., and Watson, J.D.
1989. Molecular Biology of the Cell, 2nd
edition. New York: Garland, p. 1061.
Ivi
Introduction
Figure 23- A: The anatomical arrangement of a typical reflex, in this case the knee jerk. Each cell actually represents a population
of many neurons. Information about stretch of the quadriceps femoris muscle is conveyed by afferent neurons to several loci
within the central nervous system. In the spinal cord, afferent neurons act directly on the motor neurons to the quadriceps and,
by means of inhibitory interneurons, indirectly on the motor neurons to the antagonistic muscle, the biceps. Both of these
actions combine to produce the coordinated expression of the reflex behavior. In addition, information is conveyed to higher
regions of brain to update them about the information coming into the nervous system and about the behavior that is being
generated. These higher centers, in turn, can act to modify the reflex
behavior.
B: The sequences of signaling changes that produce the reflex action.
Graded stretch of the muscle produces a graded receptor potential in the
muscle spindle of the afferent neuron that propagates passively to the trig-
ger zone at the first node of Ranvier. If the potential is sufficiently large, it
will trigger an action potential that will propagate actively along the axon
to the terminal region. At the terminal the change in membrane potential,
produced by the action potential, gives rise to a secretory potential that
leads to the release of transmitter substance. The transmitter diffuses
across the synaptic cleft and interacts with receptor molecules on the mem-
brane of the postsynaptic motor cell to initiate a synaptic potential. The
synaptic potential then propagates passively to the initial segment of the
axon, where it, in turn, initiates an action potential that propagates to the
terminals of the motor neuron. This action potential leads ultimately to a
synaptic potential in the muscle, which initiates an action potential that
causes the contraction of the muscle.
Reprinted by permission of the publisher from Kandel, E.R., and
Schwartz, J.H. Principles of Neural Science, 2nd edition, p. 20. Copyright
1985 by Elsevier Science Publishing Co., Inc.
Cell body in
dorsal root ganglion^
/ i^fii/j II I' ^,''1 1 II! (film III 1 1 /li I I ^"^^
Quadriceps
(extensor)
Patella
la afferent fiber
Pyramidal
tract
To contralateral
spinal cord
Inhibitory
Interneuron
B
Graded
receptor
potential
Action
potential
Action
potential
J
Secretory
potential
Synaptic
potential
Action
potential
Synaptic
potential
Action
potential
1 1
Muscle spindle Myelin
Dorsal root
ganglion cell
JL
1 1
Receptor — Trigger
zone
Input
Stretch
-3 E
Axon
Integrative Conductile
Synapse — Trigger
zone
Integrative
Output Input
Axon
Conductile
Action
potential
Synaptic
potential
Synapse
Muscle
Output Input
Ivii
Introduction
CLOSED
Figure 24. A local reduction of the voltage differential may
induce a sodium or potassium channel to change confor-
mation from one allosteric form to another, thus opening
the channel and leading to the free passage of ions. This
change is presumably caused by key electrically charged
amino acids in the channel protein, which shift their orien-
tation in response to the changed electrical field.
From Discovering Enzymes, by David Dressier and Hun-
tington Potter, p. 235- Copyright ® 1991 by David Dressier.
Reprinted with permission by W.H. Freeman and Company.
Iviii
i
Introduction
gene. Second, the similarities in the predicted
amino acid sequences suggested that the subunits
are probably arranged to form a central pore or
channel between them. Third, the presence in
each subunit of four hydrophobic regions, each
about 20 amino acids long, immediately sug-
gested the probable disposition of the subunits,
with four transmembrane domains (Ml, M2, M3,
and M4) and intervening intra- and extracellular
linking segments. More recent work has identi-
fied a fifth subunit type € and has established the
precise location of the acetylcholine-binding site
and the existence of a family of nicotinic AChRs
in the central nervous system; much has also been
learned about the regulation of the receptor dur-
ing muscle development and after denervation,
and about the process of receptor desensitization.
The nicotinic AChR belongs to a large class of
neurotransmitter receptors that operate by selec-
tively opening ion channels. Another, somewhat
larger class of receptors acts through second mes-
sengers. For example, the adrenergic receptors,
which are responsible for controlling a number
of vital functions such as heart rate and blood
pressure, act through the intermediary of a class
of so-called G proteins to activate the enzyme
adenylate cyclase and increase the intracellular
level of the important second messenger, cAMP.
The second messenger, in turn, usually acts by
stimulating protein kinases that modify (by add-
ing phosphate groups) other proteins, including
ion channels and proteins that regulate gene ex-
pression in the responding cell.
The regulation of gene expression by synapti-
cally mediated second messenger systems has be-
come one of the most active areas for research in
molecular neuroscience. Whereas neurotransmit-
ters usually result in changes that have a time
course measured in the millisecond to second
range, many of the most intriguing phenomena in
neuroscience are those that occur over periods of
hours, days, and even months or years. Recent
work has demonstrated that in addition to their
more or less immediate and short-lasting effects,
under appropriate conditions (such as those we
commonly associate with learning and memory)
neurotransmitters may, through second messen-
gers, activate a number of transcriptional regula-
tory proteins that "turn on" various classes of
genes. These, in turn, may regulate the expres-
sion of yet other genes and thus unleash a com-
plex cascade of events within the responding
nerve cell, modifying its growth and altering its
responsiveness to later neurotransmitter activa-
tion over long periods of time.
One of the major beneficiaries of the applica-
tion of the new genetics to the nervous system has
been the field of developmental neuroscience.
Indeed it is no exaggeration to say that since the
late 1970s this field has been transformed from
an essentially descriptive science into one in
which, for the first time, mechanistic explana-
tions are emerging to account for the growTih of
nerve cells and their processes, for the deploy-
ment of cells into peripheral ganglia and within
cortical layers or nuclear groups in the central
nervous system, for the formation of specific pat-
terns of connections, and for the elimination of
redundant cells and inappropriate connections.
Because of the complexity and inaccessibility
of the mammalian central nervous system, until
recently much of the most definitive work on
neural development has been carried out in
simpler forms such as the nematode C. elegans
and the fruit fly Drosophila. It is difficult to sum-
marize the broad sweep of this work, except to
say that it has served to clarify the genetic mecha-
nisms that determine the distinct front-to-back
and top-to-bottom organization of all developing
organisms, that determine not only which cells
will become neurons but also how many neurons
will be generated and what type they will be
(e.g., sensory cells, interneurons, or motor
cells), and that determine finally whether the
neurons that are initially formed will survive. In
some instances it is clear that the character or
phenotype of the nerve cells is determined by
their lineage; in other cases cell-cell interactions
are more important, and the nature of the signals
that developing cells transmit to their neighbors
is currently being elucidated. Of special impor-
tance are the molecules on the surfaces of cells
that enable them to recognize and aggregate with
other cells of like kind or enable them to migrate
along other cells or across territories filled with
extracellular matrix materials. While much re-
mains to be discovered, the first fruits of this har-
vest hold great promise for future progress in this
important field.
Finally, no account of progress in molecular
neuroscience would be complete without refer-
ence to the striking developments in our under-
standing of the basis of some of the major genetic
disorders that affect the nervous and related mus-
cular systems. Perhaps the most striking of these
developments has been the cloning of the gene
lix
Introdtiction
for Duchenne and Becker muscular dystrophy.
These are X-linked recessive disorders that, in the
more severe (Duchenne) form lead inexorably
from muscular weakness to muscular atrophy and
finally death. The extreme size of this gene (it
comprises about 1 percent of the X chromosome
and almost 0.1 percent of the total human ge-
nome) renders it especially vulnerable to muta-
tion, and in many of the identified mutations, the
protein encoded by the gene, dystrophin
(which appears to be critical for coupling muscle
excitation and contraction), is either absent or
markedly deficient.
Some years ago the general location of the gene
responsible for the severe neurological disorder
known as Huntington's disease (HD) was deter-
mined using RFLPs, as discussed in the section on
genetics. Although the HD gene itself has so far
eluded us, there is every reason to be optimistic
that within a year or two it will be identified and
cloned. In the meantime, the relevant RFLP has
provided a useful marker for identifying carriers
of the disordered gene. There is similarly reason
for optimism that in the near future the genetic
basis for the two major affective disorders, manic
depression and schizophrenia, will be eluci-
dated. Careful studies of family histories and of
identical twins raised apart have clearly estab-
lished that both illnesses have an important ge-
netic component, and while neither is probably
due to a single genetic mutation, RFLP analysis
and other genetic approaches should reveal the
genes involved. That such complex behavioral
Investigators in the Neuroscience Program
Adams, Paul R., Ph.D.
Aldrich, Richard W., Ph.D.
Anderson, David J., Ph.D.
Artavanis-Tsakonas, Spyridon, Ph.D.
Axel, Richard, M.D.
Corey, David P., Ph.D.
De Camilli, Pietro, M.D.
Evans, Ronald M., Ph.D.
Goodman, Corey S., Ph.D.
Horvitz, H. Robert, Ph.D.
Huganir, Richard L., Ph.D.
Hurley, James B., Ph.D.
Jahn, Reinhard, Ph.D.
Jan, Lily Y., Ph.D.
Jan, Yuh Nung, Ph.D.
Jessell, Thomas M., Ph.D.
Kandel, Eric R., M.D.
Lerner, Michael R., M.D., Ph.D.
Miller, Christopher, Ph.D.
Movshon, J. Anthony, Ph.D.
disorders might yield to this type of approach was
unthinkable only a decade ago; as noted in the
section on genetics, nothing serves to emphasize
more dramatically the power of the new genetics
or the exciting possibilities it portends.
The dramatic advances in cellular and molecu-
lar neuroscience should not obscure the fact that
the distinctive role of the nervous system in the
economy of an organism is its capacity to inte-
grate sensory information that is received (both
from within and from outside the body) and to
organize it into patterns of behavior that allow
the organism to respond to changes in its environ-
ment in appropriate ways. And for human beings,
it is through the nervous system that we learn
both from personal experience and the accumu-
lated wisdom of previous generations not only
how to survive but how to enjoy and profit from
the richness of mental experience and meaning-
ful social interactions. To understand all this we
will need to learn a great deal more about the
activities of large populations of nerve cells,
about the computational capacity of complex
neural networks, and about the extraordinary
ways in which the human brain, with its 100 bil-
lion or more neurons and its more than 1 trillion
synapses, receives and processes information. We
are at the threshold of being able to understand
the logic of the simplest organisms at both the
molecular and systems levels; the exploration of
human behavior in these same terms stands as
perhaps the greatest challenge to modern
science.
Reed, Randall R., Ph.D.
Reichardt, Louis P., Ph.D.
Rosenfeld, Michael G., M.D.
Rubin, Gerald M., Ph.D.
Sakmar, Thomas P., M.D.
Scheller, Richard H., Ph.D.
Sejnowski, Terrence J., Ph.D.
Siegelbaum, Steven A., Ph.D.
Steller, Hermann, Ph.D.
Stevens, Charles F., M.D., Ph.D.
Struhl, Gary, Ph.D.
SUdhof, Thomas C, M.D.
Tanabe, Tsutomu, Ph.D.
Tsien, Roger ¥., Ph.D.
Yau, King- Wat, Ph.D.
Yellen, Gary, Ph.D.
Ziff, Edward B., Ph.D.
Zipursky, S. Lawrence, Ph.D.
Zuker, Charles S., Ph.D.
Ix
Introduction
Structural Biology Program
The primary goal of structural biology is to un-
derstand, in atomic detail, the three-dimensional
architecture of proteins, protein assemblies, and
the complexes formed by proteins that interact
with RNA and DNA. Underlying this approach is
the belief that fundamental insights into the func-
tional role of biologically interesting molecules
can best come from understanding the forms of
the molecules themselves. As Francis Crick, the
co-discoverer of the double-helical structure of
DNA (see Figure 9), remarked: "to understand
function it is essential to study structure." It was
with this in mind that in 1985 the Institute made
a substantial commitment to develop a new Pro-
gram in Structural Biology.
At present, x-ray crystallography is the most
powerful approach for visualizing the three-
dimensional structures of large molecules (com-
monly called macromolecules) . An essential pre-
requisite for x-ray analysis is the availability of
crystals of the molecule or molecular complex
that are suitable for recording the diffraction of
x-rays. The production of crystals, in turn, re-
quires chemically homogeneous preparations.
Moreover, molecules that are not spatially uni-
form (too "floppy") must be broken down or mo-
lecularly dissected into defined and rigid com-
ponents. For example, to crystallize antibody
molecules, it was important first to cleave them
into their principal fragments, Fab and Fc, because
these pieces are normally connected by a flexible
hinge. And because of antibody diversity (de-
scribed in the section on immunology) , it became
necessary to study Fab fragments from monoclonal
immunoglobulins. A continuing challenge to
structural biologists is the development of strate-
gies for crystallizing membrane proteins — by solu-
bilizing them with detergents, dissecting them
into pieces, or altering them by mutation.
Genetic engineering has transformed struc-
tural biology. This approach, which makes it pos-
sible to produce large quantities of pure proteins,
also allows an investigator either to choose a suit-
able fragment for study or to modify genetically
the molecule to be crystallized. Other method-
ological advances in crystallography itself are
transforming the field by extending the range of
problems that can be tackled routinely.
There are essentially four stages in determining
a structure by x-ray diffraction analysis (Figures
25 and 26): 1) diffraction experiments (data col-
lection); 2) complex computations that pro-
duce, in effect, an image of the molecule(s) in
the crystal; 3) interpretation of the computed
image in terms of a molecular model; and 4) re-
finement of the model by further computation.
Synchrotron x-ray sources, which are a thou-
sand or more times stronger than conventional
laboratory x-ray generators, are making it possi-
ble to study structures that could not previously
be solved. (HHMI is currently developing a
synchrotron resource for use by the biological
community at the National Synchrotron Light
Source at Brookhaven National Laboratory on
Long Island.) Recent examples from HHMI labo-
ratories are the human class I major histocompati-
bility antigen and the DNA virus SV40. At the
same time, position-sensitive x-ray detectors have
greatly extended the applications of conven-
tional radiation sources. Lastly, novel computa-
tional methods have made the production of a
molecular image (phase determination) less
dependent on extensive ancillary data from
heavy-atom modified crystals and have made re-
finement of models less cumbersome and more
objective.
In the 1950s and 1960s, x-ray crystallography
revealed the structures of the first biologically
important molecules, including DNA, hemoglo-
bin, and insulin. In the 1970s it revolutionized
the field of enzymology by making it possible to
visualize directly the active sites of enzymes. In
the 1980s it made comparably far-reaching con-
tributions to virology, immunology, and mem-
brane biology by revealing the structures of vi-
ruses, antibodies, and a photosynthetic reaction
center. What can we expect in the 1990s? It
seems reasonable to predict the following:
1. Structures of different classes of proteins
or protein/nucleic-acid complexes. Three de-
cades of biological crystallography have left sev-
eral major areas unexplored. For example, we
have yet to know what any of the major proteins
of the cytoskeleton and of cellular motility look
like (actin, myosin, tubulin, and so forth). We
have yet to visualize any of the membrane recep-
tors referred to above, and we have yet to see an
ion channel, a ribosome, an RNA polymerase, or a
ribozyme; and, with the exception of transfer
RNAs (tRNAs), little is known of the three-
dimensional structure of most RNAs and RNA-
protein complexes. Progress toward some of
these goals is reported in this volume; others will
no doubt be achieved before long, as more and
more workers are drawn into the field and as new
techniques are developed.
Ixi
Introduction
Single
crystiil
• 4
»
1
ft •
• • J
• • i
»- — • '~
» • •
k • •
• •
• •
• •
• •
} •
» •
ft •,^M--^
'/^ i
ft • f
> • «
1 • •
\ • •
► • f
;*o'
» • <
<
I • • — i
> • •
» • i
/ • •
» • • •
r # •
• •
• •
• •
• •
1 • <
• •
M.iihemalical
analysis of
.callered X raw
Diffraction pattern
Electron-density
maps of sections
through the molecule
Electron-density map
Figure 25. The use of x-rays to determine the structure of enzymes. When an enzyme is
in a crystalline form, its molecules are arranged in regularly repeating arrays. A beam
of x-rays passed through the enzyme is diffracted to give a pattern of spots on a detec-
tor screen behind the sample. Each spot contains information about the electron den-
sity ( and hence the types of atoms: carbon, oxygen, nitrogen, etc.) in various parts of
the crystal. This information can be used to construct electron density maps whose
contours indicate the type of atoms present. The contour map shown at the right
represents a short segment of polypeptide backbone with a tyrosine side chain.
Adapted from Geoffrey Zubay, Biochemistry, Addison-Wesley, 1985.
rosine side chain
Backbone
Electron-density map
with inferred polypeptide
backbone and tyrosine
side chain fitted in
- - -<i
B
Figure 26. A: The x-ray diffraction pattern given by the enzyme chymotrypsin.
Courtesy of Thomas A. Steitz.
B: A representation of the three-dimensional structure of chymotrypsin. Carbon atoms are shown in
black, nitrogen in blue, oxygen in red, hydrogen in white, and sulfur in yellow. The diameter of the enzyme
is about 45 A (somewhat less than a millionth of an inch). The hydrophilic side chain of arginine- 145 is
clearly visible projecting outward from the right side of the molecule. The ridges and grooves on the surface
of the chymotrypsin molecule are as unique as the mountains and craters of the moon, and herein lies the
fulfillment of the lock-and-key mechanism hypothesized by Emil Fischer at the turn of the century.
Courtesy of Molecular Simulations, Waltham, MA.
Ixii
Introduction
2. Time-resolved images of events at the ac-
tive site of an enzyme. New ways of using
synchrotron x-ray radiation permit very rapid
measurements of diffraction data, so that in prin-
ciple it should be possible to follow the struc-
tural changes that occur during an enzymatic re-
action. If we understood these changes, it might
be possible to develop enzymes with usefully al-
tered properties and to synthesize enzyme inhibi-
tors with enhanced specificity.
3. The role of molecular recognition in the
regulation of cellular activity. How do proteins
that control transcription recognize specific DNA
sequences? How do cell surface proteins in the
immune system recognize and present antigens?
The answers to these questions are beginning
to emerge from crystal structures of molecular
complexes, such as the complexes formed by reg-
ulatory proteins with DNA, those formed by bind-
ing proteins with their appropriate ligands, of an-
tibodies with antigens, and of MHC molecules
with peptides. These current efforts give consid-
erable promise for understanding how hormones
or neurotransmitters trigger a cascade of events
that involves the formation and dissociation of
protein assemblies inside cells. A large and medi-
cally significant class of regulatory interactions
involves the protein products of oncogenes or
proto-oncogenes .
As our knowledge of important proteins rap-
idly increases, we can only hope that our capacity
to anticipate aspects of structures not yet deter-
mined will keep pace. The goal of accurately pre-
dicting the three-dimensional structure of any
protein from its amino acid sequence is still a
long way off. But recent advances in computa-
tional chemistry make it possible to predict the
effects of small perturbations, such as point mu-
tations, on the folding of a protein and to calcu-
late differences in binding free energies for re-
Investigators in the Structural Biology Program
Agard, David A., Ph.D.
Briinger, Axel T., Ph.D.
Burley, Stephen K., M.D., D.Phil.
Deisenhofer, fohann, Ph.D.
Fox, Robert O., Ph.D.
Harrison, Stephen C, Ph.D.
Hendrickson, Wayne A., Ph.D.
Kuriyan, John, Ph.D.
lated ligands. And systematic approaches to
designing drugs, such as antagonists or inhibitors
of enzymes, are beginning to emerge now that we
can carry out meaningful calculations on known
structures.
Nuclear magnetic resonance (NMR) meth-
ods offer an alternative route to determining the
three-dimensional structures of peptides and
small proteins. The DNA-binding domains of tran-
scriptional activators and repressors are good
candidates for this type of analysis. The past two
years have seen the determination by NMR of the
structure of a developmentally important DNA
sequence known as a homeodomain and of a
DNA-binding structure known as a zinc finger.
NMR has the great advantage that it circumvents
the need to crystallize the protein to be studied.
At the other end of the size scale, imaginative
combinations of light and electron microscopy
have begun to reveal important patterns and re-
gularities in very large structures, such as chro-
mosomes, viruses, and receptors. New methods
for recording images and enhancing contrast in
light microscopy make it possible to record in
real time the events of intracellular transport or
the process of chromosome condensation. As the
molecules that generate these large-scale intra-
cellular motions are characterized, it should be-
come possible to relate such changes to the spe-
cific molecular recognition events that control
them. It is fortunate, but not coincidental, that as
biologists have become increasingly aware of the
need to know the precise structure of the mole-
cules that mediate the phenomena they are inter-
ested in, a whole range of new experimental
methods has been developed, and a new genera-
tion of structural biologists has emerged to assist
them and to advance their understanding of the
complex relationships between structure and
function.
Matthews, Brian W., Ph.D., D.Sc.
Pabo, Carl O., Ph.D.
Quiocho, Florante A., Ph.D.
Sedat,John W., Ph.D.
Sigler, Paul B., M.D., Ph.D.
Sprang, Stephen R., Ph.D.
Steitz, Thomas A., Ph.D.
Wiley, Don C, Ph.D.
W. Maxwell Cowan, M.D., Ph.D.
Vice President and Chief Scientific Officer
Ixiii
I
I
I
I
■i
I
I
i
Electrical Activity of Nerve Cells
Paul R. Adams, Ph.D. — Investigator
Dr. Adams is also Professor of Neurobiology and Behavior and of Pharmacological Sciences at the State
University of New York at Stony Brook. He received his B.A. degree in physiology and pharmacology from
Cambridge University and his Ph.D. degree in pharmacology from the University of London. His
postdoctoral work was done with Bert Sakmann at the Max Planck Institute, Gottingen, and with
Philippe Ascher at the Ecole Normale, Paris. Dr. Adams is currently a MacArthur Fellow. He was elected
Fellow of the Royal Society.
NERVE cells are specialized to generate, trans-
mit, and receive rapid electrical messages.
Electrical impulses, called action potentials, last
about 1,000th of a second and can travel along
specialized nerve cell extensions at speeds over
100 mph. Chemical transmitter substances re-
leased onto the nerve cell by other nerve cells
control the precise timing of these electrical
pulses. We are trying to understand how these
pulses are generated and how transmitters im-
pinging on the cell control them.
Cell membranes are normally effective barriers
to the movement of ions (electrically charged
atoms) between the cell environment and the
cell interior. This insulating property allows the
inside of a nerve cell to have a different electrical
voltage from the outside or from a neighboring
cell. The electrical activities described above are
regulated by special protein molecules, called
ion channels, which are embedded in the cell
membrane. There are many types of ion channel.
Each type has a specific role, but all have in com-
mon a unique feature that allows certain ions to
travel easily through them. The protein chains that
make up an ion channel molecule are arranged to
create a minute tunnel, through which certain
types of ions — for example, sodium, potassium, or
calcium ions — can quickly move.
The direction that the ion moves is not con-
trolled by the tunnel but by the ion concentra-
tions and the transmembrane voltage. The tunnel
does, however, control the type of ion that
moves. Thus the sodium channel only allows so-
dium ions to pass. Because sodium ions are abun-
dant outside, but not inside, nerve cells, the exis-
tence of open sodium channels leads to an inward
stream of sodium ions, making the cell interior
positive. On the other hand, when potassium
channels open, potassium streams out of the cell,
making it negative. Because these tunnels are not
always open (indeed are closed most of the
time) , it is supposed that the channel must have
some sort of gate.
There are many types of potassium channel,
differing according to their molecular structure
(see the articles in this volume by Richard
Aldrich and Lily Jan), speed of opening and clos-
ing, and the way in which their gates are con-
trolled. We are studying potassium channels in
bullfrog sympathetic ganglion cells to under-
stand better how they are controlled and how
they contribute to the electrical activity of nerve
cells. We have been particularly intrigued by a
channel we call the M channel. This channel is
turned off as a result of the binding of certain
neurotransmitters (chemicals released from
nerve endings) to receptors on the cell surface.
As a result, less potassium leaves the cell, which
is therefore less negative and more able to fire
electrical pulses. Tumofif of this channel occurs
via activation of a G protein (see the article by
John Exton) . However, after the neurotransmitter
has been removed, M channels turn back on and
transiently become more numerous than initially
observed. We have shown that calcium and ara-
chidonic acid are involved in this overshooting
response. Various levels of calcium were per-
fused into nerve cells while they were visualized
with a special calcium-detecting microscope.
Neurotransmitter stimulation of receptors pro-
duces a small calcium signal that is sufficient to
increase the activity of M channels. However, this
is not seen until the concomitant G protein-
mediated suppression of the channels is termi-
nated by removing the transmitter.
M channels work in concert with many other
types of channels, which we have also character-
ized. This information can then be combined
with studies of calcium diffusion and membrane
geometry to predict completely the cell's electri-
cal output. We are making detailed quantitative
morphological measurements of cells that have
been previously characterized electrophysiologi-
cally in both bullfrog ganglia and mammalian
hippocampus and lateral geniculate. This is
achieved by automatic three-dimensional recon-
struction of dye-filled cells, which are optically
sectioned using confocal microscopy. Our voxel-
based reconstructions, developed in collaboration
with the computer science department at Stony
Brook, can then be used as a platform for Monte
Carlo simulations of the movements of single ions
in small cell structures, such as dendritic spines.
Three-dimensional computer reconstructions of nerve cells. The upper image shows
a sympathetic ganglion cell of a bullfrog; the lower image, a hippocampal pyramidal
cell of a rat.
Research of Rick Avila, Barry Burbach, Jim Monckton, Ted Carnevale, Arie Kauf-
man, and Paul Adams.
2
Three-Dimensional Macromolecular
and Cellular Structure
David A. Agard, Ph.D. — Associate Investigator
Dr. Agard is also Associate Professor of Biochemistry and Biophysics at the University of California, San
Francisco. He did his undergraduate work at Yale University with Frederic Richards, Hal Wyckoff, and
Thomas Steitz. He received his Ph.D. degree in chemical biology from the California Institute of
Technology, where he studied with Robert Stroud and began a continuing collaboration with John Sedat.
His postdoctoral work was done on high-resolution electron microscopic crystallography at the MRC
Laboratory of Molecular Biology in Cambridge, England, with Richard Henderson. There he also began
the cloning of the a-lytic protease gene with Sydney Brenner.
WE study chromosome structure in close col-
laboration with John Sedat (HHMI, Univer-
sity of California, San Francisco); hence only a
subset of these studies will be discussed here.
Our primary aim in this area is to provide a physi-
cal basis for understanding chromosome behav-
ior and function by directly determining the
three-dimensional structure of chromosomes as a
function of both transcriptional state and the cell
cycle stage. Current efforts are aimed at deter-
mining how fibers of nucleosomes are folded into
higher-order structures within the chromosome
and what role specific chromosomal proteins
play in determining these structures.
We are using intermediate voltage electron mi-
croscope (IVEM) tomography to examine higher-
order chromosome structure. In the past year sig-
nificant improvements in the quality of the
three-dimensional reconstructions have led to
the first new insights into the structure of the 30-
nm fiber (in collaboration with Chris Woodcock,
University of Massachusetts) . We are now begin-
ning to be able to trace the paths of the 30-nm
chromatin fibers within telophase chromosomes.
It is clear that the existing models of chromatin
structure are seriously flawed. In addition, we
have made dramatic steps toward our goal of fully
automating the complex task of collecting three-
dimensional IVEM tomography data. This will
simplify the arduous task of collecting 100-150
images and should dramatically reduce radiation
exposure (from 3 hours to about 5 minutes) . Sig-
nificant progress has also been made on under-
standing the mechanism of image formation for
thick specimens and on applying a powerful new
approach to the problem of electron microscopic
(EM) reconstruction. Together with a new stain-
ing approach and cryopreparation methods,
these improvements should allow us to trace the
paths of the 30-nm fibers throughout the chro-
mosome and will finally make this exciting struc-
tural approach available to the general cell
biologist.
Structural Basis of Enzyme Specificity
By combining solution kinetic analysis, x-ray
crystallographic structural analysis, and site-
directed mutagenesis, we have been able to
probe the structural basis for substrate specificity
in unprecedented detail, using a-lytic protease as
a model system. The availability of peptide bor-
onic acid inhibitors, which act as excellent
mimics of the reaction transition state, or nearby
intermediates, has allowed us to use x-ray crystal-
lography to examine the complex set of interac-
tions between enzyme and substrate that mediate
specificity.
By mutation, we have been able to alter dramat-
ically the pattern of substrate specificity while
maintaining or even increasing enzyme activity.
Approximately 40 high-resolution, extremely
well-refined crystal structures have now been de-
termined and analyzed. Detailed structural analy-
ses of three mutants as free enzymes and as com-
plexes have provided surprising insights into the
mechanism of specificity and have indicated the
crucial role that protein flexibility plays in selec-
tivity. During the past year we have made numer-
ous other mutations and examined their kinetic
and structural propenies. Of particular note is a
mutation that alters specificity indirectly by
changing active-site flexibility. This remarkable
finding is the first demonstration that residues
beyond those that directly contact the substrate
can play a significant role in determining the pat-
terns of specificity.
Not long ago we began a collaboration with
Vladimir Basus (University of California, San Fran-
cisco) to perform a complete analysis of a-lytic
protease structure in solution by nuclear mag-
netic resonance (NMR) methods. We have
now made '^N,''C doubly labeled enzyme and
have collected sufficient high-resolution three-
dimensional NMR data sets to permit complete
backbone assignment — a significant task for such
a large protein. Currently we are working on the
assignments and on the collection of data for the
complete side chain assignments. We anticipate
that the NMR data will provide insights into
correlated motions within the enzyme and be
crucial for the folding studies described below.
3
Three-Dimensional Macromolecular and Cellular Structure
Last year we had developed a new method for
predicting the energetics of protein-substrate in-
teractions. This approach (based on Ponder-
Richards rotamers combined with energetics and
solvation terms) can predict k^^JK^ with stun-
ning accuracy. We have used this method to de-
sign a new enzyme with particular properties,
and so far the results have been remarkable. This
year we have been working to extend these re-
sults to other systems and to improve their accu-
racy. We have been able to predict quantitatively
100 values of k^^JK^ for subtilisin with equiva-
lent accuracy. This approach is being expanded
to allow its use in drug design situations and for
modeling proteins based on the structure of a ho-
mologous protein.
Structural and Biochemical Probes
of Folding of a-Lytic Protease
a-Lytic protease is synthesized as a prepro-
enzyme. Experiments in Escherichia coli have
demonstrated that the 166-amino acid precursor
domain is required for the proper folding of the
198-amino acid protease domain. Furthermore,
we have shown that the pro region does not have
to be attached covalently in order to function.
The development of the in vitro folding system
has led to the unprecedented ability to trap and
purify a stable folding intermediate under non-
denaturing conditions. Although the interme-
diate is unable to fold by itself, it is rapidly con-
verted to active enzyme upon addition of the pro
region. Analysis of the folding kinetics proves
that the pro region acts essentially as a "foldase"
to stabilize the transition state for folding,
thereby speeding up folding by more than 10^.
Current efforts are focused on characterizing
the folding intermediate and analyzing the fold-
ing reaction. We plan to use a combination of
approaches — including NMR and crystallogra-
phy— to probe the structure of the intermediate
and the role of the pro region in the final stage of
folding.
Receptor-Ligand Targeting and Human
Cholesterol Metabolism
Apolipoprotein E is an important protein in
cholesterol metabolism in mammals. Responsi-
ble for targeting chylomicrons, very low density
lipoprotein (VLDL), and high-density lipoprotein
(HDL) particles to the low-density lipoprotein
(LDL) receptor, apolipoprotein E has a major role
in triglyceride and cholesterol metabolism. The
protein itself has two distinct structural and func-
tional domains: the amino-terminal 22-kDa do-
main contains the receptor-binding functional-
ity; lipid binding resides primarily with the
10-kDa carboxyl-terminal domain. In collabora-
tion with the Mahley group at the Gladstone
Foundation Laboratories for Cardiovascular Dis-
ease, we obtained crystals of the 22-kDa recep-
tor-binding domain. Last year we reported that
the protein is an unusually elongated four-helix
bundle. Although the surface is exceptionally
charged, there is a precise balance between
groups with a positive charge and groups with a
negative charge, except in what we believe is the
receptor-binding region. We have recently fin-
ished the analysis of a naturally occurring human
mutant that significantly reduces receptor bind-
ing and can lead to premature atherosclerosis.
The structure reveals that the mutant disrupts the
complex set of surface salt bridges, which then
causes a key arginine required for binding to be
recruited into a salt bridge. Current efforts are
aimed at generating a soluble fragment of the LDL
receptor-binding domain for crystallographic
study.
4
Molecular Mechanisms of Ion Channel Function
Richard W. Aldrich, Ph.D. — Associate Investigator
Dr. Aldrich is also Associate Professor of Molecular and Cellular Physiology at Stanford University. He
received his B.S. degree in biology from the University of Arizona and his Ph.D. degree in neuroscience
from Stanford. He carried out postdoctoral research at Yale University with Knox Chandler and Charles
Stevens. Before returning to Stanford, he was Assistant Professor of Molecular Neurobiology at Yale.
Among his awards are a Searle Scholars Award, the Young Investigator Award of the Society
for Neuroscience, and the Young Investigator Award of the Biophysical Society.
ION channels are the molecular units of electri-
cal signaling in cells. They are proteins that
regulate the movement of ions, such as sodium,
calcium, and potassium, into and out of cells.
They are responsible for the conversion of exter-
nal sensory signals to the electrical language of
the nervous system and for the integration of
these signals to generate appropriate behavior. In
addition, ion channels are important for the gen-
eration and regulation of the heartbeat, for con-
traction of muscles, and for the release of hor-
mones in the bloodstream. A very large variety of
ion channel types are found in the body. They are
specialized to select for certain species of ions
and to open and close in response to a number of
different stimuli, such as the binding of a neuro-
transmitter molecule or a change in the voltage
that exists across a cell's membrane. Our labora-
tory is interested in the molecular mechanisms of
ion channel function. One of our major goals is to
understand the conformational changes that oc-
cur as the channels open and close in response to
appropriate stimuli.
Voltage-gated ion channels are an important
functional class. As their name implies, they can
open in response to changes in the electrical po-
tential across the cell membrane, a property cru-
cial for the generation of electrical signals and for
the transmission of information throughout the
body. These molecules have a way to measure the
electrical potential and open accordingly. In ad-
dition, a number of these channels inactivate or
become unavailable for opening after use. In re-
cent years we have studied the molecular mecha-
nisms of inactivation of a class of potassium chan-
nels that were cloned in Drosophila. These
channels are products of the Shaker gene. They
exhibit the fastest inactivation of any potassium
channels yet cloned. William Zagotta and I began
by using single-channel recording methods to
study the gating properties of wild-type Shaker
channels in their native tissue. Such methods al-
low us to record the behavior of a single-channel
molecule as it opens and closes on a millisecond
time scale.
We determined that the conformational
changes associated with opening the channel de-
pended strongly on the membrane voltage and
therefore involved a substantial rearrangement of
an electrically charged part of the channel in the
membrane. On the other hand, the inactivation
process did not involve significant charge rear-
rangement. This result, combined with our abil-
ity to alter inactivation by internal enzymes, led
us to the hypothesis that inactivation involved a
conformational change on the inside of the mem-
brane that blocked the flow of potassium ions
through the channel.
Dr. Zagotta, Toshinori Hoshi, and I further stud-
ied this hypothesis by making altered channels
with recombinant DNA methods and expressing
the normal and altered channels in frog oocytes.
Our results are strikingly consistent with the
"ball and chain" model of inactivation originally
proposed for the voltage-dependent sodium
channel by Armstrong and Bezanilla in 1977. The
following model of the molecular mechanism of
inactivation emerges from our results. The amino
terminus of the Shaker channel acts as an inacti-
vation particle or internal plug for the channel.
When the inactivation particle is bound to the
receptor, the channel closes. We tested this
model further by applying a solution containing
free synthetic inactivation particle to the inside
face of mutant channels that did not inactivate. In
the presence of the synthetic inactivation parti-
cle, the mutant channels regained inactivation,
consistent with the ball and chain mechanism.
Ruth Murrell-Lagnado and I have used syn-
thetic peptide inactivation particles with altered
amino acid composition to determine the impor-
tant features that influence an ability to bind to
the internal mouth of the channel. The naturally
occurring amino acid sequence can be divided
into an uncharged region and a highly charged
region. The charged amino acids interact with
negative charges at or near the mouth of the chan-
nel to influence the rate of binding. The more
positive charges in the charged region, the faster
the binding rate. Negative charges in the charged
region inhibit binding. A surprising result is that
the net charge in this region seems to be the im-
portant factor for the binding rate, regardless of
5
Molecular Mechanisms of Ion Channel Function
which of the amino acids are charged. Con-
versely, the uncharged region is important for de-
termining the unbinding rate. This work was sup-
ported in part by the National Institutes of Health
and the National Institute of Mental Health.
Shaker potassium channels also exhibit a
slower inactivation process, which can be seen
both in wild-type channels and after the faster
inactivation process has been removed by muta-
genesis. The slow inactivation does not require
intact fast inactivation. Although it also does not
involve rearrangement of charge in the mem-
brane, this slower inactivation seems to involve a
mechanism different from the fast process. In col-
laboration with Kathleen Choi and Gary Yellen,
we have found that the slow inactivation is af-
fected by external agents, suggesting that the
conformational changes for this process involve
external structures. Dr. Hoshi, Jose Lopez Barneo,
and I have found that the rate of this type of inac-
tivation is profoundly affected by external ions.
Normally sodium is the predominant positively
charged ion outside the cell. When a small
amount of potassium is added to the external so-
lution, as could occur during epileptic activity in
the brain or other pathological states, the inacti-
vation is significantly slowed. We have studied
the dependence of the inactivation rate on the
type of external ion present and have determined
that the ability of an ion species to slow inactiva-
tion correlates with its permeability in the chan-
nel. Ions that get into the channel better from the
outside inhibit inactivation more effectively.
These results are consistent with a hypothesis
whereby a channel that is occupied by an ion can-
not undergo slow inactivation.
The slow inactivation process occurs by greatly
different rates in variants of the Shaker channel
with differences in structure at the carboxyl end
of the protein. We have made mutations in both
of these variants and have localized the differ-
ence responsible for the differences in slow inac-
tivation to a single hydrophobic amino acid in a
membrane-spanning region of the channel mole-
cule. Other amino acid substitutions at this posi-
tion have dramatic effects on gating, with larger
hydrophobic amino acids leading to slower
inactivation.
6
Divergent Members of the SRY Family
of Transcriptional Regulators Bind an
Insulin-Responsive Element, IRE-A
Maria C. Alexander-Bridges, M.D., Ph.D. — Assistant Investigator
Dr. Alexander- Bridges is also Assistant Professor of Medicine at Harvard Medical School and Assistant in
Medicine at Massachusetts General Hospital, Boston. She received her M.D. and Ph.D. degrees from
Harvard University Medical School, where she was a member of the Harvard-MIT Health Sciences and
Technology program, which is geared toward students interested in academic medicine. She developed an
/ abiding interest in hormonal regulation of cellular metabolism and, as a graduate student in physiology,
L investigated the mechanism of insulin- stimulated phosphorylation of cellular proteins. Dr. Alexander-
Bridges then served as an intern and resident at the Johns Hopkins University. After subspecialty training
in endocrinology at Massachusetts General Hospital, she was a postdoctoral fellow with Howard Goodman.
THE main function of insulin is to allow a
starved animal to adapt to a glucose load. In-
sulin does this by activating enzymes that pro-
mote energy storage and inactivating enzymes
that break down energy stores. Over the long
term, insulin alters the expression of these en-
zymes by regulating the transcription of their
genes. The work in our laboratory is directed to-
ward understanding the mechanism of insulin ac-
tion on expression of metabolically active genes
in tissues that modulate glucose utilization. We
use the glyceraldehyde- 3 -phosphate dehydroge-
nase (GAPDH) gene as a model gene for the ana-
bolic effects of insulin, because it is highly regu-
lated by insulin 1) in cultured 3T3 adipocytes, 2)
during nutritional manipulations such as fasting
and refeeding, and 3) during the induction and
treatment of diabetes.
Transgenic animals that express GAPDH-
growth hormone fusion genes have been made in
collaboration with Jeung Yun and Tom Wagner
(Ohio State University). Using these animals, we
have confirmed that GAPDH gene transcription is
regulated in vivo by nutritional manipulations
that lead to hyperinsulinemia. For example, fast-
ing a rat and refeeding it a high-carbohydrate,
low-fat diet increases circulating glucose and in-
sulin levels, resulting in the induction of glyco-
lytic and lipogenic enzymes and the repression of
gluconeogenic and lipolytic enzymes.
We have identified a cis-acting sequence,
insulin-responsive element A (IRE-A), in the up-
stream region of the GAPDH gene that interacts
with an insulin-responsive DNA-binding protein
(IRP-A) . Activation of GAPDH gene expression in
insulin-responsive tissues correlates with the
presence of IRP-A. Within one hour of exposure
of 3T3 adipocytes or H35 hepatoma cells to insu-
lin, the activity of this protein is increased two- to
fourfold. The activity of IRP-A is induced four- to
eightfold in liver and fat during the process of
refeeding a fasted rat a high-carbohydrate, low-fat
diet. These observations support the importance
of GAPDH gene regulation. In muscle, where
GAPDH activity is not rate limiting and is not reg-
ulated by insulin, IRP-A binding is not detectable.
The wild-type IRE-A motif was used to isolate a
clone from a rat adipocyte library, using the
Singh-Sharp Southwestern screening approach.
The cloned cDNA encodes a protein (IRE-ABP)
that binds IRE-A DNA with sequence specificity
that overlaps that of the adipocyte IRP-A nuclear
extract protein. IRE-ABP is expressed in liver and
fat but not in muscle, which provides an explana-
tion for the tissue-specific regulation of GAPDH
gene expression. Expression of IRP-A mRNA is in-
duced during the process of fasting and refeed-
ing. In contrast, one hour of insulin exposure of
cells does not appear to alter expression of the
IRP-A mRNA. IRE-ABP footprints the upstream re-
gion of genes that are inhibited and genes that are
stimulated by insulin. We presume that IRE-ABP
will regulate the transcription of metabolic genes
with diverse functions to change the phenotype
of the fat and liver cell during the switch from the
fasted to the refed state that is initiated by glucose
and insulin. Studies are in progress to determine
whether IRE-ABP mediates the effect of insulin
alone or in association with another protein.
IRE-A DNA-binding Protein Shares Binding
Specificity with the Testis-determining
Factor
Surprisingly, the HMG (high-mobility group)
box domain of IRE-ABP is 68 percent identical to
SRY, the testis-determining factor, and is 98 per-
cent identical to an autosomal gene that was iso-
lated during the process of screening a whole
mouse embryo cDNA library for 5i?F-related se-
quences. Furthermore, IRE-ABP and SRY share
DNA-binding specificity for IRE-A. Although IRE-
ABP shows markedly higher affinity for the IRE-A
motif, the nucleotides protected by these two di-
vergent family members are essentially identical.
The sequence in IRE-A that is contacted by IRE-
ABP and SRY is highly conserved: the sequence
5'-Py-ctttg(a/t)-3', previously defined by Kather-
ine Jones and her colleagues as a consensus motif,
7
Divergent Members of the SKY Family of Transcriptional Regulators Bind an
Insulin-Responsive Element, IRE-A
is contained in several T cell-specific genes that
are bound with high affinity by TCF-la (T cell
factor la). Thus diverse members of the HMG
family of proteins may modulate transcription
through a similar spectrum of sequences that
contain a core motif. Identification of such a mo-
tif may provide a clue to the identity of important
physiological targets of the IRE-ABP and SRY-like
family of transcriptional regulators.
The laboratories of Peter Goodfellow and Ro-
bin Lovell-Badge have identified 5/?Fas the testis-
determining region on the basis of genetic evi-
dence. Several patients with an XY genotype who
failed to differentiate to the male phenotype have
mutations in the HMG box domain of SRY. The
SRY locus is widely presumed to encode a se-
quence-specific DNA-binding protein because it
contains an HMG domain within its open reading
frame. The DNA-binding properties and mecha-
nism by which this protein regulates transcrip-
tion have not been defined. Certain of the muta-
tions that were found in the sex-reversed patients
were not de novo mutations, and thus it was not
possible to deduce whether these mutations were
completely unrelated to the sex-reversed pheno-
type or contributed to it. Because IRE-ABP was
isolated on the basis of its ability to bind the IRE-A
motif, it was possible to examine the effect of
these mutations on the binding affinity of IRE-
ABP and SRY derivatives. Derivatives that con-
tained the mutations associated with sex reversal
at positions 3 and 7 in the HMG box domain
showed marked impairment in their ability to
protect the IRE-A motif from DNase I digestion.
Furthermore, derivatives of IRE-ABP and 5.^y that
contain a switch between IRE-ABP and SRYat po-
sition 3 in the HMG box domain show altered
binding affinity for the IRE-A motif. Thus posi-
tions 3 and 7 appear to be important determi-
nants of binding affinity for this family.
Future Directions
Identification of the IRE-A binding protein as a
member of the 5J?Ffamily suggests many avenues
of investigation. Studies are under way to define
the binding specificity of IRE-ABP- and SRY-like
family members. Definition of the preferred bind-
ing site for these related proteins will facilitate
the identification of other insulin-sensitive genes
that are regulated by IRE-ABP and potential tar-
gets of SRY. For example, insulin simultaneously
activates and inhibits diverse metabolic pro-
cesses to alter the flux of metabolites into glyco-
gen and fat. We have located the proposed con-
sensus sequence in the upstream region of genes
that are regulated in a positive and negative direc-
tion by insulin, and we can now establish
whether IRE-ABP plays a role in regulating these
diverse metabolic processes.
The observation that the SRY protein and
IRE-ABP share binding specificity for a se-
quence located in a glycolytic gene implies that
these proteins carry out similar functions in the
specific tissues in which they are expressed. It
is clear, for instance, that spermatogenesis in
the adult testis requires high lactate produc-
tion; thus both gene products may be involved
in regulating glycolysis/gluconeogenesis in
their respective target tissues. Conversely, IRE-
ABP-like genes may play a role in regulating
processes that show sexual dimorphism in
adult tissues or promoting differentiation of its
target tissues during embryogenesis.
Studies on the regulation of IRP-A gene expres-
sion have led to an understanding of the mecha-
nism by which insulin modulates the expression
of specific genes in specific tissues involved in
the maintenance of normal glucose and lipid me-
tabolism. Ultimately we expect these studies will
lead to an understanding of the signal transduc-
tion process by which insulin modulates the ex-
pression of these genes. Understanding the hor-
monal control of lipid metabolism at a molecular
level will provide insights into two disease states
of major importance, obesity and diabetes.
8
Genetic Mechanisms Involved in the Generation
of the Antibody Repertoire
Frederick W. Alt, Ph.D. — Investigator
Dr. Alt is also Professor of Genetics and Pediatrics at the Children 's Hospital, Boston, and Harvard Medical
School. He obtained his undergraduate degree in biology from Brandeis University and his Ph.D. degree
in biological sciences from Stanford University, where he worked with Robert Schimke. He did postdoctoral
work with David Baltimore at the Massachusetts Institute of Technology, after which he was Professor
of Biochemistry and Microbiology at Columbia University College of Physicians and Surgeons.
His honors include the Irma T. Hirschl Career Scientist Award, the Searle Scholars Award,
and the Mallinckrodt Scholar Award.
WE are interested in the molecular mecha-
nisms that underlie the generation of a spe-
cific immune response. The mammalian immune
system functions through complex interactions
between various cells and their products. Cells
that effect specific immunological responses fall
into two general categories: B lymphocytes that
mediate humoral immunity (i.e., production of
antibodies against foreign antigens) and T lym-
phocytes that mediate cellular immunity (e.g.,
foreign graft rejection). During the earlier stages
of lymphocyte development, stem cells proceed
through a developmental program that ultimately
leads to the generation of a multitude of individ-
ual B or T lymphocyte clones (each clone is an
essentially identical set of cells derived from a
common parent). Each set of clonal cells ex-
presses a novel receptor on its surface that will
recognize a unique set of antigens.
The ability of the immune system to respond
specifically to a vast array of antigens results in
substantial part from the unique organization of
the genes that encode antigen receptor proteins.
Unlike most genes, antigen receptor genes are not
inherited intact from our parents. Instead, these
genes are encoded in cassettes (gene segments)
in the germline and are assembled into complete
genes only during the somatic differentiation of
lymphocytes. This lymphocyte-specific gene as-
sembly process is referred to as VDJ recombina-
tion. Because there are many individual cassettes
that encode various portions of antibodies and
because these can be put together in various com-
binations or in various ways, the body can ran-
domly assemble a vast array of different antibody
genes from a limited amount of genetic material.
Much of our work is aimed at determining the
genetic mechanisms by which antibody genes are
assembled from gene segments, the role of the
gene assembly process in the generation of anti-
body diversity, and the mechanisms that regulate
this gene assembly process and ensure that it oc-
curs only in appropriate cell types. We are also
working on the elucidation of molecular signals
that control the various steps of B lymphocyte
differentiation.
The enzymatic system involved in the assembly
of antigen receptor genes involves a variety of
different activities. Some of these activities are
expressed specifically in developing lympho-
cytes and are likely involved in the early pro-
cesses of gene assembly, including recognition of
the specific gene segments that will be joined and
cutting them away from the surrounding genetic
material . Other activities employed in VDJ recom-
bination are likely expressed in many cell types
where they may be involved in other processes,
such as replication of genetic material or repair
of lesions in genetic material (DNA repair) . The
more widely expressed activities are probably re-
cruited by the lymphocyte-specific components
of the system to carry out certain aspects of the
joining event, such as pasting together the "cut"
gene segments.
The lymphocyte-specific components of the
VDJ recombination system are likely encoded by
two genes (recently isolated by David Baltimore
and David Schatz) referred to as recombination-
activating genes 1 and 2 {RAG-1 and -2). Simulta-
neous expression of these genes was found to oc-
cur only in developing lymphocytes; artificially
induced expression of the two genes (simulta-
neously) in nonlymphoid cells confers them with
the ability to undergo VDJ recombination. How-
ever, these genes have been found to be ex-
pressed in other tissues, including the central
nervous system, fueling speculation that VDJ re-
combination activity or the activity of one or the
other of the two RAG gene products might be
involved in developmental processes in tissues
other than the immune system.
To test unequivocally the function of the RAG-
2 gene, we have used gene-targeting technology
to eliminate a copy of this gene in mouse embry-
onic stem cell lines. These cells were then intro-
duced into developing mouse embryos, where
they were incorporated into the germ cells of the
resulting chimeric mouse. Mice that carried a sin-
gle copy of the targeted mutation in their germ-
line were interbred to generate animals that com-
pletely lack the gene. Mice that lack the RAG-2
9
Genetic Mechanisms Involved in the Generation of the Antibody Repertoire
gene are viable and appear normal at birth. How-
ever, they routinely develop severe infections in
the first months of life due to the total absence of
mature T or B lymphocytes — a severe combined
immune deficiency. These animals do have large
populations of very immature B and T lympho-
cytes in their primary lymphocyte differentiation
organs; however, these cells are unable to initiate
the VDJ recombination process. Because the
.^G-2-deficient mice have no defects in any cell
type besides lymphocytes, we have concluded
that VDJ recombination activity and RAG- 2 gene
function are required only for lymphocyte devel-
opment and not for any other developmental
process.
Our group has also helped characterize the
basis of a naturally occurring murine mutation
that when homozygous leads to severe combined
immune deficiency (the scid mutation). Like
iL4G-2-deficient mice, animals homozygous for
the scid mutation lack functional B or T lympho-
cytes due to inability to assemble antigen recep-
tor gene segments correctly. However, mice ho-
mozygous for the scid mutation produce normal
RAG-1 and -2 gene products and can efficiently
initiate the VDJ recombination process. Comple-
tion of the joining process is blocked in scid mu-
tant cells. Moreover, both lymphoid and nonlym-
phoid cells of mice homozygous for the scid
defect are unable to repair certain types of lesions
in their genetic material. Thus the product of
the gene affected by the scid mutation appears to
represent an example of a component of the
VDJ recombination system that is used both in
VDJ recombination and in more general cellular
processes.
To define further generally expressed genes
potentially involved in VDJ recombination, we
provided the lymphocyte-specific components of
this system (RAG-l and -2) to mutant Chinese
hamster ovary cell lines that are defective in abil-
ity to repair breaks in their DNA. By introducing
antigen receptor gene segments into these lines,
we could ask whether the cells were capable of
performing correct VDJ recombination. By this
approach, we have now derived a series of inde-
pendent genetic mutations that affect both DNA
repair and VDJ recombination. Introduction of
specific human chromosomes into the mutant
cell lines restored both ability to repair DNA and
to undergo VDJ recombination. We are now in the
process of isolating the defective genes in these
cells. Elucidation of these genes will provide in-
sight into the VDJ recombination process and may
also provide information relevant to understand-
ing the basis for several human diseases that affect
both ability to repair genetic lesions and to gener-
ate a normal immune system.
We are also analyzing several other novel
mouse models generated to study various factors
involved in lymphocyte differentiation and the
generation of the immune response. One such
model is a mutant mouse line that cannot pro-
duce endogenous antibody molecules because it
has been genetically engineered to lack germline
gene segments necessary for forming functional
antibody genes. We are using these animals to
study the role of antibody gene products in regu-
lating antibody gene assembly and lymphocyte
development. A potential practical use of such
mice may be achieved by breeding them with
transgenic mouse lines that have been engineered
to contain functional human antibody gene cas-
settes. The hope is that the hybrid animal will
rely on the human antibody genes for its immune
system, providing a more effective method of
generating tailor-made human antibodies.
10
Control of Cell Fate During Vertebrate
Neuronal Development
David J. Anderson, Ph.D. — Assistant Investigator
Dr. Anderson is also Associate Professor of Biology at the California Institute of Technology and Adjunct
Assistant Professor of Anatomy and Cell Biology at the University of Southern California School of
Medicine. He received his A.B. degree in biochemical sciences from Harvard College and a Ph.D. degree in
cell biology from the Rockefeller University. He then did postdoctoral research in molecular neurobiology
at Columbia University. He is the recipient of an NSF Presidential Young Investigator Award, a Sloan
Foundation Fellowship in Neuroscience, and the 1989 Charles Judson Herrick Award in Neurobiology
from the American Association of Anatomy.
WE are interested in how different types of
nerve cells are generated during the devel-
opment of the vertebrate nervous system. We
have chosen to work on the peripheral auto-
nomic nervous system, which is simpler and
more accessible than the brain. Our studies have
focused on the development of two specific cell
types: the sympathetic neurons that lie in a chain
of ganglia along the spinal cord and the chromaf-
fin cells of the adrenal medulla. These two cells
are closely related, yet distinct in major respects.
The former are true neurons, with long branching
axons and dendrites that send and receive electri-
cal signals. The latter are small, round secretory
cells that release epinephrine (adrenaline) into
the bloodstream during fright or excitement.
Studies in a number of laboratories, including
our own, have established that these two cell
types derive from a common progenitor cell. This
cell arises on top of the neural tube (the develop-
ing spinal cord) , as part of a transient structure
called the neural crest. Like parachutists, the
neural crest cells peel off the neural tube in a
wave and migrate downward through the em-
bryo. Some of them arrest their migration in a
chain of small clumps along a blood vessel,
where they eventually become sympathetic neu-
rons. Others continue their migration downward
to invade the developing adrenal gland, where
they become chromaffin cells.
Using a sophisticated fluorescence-activated
cell sorter and specific monoclonal antibodies to
tag the cells, we have succeeded in isolating
chromaffin cell precursors from the fetal adrenal
gland of the rat. By manipulating the cell culture
environment, we have shown that these precur-
sors have two possible developmental fates: if
glucocorticoid hormones are added to the
growth medium, mimicking the environment of
the adrenal gland, then the precursors develop
into chromaffin cells. If, on the other hand, fibro-
blast growth factor (FGF) and nerve growth fac-
tor (NGF) are added to the medium, the precur-
sors develop into sympathetic neurons. This
suggests that the fate of these cells is determined
in large part by signals in the environments to
which they migrate. However, these precursors
seem to have lost the ability to give rise to other
derivatives of the neural crest, such as glial cells.
Therefore, while these precursor cells have a
choice of fate, it is a restricted one.
Control of Neuronal Survival
by Neurotrophic Factors
One problem in studying the chromaffin-neu-
ron precursor cell at the molecular level is the
small number of cells that can be recovered after
isolation from rat fetuses. To circumvent this
problem, wc have applied recently developed
techniques to immortalize the cells. Using a de-
fective retrovirus as a "disposable molecular sy-
ringe," we have injected the cells with a gene,
v-myc, that allows them to divide forever in the
culture dish. In this way we can generate an end-
less supply of cells that can be used for experi-
ments at any time, without the need to perform
long hours of dissection. Fortunately, these im-
mortalized precursor cell lines still appear capa-
ble of undergoing differentiation into sympa-
thetic neurons when exposed to FGF and NGF.
Sympathetic neurons, like other neurons, de-
pend on specific neurotrophic factors for their
survival. Neurotrophic factors are proteins se-
creted by the tissues innervated by the neurons;
these proteins nourish the neurons and keep
them alive. Different types of neurons use differ-
ent neurotrophic factors: NGF is the factor for
sympathetic neurons. The specificity of NGF ac-
tion is due to a specific receptor protein present
in the membrane of the nerve cell. This receptor
binds NGF and sends a signal to the cell, keeping
it alive. We are using the immortalized precursor
cell line to study how developing neurons ac-
quire NGF receptors during development. The
NGF receptor has two components (subunits);
our data suggest that different factors induce the
expresision of these two subunits. One factor that
seems to be important for the induction of func-
tional NGF receptors is electrical activity. This
suggests that the stimulation of the developing
Control of Cell Fate During Vertebrate Neuronal Development
neuron by its presynaptic partner may endow it
with the machinery to respond to key survival
factors secreted by its postsynaptic target. In this
way, the estabHshment of functional connections
between a neuron, its input, and its output target
could be coordinated.
Control of Neuron-Specific Gene Expression
As it differentiates, the chromaffin-neuron pre-
cursor expresses genes that are not expressed in
other cell types. How such specific expression is
achieved is a basic question in modern develop-
mental biology. Studies in blood and lymphoid
cells have revealed that genes such as that encod-
ing hemoglobin are specifically turned on in the
appropriate cells by activator proteins, present
only in those cells. We have studied a neuron-
specific gene, SCGIO, and found that a different
mechanism restricts its expression to developing
neurons. SCO 1 0 appears to be specifically turned
off in all tissues except neurons; this repression is
somehow relieved in neurons and not in other
cells. We have made advances in identifying a
protein involved in the repression of SCGIO. This
"silencer" protein is present in nonneuronal
cells and tissues but not in neuronal cells, consis-
tent with the idea that it shuts off SCGIO every-
where except in the nervous system. This same
silencer protein also appears to be involved in
shutting off other neuron-specific genes. This
suggests that specific de-repression may be a
common mechanism for controlling the expres-
sion of neuron-specific genes and that a common
repressor protein may silence the expression of
several such genes. Future efforts will be directed
at cloning the gene for this repressor protein and
understanding how it in turn is controlled.
Neural Development in Mammals
and Drosophila Uses Similar
Regulatory Molecules
We have used the chromaffin-neuron precursor
cell lines to isolate new genes that may be candi-
dates for controlling the development of these
cells. Our approach is based on the idea that Dro-
sophila genes controlling neural development
might also be conserved in mammals. One impor-
tant set of genes that control neuronal develop-
ment in Drosophila are those of the achaete-
scute complex (AS-C) . These scute genes encode
a group of related proteins that act by binding to
DNA, thereby controlling the activity of other
genes. We succeeded in isolating two scute-
related genes from the rat chromaffin-neuron
precursor cell line. The structures of these genes
are remarkably similar to those of their fruit fly
counterpans. Moreover, the rat scute-veliAed
genes appear to be expressed specifically in neu-
ronal precursor cells, like their counterparts in
the fly. These data indicate that there has been a
striking parallel conservation of gene structure
and cell type specificity during evolution. They
further suggest that the scwfe-related genes may
control the development of mammalian neurons.
These exciting findings suggest that the molecu-
lar mechanisms controlling nerve cell develop-
ment in vertebrate and invertebrate organisms
may be fundamentally similar. We are now in the
process of testing this hypothesis by making tar-
geted mutations in these genes in mice.
12
Cell Fate Choices in the Nervous System
and Elsewhere
Spyridon Artavanis-Tsakonas, Ph.D. — Investigator
Dr. Artavanis-Tsakonas is also Professor of Cell Biology and Biology at Yale University. He received
his M.A. degree in chemistry from the Federal Institute of Technology (Eidgenoessische Technische
Hochschule) in Zurich and his Ph.D. degree in molecular biology from the University of Cambridge,
England, for work done at the MRC Laboratory of Molecular Biology. His postdoctoral work was done
at the Biozentrum of the University of Basel with Walter Gehring and at Stanford University
with David Hogness.
A fundamental issue in the biology of multi-
cellular organisms is how a cell acquires its
specific developmental fate. The molecular rules
by which neighboring cells choose developmen-
tally distinct paths are unknown. We are particu-
larly interested in how these rules apply to the
nervous system and thus have been exploring the
molecular biology of early neurogenesis in the
fruit fly.
The central nervous system in Drosophila de-
rives from a set of precursor cells (neuroblasts)
that segregate from the epidermal precursors
(dermoblasts) in the very early ectoderm. Embry-
ological, genetic, and molecular studies have
demonstrated that neuroblast segregation de-
pends on interactions between neighboring cells.
Several genes capable of interfering with this pro-
cess have been identified, and the Notch locus is
central. As we have shown, Notch codes for a
transmembrane protein with homology to the
mammalian epidermal growth factor (EGF), im-
plying involvement in cell-surface events.
For several years we have been studying Notch
and other genetic elements in the cell inter-
actions underlying neuroblast differentiation.
Through these studies it became clear that the
mechanisms governing the segregation of the neu-
roblasts from the epidermal precursors in the
neural ectoderm are used not only in neurogene-
sis but also in a broad spectrum of tissues and cell
fate choices.
In an attempt to examine the complexity of the
genetic circuitry in which Notch is integrated
and to identify genes whose products may inter-
act directly with the Notch protein, we have been
using genetic screens designed to identify sup-
pressors or enhancers of specific Notch muta-
tions. Through this analysis, five genes, which we
operationally refer to as the '"Notch group," have
been identified, and extensive interactions
among them documented. Besides Notch, the set
includes four members that code for cell-surface,
cytoplasmic, and nuclear elements. Delta and
Serrate both code for transmembrane proteins
with EGF homologous extracellular domains; En-
hancer of split and mastermind code for nuclear
proteins; and deltex appears to be cytoplasmic.
Among the various genetic interactions within
the Notch group, those occurring between Notch
and Delta most clearly suggest that they interact
on the molecular level. To investigate this possi-
bility, we examined the effects of Notch and
Delta expression in Drosophila S2 cells. We
found that Notch- and De/to-expressing cells
formed mixed aggregates and that this occurs via
their extracellular EGF homologous domains. In-
terestingly, A'^o^c^-expressing cells exhibit vesic-
ular structures containing both Delta and Notch
proteins, implying that Delta is internalized via
Notch as a receptor. This relationship is currently
being analyzed by electron microscopy.
Deletion mutagenesis of the extracellular do-
main of Notch determined that only two of the 36
EGF repeats are sufficient and necessary to medi-
ate interactions with Delta. Thus the extracellu-
lar domain of Notch seems surprisingly modular
and could potentially serve to bind a number of
other proteins. Some of these may bind to subsets
of the EGF repeats or to parts of the extracellular
domain of Notch, while others may compete for
the same binding sites. Indeed, Serrate, which
displays homologies to the Delta protein, inter-
acts with Notch in a Delta-like fashion and, since
it binds to the same EGF repeats, may compete
with Delta for binding to Notch.
On the basis of these results, it was proposed
that Notch may function as a multifunctional re-
ceptor containing a series of ligand-binding sites,
each of which may interact with more than one
specific ligand. Such a model provides a plausi-
ble explanation for the pleiotropic action of
Notch and suggests a general function for the
gene throughout development.
If Notch acts as a receptor, how are Notch-
mediated extracellular signals transmitted to the
nucleus? This question is central to our current
work and is being addressed at several levels.
Cloning and sequencing of deltex (which dis-
plays allele-specific interactions with Notch,
Delta, and mastermind) revealed the primary
structure of its product, a 700-amino acid pro-
tein with no known homologues. Preliminary
results indicate that the deltex protein is cyto-
13
Cell Fate Choices in the Nervous System and Elsewhere
plasmic and appears to be influenced in its sub-
cellular localization by the expression of Notch.
While these observations are compatible with the
notion that Notch interacts with a cytoplasmic
protein, recent findings regarding both En-
hancer of split and the intracellular domain of
th&^Notch protein raise the possibility that Notch
could directly interact with nuclear components.
Most of the evidence supporting this hypothe-
sis stems from our analysis of human Notch group
homologues. In our attempts to dissect the mo-
lecular mechanisms involving Notch group
members, we have been exploring their involve-
ment in vertebrate development. While helping
to establish the degree to which the Drosophila
paradigm is valid for vertebrate species, this work
also permits us to make use of the vertebrate
model's experimental advantages, such as better-
defined cultured cell lines.
The Drosophila Enhancer of split locus was
shown to be defined genetically by several tran-
scripts. With one exception, the genes contain
the helix-loop-helix (HLH) motif found in a vari-
ety of transcriptional regulators, and they are par-
tially redundant functionally. The single gene
that does not contain the HLH motif displays ho-
mology to the |S-subunit of the signal-transducing
G proteins. Cloning of human homologues,
which were named TLEs (for transducin-like En-
hancer of split), revealed an extraordinary con-
servation between fly and human sequences. In
addition, there seems to be a family of TLE genes
in the human, though not the Drosophila, ge-
nome. Similarly, we found that the human ge-
nome contains at least two Notch homologous
sequences.
Comparisons among the four isolated human
TLE genes and the Drosophila counterpart
showed that they harbor a structural motif impli-
cated in nuclear-cytoplasmic transport. Referred
to as the "CcN motif," it consists of a closely
spaced combination of a nuclear localization se-
quence and potential phosphorylation sites for
both casein kinase II and cdc2 kinase. Such a se-
quence is consistent with the notion that TLEs act
as nuclear effector molecules. Indeed, immuno-
cytochemical localizations show that both hu-
man and Drosophila molecules are predomi-
nantly nuclear but can also be detected in the
cytoplasm.
Surprisingly, comparisons of the known Notch
homologues, including the two human se-
quences, reveal the existence of a CcN domain in
the intracellular part of Notch. We do not yet
have any evidence regarding the functional signif-
icance of this motif, which raises the provocative
possibility that Notch may possess some nuclear
functions.
In conclusion, work in the past year extended
our knowledge regarding the molecular rules un-
derlying the action of the Notch group genes.
Our studies — including analyses of expression
patterns, genetic interactions, and mutant pheno-
types — reinforce the notion that at least some of
these genes code for elements of a pleiotropic
cell interaction mechanism that is involved in
cell fate choices in many different tissues, in-
cluding the nervous system. Within those tissues,
however, cell fate choices can also depend on
tissue-specific interaction events. Our current
working hypothesis is that regulative events con-
trolling cell fate choices throughout develop-
ment rely on the parallel action of tissue-specific
mechanisms and on A^ofc^-mediated interactions.
A^o/^c^ activity, while apparently general, can nev-
ertheless be modulated by different ligands such
as Delta or Serrate.
Expression of Notch protein in the early stages
of wild-type Drosophila egg chambers (from
dissected ovaries ) as shown by confocal micros-
copy. Notch is highly expressed at the apical
surface of the follicle cells that surround the
developing egg yolk.
From Xu, T.. Caron, L.A., Fehon, R.G., and
Artavanis-Tsakonas, S. 1992. Development
115:913-922.
14
The Complement System
John P. Atkinson, M.D. — Investigator
Dr. Atkinson is also Professor of Medicine and of Molecular Microbiology at Washington University School
of Medicine and Physician at Barnes Hospital, St. Louis. He received his B.A. degree in zoology and M.D.
degree from the University of Kansas. His training in internal medicine was at Massachusetts General
Hospital, Boston, and at NIH. He conducted postdoctoral research at NIH and Washington University.
MANY poorly understood but relatively com-
mon and serious human diseases involve
aberrations of the immune system. Such condi-
tions range from mild forms of arthritis to life-
threatening autoimmune illnesses such as sys-
temic lupus erythematosus.
The immune system provides a powerful ar-
senal of proteins to attack and eliminate infecting
microorganisms. A major means of accomplishing
this task is through the special biologic partner-
ship of antibodies and complement. Proteins
from these two systems circulate in blood and are
either formed in response to the organism (anti-
bodies) or present at all times (complement).
The work in this laboratory centers on the com-
plement system. Its name is derived from its func-
tion of "assisting" antibody in immune reactions.
However, we now know that the complement
system also serves as an independent immune sys-
tem capable of attacking foreign elements by it-
self. This first-line defense appears to have oc-
curred early in evolution, preceding antibody.
The complement system consists of at least 30
proteins that interact with each other in reactions
resembling a cascade or waterfall. The activated
molecules that result from this scheme destroy
the invading microbe and promote the inflamma-
tory response. The complement system is a pow-
erful, swift, and highly effective means to fight
infectious organisms. As might be anticipated, an
inherited deficiency of a complement compo-
nent predisposes an individual to infections.
The production of antibodies is triggered when
foreign substances invade the body. Although
complement independently attacks the patho-
gen, the binding of antibodies to a target such as a
bacterium initiates and potentiates a series of re-
actions in which complement proteins swarm
onto the surface of the microbe. Such compo-
nents serve as ligands for complement receptors
on blood cells. A foreign particle so coated
quickly becomes adherent to and is ingested by
these blood cells. During complement activa-
tion, small fragments (peptides) that promote the
inflammatory response are liberated from the
complement proteins. These molecules dilate
blood vessels and summon scavenger cells, called
phagocytes, from the bloodstream. The phago-
cytic white blood cells, upon arrival at the site,
find organisms that are already prepared for in-
gestion; i.e., they are coated with antibody and
complement. This phenomenon is known as op-
sonization (from the Greek word opsonein, to
prepare for the table). As a result, the infection is
sequestered and eliminated via phagocytosis.
Sometimes, however, the immune system, in-
stead of synthesizing antibodies to foreign mate-
rials, seemingly makes a mistake and produces
antibodies that react with its own cells. For exam-
ple, in certain immune disorders, individuals
make antibodies to their own red blood cells. An-
tibody and complement then attach to the red
blood cells and destroy them. Such conditions are
known as autoimmune diseases, in this case au-
toimmune hemolytic anemia.
In other pathologic conditions, the immune
system does not efficiently eliminate the in-
fectious particles, and excessive quantities of im-
mune complexes form. Immune complexes con-
sist of antibodies, complement proteins, and the
foreign particle. These immune complexes are
proinflammatory and cause tissue damage if de-
posited in undesirable locations such as joints,
skin, and kidney. This leads to arthritis, dermati-
tis, and glomerulonephritis, respectively.
Thus there are two general ways in which the
humoral immune system can damage its own tis-
sue. The first is to produce antibodies to self-
components. The second is to form excessive
amounts of immune complexes.
Our goal is to understand how immune com-
plexes form and are processed. We are studying a
biologic mechanism that evolved to remove im-
mune complexes from the circulation. This pro-
cess helps prevent the pathologic accumulation
of these complexes in tissue. This system can be
likened to an "inner space shuttle." The physio-
logic vehicle for this journey is the red blood
cell. The red blood cell participates in this reac-
tion through a complement receptor protein on
its surface. These receptor proteins latch onto
complement-coated foreign particles, such as vi-
15
The Complement System
ruses and bacteria. As the circulating red blood
cell passes through the liver or spleen, its im-
mune cargo is released and metabolized. The red
blood cell then returns to the circulation ready to
ferry another load.
Our laboratory is studying the complement re-
ceptor involved in this process. We are also exam-
ining the complement proteins that swarm onto
the foreign particles. Furthermore, we are inves-
tigating certain "housekeeping" proteins of the
complement system. Because of the powerful de-
structive capabilities of the complement system,
it is perhaps no surprise that the body must keep
this system tightly regulated. Special proteins are
synthesized to protect the body's own cells from
damage by complement factors. Our laboratory
discovered a new family of genes that encode for
at least six complement receptor and regulatory
proteins. Two of these regulatory proteins occur
on almost all cells of the body. These are termed
decay-accelerating factor and membrane cofactor
protein. Recently these regulatory proteins have
been demonstrated to be expressed in relatively
high concentration on reproductive tissue, in-
cluding placental tissue and sperm. A new direc-
tion for the laboratory concerns the role of these
proteins in reproduction. Modulating the func-
tion of these regulatory proteins may also be im-
portant in improving the destruction of tumor
cells.
A further potential benefit from our research
involves the field of transplantation. Presently
the supply of donor organs is far less than the
number of patients. Many people worldwide die
while waiting for a suitable organ. Use of animal
organs has not been possible, partly because of
the complement system's attack on these trans-
plants. However, this problem could be sur-
mounted by using animal organs genetically engi-
neered with complement regulatory proteins
designed to prevent attack by the host's com-
plement system. We are investigating these
possibilities.
Finally, because infectious, autoimmune, and
immune complex-mediated illnesses result from
aberrations of the complement system, our re-
search efforts are helping to define the pathophys-
iologic basis of such diseases. In many autoim-
mune and immune complex-mediated diseases,
there is an inherited defect in the handling of
immune complexes or in the activation or regula-
tion of the complement system. Variations in the
structure, function, and expression of comple-
ment proteins are important aspects of autoim-
mune diseases. The information gleaned from
such research efforts could provide a basis for
new treatments.
16
The Molecular Biology of Smell
Richard Axel, M.D. — Investigator
Dr. Axel is also Higgins Professor of Biochemistry and Molecular Biophysics and of Pathology at Columbia
University College of Physicians and Surgeons. He received his undergraduate degree from Columbia
College and his M.D. degree from the Johns Hopkins University School of Medicine. He then came to
Columbia University as a resident in pathology at the College of Physicians and Surgeons. He held
fellowships in Columbia's Institute of Cancer Research (in Sol Spiegelman's laboratory) and the
Department of Pathology. Dr. Axel is a member of the National Academy of Sciences. Among his many
honors are the Eli lilly Award in Biological Chemistry and the Richard iounsbery Award from the
National Academy of Sciences.
PERIPHERAL neurons in vertebrate sensory
systems respond to environmental stimuli
and transmit these signals to higher sensory
centers in the brain, where they are processed to
allow the discrimination of complex sensory in-
formation. Mammals possess an olfactory sensory
system of enormous discriminatory power. Hu-
mans, for example, are thought to be capable of
distinguishing among thousands of distinct
odors. Even subtle alterations in molecular struc-
ture of an odorant can lead to profound changes
in perceived odor quality. How is this diversity
and specificity accomplished? The detection of
chemically distinct odors presumably results
from the association of odorants with specific re-
ceptors on olfactory neurons that reside in spe-
cialized epithelium in the nose. The brain must
distinguish which receptors or which neurons
have been activated to allow the discrimination
among different odorant stimuli. What mecha-
nisms have vertebrates evolved to allow the recog-
nition of this huge array of odorant molecules?
How does the brain know what the nose is smell-
ing? Insight into these problems is likely to de-
pend on the isolation and characterization of the
odorant receptors expressed in the nose.
We have recently identified an extremely large
multigene family whose members are likely to
encode a diverse family of odorant receptors. We
have determined the sequence of 30 genes and
have deduced the protein sequence of 30 differ-
ent putative olfactory receptors. The olfactory
proteins are clearly members of a superfamily of
receptors that traverse the membrane seven
times. Analysis of the proteins reveals structural
features that may render this family particularly
well suited for the detection of a diverse array of
structurally distinct odorants.
These olfactory proteins can be divided into
several different subfamilies that exhibit signifi-
cant divergence in the region of the receptor mol-
ecule thought to interact with the odorous li-
gand. These observations suggest a model in
which each of the individual subfamilies encodes
receptors that bind distinct structural classes
of odorants. Within a given subclass of odorous
ligands, the members would recognize more sub-
tle variations among odor molecules of a given
structural class. As such, this superfamily of mole-
cules would be uniquely suited to its putative
role in the fine discrimination of odor molecules
of subtly different structures.
The isolation of this large family of genes en-
coding the receptor molecules immediately pro-
vides one solution to the problem of olfactory
perception. How do we recognize so diverse an
array of odors? At one extreme, we would argue
that the recognition of diverse odorants could be
accomplished by a small number of promiscuous
receptors, each capable of interacting with sev-
eral structurally distinct odor molecules. Alterna-
tively, olfactory perception would result from
the presence of a large number of different recep-
tor molecules, each capable of interacting with
one or a small number of specific odorants. The
size of the gene family we have characterized
suggests that there are indeed a very large number
of odorant receptors, each capable of interacting
with a small number of odorous ligands. These
observations are in sharp contrast to other sensory
systems, such as vision or touch, where discrimi-
nation of sensory information is accomplished by
a rather small number of receptor modalities.
How then does the brain distinguish which re-
ceptors or which neurons have been activated to
allow the discrimination between different odor-
ant stimuli? In other sensory systems, such as vi-
sion and touch, neurons in the brain are orga-
nized in a topographic map that identifies the
position of a sensory stimulus. Thus the position
of a given neuron in the brain is used to define the
location of a sensory stimulus within the external
environment. Olfactory processing does not ex-
tract spatial features of the odorant stimulus. Re-
lieved of the necessity to encode information
about the spatial localization of a sensory stimu-
lus, the olfactory system may use spatial segrega-
tion of sensory input solely to encode the identity
of the stimulus itself. Recent data utilizing the
17
The Molecular Biology of Smell
receptor genes as molecular probes suggest a
model of olfactory processing in which sensory
neurons expressing distinct odorant receptors all
project to a restricted region within the olfactory
bulb, the first relay station in the brain. In this
instance the discrimination of odors would be a
consequence of the position of second-order neu-
rons in the olfactory bulb, such that spatially lo-
calized groups of neurons would preferentially
respond to different odorants. The molecular
identification of the genes encoding a large fam-
ily of olfactory receptors has provided initial in-
sight into the logic of olfactory processing in the
mammalian brain.
18
Mammalian Developmental Genetics
Gregory S. Barsh, M.D., Ph.D. — Assistant Investigator
Dr. Barsh is also Assistant Professor of Pediatrics at Stanford University. He received his M.D. and Ph.D.
degrees from the University of Washington, where he studied inherited diseases of collagen biosynthesis
in the laboratory of Peter Byers. Dr. Barsh's postgraduate training was in internal medicine and medical
genetics at Harbor- University of California Hospital, Los Angeles, and at the University of California,
San Francisco. His research in the laboratory of Charles Epstein focused on a molecular and genetic
characterization of recessive lethal mutations at the mouse agouti locus.
VERY little is known about the genetic control
of mammalian development. But embryo-
genesis of all mammals follows a similar plan, and
the basic rules discovered in one species are
likely to apply to others. By studying the mouse, a
species in which the early embryo can be ob-
served and manipulated, we will better under-
stand how genes control human development
and how disruption of these processes may lead
to such abnormalities as miscarriages and birth
defects.
In organisms traditionally subject to experi-
mental genetic analysis, such as fruit flies and
nematodes, mutations in a particular develop-
mental pathway can be selected in a comprehen-
sive screening experiment. In mice, however,
this approach has been hampered by the inability
to study and recover conditional mutations and
by the inefficiency of generating new mutations
through the insertion of mobile genetic ele-
ments. As a result, much of our insight into mam-
malian developmental genetics comes from the
study of preexisting mutations. We are examining
a group of previously identified genes that affect
development around the time of implantation. In
addition, we are developing a system to allow the
conditional disruption of genes with recessive
phenotypes in cell culture and in transgenic
mice.
Characterization of the Mouse agouti
and kreisler Genes
Located within a small region of mouse chro-
mosome 2 are a group of genes required for fun-
damental aspects of peri-implantation develop-
ment. The mousc^agouti {A) locus, originally
described as a coat color gene, controls the tim-
ing and type of pigment deposition in developing
hair follicles. Several A mutations, including A^,
a'^", and are lethal when homozygous, and at
least three genes required for embryonic develop-
ment have been defined at or near the A locus.
Closely linked to A is the segmentation gene
kreisler (kr), which affects formation of the in-
ner ear by interfering with the number and pat-
tern of metameric units, the so-called rhombo-
meres, in the developing hindbrain.
Toward the eventual goal of isolating these
genes, we have constructed high-resolution ge-
netic and physical maps of mouse chromosome 2,
using classical and somatic cell genetic ap-
proaches. These maps are based on a variety of
molecular markers in the area, including the in-
sertion sites of two endogenous retroviruses,
Emv-15 and Xmv-10, which have been closely
associated with the A^ and a mutations, respec-
tively. Our results have shown that neither Emv-
15 nor Xmv-10 has caused an A mutation. We
have also established the order and relative dis-
tances between many closely linked molecular
markers in the region and are beginning to lead
toward candidate cDNAs affected by the A and kr
mutations.
We have further characterized the relation-
ships among rhombomere formation, segmenta-
tion, and the kr mutation, based on a detailed
analysis of branchial arch derivatives in kr/kr
adults. In collaboration with Michael Frohman
and Gail Martin, we have studied the expression
of rhombomere-specific genes in fer/^r embryos.
Two of the rhombomeres affected by the kr muta-
tion contribute to the second and third branchial
arches, which ultimately form, among other
structures, the hyoid bone. In kr/kr adults, the
lower part of the hyoid, normally derived from
the third branchial arch, exhibits some morpho-
logic characteristics of the upper part of the
bone, normally derived from the second bran-
chial arch.
In kr/kr embryos, the expression domains of
Krox-20, Hox-2.6, Hox-2. 7, and Hox-2.9, which
probably code for transcription factors, and Int-
2, which codes for a growth factor and may play a
role in morphogenesis of the inner ear, are all
shifted rostrally toward the head. Taken together,
these results suggest that fer affects the location of
position-specific gene expression along the ros-
trocaudal axis and that the kr* gene product plays
a key role in the acquisition of positional iden-
tity. By isolating the kr gene and further charac-
terizing the fer/fer phenotype, we hope to clarify
19
Mammalian Developmental Genetics
the molecular pathways of mammalian segmenta-
tion and the role of Hox-2 genes in rhombomere
morphogenesis.
Insertional Mutagenesis
An important development in mammalian ex-
perimental embryology has been the ability to
isolate embryonic stem cells from preimplanta-
tion mouse embry^os, which can be modified in
cell culture and then used to reconstitute an in-
tact animal. Like the whole organism, these cells
are diploid and contain two copies of every auto-
somal gene. When embryonic stem cells are
mixed with a fragment of "reporter" DNA, such
as the coding regions of the neomycin resistance
gene or the /3-galactosidase gene, insertion of the
exogenous DNA provides a "gene trap, ' ' in which
expression of the reporter sequences is con-
trolled by regulatory^ elements of an endogenous
gene.
Insertion of the reporter DNA is likely to
disrupt expression of the endogenous gene, but
in most cases, expression from the uninterrupted
homologue will be sufficient to prevent pheno-
typic eff^ects. To block this expression in a single
step, and in a conditional fashion, we have con-
structed a series of gene trap vectors that contain
an inducible promoter on the strand opposite
the reporter sequences. "Captured" cell clones
are first selected by antibiotic resistance or /3-
galactosidase staining. After removal of selective
pressure and activation of the inducible pro-
moter, an antisense transcript is generated against
coding sequences of the endogenous gene. In
certain cases, this transcript will function in
trans, thereby inhibiting expression of the endog-
enous gene from the uninterrupted homologue.
Control experiments suggest that these DNA
vectors are capable of trapping endogenous
genes and that basal expression of the inducible
promoter does not reduce appreciably the fre-
quency of trapping. We are currently isolating a
panel of captured cell clones. The efficiency of
these antisense promoters in cis will be analyzed
by measuring expression of the reporter se-
quences before and after antisense induction. Po-
tential eff'ects of trans inhibition will then be
tested by examining the phenotypes of chimeric
mice that contain the mutant embryonic stem
cells. This approach will allow the phenotypic
effects of a recessive mutation to be studied in a
diploid organism by altering only one of the two
copies and may be applicable to many organisms
and developmental systems.
Ultimately we plan to place the gene trap vec-
tors in the context of retroviral packaging ele-
ments to allow their introduction into the mouse
germline by infection of developing gametes.
Theoretical advantages of using retrotransposons
as insertional mutagens include an enzymatic
mechanism of integration that conserves host
DNA sequences, accessibility to most if not all
segments of the genome, and the efficient detec-
tion and recovery of new integration sites. These
advantages, however, have been difficult to
achieve in the mouse, in part because of the low
efficiency of retrotransposition in the germline.
In the case of endogenous ecotropic proviruses
and exogenous Moloney murine leukemia virus
(MoMuLV) , a block to retroviral infection of de-
veloping gametes seems likely to be due to a com-
bination of factors, including decreased accessi-
bility of germline cells to viral particles with very
short infectious half-lives and a transcriptional
block to expression from the retroviral long ter-
minal repeat (LTR). As a first step toward over-
coming these hurdles, we have constructed, in
collaboration with Patrick Brown (HHMI, Stan-
ford University) , transgenic mice that contain the
transcriptional control sequences for the mouse
protamine 1 gene fused to the coding sequences
for MoMuLV. Because the mouse protamine 1
gene is highly expressed in postmeiotic sperma-
tids, the transgene should provide a rich source
of native retroviral particles in the vicinity of de-
veloping male gametes. Our preliminary results
show that the chimeric protamine-MoMuLV
transgene is functional and can lead to retroviral
infection of somatic and germline tissues.
20
Cell Cycle Control
David H. Beach, Ph.D. — Investigator
Dr. Beach is also Senior Staff Scientist at Cold Spring Harbor Laboratory and Adjunct Associate Professor
in the Department of Microbiology at the State University of New York at Stony Brook. He received
his undergraduate degree at the University of Cambridge, followed by a Ph.D. degree at the University of
Miami, where he worked with Marcus Jacobson. His postdoctoral studies were done with Sydney Shall
at the University of Susex and with Amar Klar at Cold Spring Harbor Laboratory.
THE cell division cycle interests an increasing
number of scientists and continues to be the
central focus of our research. The two most criti-
cal events of the cycle are the replication of DNA
that occurs during the S (synthetic) phase and the
segregation of chromosomes into daughter cells
that occurs in the M (mitotic) phase. DNA replica-
tion and mitosis are profoundly different cellular
events, both molecularly and in terms of cellular
mechanics, but both are regulated by closely re-
lated enzymes, members of a family known as the
cyclin kinases.
Each of these kinases has a catalytic subunit, of
which cdc2 is the prototype, and a regulatory sub-
unit that is a cyclin. Cyclins derive their name
simply from the property by which they were first
recognized. Their abundance oscillates in the di-
vision cycle, with the same periodicity as the cy-
cle itself. Many cyclins have been discovered, all
related in their amino acid sequence. In humans
there are at least five groups of such proteins,
called cyclins A, B, C, D, and E. Within each
group there may be further members, such as Dl ,
D2, D3 or Bl, B2. Cyclins of different classes ap-
pear to act at different stages of the cell cycle —
B-type cyclins at mitosis, A type at S phase, and
so on.
Also, in the case of the D-type cyclins, each
member of the family displays a specific tissue
distribution. D-type cyclins have attracted partic-
ular interest because they appear to act early in
the cell division pathway and can also mutate to
give rise to oncogenic forms. With cyclin Dl , the
gene is amplified or rearranged in a wide range of
human tumors, including cancers of the breast,
esophagus, bladder, and lymphatic system. It is
not yet known how overexpression or activation
of cyclin Dl contributes to the pathology of
cancer, but a clear connection has been
established.
A major activity in the Beach laboratory has
centered on identification of the regulatory mole-
cules controlling cyclin kinases, particularly the
cyclin B/cdc2 enzyme that functions in mitosis.
This enzyme is influenced by at least three phos-
phorylation events, some activating and others in-
hibiting. Tyrosine phosphorylation of cdc2 is the
best-characterized inhibitory mechanism, and
the relevant tyrosine kinases and phosphatases
have been identified both in yeast and human
cells.
The tyrosine kinases are encoded by the mikl'^
and weel^ genes of yeast, and the tyrosine phos-
phatases by the cdc25^ gene. Kinases and phos-
phatases act antagonistically. Thus, in the ab-
sence of the tyrosine kinase, cells enter mitosis
prematurely before the completion of DNA repli-
cation. Lack of the tyrosine phosphatase causes
arrest in the G2 phase of the division cycle.
A further class of negative regulators of mitosis
has recently been discovered. The piml^ (pre-
mature initiation of mitosis) gene of yeast en-
codes a homologue of a human protein called
RCCl. In the absence of piml/^CCl function,
cells also enter mitosis before the completion of
DNA replication. The gene product pirn 1 /RCCl
is an integral nucleosomal protein (present at
one copy per nucleosome) and is responsible for
signaling the state of the chromatin to the cdc2/
cyclin B enzyme. Only when the chromatin is
fully replicated can the cell proceed to division
by activation of the cyclin B/cdc2 enzyme. The
mechanism by which the pirn 1 /RCCl nucleoso-
mal protein communicates with the cyclin ki-
nases is an area of major current interest.
Many substrates of the cyclin kinases have been
discovered. They include major structural pro-
teins such as lamins and vimentin, which play a
vital role in the structural reorganization of the
cell that occurs during mitosis. Other substrates
include transcription factors and also negative
growth regulators such as the p53 and rb
proteins.
It is unclear why such different cellular activi-
ties as DNA replication and cell division are regu-
lated by a family of cyclin kinases that are clearly
related evolutionarily. The evolutionary similar-
ity implies that there was once only one cyclin
kinase and thus that this enzyme regulated only
one cell cycle event in a primitive cell. Was that
event DNA replication or cell division? It is very
difficult to imagine the primitive cell, because all
21
Cell Cycle Control
known cells are exquisitely interdependent enti-
ties. For example, the replication of DNA is de-
pendent on proteins, but the proteins are en-
coded by DNA, and so on. It is clear, however.
that the primitive cyclin kinase has evolved in
present-day cells into a very complicated family
of enzymes that regulate all stages of the cell divi-
sion cycle.
22
Genetic Control of Morphogenesis in Drosophila
Philip A. Beachy, Ph.D. — Assistant Investigator
Dr. Beachy is also Assistant Professor in the Department of Molecular Biology and Genetics at the Johns
Hopkins University School of Medicine. He did his graduate work in the Department of Biochemistry at
Stanford University School of Medicine with David Hogness. Before joining the Hughes Institute at Johns
Hopkins, he spent two years as Staff Associate at the Carnegie Institution's Department of Embryology.
He has been the recipient of several fellowships, including a Sloan Foundation Fellowship in Neuroscience.
TWO conceptually distinct processes operate
during Drosophila embryogenesis. In the
first of these, a detailed system of spatial informa-
tion is generated by a cascade of interactions
among several groups of genes and their prod-
ucts. During morphogenesis, the second phase,
interpretation of the blueprint results in the com-
plex arrangement of structures seen in the larva
and adult.
The first process is initiated during oogenesis
by maternal input of broad spatial cues. The in-
teracting genes that receive and refine this infor-
mation in the embryo have been extensively
characterized, in most cases at the molecular
level. Much less is known about how morphogen-
esis is directed, though it is generally believed to
occur through regulation of target genes by the
localized proteins generated during construction
of the blueprint. This laboratory's current inter-
ests center upon two groups of genes that operate
at the interface between these processes, the seg-
ment polarity genes and the homeotic genes.
The hedgehog Segment Polarity Gene
The segment polarity group is the latest acting
of the segmentation gene groups. It functions in
the refinement of positional information at the
level of individual cells within embryonic seg-
ments. During the earlier, syncytial period of Dro-
sophila embryogenesis, the absence of cell mem-
branes permits diff'usion between nuclei, which
can account for many of the interactions between
early acting segmentation genes and their prod-
ucts. Some segment polarity genes, however, are
expressed and function long beyond the syncytial
stages. Communication between cells for refine-
ment of positional information requires signaling
across cell membranes. Consistent with a role in
signaling, the products of many segment polarity
genes are secreted or localized to nonnuclear
compartments.
During the past year our laboratory isolated and
characterized the segment polarity gene hedge-
hog. The hedgehog locus was identified in the
exhaustive genetic screens carried out by Chris-
tiane Niisslein-Volhard, Eric Wieschaus, and their
colleagues, and was named for the bristle pattern
observed in mutant larvae. The phenotype resem-
bles that of a number of other segment polarity
genes, including wingless.
The wingless gene is expressed within each
segment in a stripe of cells adjacent to that in
which engrailed, another segment polarity gene,
is expressed. The normal function of each gene is
required within its own stripe for maintenance of
the other gene's expression within the neighbor-
ing (but not overlapping) stripe. The mutual re-
quirement of each gene for the other's continued
expression in a neighboring cell implies a mecha-
nism for communication between these cells.
Since wingless encodes a secreted peptide factor
and engrailed a homeodomain-containing tran-
scription factor, additional components must be
involved in the signaling mechanism.
Interest in this pathway has been stimulated by
recent results elsewhere indicating that targeted
inactivation of the murine homologues of wing-
less and engrailed produce specific developmen-
tal defects of the midbrain and cerebellum. Other
genes that may be involved in this pathway are
hedgehog and patched, which from genetic evi-
dence appear to interact. While patched encodes
a protein with apparent multiple membrane-
spanning domains, the product of the hedgehog
gene had not previously been characterized. We
gained entry into the hedgehog locus with the aid
of an "enhancer trap" P-transposable element in-
serted near the gene. Analysis revealed a se-
quence capable of encoding a protein of 471
amino acid residues. This includes a region near
the amino terminus suggestive of targeting to the
secretory pathway, either as a secreted polypep-
tide or as an integral membrane protein.
Either possibility would permit interaction of
the hedgehog protein with the membrane-span-
ning patched protein, in accordance with the ge-
netic prediction. Also consistent with such an in-
teraction, the hedgehog gene is expressed
coincidently with engrailed, while patched is
expressed coincidently with wingless in adjacent
cells. The physical nature of the interaction re-
23
Genetic Control of Morphogenesis in Drosophila
mains to be characterized, along with the intra-
cellular consequences of this signaling event.
Regulatory Functions of Homeotic Genes
Homeotic genes play a primary role in conver-
sion of the spatial information generated by the
earlier-acting segmentation genes into morpho-
logical structure. The products of homeotic
genes direct morphogenesis in a manner that
produces the diversity of structures distinguish-
ing the segments. This task is presumably ac-
complished by differential regulation of target
genes, though the identities of these targets
and the mechanisms by which homeotic genes
achieve their biological specificity remain
largely unknown.
We have studied in detail the DNA sequence-
recognition properties of homeodomains from
the homeotic genes Deformed and Ultrabitho-
rax. While these domains are very closely re-
lated, we have demonstrated clear differences in
DNA sequence preferences that lead to moderate
differences in affinity for individual binding
sites. Our recent work demonstrates that these
differences in sequence preference map to the
two DNA-contacting portions of the homeodo-
main. Even small differences in binding to indi-
vidual sites can lead to large differences in regula-
tory effect when multiple sites are present. This is
due to cooperative binding of homeodomain pro-
teins to multiple sites, a property we have studied
extensively for the Ultrabithorax protein. Dif-
ferences in DNA sequence preference by homeo-
tic gene products may thus account for a substan-
tial part of the biological specificity of homeotic
gene function.
24
Molecular Studies of Human Genetic Disease
Arthur L. Beaudet, M.D. — Investigator
Dr. Beaudet is also Professor in the Institute for Molecular Genetics and the Departments of Pediatrics and
Cell Biology at Baylor College of Medicine. He received his B.S. degree in biology from Holy Cross College
and his M.D. degree from Yale University. After completing his pediatric residency at the Johns Hopkins
Hospital, he performed postdoctoral research at NIH.
EXTRAORDINARY progress in the understand-
ing and diagnosis of human genetic diseases
has occurred over the past decade. Much of this
progress was made possible by the application of
recombinant DNA techniques to the analysis of
those diseases caused by an alteration in a single
gene. There are thousands of such disorders, in-
cluding well-known conditions such as cystic fi-
brosis, Duchenne muscular dystrophy, hemo-
philia A and B, sickle cell anemia, and Tay Sachs
disease. Although this progress in the diagnosis of
single-gene disorders continues unabated, the
ability to treat genetic disease virtually stands
still by comparison. There is hope that at least
some of these diseases will yield to somatic gene
therapy.
The extraordinary diagnostic ability that allows
anticipation of genetic symptoms in individuals
and their offspring raises important societal op-
portunities and concerns. These include preven-
tion of genetic disease through reproductive
planning, treatment prior to development of
symptoms, and possible discrimination (as for in-
surance or employment) on the basis of one's ge-
netic predispositions.
Abnormalities of a single gene, however, are
not responsible for most of the common medical
problems of adult life. Rather, these are caused
by differences in a number of genes in combina-
tion with environment and life-style. Problems of
this type include atherosclerosis, hypertension,
diabetes mellitus, and autoimmune disease.
Work in our laboratory involves various aspects
of precise genetic diagnosis, efforts to develop
somatic gene therapy, and attempts to understand
the complex genetic basis of some of these multi-
factorial disorders.
Cystic Fibrosis
Cystic fibrosis (CF) is a common genetic dis-
ease alfecting approximately 1 in 2,500 Cauca-
sians. Most CF patients die of progressive lung
disease during childhood or young adult life.
About 1 in 25 Caucasians carry an abnormal CF
gene, and about 1 in 625 couples are at high risk
of having an affected child. Our laboratory has
been actively involved in developing and imple-
menting DNA testing for purposes of prenatal
diagnosis and carrier detection of CF. Prenatal
diagnosis for high-risk couples who have a l-in-4
chance that a pregnancy will be affected with CF
is now a routine matter, and such diagnosis is per-
formed at many laboratories throughout the
world. Similarly, carrier detection for close rela-
tives of CF patients is routine.
Much more controversial is whether carrier
testing for CF should be offered to most or all
couples prior to reproduction. Although a single
common defect is present in the majority of CF
chromosomes, the remaining fraction of abnor-
mal chromosomes contains dozens of different
mutations. One goal of efforts in our laboratory is
to develop efficient methods for detection of
such multiple abnormalities. Currently carrier
testing identifies 80-90 percent of carriers and
would detect up to 80 percent of couples at risk
for having a CF child. Although it is unclear
whether CF carrier testing should be offered to all
couples immediately or in the very near future, it
is our view that carrier testing will be the starting
point for multiphasic DNA testing to provide in-
formation regarding reproductive and health
risks.
Precise genetic diagnosis for CF, and for other
genetic disorders, does not guarantee an effective
treatment. It would be desirable to have an ani-
mal model for easier analysis and therapeutic
trials. As one step in the process of developing an
animal model in mice, we have characterized the
sequence of the mouse CF gene. We have pre-
pared DNA clones that are suitable for disrupting
the normal mouse gene (which is similar to the
human gene) in cultured embryonic stem (ES)
cells. These altered ES cells can be used to gener-
ate mice affected with CF. The work to obtain a
CF mutant mouse is supported in part by a grant
from the Cystic Fibrosis Foundation.
Somatic gene therapy for CF focuses on the
lung, since it is involvement of the lung that
proves fatal for the vast majority of CF patients.
Treatment of the lung presents a unique circum-
stance, for therapeutic agents can be delivered by
25
Molecular Studies of Human Genetic Disease
inhalation rather than injection. We are preparing
to evaluate various methods for pulmonary gene
therapy. Liposomes containing DNA, retroviral
vectors, and adenoviral vectors will be used for
direct instillation into the lungs or for various
aerosol deliveries, beginning with normal ro-
dents and perhaps progressing to primates.
Leukocyte and Endothelial Cell
Adhesion Molecules
Leukocytes (white blood cells) are involved in
defense against infection and in most inflamma-
tory responses. Through complex mechanisms,
the various forms of leukocytes roll along the
blood vessel walls and eventually migrate
through the endothelial layer. This ability is me-
diated to a large extent by carefully regulated ad-
hesion molecules on the surface of leukocytes
and endothelial cells.
We are attempting to understand the role of
these proteins in normal biology and in multifac-
torial genetic diseases. This involves preparing
mutant mice and analyzing naturally occurring
genetic variation in humans. It is quite possible
that variation in these genes could be an impor-
tant factor in the predisposition to multifactorial
disorders such as atherosclerosis, autoimmune
disease, and a wide range of inflammatory pro-
cesses, including arthritis and diabetes mellitus.
We are focusing initially on CD 18, which is a
subunit of leukocyte integrin, and on intracellu-
lar adhesion molecule- 1 (ICAM-1), a member of
the immunoglobulin gene family. CD 18 is found
on the surface on many leukocytes, and its adhe-
siveness can be sharply increased through an acti-
vation process. ICAM-1 is found on many cells,
particularly endothelial cells, where it can bind
the CD18-containing integrins and mediate firm
cell attachment and migration of leukocytes out
of the bloodstream.
Individual genes can be altered in mice by the
technique of homologous recombination (ex-
change of DNA between related molecules) . Us-
ing this methodology in ES cells, a mutation was
introduced into the mouse CD 18 gene, and the
alteration was transmitted through the germline
so that the altered animals could be bred. The
mutant animals have a partial rather than com-
plete impairment of the CD 18 gene's function.
Leukocyte function shows significant alteration
in these animals, and they will be studied for any
differences in inflammatory responses or suscep-
tibility to atherosclerosis.
A mutation has also been introduced into the
ICAM-1 gene, and mice carrying this mutation in
the majority of their cells are being bred to obtain
transmission of the defect to the germline. Eff^orts
have also been initiated to prepare a mutation in
the P-selectin gene, whose product is expressed
on platelets and endothelial cells. Adhesion pro-
teins of the selectin family bind to sugar mole-
cules, rather than to proteins, on other cells. The
work to obtain mouse mutations in leukocyte and
endothelial cell adhesion molecules is supported
in part by a grant from the National Institutes of
Health.
A defect in CD 18 causes a rare human genetic
disease. In its severe form, this condition, known
as leukocyte adhesion deficiency (LAD), leads to
fatal susceptibility to infections. LAD is an attrac-
tive model for development of somatic gene ther-
apy for bone marrow-derived cells. A recombi-
nant retrovirus encoding the human CD 18
protein has been prepared and has been used to
transfer the gene into mouse and human cells.
Bone marrow transplantation was performed in
mice, and the infected cells express human CD 18
when reimplanted into the mouse. Cultured cells
from LAD patients were infected with retrovirus,
and the adhesion properties of the cells were re-
stored to normal. Bone marrow cells from af-
fected patients are being cultured in the labora-
tory and infected with the retrovirus in
preparation for performing human gene therapy.
Genetic variation in the adhesion molecules of
human leukocytes and endothelial cells is being
identified. In the ICAM-1 gene, for example,
there is an amino acid difference that is prevalent
in the population. It is of interest to determine
whether different forms of the ICAM-1 gene in
the human population are associated with differ-
ent susceptibilities to inflammatory diseases and
atherosclerosis. Some of the long-term goals of
this project are to identify naturally occurring ge-
netic variations that alter susceptibility to com-
mon disease processes. In addition, the mutant
mice can be used to determine if decreased func-
tion of some of these cell adhesion molecules re-
duces chronic inflammatory disease processes
(e.g., autoimmune disease, arthritis, multiple
sclerosis, diabetes mellitus). If reduced gene
function were to lessen the disease processes in
mouse models, this would provide evidence that
the function of these proteins might be blocked
by drugs as a means of slowing the analogous pro-
cesses in humans.
26
Molecular Genetics of Diabetes Mellitus
Graeme I. Bell, Ph.D. — Investigator
Dr. Bell is also Professor of Biochemistry and Molecular Biology, Medicine, and Genetics at the University
of Chicago. He received his B.S. degree in zoology and M.S. degree in biology from the University of
Calgary, Canada. He earned his Ph.D. degree in biochemistry and biophysics from the University of
California, San Francisco, where he also did his postdoctoral research. Before moving to the University of
Chicago, Dr. Bell served as Senior Scientist at the Chiron Corporation. Dr. Bell has received a number of
awards for his work, including the Outstanding Scientific Achievement Award from the American Diabetes
Association, the Rolf Luft Award from the Swedish Medical Society, and the Distinguished Alumni Award
from the University of Calgary.
DIABETES mellitus is a disorder of carbohy-
drate metabolism characterized by elevated
blood glucose levels. In the United States, an esti-
mated 6 million persons are known to have dia-
betes, and there are probably an equal number
with unrecognized disease. About 10 percent of
persons over age 65 have diabetes, and the com-
plications of the cardiovascular, kidney, visual,
and nervous systems are major causes of morbid-
ity and mortality.
Clinically diabetes is a heterogeneous dis-
order. One form, insulin-dependent diabetes
mellitus, primarily affects children and adoles-
cents. It results from immunological destruction
of the insulin-producing cells of the pancreas;
because of the absolute deficiency of insulin, pa-
tients require insulin therapy for survival. The
more common form of diabetes, non-insulin-
dependent diabetes mellitus (NIDDM), includes
about 90 percent of diabetic patients. This form
results from reduced insulin levels in some pa-
tients and a relative deficiency in others and is
due to abnormal functioning of the insulin-
producing cells or decreased responsiveness of
tissues to insulin. In these patients the blood glu-
cose levels can usually be controlled by diet or by
drugs that can be taken orally to improve insulin
secretion and action.
As with other common diseases of middle age
such as cardiovascular disease and hypertension,
genetic factors contribute to the development of
NIDDM. Our goal is to identify the genes that in-
crease the risk of developing diabetes and to de-
termine how, together with environmental and
life-style factors, they result in the elevated blood
glucose levels that define this disorder.
We have taken an approach that combines ge-
netics and molecular biology. Our working hy-
pothesis is that a relatively small number of po-
tentially identifiable major genes increase the
risk of developing diabetes and that the individ-
ual's overall genetic background, together with
environmental and life-style factors, influences
the phenotypic expression of the major suscepti-
bility genes.
In our genetic studies, we are studying diabetes-
prone families in which NIDDM has an early age-
at-onset and a clear autosomal dominant mode of
inheritance. In one such family we have identi-
fied a DNA marker on the long arm of human
chromosome 20 that is tightly linked to the
diabetes-susceptibility gene. We are presently
trying to isolate this gene and to identify the mu-
tation that impairs its normal function and leads
to failure of the insulin-secreting pancreatic ^-
cells. The identification of this gene may suggest
others whose mutation could also cause NIDDM.
Drawing on our understanding of the patho-
physiology of NIDDM, we are also cloning and
characterizing genes that might reasonably con-
tribute to diabetes susceptibility. Since the pan-
creatic /3-cell plays an important role in the patho-
genesis of all forms of diabetes mellitus, we are
particularly interested in studying the genes that
are responsible for its unique physiological char-
acteristics. These include genes encoding hor-
mones such as insulin and islet amyloid polypep-
tide (a newly discovered hormone-like peptide
of uncertain function), ion channels, and recep-
tors for hormones and other agents that regulate
insulin secretion. In addition to cloning and
characterizing genes expressed in /3-cells, we are
also identifying DNA polymorphisms in these
genes to facilitate genetic studies of their contri-
bution to development of NIDDM. Furthermore,
we are studying the effects of hyperglycemia and
other altered metabolic states on their expression
in order to determine if changes in the levels of
the proteins encoded by these genes can provide
a molecular explanation for the failure of the
cell to secrete appropriate amounts of insulin in
response to glucose in patients with NIDDM.
Recent studies of the receptor for the polypep-
tide hormone somatostatin are beginning to shed
new light into the mechanism by which this hor-
mone inhibits insulin secretion. In addition, they
are leading us into the area of neurobiology and
hormonal regulation of neuronal function. So-
matostatin is a cyclic 1 4-amino acid polypeptide
27
Molecular Genetics of Diabetes Mellitus
found throughout the gastrointestinal tract, in-
cluding the pancreas, and nervous systems. It has
many diverse functions, including inhibiting the
secretion of insulin and glucagon from the cells
of the pancreatic islets and of growth hormone
from the pituitary. In the central nervous system,
somatostatin acts as a neurotransmitter to regu-
late neuronal activity and to facilitate the release
of other neurotransmitters. It may also play a role
in centrally mediated behaviors such as move-
ment and cognition. We have recently cloned
the genes encoding three somatostatin receptors.
All are members of the seven-transmembrane-
spanning, GTP-binding protein family of recep-
tors. One of these, SSTR2, is expressed in the pan-
creatic /S-cell and is likely to be responsible for
mediating the inhibition of insulin release by so-
matostatin. The three somatostatin receptor iso-
forms that we have characterized have unique
pharmacological properties and couple to differ-
ent intracellular effector systems, thus providing
at least a partial explanation for the diverse physi-
ological effects of somatostatin. We are continu-
ing our studies of this family of receptors to gain a
better understanding of how somatostatin bind-
ing results in inhibition of insulin secretion.
In addition to genes expressed in the pancre-
atic /3-cell, we have also characterized genes for
membrane proteins involved in the transport of
glucose across the plasma membrane. Our stud-
ies have revealed unexpected functional com-
plexity that could have important implications
for the treatment of diabetes. These studies indi-
cate that facilitative glucose transport is not the
property of a single protein but rather involves a
family of at least five structurally related pro-
teins. These proteins have distinct patterns of tis-
sue distribution and different kinetic properties
and are independently regulated. These features
allow the precise disposal of dietary glucose
under varying physiological conditions. Recent
studies have shown that one of the five proteins
we have identified is the major glucose trans-
porter of neuronal cells and that the major func-
tion of another is to transport fructose across the
plasma membrane. This fructose transporter is
expressed on the luminal surface of the absorp-
tive epithelial cells of the small intestine and kid-
ney tubules and is also present in early and late
spermatids in the testes. Sperm require a fructose
transporter, because they utilize fructose in semi-
nal fluid as an energy source.
Our studies are leading to a better understand-
ing of the causes of diabetes mellitus. In addition,
they are providing new insight into the function
of somatostatin in the regulation of neuronal
function. The identification of the fructose trans-
porter in sperm may also have implications in re-
productive physiology. Our results also illustrate
how studies in one area can impact those in an-
other, because of the exquisite manner in which
the organism is able to utilize similar proteins for
different physiological functions.
28
Development of the Drosophila Peripheral
Nervotis System
Hugo J. Bellen, D.V.M., Ph.D. — Assistant Investigator
Dr. Bellen is also Assistant Professor in the Institute for Molecular Genetics, the Division of Neuroscience,
the Department of Cell Biology, and the Program in Developmental Biology at Baylor College of Medicine.
Educated in Belgium, he received a degree in commercial engineering and began research in sociometry,
but decided to pursue a career in medical science. He obtained a D.V.M. degree from the University of
Ghent and a Ph.D. degree in genetics from the University of California, Davis, where, in John Kiger's
laboratory, he studied mutations that affect behavior in the fruit fly. Later, in Basel, Switzerland, as
a postdoctoral fellow with Walter Gehring, he helped to develop the enhancer detection system.
OUR research is centered on the development
of the peripheral nervous system (PNS) of
the fruit fly Drosophila melanogaster. The re-
sults will help us analyze the development of the
nervous system of many eukaryotic species. We
have focused on the PNS of Drosophila because
its cells are relatively easy to study and because
sophisticated genetics can be applied to this
model organism.
We are presently studying the couch potato
(cpo) and neuromusculin (nrm) genes, which
are expressed at the onset of embryonic nervous
system development and later in embryogenesis
in most cells of the PNS. These genes were identi-
fied in enhancer detector screens in which regula-
tory sequences of genes are identified by means
of a ;8-galactosidase reporter. Here we describe
the preliminary characterization of these essen-
tial genes and the development of a novel tech-
nique to ablate nervous system cells.
The cpo gene was so named because homozy-
gous mutant flies are viable but have poor jump
responses, poor flight abilities, slow recovery
after ether anesthesia, and are generally hypoac-
tive. More than 10 insertional cpo mutations
were recovered from several enhancer detector
screens; none of the insertions caused a complete
loss of function of the gene. Some insertional al-
leles cause recessive embryonic lethality, but no
defects were observed in the PNS of the mutant
embryos.
A molecular analysis of cpo was initiated with
the isolation of cpo cDNAs. In situ hybridizations
to whole-mount embryos showed that the gene is
expressed in the precursor cells of the PNS as
well as glia of the PNS and CNS (central nervous
system). Sequencing data show that the gene en-
codes at least three different proteins that contain
many stretches of the same amino acids. Two of
these proteins contain a domain that may allow
binding to RNA molecules. This binding domain
is found in many proteins that are involved in
processing immature RNA in a form that can be
readily translated into a protein. The CPO protein
has most significant homology with two other
RNA-binding proteins that are found in nervous
tissue: HuD, a human brain paraneoplastic en-
cephalomyelitis antigen, and elav, the protein
from a Drosophila nervous system-specific gene
{embryonic lethal abnormal vision).
The cpo gene sequence was used to produce a
polyclonal antibody to CPO. The antibody allows
the localization of this nuclear protein in the de-
veloping embryo, in essentially the same cells as
the transcript. In addition, the antibody recog-
nizes a protein that is associated with polytene
chromosomal bands isolated from the salivary
gland of third instar larvae. This association may
be mediated through the binding of CPO to na-
scent RNA that is being synthesized from the DNA
of the giant polytene chromosomes. Thus it is pos-
sible that CPO acts as an essential PNS diff^erentia-
tion factor by controlling some aspects of RNA
maturation specific to the cells in which CPO is
expressed.
In addition to its biological functions, the cpo
gene is also a hot spot for P-element enhancer
detector insertions. Ten insertions that cause a
variety of phenotypes were mapped at the nu-
cleotide level, and molecular defects in cpo mu-
tant strains that carry an enhancer detector were
determined. All insertions recovered so far are
clustered in a 200-bp genomic fragment that con-
tains key regulatory regions of the cpo gene. Most
insertions are integrated upstream of the first nu-
cleotide of the longest cDNA but downstream of
the consensus binding sites of three known Dro-
5op^«7a transcription factors. Thus they are proba-
bly located between the enhancer-binding sites
and the promoter of cpo. Five insertions causing
different phenotypes have inserted at exactly the
same location. There is a perfect correlation be-
tween the observed defects (or their lack) in em-
bryos or adults and the size and the orientation of
these insertions — e.g., large insertions are less
detrimental than smaller insertions. These data
suggest that the distance between the binding
sites of the transcription factors and the transcrip-
tion initiation site is critical for proper cpo regu-
lation. In addition, insertions with the /«cZ gene
29
Development of the Drosophila Peripheral Nervous System
in the opposite orientation of cpo transcription
cause much less severe defects than insertions in
the same orientation. These observations provide
a unique opportunity to study the interactions be-
tween enhancers and promoters in vivo. The
study of cpo is also supported by the Muscular
Dystrophy Association.
The nrm gene is expressed early in the develop-
ment of the PNS and muscles. It maps at cytologi-
cal band 80A and was also identified in an en-
hancer detector screen. To our knowledge only a
single insertion has been identified in nrm, and
the jS-galactosidase expression pattern in em-
bryos of this strain is virtually identical to that of
cpo during early development. Flies homozygous
for this particular insertion are homozygous via-
ble. A mutagenesis experiment allowed us to re-
cover 10 recessive lethal mutations that fail to
complement each other. Although many subtle
defects have been observed in different mutant
strains, no consistent morphological defects have
been observed.
The cloned nrm gene encodes a single 4.6-kb
transcript that is expressed transiently in the sen-
sory mother cells and developing neurons and
support cells of the PNS (5-10 h). Later in embry-
onic development (14-17 h) the transcript is
also observed in most embryonic muscles. The
earliest expression of nrm in muscles coincides
with the formation of the neuromuscular junc-
tion. Sequencing data show that the nrm gene
encodes a novel protein that contains a signal
peptide and eight immunoglobulin domains. The
protein is possibly anchored in the membrane of
the cells by a secondary modification of the pro-
tein. On the basis of these molecular data and the
expression pattern of nrm, we propose that nrm
is a novel neural cell adhesion molecule that may
play an important role in growth cone guidance
in the PNS and may affect the formation of the
neuromuscular junction. These hypotheses are
presently being tested.
To study the role of glial cells in the develop-
ment of the PNS, we are developing a method that
allows temporally induced arrest of most func-
tions in specific cells. These arrests are induced
by temperature shifts of flies that express temper-
ature-sensitive (ts) forms of diphtheria toxin
(DT-A'*) under the control of cell- or tissue-
specific regulatory sequences. DT-A'* genes were
isolated in a mutagenesis screen using the yeast
Saccharomyces cerevisiae and subsequently
tested in PNS neurons, namely the R1-R6 photo-
receptor cells of transgenic fruit flies. Four DT-A'*
have been partially characterized in yeast cells,
and three have been tested in Drosophila. These
toxins show similar temperature dependence,
suggesting that they may be useful in a wide range
of species. We are presently trying to ablate glial
cells in the PNS and CNS, in an attempt to exam-
ine the role of glia in the living organism.
Expression pattern of the couch potato protein in an almost fully developed Drosophila
embryo. This protein is expressed in all cells of the peripheral nervous system and in a
subset of cells of the central nervous system. Anterior, left, dorsal, up.
Research and photograph by Sandra Kooyer and Diana D 'Evelyn in the laboratory of
Hugo Bellen.
30
Genetic Manipulation of Hematopoietic
Stem Cells
John W. Belmont, M.D., Ph.D. — Assistant Investigator
Dr. Belmont is also Assistant Professor of Molecular Genetics, Pediatrics, and Microbiology and
Immunology at Baylor College of Medicine. He received his undergraduate degree from the University of
Texas, Austin, and his M.D. and Ph.D. degrees from Baylor College of Medicine, where he worked with
Robert Rich. After internship and residency training in pediatrics at Children 's Hospital, Washington,
D. C, he completed a fellowship in medical genetics at Baylor College of Medicine.
PLURIPOTENT hematopoietic stem cells are
the "seed" for the development of all blood
cells. These cells, which normally reside in the
bone marrow, arise in early fetal development
and persist throughout adult life. They can be re-
moved from the bone marrow and transplanted
into a prepared recipient; they will then stably
reconstitute the entire blood and immune sys-
tems. We have developed techniques for the effi-
cient transfer of genes into mouse and human
stem cells, using retroviral vectors. The long-
term goal of these studies is to perfect methods
that could be used for the treatment of various
genetic and acquired diseases.
A viral vector system based on the Moloney mu-
rine leukemia virus (MoMuLV) has been chosen
because of its potential for very high gene
transfer efficiency. The unique life cycle of this
retrovirus makes it attractive for adaptation as a
vector, since the foreign genetic material is stably
integrated into the host cell genome. MoMuLV
vector particles are able to carry their genetic ma-
terial to the target cells but are incapable of repli-
cating and spreading as a live infectious agent.
We have used two model systems to investigate
properties of the stem cells. In one model the
bacterial antibiotic resistance gene neo intro-
duces distinct genetic tags into individual stem
cells. The second model uses the human enzyme
adenosine deaminase (ADA) as the molecular
marker for gene transfer. This system is particu-
larly suitable for studies of expression of genes by
the retroviral vectors. In addition, the genetic de-
ficiency of ADA causes a form of severe combined
immune deficiency, so that successful laboratory
experiments with this gene may in time facilitate
the clinical application of the gene transfer
procedures.
Our earlier work demonstrated that genes
could be introduced into hematopoietic stem
cells but that the process was much less efficient
than in the more mature cells of the marrow. In
mouse transplant experiments, only about 50
percent of the animals retained expression of the
human ADA enzyme in their blood for more than
six months. This has led to an investigation of the
conditions in cell culture that would optimally
support the survival or proliferation of the stem
cells.
In collaboration with Doug Williams (Im-
munex, Seattle), we have evaluated the efi'ects of
several recombinant hematopoietic growth fac-
tors on retroviral vector-mediated gene transfer
into stem cells. These factors included interleu-
kin-3, -6, and -7; granulocyte colony-stimulating
factor (G-CSF); and leukemia inhibitory factor
(LIF). LIP has been of special interest because
among its many biological functions, it appears to
prevent the difi'erentiation of mouse embryonic
stem cells. If it had a similar action on hematopoi-
etic stem cells, it might allow the preservation of
their developmental capacity in culture.
A novel assay using inbred transgenic mice was
used to test the activity of LIF on stem cells. These
experiments indicate that LIF preserves the stem
cells during the culture period required for gene
transfer. Inclusion of LIF in the bone marrow cul-
tures allows about 70 percent of the stem cells to
be infected with the retrovirus. Subsequently all
the mice transplanted with such cells maintain
high-level expression of human ADA in all their
blood and immune system organs for at least six
months. This finding has recently been supported
in collaborative experiments with Savio Woo
(HHMI, Baylor College of Medicine) using a re-
troviral vector that encodes the human a j -anti-
trypsin gene.
We suspect that LIF acts in concert with one or
more other growth factors in our experimental
model. The recently isolated ligand for the c-kit
cell surface receptor is a promising candidate for
a factor that causes stem cells to replicate. In an
effort to survey critical growth factor/ligand sys-
tems expressed by stem cells, we have tried a
cloning method based on the conservation of
gene sequence to identify new candidate recep-
tors. Messenger RNA extracted from stem cell-
enriched fractions of bone marrow has been used
as a template for amplification of conserved re-
ceptor tyrosine kinase sequences. Among the
clones are three that had not previously been de-
scribed. We are currently sequencing these genes
31
Genetic Manipulation of Hematopoietic Stem Cells
to determine if they are novel receptors involved
in the regulation of hematopoiesis.
To analyze better the behavior of stem cells and
their progeny in culture and after transplant, we
are testing a new method for introducing unique
genetic identifiers into individual stem cells.
This method allows identification of the stem
cells by the polymerase chain reaction (PGR),
which is especially suited for analysis of very
small numbers of cells. A family of vectors carry-
ing the bacterial gene neo have been constructed
that are identical except for small variations in
size. This size variation is conveniently distin-
guished by PGR. When mixed together these vec-
tors provide an array of potential markers that can
integrate into each stem cell. Because each stem
cell receives one to four vectors, the integration
of a particular subset of the vectors provides a
genetic fingerprint unique to each stem cell fam-
ily. Preliminary results with these vectors indi-
cate that we can mark the most primitive hemato-
poietic precursors very efficiently. We plan to
use this method to compare different culture
conditions for stem cells.
Retroviral vector genetic marking may also
prove to be very informative in clinical bone
marrow transplantation. GoUaborative studies
with Albert Deisseroth (M.D. Anderson Hospital,
Houston) investigated whether the techniques
that have been used to optimize mouse bone
marrow stem cell gene transfer could be used for
human cells. These experiments demonstrate the
importance of supporting cells in the culture
(called stromal cells) for efficient gene transfer.
The work was performed using an assay of human
stem cells that depends on their persistence in
culture. We hope that many valuable lessons will
be learned from these studies that will contribute
to the long-term goal of treating disease by the
use of therapeutic vectors.
Human bone marrow colony called an erythroid burst or BFU-E. The colony grows from a
single cell to form 3-9 clusters of red blood cells. Measurement of gene transfer into such
colonies has helped to improve gene therapy protocols.
Research and photograph by Kateri Moore in the laboratory offohn Belmont.
32
Proteins of the Spectrin-based Membrane Skeleton
G. Vann Bennett, M.D., Ph.D. — Investigator
Dr. Bennett is also Professor of Biochemistry at Duke University Medical Center. He received his M.D. and
Ph.D. degrees from the Johns Hopkins University Medical School. He completed postdoctoral training at
Harvard University in membrane protein biochemistry with Daniel Branton. Before joining Duke
University, he was on the faculty in the Department of Cell Biology and Anatomy at Johns Hopkins.
STRUCTURAL proteins in the cytoplasm and
membranes of cells provide the basis for spa-
tial organization of the diverse components of eu-
karyotic cells. These proteins thus are principal
participants in fundamental cellular activities
such as cell motility and cell-cell interactions.
Our work over the past 1 0 years has focused on
plasma membranes. We initiated these studies in
the human erythrocyte. This relatively simple
cell has provided an experimentally accessible
model system for detailed dissection of protein-
protein interactions that are responsible for the
structure and organization of the plasma
membrane.
The principal structural protein in the eryth-
rocyte membrane is the flexible rod-shaped
molecule spectrin, which is organized in a two-
dimensional network attached to the cytoplasmic
surface of the plasma membrane. Spectrin mole-
cules are attached at their ends to form a series of
hexagons and pentagons that closely resembles a
geodesic dome. The binding of spectrin to the
protein ankyrin attaches the spectrin network to
the plasma membrane. Ankyrin also interacts
with high affinity with the cytoplasmic domain of
an integral membrane protein (a protein that tra-
verses the membrane and actually has portions
exposed on both the inner and outer membrane
surfaces) . The spectrin-based membrane network
or skeleton is required for normal stability of
erythrocytes in the circulation. Abnormalities in
amounts or function of spectrin and associated
proteins result in hemolytic anemias and are the
basis for diseases such as hereditary spherocytosis
and hereditary elliptocytosis.
Proteins closely related to spectrin are present
in many vertebrate cells and are associated in
most cases with the plasma membrane. Spectrin
is present in especially high amounts in brain,
where it comprises 3 percent of the total mem-
brane protein. The spectrin-based membrane
skeleton in brain and other tissues is likely to play
an important role in providing organization of
integral membrane proteins in the plasma mem-
brane and for coupling membrane proteins to ele-
ments of the cytoskeleton. Potential physiologi-
cal consequences of these activities include
stabilization of the lipid bilayer and organization
of membrane proteins in specialized regions on
the cell surface in polarized cells.
Specific aims of this laboratory are to elucidate
the proteins in erythrocytes and other cells that
mediate interaction of spectrin with membranes,
determine how these protein interactions are reg-
ulated, and understand the cellular functions of
the spectrin skeleton.
Ankyrins in the Nervous System
Ankyrin appears to function as an adapter be-
tween certain membrane proteins and the spec-
trin skeleton. We have discovered that brain con-
tains multiple forms of ankyrin, with diversity
due to distinct genes as well as alternative splic-
ing of RNAs encoded by the same gene. We have
determined the complete amino acid sequence of
the major form of ankyrin in human brain and
have discovered an unusual alternative form of
this protein that contains a large inserted se-
quence. The large form of brain ankyrin is the
first ankyrin detected during brain development
and is targeted to neuronal processes including
unmyelinated axons.
The same gene that encodes ankyrin in erythro-
cytes also is expressed in brain. This form of an-
kyrin is localized in the plasma membranes of
certain neurons and is abundant in cerebellum,
brain stem, and spinal cord. The erythrocyte form
of ankyrin is missing in a strain of mutant mice
developed at the Jackson Laboratory. Ankyrin-
deficient mice experience degeneration of Pur-
kinje cells, a major type of neuron in the cerebel-
lum, and develop a stagger and difficulty in
walking. Neurological problems of ankyrin-defi-
cient mice may have counterparts in humans with
slowly progressive diseases due to death of nerve
cells.
Another isoform of ankyrin is highly concen-
trated along with the voltage-dependent sodium
channel at the nodes of Ranvier of nerve axons.
Nodes of Ranvier are specialized regions on the
axons of nerves where the myelin or insulation of
the axon is interrupted and where ions can enter
33
Proteins of the Spectrin- based Membrane Skeleton
or leave the axon through ion channels. Localiza-
tion of the voltage-dependent sodium channel at
nodes of Ranvier is important for normal conduc-
tion of nerve impulses. We hope to identify the
gene encoding the form of ankyrin at the node of
Ranvier and eventually to understand the role of
nodal ankyrin in organization of this specialized
membrane domain. These studies will have rele-
vance to diseases of neurons such as multiple scle-
rosis, where the myelin coating of axons is lost
and sodium channels are no longer restricted to
the nodes of Ranvier.
Diversity of ank)'rins suggests that this family of
proteins may interact with many membrane pro-
teins. Characterized membrane proteins in brain
that associate with ankyrin in in vitro assays in-
clude the voltage-dependent sodium channel and
sodium/potassium ATPase. We have used the
membrane-binding domain of the major form of
brain ankyrin to isolate ankyrin-binding proteins
and have identified a family of ankyTin-binding
proteins found in plasma membranes of neurons
and glial cells. These ankyrin-binding proteins
represent 0.3 percent of adult brain membrane
protein and appear late in postnatal develop-
ment. An important goal for future work will be
to isolate the cDNAs encoding these proteins and
determine their function in adult brain.
Ankyrin-Independent Membrane
Attachment Sites for Spectrin
Brain spectrin can also associate directly
with membrane proteins through an interaction
that is independent of ankyrin. We have discov-
ered that calcium, in concert with three differ-
ent calcium-regulated proteins (calmodulin, a
calcium-activated protease, and protein kinase
C), inhibits the direct spectrin-membrane link-
age but has no effect on spectrin-ankyrin interac-
tions. These results suggest that the spectrin skel-
eton includes both stable, ankyrin-mediated
linkages and dynamic calcium-sensitive associa-
tions that are subject to metabolic control. Iden-
tification of the spectrin "receptor" is the first
step in understanding the role of this type of
spectrin-membrane interaction in cells.
Ankyrin Structure
Ankyrin contains three independently folded
domains: one that interacts with certain mem-
brane proteins, another that associates with spec-
trin, and a third that regulates associations of the
binding domains. Surprisingly, the membrane-
binding domain of ankyrin includes an amino
acid sequence that is homologous to regions of
sequence in a group of apparently unrelated pro-
teins from flies, yeast, and even viruses. We have
recently discovered that this conserved portion
of the ankyrin sequence is responsible for the in-
teraction of ankyrin with at least one membrane
protein. Ankyrin thus contains a highly conserved
and ancient structural motif that may have a gen-
eral role in molecular recognition. We hope to
determine the three-dimensional structure of this
portion of ankyrin, with the expectation that this
structure will help us understand interactions of
ankyrin and other proteins with related sequences.
Adducin
The protein adducin is a candidate to play a
role in assembly of the spectrin skeleton in eryth-
rocytes, brain, and certain epithelial tissues. We
have found that adducin is localized at sites of
cell-cell contact in epithelial tissues. Adducin
and spectrin are colocalized at cell contact sites
and may be arranged in a structure analogous to
the spectrin network of erythrocytes. The associa-
tion of adducin with cell-cell contact sites occurs
before assembly of other types of specialized cell
junctions such as desmosomes.
Our working hypothesis is that adducin pro-
motes assembly of a stable spectrin network at
sites of cell-cell contact. A further hypothesis is
that the spectrin network is an essential precon-
dition for assembly of specialized cell junctions.
Formation of appropriate cell-cell contacts and
cell junctions is an essential event in embryogen-
esis and is one of the processes that is disturbed in
diseases such as cancer. We are excited by the
possibility that adducin and spectrin may partici-
pate in such a fundamental activity of cells. We
have determined the complete protein sequence
of both adducin subunits and are in the process of
determining a physical model for organization of
domains and subunits in the adducin molecule.
Future experiments will evaluate the role of ad-
ducin and spectrin in formation of junctions be-
tween cells.
34
TNF and the Molecular Pathogenesis of Shock
Bruce A. Beutler, M.D. — Associate Investigator
Dr. Beutler is also Associate Professor of Internal Medicine at the University of Texas Southwestern Medical
Center at Dallas. After receiving his M.D. degree from the University of Chicago (Pritzker School of
Medicine), he served as an intern and resident at the Southwestern Medical Center. His postdoctoral
fellowship with Anthony Cerami was completed at the Rockefeller University, which he left as an assistant
professor to assume a position at the Southwestern Medical Center.
THIS laboratory studies basic mechanisms that
lead to septic shock, a serious condition aris-
ing as a result of many types of infection. We have
learned that the final common pathway leading to
shock involves the production of certain cyto-
kines, particularly tumor necrosis factor (TNF),
by host cells known as macrophages and by other
cell types as well.
Once TNF has been produced, it alters the me-
tabolism of cells throughout the body, triggering
a breakdown of protein and fat stores. If TNF is
chronically secreted at low levels, a state of wast-
ing called cachexia will develop. This condition
is seen in cancer and many other forms of chronic
illness. On the other hand, if massive quantities
of TNF are released over a short time, as in wide-
spread injury, the protein activates neutrophils
and endothelial cells in such a way that shock
occurs.
Because TNF is a critically important molecule
in various human disease processes, we have
sought to understand how its biosynthesis is con-
trolled. Probably the most potent stimulus for
TNF release is a molecule known as lipopolysac-
charide, or endotoxin. This molecule is pro-
duced by gram-negative bacteria, which have a
remarkable tendency to cause shock. In the
course of a gram-negative infection, endotoxin is
released into the bloodstream. It is harmless to
most cells but is a powerful activator of mono-
cytes and macrophages, triggering their release of
TNF with all of its attendant consequences.
By studying different portions of the TNF gene,
we have shown that endotoxin causes two sepa-
rate responses within the macrophage: 1) It
causes increased transcription of the TNF gene,
leading to a marked accumulation of TNF mRNA
within macrophages. 2) It causes far more effi-
cient translation of the mRNA — i.e., increases the
speed with which the mRNA is read to produce
TNF protein. Acting in concert, these two effects
are responsible for a 10,000-fold increase in the
rate of TNF biosynthesis and thus a massive net
effect.
For a number of technical reasons, it has been
very difficult to demonstrate the major sources of
TNF in living animals. It is not clear, for example,
whether it is made by normal tissues in healthy
animals or whether such "baseline" production
is important for maintenance of physiological or
immunological processes. Similarly, it is not
clear whether the TNF that arises in cancer is de-
rived from cells of the tumor or from host cells
that act in response to the tumor. TNF is believed
to be made in a variety of autoimmune or allergic
diseases, but again, the principal source of the
protein remains uncertain.
To address these questions, our laboratory has
produced transgenic mice that express a reporter
construct in which an easily measurable enzyme
(chloramphenicol acetyltransferase, or CAT) is
employed as a marker for TNF. In other words,
CAT synthesis also occurs in cells that produce
TNF. CAT remains confined, however, to the cell
of origin, whereas TNF is secreted and becomes
widely dispersed in the organism. Using these an-
imals, we have found that during normal develop-
ment TNF is made by cells of the thymus. Other
investigators have further reported that thymic
production of TNF is essential for normal devel-
opment. Although the protein does not appear to
be produced elsewhere in healthy animals, it is
readily induced by administration of lipopolysac-
charide or by various authentic infections.
A second tissue in which CAT synthesis marks
TNF production is the placenta, the organ that is
formed in pregnant mammals to nourish the de-
veloping fetus. The placenta contains cells of two
separate origins: part of the placenta, the deci-
dua, is derived from the mother. The remainder
of the placenta, the trophoblast, is derived from
the fertilized egg. Studies utilizing the reporter
construct have effectively proved that the TNF
gene is constitutively expressed by the tropho-
blast. TNF may also be constitutively expressed
by trophoblastic tumors. Its function within the
placenta remains to be established; however, it is
interesting to consider that the protein may play
an important role in normal pregnancy.
Our laboratory has also made progress in un-
derstanding the mechanism of action of drugs
that inhibit TNF biosynthesis and in devising mol-
ecules that block the action of TNF once it has
35
TNF and the Molecular Pathogenesis of Shock
been released. These studies might lead to better
therapies for shock and other disorders. Gluco-
corticoid hormones (e.g., prednisone, dexameth-
asone, and Cortisol) have long been used as anti-
inflammatory drugs. One of their principal
effects appears to be a blockade of TNF biosynthe-
sis.,, which depends upon inhibition of both TNF
gene transcription and mRNA translation. Other
drugs of a class known as phosphodiesterase in-
hibitors (e.g., theophylline, caffeine, and pen-
toxifylline) also block TNF biosynthesis, achiev-
ing their effect by preventing TNF mRNA
accumulation. They appear to function at a dif-
ferent locus than do glucocorticoids. We have
shown that the two classes of drugs when com-
bined exert a synergistic effect.
Recently the cell-surface receptor for TNF was
cloned in a number of laboratories. We have engi-
neered a recombinant molecule in which the TNF
receptor is attached to a portion of a normal anti-
body, yielding a new protein molecule in which
two TNF-binding sites are expressed. This biva-
lent TNF-binding protein strongly inhibits the bi-
ological effects of TNF, is highly stable in vivo,
and may be produced in large quantities by re-
combinant techniques. We anticipate that this
type of molecule will allow a thorough investiga-
tion of the many effects of TNF in health and dis-
ease and may also be useful as a therapeutic tool.
Of particular interest will be its use in studies of
TNF production and action in the thymus and in
the placenta.
36
Cytotoxic T Lymphocyte Recognition
Michael J. Sevan, Ph.D. — Investigator
Dr. Bevan is also Professor of Immunology at the University of Washington, Seattle. He received his Ph.D.
degree for work performed at the National Institute for Medical Research, Mill Hill, london. He did his
postdoctoral work at the Salk Institute in the laboratory of Melvin Cohn, after which he was a faculty
member in the Center for Cancer Research and the Department of Biology at the Massachusetts Institute
of Technology. He later conducted research in immunology at the Scripps Research Institute before moving
to the University of Washington. Dr. Bevan was elected Fellow of the Royal Society of london.
TWO functionally different types of T lympho-
cytes mature in the thymus and populate the
peripheral lymphoid organs. Helper T lympho-
cytes, the first type, respond to antigen by releas-
ing lymphokines, which activate macrophages or
augment the response of antibody-producing B
cells. Cytotoxic T lymphocytes, on the other
hand, when induced with antigen, specifically
lyse target cells expressing the antigen and re-
lease interferon-7. Their function is thought to be
crucial in the response to intracellular patho-
gens, such as viruses and bacteria, and they may
eliminate some tumor cells.
T lymphocytes can only recognize antigen pre-
sented in association with cell surface glycopro-
teins encoded by the major histocompatibility
complex (MHC) . Recent work has shown that cy-
totoxic T cells recognize short (usually nona-
meric) antigenic peptides presented in the
groove of class I MHC molecules.
Since class I molecules are expressed on vir-
tually all tissues, any cell type can be a target for
cytotoxic cells. The peptides presented by these
MHC molecules derive from degraded intracellu-
lar proteins, so all, or most, of a cell's normal
components can provide class I-binding pep-
tides. Normally, however, the T cells tolerate
these self-peptide/self-MHC complexes. Viral in-
fection, on the other hand, leads to the produc-
tion of new proteins in the cell. These may pro-
vide nonamer peptides that can combine with
class I and be presented on the surface. If cyto-
toxic T lymphocyte surveillance works well,
these foreign-peptide/self-MHC complexes will
be recognized and the virus-infected cell
destroyed.
The class I MHC proteins expressed on the cell
surface have three components: class I heavy
chain, (82microglobulin, and a tightly bound
peptide. This trimolecular complex is assembled
shortly after synthesis of the heavy and light
chains in a pre-Golgi compartment.
Many of the peptides that are bound to class I
actually derive from cytosolic and nuclear pro-
teins. It seems likely that cytosolic proteins are
degraded in the cytosol by complex structures
called proteosomes and that the peptide degrada-
tion products pass into the endoplasmic reticu-
lum (ER). Some of the components of proteo-
somes are encoded within the MHC in close
proximity to a pair of genes that belong in the
superfamily of ATP-dependent transport pro-
teins. These gene products may mediate the
translocation of peptide antigen from the cyto-
plasm into the ER lumen.
Antigen Presentation-Defective Variants
A number of mouse and human cell lines have
been described that synthesize normal class I
heavy and /32-microglobulin chains but neither as-
semble these chains for surface expression nor
present endogenous antigens to cytotoxic T cells.
We provided evidence that the defect in the
mouse cell line RMA-S maps to chromosome 17,
the chromosome encoding the MHC. When this
cell line was fused with a wild-type partner, ex-
pression of MHC antigens was restored. When the
wild-type chromosome 17 was selected against,
expression of the MHC antigens was lost. John
Monaco (Virginia Commonwealth University,
Richmond) had previously identified and cloned
Ham l and Ham-2, two ABC transporter genes
from the mouse MHC. In collaboration with his
group and with James Forman and Kirsten Fischer
Lindahl (HHMI, University of Texas Southwest-
ern Medical Center at Dallas), we were able to
show that transfection of the Ham-2 gene into
RMA-S restored surface expression of class I MHC
antigens, as well as the ability to present five en-
dogenous antigens on the surface for recognition
by cytotoxic T lymphocytes. Thus a defect in the
Ham-2 gene leads to loss of surface class I ex-
pression. In conjunction with previous experi-
ments with the human defective cell lines, this
suggests that both Ham- 1 and Ham-2 may be es-
sential for class 1 surface expression, and in fact
they may operate as heterodimers.
In other experiments we have shown that the
RMA-S cell line expresses a subset of endogenous
peptides on the surface with class 1. For example,
the octameric peptide derived from the vesicular
stomatitis virus nucleoprotein is presented to cy-
37
Cytotoxic T Lymphocyte Recognition
totoxic T lymphocytes by RMA-S. After viral infec-
tion or transfection of the nucleoprotein gene
alone, this epitope appears on the cell surface. By
moving this sequence of eight residues around,
we hope to be able to discover what makes these
escape peptides special — is it the peptides them-
selves or surrounding sequences? The leakiness
in this cell line also suggests that a Ham- 1 homo-
dimer may function inefficiently in this mutant
cell line.
Cytotoxic Responses
to Intracellular Bacteria
Listeria monocytogenes is a frequent food-
borne pathogen that causes severe disease in
immunocompromised individuals. It is a gram-
positive facultative anaerobe. Following uptake
by macrophages. Listeria escapes from the hos-
tile environment of the phagolysosome by secret-
ing a hemolysin, listeriolysin, which ruptures the
phagosome membrane. Once into the cytoplasm,
the bacterium replicates and can push itself into a
neighboring cell by polymerizing actin. Previous
workers had demonstrated a cytotoxic T cell re-
sponse in mice to Listeria infection. Our main
priority was to determine which of the approxi-
mately 4,000 proteins made by Listeria were
providing class I-associated peptide epitopes for
cytotoxic T lymphocytes.
We determined that the secreted listeriolysin
molecule itself provided a strong class I-re-
stricted epitope in the mice we studied. To pin-
point the presumed nonamer within the 600
amino acid residues of the listeriolysin molecule,
we used class I peptide-binding motifs that had
predicted tyrosine at position 2 and leucine at
position 9. The listeriolysin contained three non-
amers that fit this criteria. These and others were
made as synthetic peptides, one of which tar-
geted our cytotoxic T lymphocytes at picomolar
concentration. Thus we have used these motifs
for the first time and shown that they are valid in
pinpointing cytotoxic T lymphocyte epitopes.
In more recent studies we have been able to
show that cytotoxic T lymphocyte immunity to
this short 9-residue peptide presented by a class I
molecule is sufficient to confer adoptive immu-
nity to the whole bacterium. Lines of cytotoxic T
lymphocytes with specificity only to this epi-
tope, when transferred into syngeneic mice,
confer protection against lethal doses of Listeria
monocytogenes.
The listeriolysin epitope of Listeria that in-
duces conventional cytotoxic T cell immunity is
restricted by a classical class I molecule. "Classi-
cal" refers here to the highly polymorphic H-2K,
D, or L loci in mice and their equivalent HLA-A, B,
and C loci in humans. We have evidence that
other peptides derived from this bacterium are
presented by nonclassical, nonpolymorphic class
I molecules. We know that class I is involved in
presenting these peptides, because cell lines that
lack j82"r"icrogIobulin are unable to present the
peptides. Furthermore, we know that the class I
molecule maps just outside the MHC, as revealed
by experiments with congenic mice differing
only in this region.
But unlike the classical class I molecules, this
I'estricting element is rather nonpolymorphic:
i.e., cells from most mouse strains are able to
bind and present the peptide. We have yet to
identify the peptide derived from Listeria or the
class I molecule involved. It will be interesting in
the future to determine whether cytotoxic T lym-
phocytes directed to these peptides, seen in the
context of nonpolymorphic MHC molecules,
confer protective immunity.
38
Vascular Endothelium in Inflammation
and Metastasis
Michael P. Bevilacqua, M.D., Ph.D. — Associate Investigator
Dr. Bevilacqua is also Associate Professor of Pathology and a member of the division of Cellular and
Molecular Medicine at the University of California, San Diego. He received his undergraduate degree in
biology from the University of Pennsylvania and his M.D. and Ph.D. degrees in pathology from the State
University of New York at Brooklyn. As a postdoctoral research fellow with Michael Gimbrone, Jr., and
Ramzi Cotran at Brigham and Women 's Hospital and Harvard Medical School, he initiated work on
endothelial leukocyte adhesion molecules. He continued these studies as a faculty member at Harvard
Medical School before assuming his present position. Dr. Bevilacqua is a Pew Scholar in the Program in
the Biomedical Sciences.
VASCULAR endothelial cells form a cobble-
stone-like lining of blood vessels throughout
the body. Despite accurate predictions by certain
investigators in the 18th century, the highly re-
sponsive and changeable nature of vascular endo-
thelium was not recognized until the early
1980s. Endothelium was widely regarded as a
"nonstick" surface that prevented blood-clotting
and resisted the adhesion of circulating leuko-
cytes. It is now apparent that the simple struc-
tural features of vascular endothelium belie a
complex functional nature.
Our research has focused on the mechanisms
by which the vascular endothelium can regulate
inflammation, immunity, thrombosis, and tumor
metastasis. It was demonstrated that soluble pro-
tein mediators known as cytokines could act di-
rectly on endothelial cells to increase the expres-
sion of prothrombotic activities and promote the
adhesion of blood leukocytes. Investigation into
the mechanisms of leukocyte-endothelial adhe-
sion led to the identification and characterization
of two endothelial cell-surface glycoproteins that
were designated endothelial leukocyte adhesion
molecule- 1 (ELAM-1) and inducible cell adhe-
sion molecule- 110 (INCAM-110). This discus-
sion focuses on endothelial adhesion molecules
and their potential roles in human disease
processes.
Endothelial Adhesion Molecules
in Inflammation
The study of inflammation has one of the
richest histories in biomedical research. Particu-
larly noteworthy is the work of Julius Conheim
(1839-1884), which provided one of the first
(and best) microscopic descriptions of the pro-
cess. He carefully observed blood vessels after a
local injury in transparent membranes such as the
tongue of a frog and described the vasodilation,
edema, and leukocyte emigration with great accu-
racy. He also suggested that the blood vessel lin-
ing in the area of inflammation became sticky for
leukocytes. This concept was largely lost during
the next century, in which a theory of leukocyte
response to soluble mediators (chemotactic fac-
tors) dominated.
Until the early 1980s endothelial cells were
thought to move politely aside or be destroyed as
leukocytes exited blood vessels to fight bacterial
infections or contain foreign substances. By 1985
the view of vascular endothelium had begun to
change. Recombinant cytokines were found to
act directly on endothelial cells to increase dra-
matically the adhesion of blood leukocytes, and it
became apparent that the endothelium produces
a "sticky" surface. By 1987 the nature of that
surface was being elucidated, with the identifica-
tion of two cytokine-inducible endothelial adhe-
sion molecules designated ELAM-1 and ICAM-1
(intracellular adhesion molecule-1). Studies on
human tissues demonstrated that these mole-
cules are expressed on endothelium at sites of
inflammation.
A solid working hypothesis emerged: foreign
substances such as bacterial products introduced
into tissues would stimulate local cells such as
macrophages to synthesize and secrete cytokines.
The cytokines would then act on vascular endo-
thelium to promote the expression of adhesion
molecules that would bind passing leukocytes.
Leukocytes could then exit the blood vessel and
migrate toward the foreign substances following
chemotactic gradients. By 1990, through the ef-
forts of multiple laboratories, five endothelial
cell-surface molecules that participate in leuko-
cyte adhesion and inflammation had been identi-
fied, cloned, and characterized. These molecules
fall into two families based on their structures.
A newly described family, the vascular selec-
tins, contains three related molecules, two of
which can be found on endothelial cells: E-
selectin (a new name for ELAM-1 ) and P-selectin
(previously known as GMP-140 or PADGEM).
The second group of endothelial leukocyte adhe-
sion molecules are part of a much larger family
known as the immunoglobulin superfamily.
The pattern of expression of endothelial adhe-
sion molecules as well as the specificity of their
binding interactions with leukocytes contributes
39
Vascular Endothelium in Inflammation and Metastasis
importantly to the control of inflammatory pro-
cesses. Acute inflammation is of relatively short
duration, lasting as little as a few minutes or as
long as 1 - 2 days. It is characterized by the exuda-
tion of plasma proteins and the emigration of leu-
kocytes, especially neutrophils. Chronic inflam-
mation is of longer duration and is associated
with the extravasation of blood lymphocytes and
macrophages and with the morphological changes
in blood vessels and connective tissues.
In most cases the pattern of E-selectin expres-
sion correlates well with acute inflammation.
This observation is consistent with the demonstra-
tion that E-selectin supports the adhesion of es-
sentially all neutrophils, but only a small portion
of lymphocytes. In contrast, INCAM-1 10 (known
also as VCAM-1) supports the adhesion of lym-
phocytes but not neutrophils, and its pattern of
expression is consistent with its primary role in
chronic inflammatory processes.
The inflammatory process is essential for host
defense. In certain settings, however, it can also
contribute to debilitating and life-threatening
diseases. Examples of human diseases with signifi-
cant inflammatory/immunological components
are adult respiratory distress syndrome (ARDS),
myocardial reperfusion injury, rheumatoid ar-
thritis, and various autoimmune diseases. Un-
derstanding the mechanisms of endothelial-
leukocyte adhesion may allow us to enhance our
defense mechanisms, or dampen their responses
when appropriate.
Vascular Selectins and Their
Carbohydrate Ligands
Each of the three selectins is a cell-surface gly-
coprotein with a mosaic structure. The portion of
the molecule that is external to the cell, and
hence reaches out into the bloodstream, contains
a lectin domain. Lectins are protein molecules
that bind to carbohydrates (sugars) . Our labora-
tory is utilizing the recombinant E- and P-selectin
proteins and soluble synthetic carbohydrates to
determine the structures involved in recognition.
Early studies had suggested that E-selectin me-
diates the adhesion of blood leukocytes through
the binding of a carbohydrate known as sialyl-
Lewis X. This structure contains a terminal
sialic acid as well as a fucose group bound to
a carbohydrate backbone of galactose and A'^
acetylglucosamine. Surprisingly, we demon-
strated that a soluble form of sialyl-Lewis X was
only a modest inhibitor of E-selectin-dependent
adhesion, whereas a related carbohydrate called
sialyl-Lewis A, difi'ering only in the linkage of the
backbone sugars, was much more effective. Sialyl-
Lewis A is not found on blood neutrophils but is
expressed by a variety of cancers, including those
of the colon. The relative blocking activities of
sialyl-Lewis X and sialyl-Lewis A were reversed
for P-selectin.
In addition, a third carbohydrate compound
designated CD65, which is found on leukocytes
and some cancer cells, was shown to be an efl'ec-
tive blocker of adhesion to both E- and P-selectin.
Subsequent studies using modified carbohydrates
have demonstrated that the presence and precise
arrangement of sialic acid and fucose are impor-
tant in the recognition by the two selectins. Fur-
thermore, minor modifications in the carbohy-
drate structures can result in greatly increased
activity in blocking leukocyte adhesion.
A Possible Role in Cancer Metastasis
If cancers grew and invaded tissues locally but
failed to metastasize, most of them could be
cured by surgery. Indeed, the lethal effects of a
cancer are generally ascribed to its metastatic
abilities. Hematogenous metastasis involves re-
lease of cells from a primary tumor into blood
vessels and transport to new tissues. Substantial
evidence indicates, however, that the blood is
hostile to cancer cells and generally destroys
them. Thus most cancer cells must exit the
bloodstream in order to survive. Our laboratory
has focused on the mechanisms by which tumor
cells bind the vessel wall and extravasate. We
now appreciate that many tumor cells can inter-
act with the vascular endothelium through the
same molecules blood leukocytes use.
As noted above, colon cancers can express car-
bohydrate ligands for E-selectin. In a parallel line
of investigation, we have demonstrated that hu-
man melanomas can express the integrin a4l3l,
which is a receptor for the endothelial cell-
surface molecule INCAM-1 10 or VCAM-1. These
observations suggest the possibility of novel ther-
apeutic approaches to block the metastatic
spread of cancer.
In summary, vascular endothelium can exist in
different functional states. These states are deter-
mined by stimuli such as cytokines and bacterial
products. Activated endothelial cells express ad-
hesion molecules for circulating leukocytes and
can thereby regulate inflammatory processes. In
addition, it appears that cancer cells can utilize
these same adhesion molecules during the pro-
cess of hematogenous metastasis.
40
Structural Studies of Molecules Involved
in the Immune Recognition of Infected Cells
Pamela J. Bjorkman, Ph.D. — Assistant Investigator
Dr. Bjorkman is also Assistant Professor in the Division of Biology at the California Institute of
Technology and Adjunct Professor of Biochemistry at the University of Southern California School of
Medicine, Los Angeles. She received a B.A. degree in chemistry from the University of Oregon and a Ph.D.
degree in biochemistry and molecular biology from Harvard University, where her thesis advisor was Don
Wiley. She completed a low-resolution structure of a human histocompatibility molecule for her thesis
and then stayed on to finish the work. She continued her postdoctoral training at Stanford University in
the laboratory of Mark Davis before joining the staff at Caltech. Dr. Bjorkman received the William B.
Coley Award for Distinguished Research in Fundamental Immunology from the Cancer Research Institute.
THE technique of x-ray crystallography allows
visualization of the three-dimensional struc-
tures of proteins in atomic detail. In other words,
we get a picture of the protein that shows the
location of all the atoms and how they interact.
The shape of a protein and the location of individ-
ual atoms with respect to one another are impor-
tant for determining how the protein functions.
With such knowledge it is often possible to de-
sign compounds that modify the protein for medi-
cal intervention.
The proteins that we are studying structurally
are those that mediate the immune response
against viruses and other pathogens. The immune
system has evolved so that highly specific mole-
cules on the surfaces of lymphocytes can recog-
nize a virally infected cell. In the infected cell,
pieces of viral proteins are fragmented and bound
to a cellular protein called a histocompatibility
molecule. If the complex formed between the
histocompatibility molecule and the viral frag-
ment is recognized by a protein on the lympho-
cyte, the infected cell is destroyed. The lympho-
cytes that bear the recognizing proteins are T
cells, and the proteins on their surface are T cell
receptors.
The three-dimensional structure of a histocom-
patibility protein gives us a picture of how and
where viral molecules bind and how T cell recep-
tors might bind to the complex formed between
the viral and histocompatibility molecules. My
laboratory now seeks to determine a three-
dimensional structure for a T cell receptor, in
order to understand the atomic details of its inter-
action with the surface of an infected cell. (This
work is being done in collaboration with Mark
Davis, HHMI, Stanford University.) We have crys-
tallized a T cell receptor and now seek to make
crystals of a complex between a T cell receptor
and histocompatibility molecule. An understand-
ing of how T cell receptors interact with histo-
compatibility molecules complexed to viral frag-
ments should increase our understanding of how
the immune system distinguishes normal, healthy
cells from virally infected cells that need to be
destroyed.
Our laboratory is also using protein expression
systems in mammalian and bacterial cells to pro-
duce the large quantities of proteins necessary for
crystallization and structural studies. Using mo-
lecular biological techniques, it is possible to
transfect a protein-encoding gene into a cell in
which it is not normally found, thus persuading
that cell to manufacture that protein. Many pro-
teins normally occur in small quantities; the use
of such protein expression systems allows the
isolation of much more of the protein than would
be otherwise possible. We have expressed histo-
compatibility proteins in bacteria and formed
complexes between these proteins and antigenic
peptides.
We have also used a similar system to make an
Fc receptor, a protein that binds to the Fc portion
of antibody molecules. This particular Fc recep-
tor is found in the intestine of newborn mammals
and binds immunoglobulin found in mother's
milk, thus transferring partial immunity from
mother to infant. The amino acid sequence and
structural organization of this molecule show sim-
ilarities to histocompatibility molecules. We
have made a soluble (non-membrane-bound)
version of this molecule, which is still capable of
binding to Fc molecules. We have recently puri-
fied and crystallized this Fc receptor and initiated
a three-dimensional structure determination. We
have also crystallized a complex between the re-
ceptor and the Fc portion of an antibody; this will
ultimately allow us to obtain a picture of how the
two proteins interact. A comparison of the Fc re-
ceptor structure and its interaction with antibod-
ies should increase our understanding of the
structurally related histocompatibility proteins
and perhaps reveal reasons for the evolution of
this type of structure in the immune system. We
also have crystals of a structurally unrelated Fc
receptor and are interested in comparing how the
two types of receptors bind to their common
ligand.
41
Mechanisms of Insulin Action
Perry J. Blackshear, M.D., D.Phil. — Investigator
Dr. Blackshear is also Professor of Medicine and Assistant Professor of Biochemistry at Duke University
Medical Center. He received his D.Phil, degree in biochemistry from Trinity College, Oxford University,
and his M.D. degree from Harvard Medical School. Before moving to Duke University, he was Assistant
Professor of Medicine at Harvard Medical School. Dr. Blackshear has received the Young Investigator
Award for Clinical Research from the American Federation for Clinical Research.
OUR laboratory is mainly interested in the mo-
lecular mechanisms of action of insulin and
polypeptide growth factors. The studies involv-
ing insulin action are particularly relevant to
common clinical disorders of insulin resistance,
such as type II (adult-onset) diabetes and obesity,
in which the locus of the resistance is thought to
be at a "postreceptor" step within the cells of
muscle, liver, and adipose tissue. Our work is
aimed at understanding the biochemical steps in-
volved in insulin action in these tissues, with the
ultimate hope of identifying the abnormal steps
in these insulin-resistant states and possibly using
this knowledge to develop novel drugs aimed at
correcting the abnormalities.
Another major area of interest is the molecular
steps involved in mediating the effects of a wide
variety of hormones, neurotransmitters, and
drugs on their target cells. The common denomi-
nator of these agents is that the mechanism of
action involves the stimulated breakdown of cer-
tain membrane lipid compounds, leading to the
generation of intracellular lipid mediators known
as diacylglycerols. These in turn can activate an
important cellular enzyme, protein kinase C (PKC).
Over the past several years, a large amount of
information has accumulated about the way in
which hormones stimulate the breakdown of
these membrane lipids and about the molecular
biology and biochemistry of the PKC family of
enzymes. However, almost nothing is yet known
about the proteins that this kinase phosphorylates
in cells and tissues, or about the involvement of
these phosphorylated proteins in mediating a vari-
ety of cellular effects. To understand the role of
these PKC substrates in the cell is a major goal of
our laboratory. Within the past year we have
made several advances in each of these areas.
With regard to signal transduction, we are in-
vestigating the molecular nature of a family of
PKC substrates known by the acronym MARCKS
(myristoylated alanine-rich C-kinase substrates).
These are widely distributed cellular proteins
that are phosphorylated within seconds of PKC
activation in intact cells, and it is presumed that
this phosphorylation serves some function in me-
diating the effects of the activated kinase. Our
group has established that these proteins are ex-
cellent substrates for PKC, and that phosphoryla-
tion leads to changes in their properties that may
influence their behavior in the cell. For example,
phosphorylation of the protein disrupts its associ-
ation with model cellular membranes, perhaps
allowing for phosphorylation-dependent changes
in the protein's intracellular location.
Similarly, phosphorylation of the protein
disrupts its ability to form a tight complex with
calmodulin, the ubiquitous calcium-dependent
regulator of a wide variety of cellular enzymes.
We postulate that such disruption leads to an in-
crease in the cellular content of free calmodulin,
which might then lead to concomitant activation
of calmodulin-sensitive enzymes. In this way
phosphorylation of the MARCKS protein and its
relatives by PKC might lead to increases in the
activity of calcium/calmodulin-dependent pro-
cesses, a synergistic relationship that has long
been postulated but not proved.
The overall goal of these studies is to try to
determine the role that the MARCKS protein fam-
ily plays, if any, in mediating the actions of PKC
in various tissues. One important series of studies
now under way involves attempts to render cells
and intact animals deficient in these proteins, us-
ing the techniques of antibody microinjection,
antisense RNA expression, and ultimately gene
disruption. It is hoped that within the next year
or two, such deficient cells and animals will be
available to study the potential role of this family
of proteins in development, particularly of the
nervous system, in which the MARCKS proteins
and their relatives are highly expressed.
Considerable progress has also been made in
the studies of the molecular mechanisms of insu-
lin action. In one group of studies involving two
genes whose transcription is rapidly stimulated
by insulin, we have made progress in identifying
regions of the DNA that are implicated in the in-
sulin reaction. We have also identified proteins
that bind to these regions and have demonstrated
that they are modified in some way by rapid insu-
lin exposure of the cells. Current studies in the
laboratory are attempting to clone and sequence
43
Mechanisms of Insulin Action
these binding proteins and to determine the na-
ture of the insulin-stimulated modification.
Another group of studies involves the rapid
insulin-stimulated translation of messenger RNA,
using the enzyme ornithine decarboxylase in a
rnodel system. Within the past year, we have de-
termined that insulin can rapidly stimulate the
translation of certain types of mRNA that contain
extremely extensive and convoluted secondary
structures at their 5'-untranslated ends. We have
proposed that insulin does this by means of a spe-
cific ability to activate translation initiation fac-
tors that can unwind or "melt" the RNA second-
ary structure, and have determined that insulin
can rapidly stimulate the phosphorylation and
presumably the activation of these initiation fac-
tors in intact cells.
Current studies are focusing on the protein ki-
nases and phosphatases involved in this initiation
factor phosphorylation, as well as on other rap-
idly stimulated, insulin-activated protein kinases
that could play a role in the phosphorylation of
the transcription factors alluded to above. In all
of these cases, the ultimate goal is the elucidation
of the biochemical steps between insulin binding
to its receptor and its ultimate intracellular ef-
fect, such as gene transcription, with the hope
that steps in this pathway that are abnormal in
states of insulin resistance can be identified, lead-
ing to the development of specific therapies.
44
Intracellular Protein Traffic
and Nuclear Organelles
Gunter Blobel, M.D., Ph.D. — Investigator
Dr. Blobel is also Professor of Cell Biology at the Rockefeller University. He received bis M.D. degree from
the University of Tiibingen and his Ph.D. degree with Van Potter in oncology from the McArdle Laboratory
at the University of Wisconsin-Madison. Thereafter, he did postdoctoral work with George Palade at
Rockefeller. Dr. Blobel is a member of the National Academy of Sciences and of several other distinguished
societies. He has received many honors, including the Gairdner Foundation Award.
NUMEROUS structurally and functionally di-
verse proteins can be translocated across a
few distinct cellular membranes. It is now estab-
lished that targeting to these membranes and
translocation across them is specified by a mem-
brane-specific "signal" sequence that is a tran-
sient (or permanent) part of the protein to be
translocated. The primary structure of numerous
representatives for such a sequence has been es-
tablished in our laboratory and by others. Present
work focuses on the identification and character-
ization of machinery involved in the recognition
of a signal sequence, in its targeting to the proper
membrane, and in protein translocation. Collec-
tively, these components comprise what has been
termed a protein translocon.
Our laboratory is working on translocons for
four distinct cellular entities: 1 ) the endoplasmic
reticulum (ER) of animal (and yeast) cells,
which is able to translocate proteins from the cy-
tosol to the ER lumen; 2) bacterial plasma mem-
branes (gram-negative bacteria), able to translo-
cate proteins from the cytoplasm to the
periplasmic space; 3) yeast mitochondria, able to
translocate protein from the cytoplasm to the mi-
tochondrial interior ("matrix") across outer and
inner mitochondrial membranes; and 4) plant
cell (pea and spinach) chloroplasts, able to trans-
locate proteins from the cytoplasm to the chloro-
plast interior ("stroma") across outer and inner
chloroplast membranes.
From studies on these four translocons so far,
but especially from studies on the ER translocon,
it is likely that a translocon is composed of at least
four entities: 1) a soluble signal-recognition
factor (SRF), 2) a homing receptor, 3) a signal
sequence-gated protein-conducting channel,
and 4) a signal-removing peptidase. The SRF has
two functions: recognition of the signal sequence
and targeting to the homing receptor, which is
restricted in' its localization to a translocon-
specific membrane. Interaction of the signal se-
quence-SRF complex with the homing receptor
leads to dissociation of the signal sequence and
its presentation to a signal sequence receptor,
which might be a subunit of the protein-
conducting channel. This channel would close
immediately following completion of transloca-
tion, only to open again after presentation of an-
other signal sequence. The signal peptidase
would, in most cases, remove the signal sequence
either during or shortly after translocation.
In the case of the ER, the SRF was isolated and
shown to be a ribonucleoprotein particle. Re-
ferred to as a signal-recognition particle (SRP),
this consists of one 7S RNA molecule and six dif-
ferent proteins. Likewise, a homing receptor (re-
ferred to as the SRP receptor) and a signal -remov-
ing peptidase, a complex of five proteins, were
isolated. More recently we were able to demon-
strate the existence of a protein-conducting
channel, using electrophysiological methods.
The protein constituents of this channel remain
to be identified. We have been able to solubilize
the ER membranes by detergent and to reconsti-
tute translocation-competent vesicles. Using this
method, it should be possible to identify the
channel proteins.
We recently identified components of other
translocons. An SRF was isolated for signal se-
quence targeted to the bacterial plasma mem-
brane. Moreover, we identified signal sequence-
binding subunits for the mitochondrial and
chloroplast translocons. These integral mem-
brane proteins are located in contact sites be-
tween outer and inner organelle membranes and
are candidates for subunits of a protein-conduct-
ing channel in the outer organelle membrane
linked to a protein-conducting channel in the in-
ner membrane.
Our other major research effort focuses on the
organelles associated with the cell's nuclear en-
velope membranes. These organelles are thought
to organize the large amount of information in
the linear structure of the DNA into numerous
structurally and functionally distinct three-
dimensional superstructures, allowing only a lim-
ited amount of that information to be expressed.
Characterization of these organelles should ad-
vance understanding of such fundamental pro-
cesses as differential gene expression, cell differ-
entiation, and development.
Our efforts focus on the structural and func-
tional characterization of two morphologically
45
Intracellular Protein Traffic and Nuclear Organelles
distinct structures. One, the nuclear pore com-
plex (NPC), is located in the nuclear envelope.
We speculate that this organelle is involved in
"gene gating"; i.e., each pore complex is at-
tached to a set of actively transcribed genes. We
have identified and isolated several proteins of
the NPC of mammalian cells. So far we have es-
tablished the primary structure of 3 of the more
than 100 proteins of this large organelle. Re-
cently we succeeded in isolating NPCs from
yeast. These genetically amenable cells should fa-
cilitate the structural and functional analysis of
the many NPC proteins. We have also established
the primary structure of two membrane proteins
that are specifically located in the pore mem-
brane domain of the nuclear envelope mem-
brane. We are using cell-free systems for protein
uptake into the nucleus to isolate the hypotheti-
cal SRF, the homing receptor, and the so-called
transporter of the pore complex.
The other morphologically distinct structure
associated with the nuclear envelope that we are
studying is the nuclear lamina, a fibrous mesh-
work associated with the nuclear side of the inner
nuclear envelope membrane. The lamina consists
of three proteins, which we have termed lamins
A, B, and C. We have speculated that the nuclear
lamina is involved in the three-dimensional orga-
nization of nontranscribed chromatin.
By molecular cloning and cDNA sequencing of
the three lamins, we and others showed recently
that these proteins are members of the interme-
diate filament protein family. We demonstrated
that lamin B binds to the carboxyl-terminal por-
tion of cytoplasmic intermediate filament pro-
teins and that ankyrin, a protein associated
with the plasma membrane, binds to the amino-
terminal portion of cytoplasmic intermediate fila-
ment proteins. Recently we have investigated the
interaction of such a protein with nuclear lamin B
in more detail and have localized the interacting
regions to the near carboxyl-terminal segment of
these proteins. Interestingly, a synthetic peptide,
representing this lamin B-binding site of the cy-
toplasmic intermediate filament protein, when
microinjected into cells, led to the detachment of
intermediate filaments from the nucleus.
These data indicate a direct connection be-
tween the plasma membrane skeleton (ankyrin) ,
the cytoskeleton (intermediate filaments), and
the peripheral nucleoskeleton (lamina) through
the NPC. Moreover, we recently identified a la-
min B receptor in the inner nuclear membrane.
Its primary structure has been determined by mo-
lecular cloning and cDNA sequencing. This re-
ceptor has a number of interesting sites, suggest-
ing that it may not only interact with lamin B but
also with DNA.
46
Immunity and Pathogenesis of Third World
Diseases: Leprosy and Tuberculosis
Barry R. Bloom, Ph.D. — Investigator
Dr. Bloom is also Weinstock Professor of Microbiology and Immunology at Albert Einstein College of
Medicine. He received his B.A. degree and an honorary Sc.D. degree from Amherst College and his Ph.D.
degree from the Rockefeller University. He is active as an advisor to the World Health Organization in the
areas of tropical diseases and vaccine development. Dr. Bloom also serves on the Board of Science
and Technology in Development of the U.S. National Research Council and the National Vaccine
Advisory Committee. He is a member of the National Academy of Sciences, the Institute of Medicine,
and the American Academy of Arts and Sciences.
THE commitment of our laboratory is to inves-
tigate basic scientific problems that have par-
ticular relevance for health in the Third World.
Three-quarters of the world's population lives in
the Third World, and one-fourth of that popula-
tion suffers from malnutrition and disease. The
premises of our research are that the advances in
molecular biology and immunology have a great
deal to offer for understanding infectious diseases
afflicting people in developing countries and, re-
ciprocally, that the study of some of those dis-
eases can provide insights into fundamental im-
munological and pathogenetic mechanisms of
relevance to people in the industrialized world.
The Importance of Mycobacterial Diseases
Tuberculosis and leprosy are both caused by
mycobacteria. Leprosy afflicts 4-6 million peo-
ple in the world and produces deformity in 30
percent if untreated. Throughout time and in all
cultures, leprosy has engendered a unique fear
and stigma. Although Mycobacterium /eprae was
the first major human bacterial pathogen de-
scribed, it remains one of the few that has never
been cultivated in the test tube. However, its an-
tigens can be produced and studied vicariously in
Escherichia coli by means of recombinant DNA
technology.
Tuberculosis is the major cause of death from a
single infectious disease in the world today. Each
year there are 8 million new cases of tuberculosis
and 3 million deaths, afflicting primarily the
most productive element of society — young
adults. Infection with HIV (human immunodefi-
ciency virus) causes a breakdown of resistance to
tuberculosis; this has produced a grave increase
in the disease, in both the developing and the
industrialized countries. In 1985, following a
32-year decline in the number of cases, the inci-
dence of tuberculosis in the Unites States began
to increase, reaching 25,701 cases reported in
1990.
Resistance to M. tuberculosis
The devastatingly heightened susceptibility of
HIV-infected individuals to a galloping course of
tuberculosis is compelling evidence that there
are powerful mechanisms of immunity in im-
munocompetent hosts. The cellular and molecu-
lar mechanisms mediating that immunity remain
enigmatic. We have, however, established that
the most common microbicidal product of acti-
vated macrophages, oxygen radicals, is ineffec-
tive at killing virulent human tubercle bacilli,
but reactive nitrogen intermediates (particularly
nitric oxide) can inhibit growth and kill M. tu-
berculosis in vitro.
Immunologic Unresponsiveness
and Leprosy
One fundamental issue in immunology is the
nature of immunological tolerance, i.e., the
mechanisms by which cells in the immune sys-
tem discriminate between foreign antigens and
self-antigens and prevent responses to self. A
breakdown of tolerance to self-antigens leads to
autoimmune diseases, such as rheumatoid arthri-
tis, juvenile diabetes, and perhaps multiple scle-
rosis. The principal mechanism for developing
tolerance is thought to be deletion of clones of
potentially autoreactive T cells in the thymus
during neonatal life. Clearly, however, not all
such clones can be deleted in the thymus; there
must be additional mechanisms by which self-
reactive cells can be rendered unresponsive after
birth.
Leprosy provides a unique model with which
to study immunoregulation and unresponsive-
ness. The disease comprises a spectrum of clini-
cal entities. In the tuberculoid form, strong cell-
mediated immunity kills the organism but
damages nerves in the process. In the leproma-
tous form, at the other end of the spectrum, pa-
tients are unable to respond immunologically to
M. leprae antigens. Because infection occurs
after birth, there is little evidence of clonal dele-
tion of T cells capable of reacting to this organ-
ism. Therefore understanding the mechanisms of
that unresponsiveness is relevant to preventing
47
Immunity and Pathogenesis of Third World Diseases: Leprosy and Tuberculosis
autoimmune disease and increasing transplant
survival.
We have learned that the unresponsiveness in
leprosy is antigen-specific-, lepromatous leprosy
patients unable to respond to antigens of M. le-
prae usually respond to those of M. tuberculosis,
which is a closely related mycobacterium. How is
it possible for T cells to recognize antigens in the
tubercle bacillus and yet be unable to recognize
the same or closely related antigens associated
with the leprosy bacillus? We proposed that there
might be one or a few unique antigens associated
with M. leprae that induce active T cell suppres-
sion of potentially reactive T cell clones. Sup-
pressor cells in immunology have been a contro-
versial subject, but the idea that some T cells can
down-regulate immune responses, particularly
self-destructive ones, is compelling. About 85
percent of patients with lepromatous leprosy
have a subset of T cells capable of being triggered
specifically by leprosy antigens to suppress re-
sponses of immune T cells. Although they repre-
sent a minor subset of T cells in the blood, they
are the major lymphocyte subset in lepromatous
lesions. By establishing long-term T cell clones
directly from the lesions and blood, we found
that the suppressor cells have a pattern of antigen
recognition different from other cytotoxic or
lymphokine-producing T cells. They carry the
surface marker CDS and recognize foreign anti-
gens in association with the HLA-DQ region of the
human major histocompatibility complex (MHC)
class II. We speculate that presentation of anti-
gens by this MHC subset predisposes the immune
responses toward negative rather than positive
responses. Our studies suggest that functionally
distinct subsets of T cells are characterized by
distinct patterns of lymphokines produced upon
antigenic stimulation.
New Vaccines from Old — Recombinant BCG
as a Multivaccine Vehicle
Vaccines represent the most cost-effective med-
ical intervention. Yet three general problems
limit the use of many current vaccines: 1) they
require multiple booster shots to be effective; 2)
they cannot be given for 6-12 months after birth,
because of transferred maternal antibodies that
inactivate them; and 3) the cost. BCG (bacille
Calmette-Guerin) , the most widely used vaccine
in the world, is a live, attenuated bovine tuber-
cle bacillus given to protect children against
tuberculosis.
BCG has been given to more than 2.5 billion
people and has a very low incidence of serious
side effects. It is one of only two childhood vac-
cines that can be given at birth or any time there-
after. It is a single-shot vaccine that engenders
long-lasting cellular immunity and costs only
$0.10 a dose.
The unique attributes of BCG suggested to us
that, if it could be genetically engineered to ex-
press a variety of foreign antigens protective for
different pathogens, a single immunization might
be capable of engendering protective responses
to multiple pathogens simultaneously. One prob-
lem, however, was the paucity of molecular ge-
netic information about the Mycobacteria. In col-
laboration with William Jacobs (HHMI, Albert
Einstein College of Medicine), we developed ge-
netic systems for introducing and expressing for-
eign genes in mycobacteria, particularly BCG
strains. We developed a shuttle strategy in which
mycobacterial DNA could be genetically cloned
and manipulated in E. coli and then transferred
into mycobacteria. Our first approach was to use
mycobacteriophages (viruses that infect bacte-
ria) as vectors to target foreign genes to a specific
site in the bacterial chromosome. This enabled us
to introduce foreign DNA into BCG for the first
time. Recently we have developed shuttle plas-
mid vectors that are able to produce many copies
of foreign genes in BCG.
With several collaborators at Medlmmune,
Inc., and the University of Pittsburgh, we have
developed the first experimental recombinant
BCG vaccines. These express protective antigens
from M. leprae, schistosomes, malaria, measles
virus, leishmania, and HIV. Initial experiments in
mice indicate that three major types of protective
immune responses can be generated in vivo —
namely immunoglobulin antibodies, T cell lym-
phokines, and cytotoxic T lymphocytes. Continu-
ing efforts will be made to define antigens that
will engender, through recombinant BCG, pro-
tective immunity against a variety of viral, bacte-
rial, and parasitic pathogens.
48
Molecular Biology of the Extracellular Matrix
Jeffrey p. Bonadio, M.D. — Assistant Investigator
Dr. Bonadio is also Assistant Professor of Pathology and Member of the Program in Bioengineering at
the University of Michigan Medical School. He received his bachelor's degree in biology from Marquette
University and his M.D. degree from the Medical College of Wisconsin, Milwaukee. He studied anatomical
pathology with Bruce Beckwith and medical molecular genetics with Peter Byers at the University
of Washington, Seattle.
THE long-term goal of our research is to under-
stand how extracellular matrix proteins con-
tribute to skeletal structure and function. Quanti-
tative and qualitative changes in these proteins
occur during morphogenesis and as part of the
wound healing process. These observations sug-
gest that both the organization and protein com-
position of the matrix are precisely regulated. It
is clear that this regulation occurs in part at the
level of gene expression and in part at the level of
the assembly of proteins into a matrix-like
configuration.
I have chosen to focus for the most part on the
matrix molecule type I collagen. This collagen is
a polymer of two related proteins whose se-
quence has been determined. Moreover, the mul-
tidomain structure of the molecule and a general
outline of collagen biosynthesis are known, and
the molecule is recognized to be distributed
widely within tissues such as bone, tendon, liga-
ment, tooth, dermis, and sclera. Previous studies
have implied that type I collagen makes an im-
portant contribution to the structure, integrity,
and normal homeostasis of these tissues. Over the
past year we have continued our work to establish
model systems that would allow us to study this
contribution at the molecular level.
One system is designed to investigate the intra-
cellular assembly of the collagen molecule. In
general, this work involves site-specific mutagen-
esis and assays that quantify the effects of muta-
tion on the assembly process. These effects are
studied at two levels. First, we have established
conditions that allow synthetic peptides to fold
into a collagen-like triple helix. Peptide folding
is slow enough that the process can be character-
ized by methods such as circular dichroism. In
addition, the triple helix formed in vitro is stable
enough that its structure can be characterized by
nuclear magnetic resonance (NMR) techniques.
Therefore the effect of a given mutation can be
quantified by directly comparing the behavior of
a normal peptide with that of mutant peptide.
Second, cellular transfection methods have been
developed to express and assemble collagen mol-
ecules in vitro. Again, the effect of mutation on
the assembly process can be quantified by di-
rectly comparing the behavior of normal and mu-
tant molecules.
In our initial mutagenesis experiments, we
characterized a highly conserved region of the
triple-helical domain and demonstrated that it
made an important contribution to the assembly
of collagen molecules into a thermodynamically
stable conformation. We speculate that this re-
gion was conserved during evolution because it
plays an important role in collagen biosynthesis,
i.e., in folding the collagen molecule into its
correct conformation. In the future, we hope to
use this model system to define further the nor-
mal contribution made by other collagen do-
mains to the assembly process. In addition, we
are interested in characterizing those regions of
the molecule that mediate interactions between
collagen and other matrix molecules such as
fibronectin, heparin sulfate proteoglycan, and in-
tegrins. These interactions are important because
they represent a molecular basis for the assembly
of collagen within the matrix.
A second system is designed to investigate the
function of type I collagen at the level of connec-
tive tissue. Our initial set of experiments utilized
a transgenic mouse strain that expressed only half
the normal amount of type I collagen. We demon-
strated that the mutation adversely affected the
connective tissue of bone and skin dermis. In ad-
dition, the mutant mice were profoundly deaf.
We utilized biomechanical tests to quantify the
effect of the collagen deficiency at the tissue
level, and these studies demonstrated that the
major role of type I collagen is to provide con-
nective tissue with a high degree of resiliency.
More recently, we also demonstrated that the
skeleton of these transgenic mice is able to adapt
to the inherited collagen deficiency. This adapta-
tion involves a thickening of cortical bone and
results from the synthesis of new bone matrix.
Particularly intriguing was our observation that
the adaptation was associated with a significant
improvement in bone strength. This result is im-
portant because it suggests the basis for a strategy
to strengthen the fragile skeleton.
49
Functional Heterogeneity in CD4-bearing
T Lymphocytes
H. Kim Bottomly, Ph.D. — Associate Investigator
Dr. Bottomly is also Associate Professor in the Section of Immunobiology and the Department of Biology
at Yale University School of Medicine. She received her Ph.D. degree from the University of Washington,
Department of Biological Structure, where she studied with Roy Schwartz. Her postdoctoral training
was received in the field of immunology with Don Mosier at NIH.
LYMPHOCYTE interactions during an immune
response are necessary for the induction of
antigen-specific lymphocytes. The resulting ef-
fector phase of an immune response is described
as either humoral or cell mediated: each of these
phases combats different types of microorgan-
isms. Although these effector mechanisms are
well characterized, the precise mechanism by
which the response to a given antigen or in-
fectious agent is directed into the humoral or
cell-mediated mode is not known. What is clear,
however, is that both types of immunity depend
on the activation of CD4-bearing T lymphocytes,
which in turn induce other cell types to respond
to the foreign antigen. These responses include
activation of B cells to proliferate and to secrete
antibody, induction of delayed-type hypersensi-
tivity reactions, activation of CDS cytolytic T
cells, and activation of macrophages.
Two questions were then asked. 1 ) Could the
same CD4 T cell activate all these target cells,
therefore mediating both humoral and cell-
mediated responses? 2) Is the CD4 T cell popula-
tion functionally heterogeneous; i.e., do some
CD4-bearing T cells activate B cells and play a
primary role in the induction of humoral immu-
nity, and do some activate macrophages and CD8-
bearing T cells and play a primary role in cell-
mediated immunity?
For several years we have focused on the hetero-
geneity in CD4 T cell function and the activation
conditions that lead to it. These studies have suc-
cessfully shown that monoclonal CD4 T lympho-
cytes, obtained by T cell cloning and expansion
in tissue culture, belong to two distinct subsets of
T cell. One subset can help B cell antibody secre-
tion but cannot activate macrophages; the other
set activates macrophages but is a poor B cell acti-
vator. To reflect their main functions, these two
subsets are called helper T cells (Th2) and inflam-
matory T cells (Th 1 ) . The distinct functional abil-
ities of Th2 and Thl subsets are reflected in their
release, upon activation, of distinct panels of cy-
tokines. Several cytokines are produced selec-
tively by one or the other subset. In particular,
interIeukin-2, interferon-7, and lymphotoxin are
produced by the Thl but not Th2 cell subset.
Interleukin-4 is produced by the Th2 but not Th 1
cell subset.
Thus there is a correlation between cytokine
production and function. Interleukin-4 is a po-
tent B cell activator involved in B cell prolifera-
tion and secretion of immunoglobulins IgGl and
IgE. By contrast, interferon-7, lymphotoxin, and
interleukin-2 are associated with responses in-
volving the activation of macrophages, lysis of
target cells, and induction of cytolytic T cells,
which is consistent with the known function of
these cytokines. One can conclude from these
studies that T cells are committed to the release
of a distinct panel of lymphokines when acti-
vated, with the released effector molecules de-
termining their effector function.
Recent studies in this laboratory have focused
on determining whether functionally distinct
subsets of CD4 T cell exist in vivo and whether
the selective activation of one subset or the other
has the expected functional consequences. Nu-
merous clinical studies have suggested that
various immunization schemes induce primarily
a humoral or cell-mediated immune response. It
is critically important to determine if such major
differences in protective immunity reflect differ-
ences in the proportion of CD4 T cell subsets
activated. If this is true, one might propose that
the form of the antigen or the antigen presenta-
tion must somehow direct which CD4 T cell will
be preferentially activated. To test this possibil-
ity, several questions about normal CD4 T cells
have been asked.
First, is there a separation of resting CD4 T cells
into subsets? To answer this, analysis of the re-
sponses to antigens that give rise to primarily hu-
moral or cell-mediated immunity has been per-
formed. When humoral immunity dominates,
Th2-like CD4 T cells are activated. By contrast,
when cell-mediated immunity dominates, Thl-
like CD4 T cells are activated. Thus during the
course of an immune response, CD4 T cells may
become specialized in their functional capabili-
ties, and these T cells in vivo resemble, in their
activities, Thl and Th2 cloned lines.
51
Functional Heterogeneity in CD4-bearing T Lymphocytes
Second, when during development or activa-
tion do subsets of CD4 -bearing T cells arise? A
particularly interesting question is whether the
commitment of CD4 -bearing T cells to the Thl
and Th2 subsets is a consequence of antigen
priming. Analysis of memory and virgin CD4-
bearing T cells indicates that the functional spe-
cialization characteristic of Thl and Th2 subsets
resides mainly in the memory population. This
suggests that contact with foreign antigen initi-
ates a commitment of CD4-bearing T cells to a
particular effector response.
The form an immune response takes must be
appropriate to the microorganism that is causing
the disease. This is seen clearly in leprosy, where
one response produces abundant antibody and, at
the same time, overwhelming growth of the lep-
rosy bacterium in macrophages. This is probably
an example of selective activation of the helper
(Th2), rather than the inflammatory (Thl), sub-
set of CD4 T cells. What controls the form such a
response will take? Model systems to explore
these questions are needed, so that an inappropri-
ate immune response can be redirected to be-
come curative. Future research in this laboratory
will focus on this question of the control of CD4
T cell subset activation and regulation.
Studies analyzing those factors important in the
generation of memory and effector CDA^ T lym-
phocytes and those analyzing the selective activa-
tion of CD4 T cell subsets are supported by grants
from the National Institutes of Health.
52
Retroviral Replication
Patrick O. Broum, M.D., Ph.D. — Assistant Investigator
Dr. Brown is also Assistant Professor of Pediatrics and of Biochemistry at Stanford University School of
Medicine. He received his B.A. degree in chemistry from the University of Chicago. His graduate work with
Nicholas Cozzarelli at the University of Chicago was focused on the mechanisms of DNA topoisomerases.
He received his Ph.D. and M.D. degrees from the University of Chicago, completed a pediatrics residency
at Children's Memorial Hospital in Chicago, and then joined Michael Bishop's laboratory at the University
of California, San Francisco. There he began to investigate the mechanism of retroviral integration,
which has continued to be the major focus of his research.
RETROVIRUSES are an important cause of dis-
ease in most vertebrate species. In humans,
retroviral infections can lead to AIDS (acquired
immune deficiency syndrome), leukemia, lym-
phoma, and degenerative diseases of the central
nervous system. Millions of people are infected
with the human immunodeficiency virus, HIV,
and will likely succumb to AIDS unless an effec-
tive treatment is developed.
The retroviral genes are carried in the virus
particle as RNA molecules. When the virus infects
a cell, it transcribes these molecules, its RNA ge-
nome, into a double-stranded DNA molecule and
inserts this into a chromosome in the nucleus of
the host cell. Thus the viral genome, then called a
provirus, becomes an integral part of the cell's
genome. Integration of a provirus into its host
cell's DNA is essential for retroviral reproduc-
tion. This distinctive feature of the retroviral life
cycle accounts for many of the characteristics as-
sociated with retroviral infection, including in-
sertional mutagenesis, induction of tumors, and
the latent and persistent nature of many retroviral
infections. Moreover, the fact that retroviruses
are designed to introduce foreign genes into cel-
lular DNA makes them exceptionally useful as
tools for genetic engineering.
How does a retrovirus get its DNA into a cell's
nucleus and integrate it into the cell's DNA, and
how does the cell regulate these processes?
To investigate the molecular mechanism by
which a retrovirus inserts its DNA into that of the
infected cell, we have developed a variety of
methods for studying the retroviral integration re-
action in a test tube. We have used this approach
to define several discrete steps in the joining of
viral to cellular DNA and to determine the re-
quirements for the reaction. The enzymatic ma-
chinery that carries out integration can be iso-
lated from infected cells in a stable complex with
the unintegrated viral DNA molecule. We are
currently using electron microscopy as well as
biochemical methods to define the structure of
this viral replication intermediate.
To study the enzymology of integrase, the viral
protein that actually catalyzes integration, we
have constructed genetically engineered bacte-
rial strains that produce abundant quantities of
the integrase proteins from HIV and murine leu-
kemia virus and have developed simple purifica-
tions of these proteins. Using small synthetic DNA
molecules as model substrates, we can now
readily study their catalytic activities, which in-
clude the sequence-specific processing of the
ends of the viral DNA and the joining of these
ends to a target DNA molecule. We have made
progress in the past year toward understanding
the organization of integrase and how it binds the
viral and target DNA substrates. For example, we
have identified DNA substrates that function pref-
erentially either as donors or as targets in a DNA
joining assay. By studying the competition of
these DNAs for binding to the enzyme, we have
identified two distinct sites that bind viral and
target DNA, respectively. By analyzing the rever-
sal of the usual DNA joining reaction, we have
discovered that integrase has a previously unrec-
ognized DNA splicing activity. We are investigat-
ing the possibility that this new activity may play
a role in integration and in the high-frequency
recombination that occurs between viral ge-
nomes during replication.
The structure of integrase is clearly of central
importance to understanding integration. Be-
cause efforts to obtain conventional crystals of
integrase have been unrewarding, we are prepar-
ing to use electron diffraction methods to deter-
mine a structure from two-dimensional crystals.
By constructing specific mutants and defining
their biochemical defects, we have begun to
identify the functions of specific protein regions.
To extend this approach, we are developing a
new genetic system that we hope will enable us
to screen for functional defects among millions of
mutant integrases, based on their ability to carry
out recombination in bacteria. This system
should also facilitate our efforts to develop genet-
ically altered integrases with properties more fa-
vorable for therapeutic applications. For exam-
ple, we would like to develop an integrase that
53
Retroviral Replication
can selectively integrate the viral DNA into pre-
determined sites in the target DNA.
The ultimate goal of our work on integration is
to understand in molecular detail how the pro-
teins of the integration machinery recognize the
viral DNA, assemble into an active complex, rec-
ognize the target DNA, and finally catalyze the
DNA breakage and joining reactions that lead to
integration of the provirus. It is hoped that this
understanding will lead to the development of
new agents for inhibiting the replication of patho-
genic retroviruses and to improved systems for
the therapeutic introduction of genes into mam-
malian cells.
It has been recognized for many years that es-
tablishment of a retroviral provirus proceeds
much more readily in actively dividing cells than
in their resting counterparts. However, the basis
for this phenomenon remains obscure. To bring
this observation into clearer focus, we have in-
vestigated the dependence of specific steps in the
life cycle of the murine leukemia virus on the
host cell's stage in its own division cycle. Using
drugs or a mutation in the cell-cycle control gene
cdc2 to regulate the cell cycle of the host cells,
we have found that integration in vivo depends
on mitosis. Yet synthesis of viral DNA and accu-
mulation of integration-competent intermediates
occur normally in cells blocked from entering
mitosis. The intermediates accumulate in the cy-
toplasm and remain stable for hours, awaiting mi-
tosis. Current investigations are aimed at under-
standing why mitosis is required for integration.
Preliminary evidence suggests that disassembly
of the nuclear envelope at mitosis may provide a
route of entry for viral replication intermediates.
Understanding how cellular functions can deter-
mine the fate of an infecting retrovirus may lead
to new approaches to antiviral therapy and to im-
provements in the use of retroviruses as vectors
for gene therapy.
Our work on retroviral integration is supported
by a grant from the National Institutes of Health.
New Methods for Linkage Mapping
in Complex Genomes
A major impediment to defining and character-
izing the genes that influence complex human
traits has been the difficulty of collecting suitable
large families in which the trait segregates. Such
families are generally needed to find genes by
conventional linkage mapping. The same genes
could in principle be mapped more easily by an
alternative strategy that involves collecting and
analyzing pairs of relatives that share a trait of
interest. However, linkage mapping with small
sets of relatives generally requires analysis of a
large number of closely spaced and highly poly-
morphic genetic markers, which makes this strat-
egy impractical with current technology.
We are developing a new set of genetic tools
that will allow widespread application of these
highly efficient linkage mapping methods. Ex-
periments are in progress to test these new meth-
ods in model systems. In parallel with our efforts
to develop new biochemical methods, we are
working to find optimal statistical methods, using
high-resolution maps of genetic identity between
pairs of relatives. Our aim is to apply this technol-
ogy to map genes for complex human traits.
This work is supported by a grant from the Na-
tional Institutes of Health.
54
Regulation of Cellular Processes
by Protein-Tyrosine Phosphorylation
Joan S. Brugge, Ph.D. — Investigator
Dr. Brugge is also Professor of Microbiology at the University of Pennsylvania School of Medicine. She
received her B.A. degree in biology from Northwestern University and her Ph.D. degree in virology from
Baylor College of Medicine. Her postdoctoral research was done with Raymond Erikson at the University
of Colorado School of Medicine, Denver. Before moving to the University of Pennsylvania, Dr. Brugge
was a member of the Department of Microbiology at the State University of New York at Stony Brook.
THE modification of intracellular proteins by
the reversible addition of phosphate groups
(phosphorylation/dephosphorylation) is the most
common mechanism for regulating their activity.
Enzymes that specifically transfer phosphate to
tyrosine residues of proteins (protein-tyrosine ki-
nases) play critical roles in intracellular signal
transduction events — biochemical processes
that transform interactions at the surface of the
cell into an intracellular response. Our labora-
tory is interested in investigating the role of tyro-
sine phosphorylation in these processes. We have
chosen for these studies two model systems that
offer unique advantages: platelets, which are ide-
ally suited for studies of cell adhesion, secretion,
and cytoskeletal rearrangements, and PCI 2 cells,
which allow dissection of events that are in-
volved in the differentiation of neuronal cells
after treatment with nerve growth factors.
Tyrosine Phosphorylation in Platelets
Platelets are small, anucleate peripheral blood
cells that contain many intracellular vesicles
whose components are released upon activation
by cellular hormones. Platelet aggregation and
the released products from platelets are responsi-
ble for the formation of blood clots and wound
healing. We have found that hormones that acti-
vate platelet aggregation and secretion cause
rapid changes in the phosphorylation of multiple
proteins on tyrosine. In collaboration with San-
ford Shattil (University of Pennsylvania School of
Medicine) , we have shown that the phosphoryla-
tion of several of these proteins requires platelet
aggregation.
This process is mediated by interactions be-
tween the serum adhesion protein fibrinogen and
its receptor, GP Ilb-IIIa, on the platelet surface.
GP Ilb-IIIa is a member of the integrin family of
receptors that bind to extracellular matrix and
adhesion proteins. These receptors are believed
to be important in causing changes in cell behav-
ior that are mediated by cell adhesion. Our stud-
ies demonstrating that inhibition of GP Ilb-IIIa-
induced cell aggregation prevents tyrosine
phosphorylation of cellular proteins suggest that
this receptor may regulate the activation of tyro-
sine kinases, which play a role in aggregation-
dependent events.
We have recently found that a cytoplasmic tyro-
sine protein kinase, pi 25-FAK, is phosphorylated
on tyrosine and activated following platelet ag-
gregation. This kinase was first identified by Tom
Parsons (University of Virginia School of Medi-
cine) as a substrate phosphorylated in Rous sar-
coma virus-transformed cells. The enzyme is lo-
calized in focal adhesion plaques — sites where
integrin receptors couple with extracellular ma-
trix proteins and intracellular microfilaments. It
is not phosphorylated on tyrosine or activated in
platelets from patients with Glanzmann's throm-
bocytopenia, a disorder caused by the absence of
a functional GP Ilb-IIIa. These results suggest
that pi 25-FAK may be activated by events follow-
ing GP Ilb-IIIa-dependent platelet aggregation.
It is also phosphorylated on tyrosine in fibro-
blasts following activation of other integrin re-
ceptors and could play a critical role in integrin-
mediated activation of intracellular signal
transduction events mediated by cell-cell and
cell-matrix interactions.
GP Ilb-IIIa may also be important for the assem-
bly of intracellular cytoskeletal complexes that
follow thrombin-induced platelet aggregation.
Together with Joan Fox (Gladstone Foundation),
we have found that the assembly of these cytoskel-
etal proteins with enzymes involved in signal
transduction is dependent on GP Ilb-IIIa and
platelet aggregation. We have found that four
members of the Src family of protein-tyrosine ki-
nases— Src, Fyn, Lyn, and Yes — are part of these
assemblies, as are the cytoskeletal proteins actin,
vinculin, talin, and GPIV. In platelets from
Glanzmann's thrombocytopenia patients, these
proteins do not associate with the detergent-
insoluble cell fraction, indicating that GP Ilb-IIIa
is critical for these cytoskeletal rearrangements.
Thus platelet aggregation mediated by fibrinogen
binding to GP Ilb-IIIa appears to play an impor-
tant role in nucleating the assembly of cytoskele-
tal signaling complexes that may be important for
activation of intracellular processes, and these
55
Regulation of Cellular Processes by Protein- Tyrosine Phosphorylation
studies raise the possibility that integrin recep-
tors in other cell types play similar roles in cou-
pling with cytoplasmic tyrosine kinases.
We have also found that another platelet mem-
brane receptor, GPIV, or CD36, is tightly cou-
pled with three other protein-tyrosine kinases.
These kinases, Fyn, Lyn, and Yes, are members of
the Src family of protein-tyrosine kinases, which
have been shown to be linked with transmem-
brane receptors in other cell types. The ligand
that binds to GPIV/CD36 has not been identified;
however, the evidence that this receptor is asso-
ciated with protein-tyrosine kinases strongly im-
plicates it is an intracellular signal transducer.
The Fc receptors for the complement-binding
domain of antibody molecules are also found on
platelets. Activation of these receptors causes
platelet aggregation and secretion, and the in-
duction of tyrosine phosphorylation of the same
proteins that are phosphorylated in thrombin-
treated platelets. However, unlike other platelet
receptors, the FC-7RII receptor is itself phos-
phorylated on tyrosine. The site(s) of tyrosine
phosphorylation of this receptor lies within a
peptide motif that is shared with several other
lymphocyte receptors in T cells, B cells, natural
killer cells, mast cells, and basophils. This motif
appears to be involved in coupling these recep-
tors with protein-tyrosine kinases. Tyrosine phos-
phorylation of the conserved tyrosine residues in
this motif may be important for these coupling
interactions.
Tyrosine Phosphorylation in Neuronal
PCI 2 Cells
We are interested in defining the role of tyro-
sine phosphorylation in mediating neuronal dif-
ferentiation. PCI 2 cells, derived from a rat
pheochromocytoma, provide a useful model sys-
tem, since these cells differentiate into cells re-
sembling sympathetic neurons after treatment
with nerve growth factor (NGF) or fibroblast
growth factor (FGF) . Both of these receptors are
transmembrane proteins whose cytoplasmic do-
mains contain protein-tyrosine kinase activity.
Growth factor binding to these cells causes a
burst of tyrosine phosphorylation of multiple
cellular proteins.
Recently, in collaboration with Simon Ha-
legoua, we have found that tyrosine phosphoryla-
tion of two serine/threonine protein kinases,
p42MAPK p44MAPK jg dependent on the small
GTP-binding protein Ras. Expression of a mutant
of Ras that interferes with the activity of the en-
dogenous Ras protein blocks tyrosine phosphory-
lation of these proteins, and expression of a con-
stitutively activated form of Ras leads to their
tyrosine phosphorylation. These mutant forms of
Ras have no effect on the activity of the NGF re-
ceptor itself or on the phosphorylation of pro-
teins that are direct substrates of this kinase.
The p42 and p44 MAP kinases are referred to as
"switch kinases." They are activated by phos-
phorylation on tyrosine but phosphorylate other
proteins on serine and threonine. MAP kinases are
activated by a variety of mitogens and growth fac-
tors, and are believed to serve as critical compo-
nents of intracellular signaling pathways by inte-
grating signals from a diverse array of receptors.
Our studies suggest that tyrosine phosphoryla-
tion of MAP kinase is dependent on pathways reg-
ulated by the Ras GTP-binding protein, and that
this event can thus be distinguished from phos-
phorylation events that are directly mediated by
the NGF receptor itself. Since Ras activity is es-
sential for NGF- and FGF-induced neuronal dif-
ferentiation of PC 12 cells, these results raise the
possibility that MAP kinases are also essential
components of this process.
Dr. Brugge is now Scientific Director at Ariad
Pharmaceuticals, Cambridge, Massachusetts.
56
Computational Structural Biology
Axel T. Brunger, Ph.D. — Assistant Investigator
Dr. Brunger is also Associate Professor of Molecular Biophysics and Biochemistry at Yale University. He
was born in Leipzig, Germany. He received his diploma in physics at the University of Hamburg and his
Ph.D. degree from the Technical University of Munich. He held a NATO postdoctoral fellowship and
subsequently became a research associate with Martin Karplus in the Department of Chemistry at Harvard
University before joining the faculty at Yale. His research has focused on molecular dynamics studies
of protein structure and function and on methods in protein crystallography and nuclear magnetic
resonance spectroscopy.
OUR research lies at the interface between
theory and experiment in the area of struc-
tural biophysics. The research tools are simula-
tion methods of computational chemistry adapted
to the requirements of macromolecular systems.
Macromolecular simulations are an important ad-
dition to the arsenal of methods available to struc-
tural biologists working with x-ray crystallo-
graphic or nuclear magnetic resonance (NMR)
spectroscopic data. In one class of projects, we
are trying to understand the detailed microscopic
interactions that govern stability and recognition
in biological systems and to test the reliability of
the theoretical methods as tools for this purpose.
In another class of projects, we are directly com-
bining macromolecular simulation with experi-
mental data in order to make data analysis possi-
ble or more efficient.
Accuracy of Crystal and Solution
NMR Structures
As methods for determining macromolecular
three-dimensional structure continue to become
more powerful and are being applied to many
biologically interesting systems, concern has
been raised about the verification of final atomic
models. A common problem arises when models
are fitted against preliminary experimental data
of mediocre quality. The recent revision of a num-
ber of published structures, both x-ray and solu-
tion NMR, illustrates the need to develop im-
proved tools for checking the accuracy of the
final atomic models.
Structure determination of macromolecules by
crystallography involves fitting atomic models to
the observed diff'raction data. The traditional
measure of the quality of this fit, and presumably
the accuracy of the model, is the R value. Despite
stereochemical restraints, it is possible to overfit
or "misfit" the diffraction data: an incorrect
model can be refined to fairly good R values, as
several recent examples have shown. We recently
proposed a reliable and unbiased indicator of the
accuracy of such models.
In analogy to testing statistical models by cross-
validation, we defined a statistical quantity, Rjee.
that measures the agreement between observed
and computed structure factor amplitudes for a
"test" set of reflections that is omitted in the mod-
eling and refinement process. As examples show,
there is a high correlation between Rj^e and the
accuracy of the atomic model phases. This is use-
ful, since experimental phase information is
usually inaccurate, incomplete, or unavailable.
The enhanced sensitivity of 11^,^^ with respect to
model errors was illustrated for the crystal struc-
ture of the plant ribulose-1 ,5-bisphosphate car-
boxylase/oxygenase (RuBisCO) (David Eisen-
berg. University of California, Los Angeles). A
partially incorrect model for RuBisCO, which es-
sentially had the small subunit traced backward,
showed only a 4 percent conventional R value
difference from the correct model. On the other
hand, RX^g showed a 13 percent difference, sug-
gesting that the incorrect model had been overfit.
We concluded that Rj^g represents a reliable
and unbiased parameter by which to evaluate the
information content of a model produced by x-
ray crystallography. It is not restricted to high-
resolution diffraction data. The observation that
Rjee cai^ distinguish between a random distribu-
tion of scatterers and a distribution close to the
protein suggests applications to ab initio phas-
ing. Presently we shall apply this method to as-
sess models of thermal motion and disorder,
time-averaging, and bulk solvent in protein
crystals.
A similar approach might be useful for the
three-dimensional structure determination by so-
lution NMR. Using molecular dynamics refine-
ment, we have just succeeded in implementing
the free R approach. At present we are testing it
for a number of model systems. The question of
accuracy is an even more fundamental problem
for solution NMR structures because of the ad-
verse observable-to-parameter ratio.
This work is also supported by the National
Science Foundation.
Predictions of Helix-Helix Association
and Stability
Prediction of the three-dimensional structure
^ 7
Computational Structural Biology
of proteins based on their sequence remains im-
possible. This fundamental problem of structural
biology (the "folding problem") is still un-
solved, despite improvements in computational
techniques for macromolecular simulation and
CQjnputer hardware. Nevertheless, macromolecu-
lar simulation has been successful in predicting
localized conformations if sufficient experimen-
tal constraints or restraints (e.g., in the form of an
x-ray structure) are available. It is therefore con-
ceivable that other more global predictions are
possible if appropriate experimental information
is available. We have embarked on trying to pre-
dict the association and stability of helices that
form coiled coils. Conformational search strate-
gies are being employed with empirical energy
functions, using molecular dynamics and energy
minimization.
Presently we shall apply this approach to the
family of leucine zipper proteins, which are se-
quence-specific DNA-binding proteins that regu-
late gene expression in certain mammalian cells.
We have successfully predicted the structure of
the dimerization domain of GCN4, for which a
high-resolution x-ray has become available
(Thomas Alber, University of Utah).
Macromolecular Simulation of Free-Energy
Differences
We are involved in a number of projects
aimed at simulating free-energy differences be-
tween two states of a biological system, using
the so-called free-energy perturbation tech-
nique. The goal is to investigate microscopi-
cally the structure and stability of protein sec-
ondary structural elements and protein-peptide
complexes. Furthermore, we would like to eval-
uate the reliability of free-energy calculations
and molecular dynamics simulations as tools
for this purpose.
One project concerns the complexes of bovine
pancreatic ribonuclease S and a number of mu-
tants of the S-peptide for which x-ray crystal
structures, binding free energies, and enthalpies
have been obtained by Frederic Richards and Ju-
lian Sturtevant (Yale University) . Another project
involves the study of a number of site-directed
mutants of |8-turns in staphylococcal nuclease in a
joint project with Robert Fox (HHMI , Yale Univer-
sity) for which x-ray crystal structures and infor-
mation about the cis to trans equilibria of a pro-
line side chain in the turn have been obtained in
Dr. Fox's laboratory.
Illustration of hulk solvent regions in protein crystal struc-
tures. The regions are indicated by blue dots in this unit cell of a
crystal of penicillopepsin from Penicillium janthinellum, whose
structure was solved in Michael fames' laboratory (Alberta,
Canada) at 1.8- A resolution. The protein backbone atoms ap-
pear as yellow lines; ordered water molecules, as pink dots.
Bulk solvent typically constitutes 40-60 percent of the unit cell
contents, yet its physical characteristics and biological func-
tion are poorly understood.
Research and photograph by Axel Brunger, using a graphics
interface developed by Warren Delano.
58
Biophysical Studies of Eukaryotic Gene Regulation
and Molecular Recognition
Stephen K. Burley, M.D., D.Phil. — Assistant Investigator
Dr. Burley is also Assistant Professor and Co-Head of the Laboratory of Molecular Biophysics at the
Rockefeller University. He received a B.Sc. degree in physics from the University of Western Ontario, a
D.Phil, degree in molecular biophysics from Oxford University, and an M.D. degree from Harvard Medical
School in the Harvard MIT Joint Program in Health Sciences and Technology. While a medical student,
he carried out research in protein crystallography with Gregory Petsko. During his clinical training
at Brigham and Women 's Hospital, he also conducted postdoctoral research in protein crystallography
with William Lipscomb at Harvard University, where he solved the three-dimensional structure
of leucine aminopeptidase.
WE are interested in developing a detailed
understanding of the physical principles
that govern the general problem of molecular rec-
ognition in biological systems. Our approach is
to use x-ray crystallography and complementary
biophysical methods to determine and character-
ize the three-dimensional structure and function
of biological macromolecules and their com-
plexes with DNA, proteins, or smaller ligands.
These structures contain a wealth of atomic detail
that we can analyze with biochemical, molecular
genetic, and theoretical methods to provide a
functional description of the intra- and intermo-
lecular interactions responsible for stabilizing
macromolecular complexes.
In the long term, we hope that our biophysical
studies and analyses will allow us to exploit the
powerful formalism of physics to classify system-
atically the interactions between individual
atoms that are responsible for molecular recogni-
tion in biological systems. We believe that such a
quantitative understanding will ultimately per-
mit us to harness the machinery of molecular rec-
ognition and, thereby, make defined interven-
tions into important biochemical processes such
as disease states.
Eukaryotic Gene Regulation
We are examining the problem of eukaryotic
gene regulation, with the goal of improving our
understanding of the structural and physical
bases of transcriptional control of genes. Three
distinct classes of proteins are active in transcrip-
tion, and we are studying representative mem-
bers of each class. First, we are collaborating with
Robert Roeder (Rockefeller University) on x-ray
crystallographic and complementary biophysical
studies of transcription factor IID (TFIID) and
other components of the basic transcription ma-
chinery. These proteins form a stable multipro-
tein, or preinitiation, complex with DNA se-
quences found immediately upstream of the
transcription start site, where they mediate tran-
scription by RNA polymerase II. We have pu-
rified, characterized, and crystallized TFIID,
which begins preinitiation complex formation by
binding to the TATA consensus sequence. In ad-
dition, we have started work on TFIIB, the second
protein to be recruited to the preinitiation
complex.
Second, we are also collaborating with Dr.
Roeder on studies of upstream stimulatory factor
(USF), a member of the c-mjr-related family of
DNA-binding proteins that contains both a helix-
loop-helix motif and a leucine repeat. We have
used various biophysical methods to purify and
extensively characterize USF and its mechanisms
of action. Moreover, we were recently able to
grow small cocrystals of USF and DNA. During
transcription, TFIID, the other basic factors, and
USF bind to DNA in close proximity and interact
with one another to enhance both DNA binding
and transcription. After determining the three-
dimensional structures of each of these proteins
and their complexes with their respective pro-
moter DNA elements, we hope to determine the
structures of some biologically relevant multi-
protein-DNA complexes.
Third, we are collaborating with Eseng Lai (Me-
morial Sloan-Kettering Cancer Center) on struc-
tural studies of human hepatocyte nuclear factor
3. This transcriptionally active protein belongs to
a gene family in mammals that is homologous to
the Drosophila homeotic gene fork head. These
diverse proteins share a highly conserved DNA-
binding region and influence transcription by
binding to DNA elements, known as enhancer se-
quences, that are located far from the transcrip-
tion start site. During the past year, we have co-
crystallized a member of this family with DNA
and are proceeding with a three-dimensional
structure determination. Detailed structural and
biophysical studies of these three distinct classes
of participants in eukaryotic transcription should
provide insights into the precise role molecular
recognition plays in gene regulation.
Molecular Recognition
During the past year, we have also begun
to study other biological systems that function
59
Biophysical Studies of Eukaryotic Gene Regulation and Molecular Recognition
via molecular recognition, including steroid- minations of these proteins will be pursued once
binding proteins, polypeptide-binding proteins, they are fully characterized and suitable crystals
and enzymes. Three-dimensional structure deter- and cocrystals become available.
F-Actin
Proposed model of the dystrophin-glycoprotein complex. Dystrophin dimers link the cyto-
skeleton with the membrane glycoprotein complex by binding to actin at the amino ter-
minus (N) and to the 50-, 43-, and 35-kDa transmembrane proteins at the carboxyl ter-
minus (C).
From Ervasti, J.M., and Campbell, K.P. 1991. Ceil 66:1121-1131- Copyright© 1991 by
Cell Press.
60
Molecular Studies of Calcium Channels
and the Dystrophin-Glycoprotein Complex
Kevin p. Campbell, Ph.D. — Investigator
Dr. Campbell is also Professor of Physiology and Biophysics at the University of Iowa, Iowa City. He
received his B.S. degree in physics from Manhattan College, his master's degree from the University of
Rochester School of Medicine and Dentistry, and his Ph.D. degree from the Department of Radiation
Biology and Biophysics at the University of Rochester. He did postdoctoral studies in the laboratory of
David MacLennan at the Banting and Best Department of Medical Research, University of Toronto,
before coming to Iowa.
CALCIUM ion (Ca^^) functions as a ubiquitous
intracellular messenger regulating a wide va-
riety of cellular responses, including contrac-
tion, secretion, and cell proliferation. Two com-
mon pathways by which intracellular Ca^"^
transients can be triggered have been identified.
The first involves the influx of Ca^"^ into the cyto-
plasm from the extracellular medium through
specific plasma membrane channels, and the sec-
ond involves the release of Ca^"^ into the cyto-
plasm from intracellular stores through specific
release channels.
Over the past 10 years, we have investigated
the structure and function of the membrane
components involved in Ca^"*^ fluxes across mem-
branes. In particular, we have focused on identi-
fying, purifying, and characterizing the mem-
brane proteins that function as surface Ca^^
channels and intracellular Ca^"^ release channels
in excitable cells.
Neuronal Calcium Release Channels
A major area of research in my laboratory con-
cerns the structure and function of intracellular
Ca^^ release channels. In skeletal muscle, Ca^"^
release from the sarcoplasmic reticulum initi-
ates contraction. We previously purified a high-
affinity ryanodine receptor from skeletal muscle
sarcoplasmic reticulum and showed it to be iden-
tical to the sarcoplasmic reticulum Ca^^ release
channel.
In neurons, inositol 1 ,4,5-trisphosphate (IP3)
is an important second messenger involved in
Ca^"^ release. However, physiological and pharma-
cological evidence has indicated the presence of
non-IPj-gated Ca^^ pools in neurons. Recently
my laboratory also identified and purified a neu-
ronal ryanodine receptor. The purified receptor
was shown to be a homotetramer composed of
protein subunits of approximately 500 kDa. Re-
constitution of the purified ryanodine receptor
into lipid bilayers has demonstrated that it func-
tions as a caffeine- and ryanodine-sensitive Ca^^
release channel that is distinct from the brain IP3
receptor. Immunoblotting experiments indicate
that the brain receptor is more like the cardiac
receptor than the skeletal receptor. Thus we be-
lieve that the brain ryanodine receptor may oper-
ate as a Ca^^-induced release channel for intracel-
lular Ca^^ pools in neurons. In the upcoming year
further studies of this receptor should provide
insights into its role in neuronal Ca^^ homeostasis
and its distribution within the central nervous
system.
Voltage-gated Calcium Channels
A second major area of research concerns volt-
age-gated Ca^^ channels in excitable cells. We have
studied in skeletal muscle the dihydropyridine-
sensitive Ca^^ channel, which is essential to exci-
tation-contraction coupling. The skeletal muscle
dihydropyridine receptor has been purified and
consists of four subunits (aj, and 7).
In neurons, voltage-gated Ca^^ channels exist
as several types (L, N, T, and P) with different
kinetic and pharmacological properties. Dihy-
dropyridines bind specifically to L-type Ca^"*"
channels and alter their channel activity. For N-
type channels, which are likely responsible for
triggering neurotransmitter release at synapses,
oj-conotoxin is largely specific. We are now using
antibodies and cDNA probes to the various sub-
units of the dihydropyridine-sensitive channels
to study oj-conotoxin-sensitive Ca^^ channels.
We have demonstrated that the brain
conotoxin-sensitive Ca^^ channel contains a
component homologous to the 18-subunit of the
dihydropyridine-sensitive Ca'^'^ channel of skele-
tal muscle. We have also isolated a cDNA clone
from a brain cDNA library encoding a protein
with high homology to the /8-subunit of the skele-
tal muscle dihydropyridine-sensitive Ca^^ chan-
nel. This brain /3-subunit cDNA encodes numer-
ous consensus phosphorylation sites, suggesting
a role in Ca^^ channel regulation.
We are now purifying the w-conotoxin-sensi-
tive Ca^^ channel, using affinity chromatography,
in order to analyze its subunit composition and to
demonstrate that it is identical to the N-type Ca^^
channel. Experiments are also in progress to
coexpress the brain /3-subunit with the aj-subunit
61
Molecular Studies of Calcium Channels and the Dystrophin-Glycoprotein Complex
in order to study how the ;8-subunit regulates
Ca^""" channel activity.
Dystrophin-Glycoprotein Complex
A third major project in my laboratory is aimed
at-understanding the function of dystrophin in
normal muscle and determining how the absence
of dystrophin leads to Duchenne muscular dys-
trophy (DMD). Dystrophin is localized to the in-
ner surface of the sarcolemma in normal muscle
but is absent in skeletal muscle of DMD patients
and mdx mice. We previously showed that dys-
trophin is tightly linked to a large oligomeric
complex of sarcolemmal glycoproteins. We have
isolated a dystrophin-glycoprotein complex and
have shown that it consists of cytoskeletal, trans-
membrane, and extracellular components. These
data have allowed us to propose a model for the
organization of the dystrophin-glycoprotein
complex (see figure). The membrane organiza-
tion of the dystrophin-glycoprotein complex and
the high density of dystrophin in the sarcolemma
membrane suggest that this complex could have
an important structural role in skeletal muscle.
In the past year we have investigated the rela-
tive abundance of each of the components of the
dystrophin-glycoprotein complex in skeletal
muscle from normal and mdx mice. Our results
demonstrate that all of the dystrophin-associated
glycoproteins (DAGs) are significantly reduced
in mdx skeletal muscle and suggest that the loss
of DAGs is due to the absence of dystrophin and
not to secondary effects of muscle fiber degrada-
tion. Furthermore, we recently showed that the
absence of dystrophin in skeletal muscle from
DMD patients leads to a dramatic loss of all the
components of the dystrophin-glycoprotein com-
plex. Thus the abnormal expression of the DAGs
may play a crucial role in molecular pathogenesis
in DMD.
In order to identify the normal function of
the DAGs, we recently established, by cDNA
cloning, the primary sequence of two compo-
nents of the dystrophin-glycoprotein complex.
The 43- and 1 56-kDa DAGs are encoded by the
same mRNA, and post-translational modifica-
tion of a 97-kDa precursor protein results in
two mature proteins: the transmembrane 43-
kDa DAG and the extracellular 1 56-kDa DAG.
In addition, we have shown that the 1 56-kDa
DAG binds laminin, a well-characterized com-
ponent of the extracellular matrix. Our results
demonstrate that the 43/1 56-kDa DAG (named
dystroglycan) is a novel laminin-binding glyco-
protein and suggest that the function of the
dystrophin-glycoprotein complex is to provide
a linkage between the subsarcolemma cytoskel-
eton and the extracellular matrix. Our findings
strongly support the hypothesis that in DMD a
dramatic reduction in a 1 56-kDa DAG leads to
the loss of a linkage between the sarcolemma
and extracellular matrix. This may render mus-
cle fibers more susceptible to necrosis or may
disrupt the integrity of muscle.
Our goal for the next year is to clone the other
DAGs in order to express the entire complex in
nonmuscle cells to study its function. We also
plan to examine the possible involvement of the
DAGs in other muscular dystrophies.
This work was also supported by the Muscular
Dystrophy Association.
62
Gene Targeting
Mario R. Capecchi, Ph.D. — Investigator
Dr. Capecchi is also Professor of Human Genetics at the University of Utah School of Medicine and
Professor of Biology at the University of Utah. He received his B.S. degree in chemistry and physics from
Antioch College and his Ph.D. degree in biophysics from Harvard University, where he worked with James
Watson. Dr. Capecchi remained at Harvard as a Junior Fellow of the Society of Fellows and then joined
the Harvard faculty. Before moving to the University of Utah, he was Associate Professor of Biochemistry
at Harvard School of Medicine. Dr. Capecchi is a member of the National Academy of Sciences.
GENE targeting — homologous recombination
between DNA sequences in the chromo-
somes of mouse embyro-derived stem (ES) cells
and newly introduced, exogenous DNA se-
quences— has been used to create mice with null
(ablating) mutations in members of two families
of genes. The first set of genes are members of the
int proto-oncogene family (int l, int-2, etc.).
This set of genes is believed to be involved in
localized developmental decisions mediated
through cell-cell signaling. Their protein prod-
ucts resemble growth factors, and int-2 is a
member of the fibroblast growth factor family.
These genes were initially identified through
their pathological role in the genesis of mouse
mammary carcinomas. We are addressing the ques-
tion of their normal role during embryogenesis.
Targeted disruption of int-1 (wnt-l) resulted
in mice with a range of phenotypes from death at
birth to survival to adulthood. Those mice surviv-
ing exhibited severe ataxia (loss of balance and
coordinated movement). Those int- 1~ / int- 1~
mice that died at or near birth showed severe ab-
normalities in the formation of the entire cerebel-
lum and the major portion of the midbrain,
whereas the defect in the survivors was restricted
to the formation of the anterior region of the cere-
bellum. We have shown that a pre-existing mu-
tant mouse, identified by its ataxic behavior, con-
tains a frameshift mutation in the int-1 {wnt-1)
gene. This allele also appears to be a null muta-
tion and exhibits the same range of variation in
expressivity as our targeted null allele.
We have created mice with null mutations in
int-2. These mice also exhibit a variable pheno-
type from death at birth to survival to adulthood.
The survivors are even fertile. The int-2'' /int- 2~
mice have defects in the formation of the vesti-
bule and cochlea, resulting in loss of balance and
deafness. However, the degree of malformation
of these inner ear compartments varies.
The second set of genes we are analyzing con-
stitute part of the developmental program that
specifies positional information along the antero-
posterior axis of the early embryo. These genes,
collectively known as the Hox genes, code for
transcription factors in both the human and
mouse. The 38 Hox members are present on four
linkage groups of four separate chromosomes.
The four linkage groups are believed to have
arisen early in chordate evolution, as a result of
quadruplication of a single ancestral group com-
mon to both vertebrates and invertebrates. Ex-
pansion of this gene complex may have had a
critical role in the progression from invertebrates
to vertebrates, by supplying the necessary com-
plexity to this network of genes to accommodate
the development of our complex body plan.
In Drosophila these genes act as master
switches directing the course of morphogenic de-
velopment of each parasegment. A mutation in
one gene can result in dramatic homeotic trans-
formations of one body part into another, such as
the conversion of antennae to legs. Determining
the function of the corresponding genes in mam-
mals is just beginning.
We have initiated a systematic genetic analysis
of the Hox genes in mice. First, we are creating
mice with null mutations in each of these genes
to define their individual functions. From such an
analysis, patterns should emerge that define the
zones governed by these genes. Second, through
epistasis and molecular analysis of combinations
of hox mutations, we hope to define how this set
of genes functions as a network to specify posi-
tional information along the body axis of the em-
bryo. To date we have analyzed mice containing
targeted disruptions in the two closely linked
Hox genes, hox- 1.5 and hox- 1.6.
Disruption of hox- 1.5 or -7.6 resulted in mice
with complex but regionally restricted develop-
mental defects. Interestingly, the sets of defects
associated with disrupting these two genes are
distinct and non-overlapping. The hox-1.5~/
hox- 1.5'^ mice are athymic, aparathyroid, and
have reduced thyroid and submaxillary tissue.
They also exhibit a wide spectrum of throat abnor-
malities (including shortened necks, abnormal
larynx, truncated soft palate, and poor organiza-
tion of throat musculature) and defects of the
heart and major arteries. This collection of defi-
ciencies is remarkably similar to those afflicting
humans with the congenital DiGeorge syndrome.
On the other hand, hox- 1 .6~ /hox- 1 .6" mice
63
Gene Targeting
executing each of those pathways. The non-
overlap of phenotypes associated with disrupting
these two genes argues for their independent
character. For example, if a nonredundant com-
binatorial code of box genes is used to specify
each cell type in a given region, then the com-
bined presence of both of these gene products
does not specify any tissue.
Disruption of the hox-1.6 gene results in regionally restricted developmental defects, including
formation of the ear, cranial nerves, and ganglia, as well as the brain stem. This is illustrated by
these embryos of El 0.5 mice immunoreacted with a monoclonal antibody against the 155-kDa
neurofilament protein, a) cowfro/ hox-1 .6"^/hox-l .6^ embryo; b, c) two mutant hox-l .6~/hox-
1.6" embryos.
Reprinted by permission from Chisaka, O., Musci, T.S., and Capecchi, M.R. 1992. Nature
355:516-520. Copyright© 1992 Macmillan Magazines Limited.
exhibit profound defects in the formation of the
ear, hindbrain nuclei, and cranial nerves and gan-
glia. The glossopharyngeal and vagus nerves are
poorly developed, and the preganglionic connec-
tions with the brain stem are not formed.
In both hox-l. 5 and -7. 6 mutants, the affected
tissues are formed by very different embryonic
pathways, yet the two hox genes are involved in
64
Genetic Control of Pattern Formation
in Drosophila
Sean B. Carroll, Ph.D. — Assistant Investigator
Dr. Carroll is also Associate Professor of Molecular Biology, Genetics, and Medical Genetics at the
University of Wisconsin-Madison. He obtained his B.A. degree in biology from Washington University in
St. Louis and his Ph.D. degree in immunology from Tufts University School of Medicine in Boston. He
received postdoctoral training in developmental genetics working with Matthew Scott at the University
of Colorado. In addition to his central work on pattern formation in Drosophila, Dr. Carroll has
also conducted basic research on new types of snake antivenoms, which are now under evaluation
as potential pharmaceuticals. His honors include the NSF Presidential Young Investigator Award.
INSECTS are the dominant group of animals on
earth today. Nearly 1 million species have been
classified among an estimated 20 million species
extant. The innumerable sizes, shapes, and colors
of insects have evolved from a basic segmented
body plan consisting of three broad divisions —
head, thorax, and abdomen — with three pairs of
ambulatory legs and, in some orders, wings.
One insect, the fruit fly Drosophila melano-
gaster, has emerged as a key model for investiga-
tions into the genetics and molecular biology of
animal development. An enormous amount of in-
formation has been gained about the organization
of the Drosophila egg, the dynamic regulatory
mechanisms guiding embryonic development,
and the cellular and molecular processes in-
volved in tissue differentiation and organogene-
sis. By addressing the intricacies of Drosophila
development at the genetic, cellular, and molecu-
lar levels, one of the central puzzles of biology is
being deciphered: How do complex animals
form from a simple egg?
In addition, the stage is being set for compara-
tive studies that will integrate the growing knowl-
edge of Drosophila development with contempo-
rary views of animal evolution. Efforts in our
laboratory are aimed at both a detailed under-
standing of Drosophila embryology as a model
for animal development and comparative studies
of other insects (e.g., beetles and butterflies) as
models for evolution.
Pattern Formation in Drosophila Embryos
The genetic control of pattern formation can be
broken down conceptually into at least three
phases. The first consists of a molecular prepat-
tern, revealed as chemical changes that take place
in different regions of the animal and foreshadow
the cellular events to follow. For example, cer-
tain key regulatory proteins in Drosophila come
to be expressed in stripes of cells encircling the
embryo — stripes representing future segmental
divisions. The second phase involves the specifi-
cation of groups of precursor cells that populate
the different tissues of the animal. For example.
genes such as those of the achaete- scute complex
(AS-C) are activated in clusters of cells that give
rise to the central and peripheral nervous sys-
tems. Finally, in the third phase, these precursor
cells divide and differentiate, giving rise to the
full complement of specialized cells that make
up different tissues and organs and express dis-
tinct structural genes to carry out their individual
tasks.
From molecular prepattern to the formation of
stem cells to the differentiation of their progeny,
there is a flow of genetic information. The pre-
pattern specifies the spatial domains of genes that
are activated in stem cells, and these genes in turn
regulate cell type-specific gene expression. Our
laboratory is interested in the genetic basis of this
information flow and focuses on three aspects of
pattern formation.
First, we are studying the molecular regulation
and function of the pair-rule genes, the first genes
expressed in a segmentally repeating prepattern.
Second, we are studying how the early prepattern
genes regulate the expression of the stem cell-
specific proneural genes of the AS-C. Third, we
are trying to identify some of the molecular pre-
patterns that govern pattern formation in imagi-
nal discs — the distinct pouches of cells that are
set aside during embryogenesis, proliferate exten-
sively during larval development, and give rise to
adult structures such as the wing, leg, and eye.
In addition, we have recently initiated studies
of other insects (guided by our knowledge of Dro-
sophila genes) to compare the molecular aspects
of segmentation in species that exhibit different
styles of oogenesis and early development and to
explore the genetic basis of wing development
and evolution among the flying insects.
The Pair-Rule Segmentation Genes
During early Drosophila development, small
batteries of genes are expressed in rapid succes-
sion to establish the segmental prepattern. The
first genes expressed in a periodic pattern are the
pair-rule genes, deployed in one stripe every two
segments. The striped patterns of certain pair-
65
Genetic Control of Pattern Formation in Drosophila
rule genes are generated by the aperiodic pat-
terns of gap proteins, a handful of transcription
factors expressed in broad, partially overlapping
regions of the early embryo.
Three aspects of pair-rule gene regulation and
function are under study. First, we wish to deter-
mine the molecular mechanisms of gap gene reg-
ulation of the pair-rule hairy gene. Second, we
want to know how the pair-rule genes regulate
the spatial expression of proneural genes during
early formation of the nervous system. And third,
we are investigating whether the same sort of seg-
mentation genes function in insects with differ-
ent styles of early development (e.g., beetles and
butterflies) . To this end, we are now isolating gap
and pair-rule genes from other insects and arthro-
pods to study their embryonic function.
Proneural Genes and Early Neurogenesis
Once the embryo is subdivided by the segmen-
tation genes, the formation of different tissues be-
gins. One of the earliest events in embryogenesis
is the segregation of the central nervous system
precursor cells (neuroblasts) from the overlying
ectoderm. The proneural genes, which are
named for their role in promoting the neural over
the epidermal pathway in the insect embryonic
ectoderm, are expressed in segmentally repeat-
ing clusters of 4-6 cells. A single proneural-
expressing neuroblast will segregate from each
cluster.
We have recently determined that the estab-
lishment of proneural gene expression in these
clusters is almost entirely regulated by up to
eight pair-rule genes (all of which appear to be
nuclear regulatory proteins) and an unknown
number of genes acting along the dorsoventral
axis. This demonstrates that the pair-rule genes
are not simply involved in regulating the expres-
sion of other segmentation genes, but also direct
the expression of genes that govern the behavior
of small groups of cells. They serve to integrate
the global regulatory system governing segmenta-
tion with the local regulatory system that speci-
fies tissue architecture.
The pair-rule genes are not sufficient to specify
individual cell fates, only the position of pro-
neural cell clusters. Interactions between cells
within each cluster determine their neural or epi-
dermal fate. We have shown that the so-called
neurogenic genes — a handful of loci that encode
a variety of different proteins involved in cell sur-
face interactions, signal transduction, and gene
regulation — are required to single out just one
neuroblast from the proneural cluster and pre-
vent the other cluster cells from entering the
neural pathway.
Development and Evolution of Insect
Appendages
One of the best known but least understood
aspects of Drosophila development concerns the
morphogenesis of adult structures from the larval
imaginal tissues. The eyes, wings, legs, antennae,
and other structures are derived from sacs of cells
that undergo extensive growth and morphogene-
sis during the larval and pupal stages. We are
most interested in the development of the wing.
We have recently shown that the product of the
vestigial gene is precisely expressed in those
cells of the imaginal wing disc that will form the
actual flight appendage and is not expressed in
the other disc cells that will form structural com-
ponents of the thorax. In the absence of vestigial
gene function, fruit flies are wingless, suggesting
that vestigial plays a special role in the develop-
ment of wings from imaginal tissues. Further stud-
ies into the regulation of vestigial expression in
Drosophila may provide some interesting clues
to the evolution and function of the vestigial
gene in winged and wingless insects.
66
The WMc/ear vestigial gene product controls wing and haltere development in the Drosophila em-
bryo. Vestigial protein is specifically expressed in cells of the imaginal discs that will form the
flight appendages — wing (top left) and haltere (bottom left). The nonexpressing cells of these
discs will form structural parts of the thorax. During morphogenesis the vestigial-expression
region everts (top right) and forms the wing blade (bottom right). In mutants lacking yestigidA
protein, no wing or halteres form.
Research and photography by Jim Williams and Steve Paddock in the laboratory of Sean
Carroll. From Williams, J., and Carroll, S.B. 1991. Genes Dev 5:2481-2495.
67
Human Disease Gene Identification
and Correction
C. Thomas Caskey, M.D. — Investigator
Dr. Caskey is also Professor of Molecular Genetics, Biochemistry, Medicine, and Cell Biology at Baylor
College of Medicine. He received his M.D. degree at Duke University. His internship and residency training
were in internal medicine, also at Duke; his postdoctoral training was at NIH, under the supervision
of Marshall Nirenberg. Dr. Caskey is a past president of the American Society for Human Genetics
and was named Distinguished Service Professor by the Board of Trustees of Baylor.
MOLECULAR genetics offers unprecedented
opportunities for the discovery of disease
genes, the development of simple DNA-based
diagnostics, and correction of single-gene de-
fects. Recent isolation of the gene responsible for
the fragile X syndrome has led to improved diag-
nostic procedures and discovery of a novel mech-
anism for the occurrence of genetic disease. Fur-
ther knowledge regarding the genetic mechanism
and pathology of two other X chromosome dis-
orders has also been gained. In addition, our labo-
ratory has made significant progress in the devel-
opment of gene replacement therapies for three
diseases.
Disease Gene Identification and Diagnosis
The gene that contains the breakage point in
fragile X syndrome has been isolated as the result
of a collaboration with Stephen Warren (HHMI,
Emory University) and Ben Oostra (Erasmus Uni-
versity). This gene contains a three-nucleotide
sequence repeated in tandem, and the repeat re-
gion forms the fragile site itself. The number of
repeats is variable and correlates with occurrence
of the fragile X disorder. Normal individuals have
approximately 6-50 repeats; affected individ-
uals, over 200 repeats; and those with an inter-
mediate number appear to harbor a "premuta-
tion" that is very likely to give rise to the full
mutation by inheritance through females. This
phenomenon of genetic disease caused by se-
quence amplification was previously unknown,
but may prove to play a part in other inherited
disorders.
Since the length of this particular region of
DNA is an indicator of the affected or carrier sta-
tus of the individual, DNA diagnosis of fragile X
syndrome can be performed by Southern analysis,
and more recently by the polymerase chain reac-
tion. These DNA-based methods improve on cyto-
genetic diagnosis, particularly in the detection of
female carriers and unaffected males who never-
theless can transmit the disorder.
Study of Genetic Disorders
The central region of the dystrophin gene is a
"hot spot" for the deletion end-points causing
Duchenne and Becker muscular dystrophies. This
region was studied in greater detail to determine
possible mechanisms that could explain why de-
letions were so frequent. In two independent pa-
tients with Duchenne muscular dystrophy, the
ends of deletion lay within a transposable ele-
ment (a portion of DNA that can move around the
genome) in the dystrophin gene. It is likely that
such transposable elements are involved in other
deletions in this and other genetic disorders.
Lesch-Nyhan syndrome in humans results from
lack of the enzyme hypoxanthine guanine phos-
phoribosyltransferase (HPRT). The neurological
aspects of this incurable disorder are poorly un-
derstood. Study of the syndrome and develop-
ment of therapeutic measures would greatly ben-
efit from the generation of an animal model.
Transgenic mice have been engineered to lack
the uricase enzyme (which is absent in humans),
and these are being bred with mice lacking the
HPRT enzyme. This strategy has potential for gen-
erating a mouse model of the human Lesch-Nyhan
syndrome.
Genetic Correction of Inlierited Disease
Human genes can now be cloned, placed into
defective (safe) viral vectors, and transferred into
other cultured cells, embryonic cells, and ani-
mals. These encouraging developments increase
the likelihood of successful gene replacement
therapy. Our laboratory is developing technology
toward that objective for three heritable diseases.
Each disease offers different technical and strate-
gic challenges.
Adenosine deaminase (ADA) deficiency is an
inherited autosomal recessive disease. Bone
marrow transplantation is curative, but carries
the risk of a graft rejection . Administration of PEG
ADA (the enzyme ADA attached to polyethylene
glycol) on a continuing basis has provided im-
provement in the immunologic function of pa-
tients, but is not a cure.
The goal of this laboratory is the development
of a clinical gene therapy protocol for ADA defi-
ciency. This will require the treatment of human
69
Human Disease Gene Identification and Correction
cells with supernatant containing vector viruses
(and no cells) , rather than with virus-producing
cells, because the former method is considered to
be safer. Supernatant viral infection of bone
marrow cells resulting in the acquisition of ADA
activity can be achieved, but high efficiency re-
quires the presence of other supporting cells,
which is impractical in the context of an autolo-
gous bone marrow transplant. Attempts are under
way to purify bone marrow stem cells, which are
responsible for the regeneration of the entire he-
mopoietic system, since targeting this group of
cells would provide a more efficient approach to
the permanent repopulation of bone marrow
with genetically corrected cells.
Ornithine transcarbamylase (OTC) deficiency,
the most common urea cycle defect in humans, is
inherited in an X-linked recessive manner. Coma,
seizures, and retardation are the result of hyper-
ammonemia secondary to the enzyme deficiency.
Liver transplantation provides a once-per-lifetime
cure, although again with the potential compli-
cations of a graft-versus-host reaction. This condi-
tion is poorly managed through dietary protein
restriction and medical therapy, making the dis-
order another excellent candidate for gene
therapy.
Two mouse models of the human disorder are
available for developing gene therapy. In this lab-
oratory, the OTC deficiency in one of the strains
of mice has been corrected by retroviral delivery
of the human OTC gene to the small intestine. In
current studies toward the development of hu-
man gene therapy for OTC deficiency, retrovi-
ruses and defective (safe) adenovirus vectors are
used to transfer the human gene to cultured liver
cells.
Duchenne muscular dystrophy (DMD) is a se-
vere disorder, also inherited in an X-linked reces-
sive manner. Deficiency of the protein dystro-
phin leads to multiple muscle abnormalities and
eventually death. The gene is large and complex.
Several mouse models of DMD are available for
study, and the dystrophin deficiency in one of
these strains has been corrected by introduction
of a construct expressing mouse dystrophin.
In collaboration with Helen Blau (Stanford
University), myoblasts (precursor muscle fiber
cells) are being isolated from very young patients
with DMD. Different constructs will be tested for
the transfer of the dystrophin gene and for creat-
ing gene-corrected myoblasts that can be re-
turned to the patient. The gene is so large that
currently used viral vectors will not accommo-
date it. Approaches being used to overcome this
problem include modifying the vectors, truncat-
ing the gene, and using physical methods (that do
not present a size limitation) for transferring the
gene, such as electroporation or ballistic gene
transfer (using the Du Pont "gene gun").
70
Enzymatic RNA Molecules and the Structure
of Chromosome Ends
Thomas R. Cech, Ph.D. — Investigator
Dr. Cech is also American Cancer Society Professor at the University of Colorado at Boulder and Professor
of Biochemistry, Biophysics, and Genetics at the University of Colorado Health Sciences Center, Denver.
He received his B.A. degree in chemistry from Grinnell College and his Ph.D. degree in chemistry from the
University of California, Berkeley. His postdoctoral work in biology was conducted in the laboratory of
Mary Lou Pardue at the Massachusetts Institute of Technology. Dr. Cech is a member of the National
Academy of Sciences. Among his many honors are the Lasker Award and the 1989 Nobel Prize in chemistry.
THE nucleic acids, DNA and RNA, are best
known as the information-carrying molecules
of a living cell. For example, a molecule of DNA
or RNA might carry the instructions to build myo-
sin, a protein involved in muscle movement,
or pepsin, a protein enzyme that helps digest
food. Our laboratory is investigating the non-
informational roles of nucleic acids, situations in
which a nucleic acid molecule has an important
cellular function other than encoding a protein.
In the area of RNA catalysis, we wish to under-
stand how a folded RNA structure can have enzy-
matic activity. In the area of chromosome func-
tion, we are characterizing the DNA and
associated protein necessary for proper mainte-
nance of a chromosome end. In both projects,
chemical and biological approaches are com-
bined for fuller analysis of a biochemical
problem.
RNA Catalysis
A cell must orchestrate thousands of chemical
reactions in order to live, grow, and respond to its
environment. These chemical reactions, rarely
spontaneous, are usually catalyzed by macromole-
cules called enzymes. It was long thought that all
enzymes were proteins. More recently we and
others have found that a nucleic acid, RNA, can in
some cases act as an enzyme.
The finding of RNA catalysis has several impli-
cations. First, it means that RNA is not restricted
to being a passive carrier of genetic information
but can participate actively in directing cellular
biochemistry. In particular, many RNA-process-
ing reactions are at least partly catalyzed by RNA.
Second, the study of RNA enzymes, called ribo-
zymes, may reveal hitherto unknown mecha-
nisms of biologic catalysis. Third, ribozymes can
be used as sequence-specific RNA cleavage agents
in vitro, providing useful tools for study of RNA
biochemically. Finally, on a more speculative
note, RNA catalysis has the potential of providing
new therapeutic agents. For example, ribozymes
efficiently cleave and thereby destroy viral RNAs
under controlled laboratory conditions, suggest-
ing that they might be able to inactivate viruses in
a living organism.
Many of our studies of RNA catalysis concern
the Tetrahymena ribozyme, named for the sin-
gle-celled animal from which it was originally
isolated. This RNA enzyme is capable of cleaving
other RNA molecules (substrates) in a sequence-
specific manner. One of our objectives is to un-
derstand the mechanisms by which this RNA mol-
ecule acts as a catalyst. A second goal, in the area
of structural biology, is to obtain a detailed pic-
ture of the active site of this ribozyme.
Last year we demonstrated that this ribozyme
uses a novel mode of RNA recognition to bind its
RNA substrate. In addition to the well-established
mode of binding by formation of base pairs (as in
the "ladder" of the DNA double helix), the ribo-
zyme also binds two of the sugar groups that form
the "backbone" of the RNA substrate chain. We
have now located the other partner in this inter-
action, a specific adenine base within the RNA's
catalytic core. In addition to support from HHMI,
a grant from the National Institutes of Health
supported a graduate student working on this
project.
In a separate study, nucleic acid chemistry was
used to change individual oxygen atoms to sulfur
atoms near the cleavage site in the RNA substrate.
With one of the sulfur-substituted RNAs, we ob-
served a change in metal-ion specificity in the
cleavage reaction. Thus we believe we have lo-
cated one of the long-postulated active-site metal
ions and have established that this RNA, like some
protein enzymes, is a metalloenzyme.
In the structural area, we have developed a
new technique for identifying parts of the folded
RNA molecule that are in proximity to some
known site. We attach a controllable agent of de-
struction to a small molecule known to dock in
the ribozyme's active site, allow it to bind, initi-
ate the chemical reaction, and map the sites of
damage. The method has revealed convincing in-
formation about the higher-order folding of this
RNA catalyst.
Telomere Structure
Unlike the circular chromosomes of bacteria,
the chromosomes found in the nuclei of higher
71
Enzymatic RNA Molecules and the Structure of Chromosome Ends
organisms are linear DNA molecules. The ends of
linear chromosomes, called telomeres, must be
protected from degradation, and special features
are required to ensure their replication. We are
studying telomere structure and function, with
special emphasis on the protein that caps off the
ends of each chromosome.
Most cells have only a few dozen chromosomes
and therefore not many telomeres. We chose to
work with a ciliated protozoan, Oxytricha nova,
because it has 26 million miniature chromo-
somes per cell. This facilitated our earlier purifi-
cation of the telomeric protein and our cloning
and sequencing of the genes that encode it. Simi-
larities in DNA sequences in the telomeres of Oxy-
tricha and those of higher cells, including human
cells, give us reason to believe that our findings
in Oxytricha will be of some generality.
Last year we used genetic engineering methods
to produce the two subunits of the telomere-
binding protein in bacteria. This has now allowed
us to assess the individual contributions of the
two subunits to the protein-DNA complex (work
supported mainly by a grant from the National
Institutes of Health). It has also allowed the puri-
fication of large amounts of the protein, enabling
attempts to crystallize this complex, in collabora-
tion with Steve Schultz's research group (Univer-
sity of Colorado). If successful, crystallization
could lead to structure determination via x-ray
diffraction, providing an atomic-resolution pic-
ture of the end of a chromosome.
72
Molecular and Cellular Physiology of Acute
Inflammatory Cytokines
David D. Chaplin, M.D., Ph.D. — Associate Investigator
Dr. Chaplin is also Associate Professor of Medicine, Genetics, and Molecular Microbiology at Washington
University School of Medicine and Assistant Physician at Barnes Hospital, St. Louis. He received his A.B.
degree in biochemistry from Harvard University and his M.D. and Ph.D. degrees in cellular and
developmental biology from Washington University. Following a residency in internal medicine at
Parkland Memorial Hospital, Dallas, he received postdoctoral training in genetics at Harvard Medical
School with Jonathan Seidman. He then returned to Washington University as a faculty member.
PROPER regulation of immune defenses re-
quires the coordinate action of an array of
different host responses. Many of these responses
depend for their regulation on the interactions of
several different types of immune cells. Some of
these interactions are mediated by physical con-
tact between the participating cells. Others, how-
ever, are mediated by soluble proteins that are
released from one cell to act by binding to spe-
cific receptors on other cells. These soluble medi-
ators have been designated cytokines.
Three cytokines appear particularly important
for the initiation of acute inflammatory and im-
mune responses: interleukin-1 (IL-1), tumor ne-
crosis factor (TNF), and interleukin-6 (IL-6).
IL-1, TNF, and IL-6 are pleiotropic cytokines.
They are produced by a wide variety of cells, and
they act on many tissues, showing a broad array of
activities. In general, these activities are proin-
flammatory, favoring first the recruitment of
white blood cells to sites of inflammation, then
activation of these cells to express their full ef-
fector potential. This recruitment and activation
is absolutely necessary for host defense against
invading microbes; however, when it occurs inap-
propriately, it can lead to acute and chronic dam-
age to self tissues.
Although a great deal of information has been
accumulated concerning the activities of these
cytokines when added to cultured cells or when
injected into live animals, little is known about
their true physiological function in normal host
responses. In the case of IL-1, this uncertainty is
underscored by studies of the cytokine's struc-
ture. First, biochemical and molecular genetic
analysis of IL- 1 has established that there are actu-
ally two forms of IL- 1 (IL- 1 a and IL- 1 /?) , each the
product of a discrete gene. The relative produc-
tion of each form is variable, with some types of
cells producing primarily IL- 1 a and some primar-
ily IL-1/3. Both forms are synthesized in the IL-1-
producing cell as a larger precursor molecule
(pro-IL-la and pro-IL-1/3), and both are cleaved
to a smaller mature form that is found outside the
cell. In vitro, mature IL-1 a and -1|8 show appar-
ently identical activities. Surprisingly, the two
cytokines share only 25 percent amino acid se-
quence identity. Early studies from our labora-
tory showed that the genes encoding the two IL- 1
proteins were generated by duplication of an an-
cestral IL-1 -like gene, with subsequent extensive
mutation to produce the existing IL-1 a and -1|8
genes. We interpret the separate lineages of the
two genes to indicate that there has been selec-
tive pressure for the independent evolution of
each gene. Although this pressure probably acts
at the level of unique functions for each of the
two IL-1 molecules, in vitro studies have not re-
vealed any functional diff^erences between the
two cytokines. This indicates to us that existing
in vitro models do not adequately represent the
normal physiology of IL-1 and that new models
must be explored to disclose the functional dif-
ferences that must exist in vivo.
Just as there are fundamental questions con-
cerning the unique functions of each of the IL- 1
proteins, there are basic questions concerning
the mechanism by which these proteins exit the
cell to perform their extracellular functions.
Normally proteins that are destined to act outside
the cell contain a characteristic signal peptide, a
short amino acid sequence that directs them into
the cellular secretory pathway. Neither IL- 1 pre-
cursor protein contains a secretion signal pep-
tide. Additionally, unlike conventionally se-
creted proteins, pro-IL-la and -1/5 reside in the
cellular cytoplasm, not associated with any mem-
branous cellular compartment. Consequently it
has been suggested that the IL- 1 proteins are re-
leased via a novel secretory pathway.
Recent studies from our laboratory have
yielded new insights into the mechanism of lL-1
release and maturation. In these studies we have
shown that release of IL-1 a and -1/8 from acti-
vated cells does not involve a unique secretory
mechanism but occurs as a consequence of injury
to the IL-1 -producing cell. This conclusion is
based on the observation that IL-1 release is ac-
companied by parallel release of other cytoplas-
mic proteins, such as the enzyme lactate dehydro-
genase (LDH). Agents that enhance IL-1 release
either induce increased IL- 1 synthesis or increase
73
Molecular and Cellular Physiology of Acute Inflammatory Cytokines
the degree of cellular injury and cause greater
release of LDH. Conversely, agents that reduce
IL-1 release either reduce IL-1 synthesis or en-
hance cell viability.
Although the release of IL- 1 from IL- 1 -produc-
ing cells is not via a specific release pathway, the
outcome of release can be modulated. This is
demonstrated by studies in which IL-1 -produc-
ing cells are deliberately injured. In the most gen-
eral terms, there are two types of cellular injury.
One type occurs when vital cellular functions are
interrupted, leading to cellular necrosis. This oc-
curs as a consequence of many stress injuries, in-
cluding hypoxia, alterations in local pH, ex-
tremes of temperature, and physical attack by
infectious lytic viruses. Morphologically, ne-
crotic cells exhibit disruptions in their plasma
membrane and disintegration of the intracellular
organelles. The consequence of necrosis is the
release into the extracellular space of all of the
soluble cellular components. As expected, when
an IL-1 -producing cell sustains a necrotic injury,
the IL- 1 precursors are released along with all of
the other cellular components; however, there is
failure of processing of the IL-1|8 precursor to its
mature form. This failure is important, because
maturation is required for the IL-l/? precursor to
acquire receptor-binding activity (and, conse-
quently, for the acquisition of all of the known
IL-1 functional activities).
In contrast to cellular necrosis, cell death can
occur in a regulated fashion. Normal develop-
ment and tissue remodeling frequently require
removal of otherwise viable cells. This occurs by
activating the pathway of programmed cell death
leading to apoptosis. Morphologically, apoptosis
is very different from necrosis, with apoptotic
cells showing retention of the integrity of their
plasma membranes and maintenance of the struc-
ture of intracellular organelles. Portions of the
cell are released as membrane-coated blebs and
are subsequently ingested by phagocytic cells
and digested. Apoptosis permits the termination
of the life of a cell with minimal release of cellu-
lar debris into the extracellular environment.
This pathway is activated during many stages of
embryological development and to initiate nor-
mal cell turnover. In the immune system, re-
moval of self-reactive T lymphocytes in the thy-
mus occurs by apoptosis, as does the death of
target cells attacked by cytotoxic lymphocytes.
When this pathway of programmed cell death is
activated in macrophages expressing the IL-1
precursors, there is striking, rapid, and complete
maturation and release of both forms of IL- 1 .
Thus, in cells expressing only IL-l/? (such as
blood monocytes and certain cells in the spleen
and thymus), apoptosis — in contrast to necrosis
— affords an efficient method for generating
bioactive IL-1 .
These results have given us a new appreciation
of IL-1 as a physiological signal of cell and tissue
injury. IL- 1 does not appear to be released from
the cell that produces it unless a significant cell
injury occurs. If the injury is necrotic in nature,
then IL-1/3 is released in only its inactive precur-
sor form. If, on the other hand, the injury leads to
apoptosis, then mature bioactive IL-1/3 is released
in a form ready to activate the systemic responses
that lead to tissue repair.
74
Hormonal Regulation of Gene Expression
William W. Chin, M.D. — Investigator
Dr. Chin is also Associate Professor of Medicine at Harvard Medical School and Senior Physician at
Brigham and Women's Hospital, Boston. He obtained his undergraduate degree in chemistry from
Columbia College and his M.D. degree from Harvard Medical School. His postdoctoral work was performed
with Jacob Maizel and Philip Leder at NIH and with Joel Habener at Massachusetts General Hospital,
Boston. His awards include the Bowditch Lectureship Award of the American Physiological Association,
the Van Meter- USV Award of the American Thyroid Association, and the Outstanding Investigator's Award
from the American Federation for Clinical Research.
HORMONES, key players in the endocrine and
nervous systems, are produced by special-
ized tissues in the body, effect extracellular com-
munication, and regulate cellular function. Our
recent studies have considered the molecular
mechanisms of gene regulation by a specific hor-
mone, thyroid hormone.
Thyroid Hormone Action
Our early work has focused on the regulation
of the thyrotropin (thyroid-stimulating hormone
[TSH]) subunit genes by thyroid hormones. TSH is
a polypeptide hormone that is produced and se-
creted by a single cell type in the anterior pitu-
itary gland. It comprises two different sugar-
containing polypeptide subunits, a and /?, which
are encoded by genes located on different chro-
mosomes. Importantly, TSH stimulates the thy-
roid gland to produce the thyroid hormones, T3
and T4. These modified amino acids regulate me-
tabolism and gene expression in almost every cell
of the body. To maintain a constant level of T3 and
T4 in the bloodstream, these hormones act back
on the pituitary to decrease TSH production, in-
cluding subunit gene expression, and secretion.
Hence TSH and thyroid hormones are involved in
a classic negative-feedback regulatory system.
To understand further the molecular mecha-
nisms involved in the negative regulation of TSH
synthesis by thyroid hormones at the transcrip-
tion level, we have isolated and analyzed the
genes encoding the a- and /5-subunits of TSH in
the rat and identified putative thyroid hormone-
responsive elements (TREs). These TREs, which
mediate a negative regulation by the thyroid hor-
mones and thyroid hormone receptors (TRs),
contain several consensus TR-binding half-sites
in a direct repeat orientation and are located im-
mediately upstream of the TATA box (a-subunit)
or near the start of transcription (TSH;3 subunit) .
Multiple Thyroid Hormone Receptors
and Other Nuclear Factors
Thyroid hormones, in general, act at the cellu-
lar level by entering the cell either as Tj or T4. T4
may be converted to T3 in certain tissues, such as
the pituitary. T3 then enters the nucleus, where it
interacts with the TR, a protein encoded by the
proto-oncogene c-erbA. The T3-TR complex then
binds directly to cis-DNA elements within thyroid
hormone-responsive genes to activate the appro-
priate responses.
The TRs are encoded by two genes, a and /3,
each expressing at least two related molecules
obtained by alternative promoter choice and
splicing. The rat TRa gene encodes a bona fide
TR, TRal , and a related but non-Tj-binding form,
c-erhAa2. The rat TRjS gene encodes two func-
tional TRs, TRjSl and TR/32, which are identical
except for different regions amino terminal to the
DNA-binding domain. Remarkably, TR/32 was
most highly expressed in the pituitary gland. This
observation has potential physiological relevance
because of the major role of thyroid hormones in
regulating important hormone genes in the pitu-
itary, including TSH. The finding also stands in
contrast to the general tissue distribution of the
other receptor forms.
We recently demonstrated the existence of an-
other nuclear protein(s) that can augment the
binding of TRs to various TREs. This ubiquitous
60- to 65-kDa protein (TRAP; TR auxiliary pro-
tein) interacts with TR to form a heterodimer and
also binds specific sequences within the TRE. Sev-
eral regions of the carboxyl-terminal or ligand-
binding domain of the TR are important for this
heterodimerization. Preliminary results show
that the TR-TRAP interaction may be critical in
thyroid hormone-mediated transactivation. Fur-
thermore, the recent identification of the reti-
noid X receptor as a potential TRAP by several
groups fuels additional excitement in this area.
In summary, the thyroid hormone receptor fam-
ily is complex. At least three biologically active
forms are expressed in a tissue-specific fashion,
and another form may have an important role in
modulating the effects of the others. Knowledge
of the functions of these receptor isoforms and
their interactions with other nuclear factors will
be critical for our full understanding of thyroid
hormone action.
75
Hormonal Regulation of Gene Expression
Thyroid Hormone Action
at the Post-transcriptional Level
In addition to the well-known transcriptional
effect of thyroid hormone on TSH synthesis, we
have recently examined the effect of thyroid hor-
mone on the half-lives of TSH subunit mRNAs in
vitro and in vivo. The turnover rate of the TSH|8
mRNA was greater in the presence of thyroid hor-
mone. This phenomenon was associated with a
shortening of the poly(A) tail of the mRNA. These
events occurred in a dose- and time-dependent
fashion. Thus thyroid hormone reduces the half-
life of the TSH/3 mRNA and causes a concomitant
decrease in its poly (A) size. These data indicate
that thyroid hormones may also exert their effects
on gene expression at the post-transcriptional
level.
Summary
Our work has focused on the molecular mecha-
nisms of thyroid hormone action, with a focus on
the regulation of TSH gene expression by thyroid
hormones, including the cis-elements and trans-
factors that are involved in these processes. The
effects of these hormones are manifest at both the
transcriptional and post-transcriptional levels.
We hope that our efforts will provide insight into
the hormonal regulation of gene expression and
cellular function in normal and pathological en-
docrine states and in cancer.
76
Technologies for Genome-sequencing Projects
George M. Church, Ph.D. — Assistant Investigator
Dr. Church is also Assistant Professor of Genetics at Harvard Medical School. He received his B.A. degree in
zoology and chemistry from Duke University and his Ph.D. degree in biochemistry and molecular biology
from Harvard University. Before moving to Harvard Medical School, Dr. Church was a scientist at Biogen
Research Corporation and a Life Sciences Research Foundation Fellow in the Department of Anatomy
at the University of California, San Francisco.
THE study of the linear sequence of bases in
genomic DNA and messenger RNA is steadily
gaining recognition, in part as a result of the in-
creasing ease and advantage of using shared com-
puter databases to find connections among dis-
tant concepts and distant biological systems. For
example, connections have been found between
human oncogenes and yeast transcription factors,
between differentiation antigens and bacterial
chaperone proteins, and between developmental
regulatory genes and bacterial DNA-binding
proteins.
Unfortunately, these database searches are fre-
quently unsuccessful, because not all classes of
genetic elements are represented. The complete
sequence of a few small genomes should rectify
this. Sequencing projects have begun for ge-
nomes of various bacteria {Mycoplasma, Myco-
bacteria, Escherichia, and Thermococcus) , a
yeast (Saccharomyces) , a plant (Arabidopsis) ,
and a worm (Caenorhabditis} , chosen for their
well-studied genetics, their small genome sizes,
and their representation of all major branches of
the evolutionary tree. The genome closest to
completion is that of Escherichia, with about 38
percent of its 4.7 million base pairs already in the
database, through the effort of 2,000 researchers.
To improve the accuracy and efficiency of
these projects, we have developed new sequenc-
ing technologies. One, called multiplex se-
quencing, is a way of keeping a large set of DNA
fragments as a precise mixture throughout most
of the steps of sequencing. Because each mixture
can be handled with the same effort as a single
sample in previous methods, more fragments can
be handled.
The mix is deciphered by strategically tagging
the fragments at the beginning with unique bits
of DNA and then, at the end, hybridizing to the
sequencing reactions complementary bits of DNA
that have been spread out by size and immobi-
lized on large membranes. This method also im-
proves the accuracy, since the mixtures contain
internal standards of known sequence that help
in the computer analysis of the film data.
The number of probings obtainable per mem-
brane represents the increased efficiency factor
of this method. This number exceeds 50 now
(the higher the better) and is likely to increase.
We have designed and tested devices to facilitate
most of the steps in multiplex sequencing, in-
cluding DNA preparation, sequencing reactions,
gel loading, hybridization, film exposure, and
film reading. All of these devices have been ap-
plied to collect over 1 million bases of raw data
and are undergoing further development. Multi-
plexing has also allowed chemiluminescent de-
tection to replace the radioactivity normally used
in DNA sequencing, reducing exposure times 1 0-
fold. To fill in specific gaps in the sequences, we
have devised multiplex oligonucleotide synthe-
sis for use in multiplex DNA sequence walking
strategies.
Toward the goal of modeling cell structure and
gene expression, we have searched for abundant
cellular proteins that have nonetheless eluded
the extensive biochemical and genetics studies of
Escherichia coli. This has been done by systemati-
cally correlating amino-terminal protein se-
quence data obtained from two-dimensional
gel spots with the DNA sequence and two-
dimensional gel databases. Of 300 sequences de-
termined so far, over 50 are candidates for such
major novel proteins.
We have extended our methods for detecting
in vivo molecular interactions by analyzing the
protection of individual DNA bases from enzy-
matic methylation. DNA protein interactions in-
volved in cAMP and pyrimidine feedback regula-
tion have been studied in this way.
In the future, with new sequencing technolo-
gies such as automated multiplex sequencing,
with examples of most of the basic genetic mod-
ules, and with an eye to sequence elements con-
served among species, the analysis and modeling
by investigators worldwide of human sequences
and genetics should become more manageable.
77
Molecular Genetics of Limb Development
in Drosophila
Stephen M. Cohen, Ph.D. — Assistant Investigator
Dr. Cohen is also Assistant Professor in the Department of Cell Biology and in the Institute for Molecular
Genetics at Baylor College of Medicine. He obtained his Ph.D. degree at Princeton University, working
with Malcolm Steinberg. He did postdoctoral work with Harvey Lodish at the Whitehead Institute
and with Herbert Jackie at the Max Planck Institute for Developmental Biology, Tiibingen,
and at the University of Munich.
MY laboratory is interested in the processes
by which genetic information can be used
to organize the three-dimensional pattern of the
body during embryonic development. The partic-
ular problem on which we have focused involves
the organization of the limbs in the fruit fly Dro-
sophila. We would like to understand the genetic
and molecular mechanisms that the fly embryo
uses to determine where the limbs will develop
with respect to the rest of the body pattern. We
are investigating how the appropriate embryonic
cells become committed to develop as limb pre-
cursors and subsequently how these cells cooper-
atively organize the spatial pattern of the leg.
Specification of Limb Cell Identity
Using classical genetics, we have identified the
gene Distal-less, which seems to function as a
critical genetic switch that initiates limb develop-
ment in the embryo. Embryos that lack Distal- less
gene function do not develop larval or adult
limbs. These observations indicate that Distal-
less activity is critically required but do not tell
us what the gene is actually doing. Using molecu-
lar probes, we can visualize the RNA product of
the Distal-less gene in the embryo and in the de-
veloping adult limbs. All of the cells that will go
on to form limb structures express Distal-less. An
additional genetic experiment indicates that
these cells need the Distal- less gene product to
be limb cells. If the gene is removed from a group
(or clone) of cells that were already committed
to develop into limb structures, these cells in-
stead become body wall cells. Distal-less there-
fore specifies their identity as limb cells. Distal-
less is a member of a family of important
regulatory genes that encode a sequence-specific
DNA-binding motif called the homeodomain.
Distal-less therefore can be presumed to define
cell identity by regulating the expression of other
genes. We wish to identify the transcriptional tar-
gets through which Distal-less acts.
How Does the Embryo Position Its Limbs?
The critical first event in limb development is
determining the position at which the legs will
develop. We are using a combination of genetic
and molecular methods to identify the source of
the instructions specifying the identity of the leg
cells and the nature of the response that this sig-
nal elicits. In the Drosophila embryo, the legs
begin as clusters of cells. These clusters are estab-
lished at a well-defined location within each tho-
racic segment, with respect to both anterior-
posterior and dorsal-ventral position. One way to
mark a unique point is to draw two intersecting
lines. This rather simple solution appears to de-
scribe what the embryo does to identify the cell
clusters. Thus the spatial cues that the embryo
uses to position its legs appear to be stripes of
cells that act as sources of secreted intercellular
signals. The earliest detectable consequence of
this signaling process is that presumptive leg
cells express the Distal-less gene.
Our previous work identified the segmentally
repeated stripes of cells expressing the wingless
gene as one of the signaling centers responsible
for initiating leg development. Stripes of cells ex-
pressing wingless bisect the nascent leg primor-
dia. The wingless gene encodes a secreted inter-
cellular signaling molecule related to the
vertebrate oncogene INT- 1 . The secreted wing-
less protein is required to signal nearby cells to
express Distal- less but is not sufficient to do so.
An intersecting stripe of information is required
to specify the cluster. The decapentaplegic gene
is a good candidate to encode this second signal.
A stripe of cells expressing decapentaplegic in-
tersects the wingless stripe at precisely the posi-
tion where the leg is formed. The decapentaple-
gic gene encodes a product homologous to the
transforming growth factor-/3 (TGF-|5) class of se-
creted signaling molecules. A group of cells that
lies near the intersection of these two stripes
should receive both signals. Although we are not
yet certain whether decapentaplegic plays a di-
rect role in this process, the simultaneous receipt
of the wingless and TGF-/3 signals should, in prin-
ciple, be sufficient to specify the identity of the
leg precursor cells.
79
Molecular Genetics of Limb Development in Drosophila
why Do Insects Have Six Legs?
Insects develop legs only in their thoracic seg-
ments, while more primitive arthropods may de-
velop abdominal legs as well. A class of genes
known collectively as homeotic genes distin-
guish segments from one another. The homeotic
genes of the Bithorax complex act as negative reg-
ulators of thoracic segment identity. The Bithorax
genes repress Distal- less expression and block
leg development in abdominal segments. Accord-
ing to our model, the intercellular signal re-
sulting from the intersection of wingless- and
decapentaplegic-express'mg cells should in prin-
ciple be capable of eliciting limb development in
the abdomen. The signaling sources are ex-
pressed in precisely the same pattern in thoracic
and abdominal segments. We have found that this
response is blocked by the Bithorax genes. Re-
moving Bithorax genes permits Distal-less ex-
pression in the abdominal segments and leads to
ectopic limb development in the abdomen.
Therefore, the signal to specify leg primordia in
the abdominal segments exists, but the conse-
quences of signaling are overridden.
Molecular analysis of the control elements that
regulate Distal-less expression in the embryo
provides compelling evidence that the intercel-
lular signal is actually received and correctly in-
terpreted by cells in the abdominal segments, just
as in the thorax. We have identified a small piece
of DNA, known as an enhancer, that acts as the
target of this signal. By functional dissection of
this enhancer we have identified a regulatory ele-
ment that mediates the repression of limb devel-
opment in the abdominal segments. Removing a
small piece of DNA renders the enhancer element
insensitive to the repressor genes of the Bithorax
complex but has no effect on its response to the
activation signal. Gene expression driven by this
mutant version of the enhancer element is there-
fore constitutive ly "de-repressed" in the abdo-
men. These observations demonstrate that the
wingless- and decapentaplegic-dependent acti-
vation signal is both received and correctly inter-
preted by the would-be abdominal leg cells. Re-
pression of leg development must therefore lie in
the prevention of Distal- less expression by these
cells. These observations strongly suggest that
turning on the Distal-less gene in these cells is
the only signal necessary to specify their identity
as leg cells.
Proximal-Distal Pattern Formation
in the Leg
As mentioned above, flies that lack Distal- less
gene activity do not develop any leg structures.
Flies in which Distal-less gene activity is im-
paired, but not eliminated, develop abnormal
legs. Distal-less mutant legs are foreshortened
along the proximal-distal axis. The characteris-
tics of different Distal-less mutations tell us that
the amount of activity of the gene is important in
controlling the range of structures that the leg
can develop. Distal parts of the leg require more
Distal-less gene activity than do proximal parts to
develop normally. These observations suggest
that Distal- less may play an important role in or-
ganizing the proximal-distal axis of the leg.
Early in development, presumptive leg cells
express Distal-less, whereas presumptive body
wall cells do not. As the leg matures, this simple
pattern transforms into a graded distribution of
Distal-less RNA across the developing limb. The
distal-most region of the limb expresses a high
level of the gene product, intermediate levels ex-
press lower levels of Distal- less RNA, and proxi-
mal regions express little or no Distal- less RNA.
These observations are particularly intriguing in
view of the regional differences in the require-
ment for the activity of the gene along the leg. We
are interested in understanding the transition
from the initial simple, two-state system of the
embryo to the complex, graded distribution of
the gene product that we see at later stages.
Understanding pattern formation in the devel-
oping fly leg will teach us about fundamental
mechanisms important for the development of
vertebrate embryos. We are studying a system in
which intercellular signaling molecules are used
to assign cells to functional developmental units.
Although the particular details of the systems
vary, the fundamental principles and molecular
mechanisms will prove to be of wider applicabil-
ity. In this context we are intrigued by the role
that Distal-less plays as a pattern organizer at the
genetic level and its prospective role as a regula-
tor of gene expression at the molecular level.
80
Tracking Genes That Cause Human Disease
Francis S. Collins, M.D., Ph.D. — Investigator
Dr. Collins is also Professor of Internal Medicine and Human Genetics at the University of Michigan
Medical School. He received his Ph.D. degree in physical chemistry from Yale University and his M.D.
degree from the University of North Carolina. After completing his internship and residency in internal
medicine at the North Carolina Memorial Hospital, he went on to a fellowship in human genetics at Yale.
He recently received the Young Investigator Award of the American Federation of Clinical Research
and the National Medical Research Award of the National Health Council and was elected to the Institute
of Medicine and the American Association of Physicians.
THE theme of our laboratory is the study of
human genetic disease at the molecular
level. Our goal is to identify genes involved in
specific genetic disorders, to define their struc-
ture and function, to understand the control of
their expression, and to use this information to
develop potential new therapies.
Over 4,000 genetic disorders are listed in the
most recent edition of Mendelian Inheritance in
Man. For the majority of these, the normal func-
tion of the gene involved is not known. The iden-
tification of disease-causing genes without knowl-
edge of their protein product or its normal role,
now referred to as positional cloning, is a major
endeavor in our research.
Only recently has it become possible to iden-
tify such genes, and the process is still laborious.
First the gene must be mapped to a specific hu-
man chromosome, using a process known as link-
age analysis. This involves identification of fami-
lies with the disorder and analysis of DNA from
these families with a panel of probes from all
parts of the human genome. The probes are used
to establish the chromosomal location of a DNA
sequence that may have been inherited in associa-
tion with a disease gene that will probably lie
close to it. Such sequences, or "markers," can
pinpoint the chromosome on which the disease
gene resides. Additional probes from that chro-
mosome can then be tested to identify markers
that are even closer. It is often possible to narrow
the responsible region to about 1 percent of a
particular chromosome. Although this is a major
achievement, such a region may yet contain 30 to
50 genes, only one of which is responsible for the
disease. Thus additional refinements must be
made before candidates for the responsible gene
can be identified.
In 1989, as part of a collaborative effort with
investigators at the Hospital for Sick Children in
Toronto, we were successful in identifying the
gene for cystic fibrosis (CF), a common severe
genetic disease characterized by lung infections,
pancreatic insufficiency, and elevations in sweat
chloride concentration.
Although one mutation (AF508) is responsible
for about 75 percent of CF chromosomes in the
United States, numerous other mutations have
now been found in other individuals with the dis-
ease. This year we identified one mutation that
gives rise to such a mild form of the disorder that
it was not recognized as being caused by this
same gene, because the affected individuals had
normal sweat chloride. This observation indi-
cates that mutations in the CF gene may have a
broader clinical spectrum than previously
suspected.
Another important observation in CF this year
has been the demonstration by our group that the
AF508 mutation does not completely inactivate
the gene. In fact, using a frog oocyte system, we
were able to show that the AF508 protein can be
activated to almost normal levels by using drugs
that raise levels of cAMP, an intracellular second
messenger. This suggests that drug therapy aimed
at this second messenger might benefit patients
with the disease by activating their defective CF
protein.
Another major project in our laboratory is an
investigation of von Recklinghausen neurofibro-
matosis (NFl), a common genetic disorder some-
times incorrectly referred to as the Elephant Man
disease. Having identified this gene in 1990, we
are now intensively attempting to identify muta-
tions that cause the disease and to determine the
normal function of the protein product of this
gene. Previous studies have demonstrated that
this protein interacts with another class of pro-
teins (encoded by the ras genes) involved in the
regulation of cell growth. Recent data from our
group have shown that the NFl protein also inter-
acts with the cytoskeleton and can be seen to co-
localize with microtubules using immunofluores-
cence techniques. This unexpected observation
suggests that the protein product of the NFl gene
may play a crucial role in the regulation of cell
division. In the longer term, it is hoped that this
improved understanding of the biology of the dis-
ease will lead to improved therapies.
The Huntington disease (HD) gene has been
81
Tracking Genes That Cause Human Disease
localized by a collaborative group, which in-
cludes our laboratory. It is now known to lie in a
region near the tip of the short arm of chromo-
some 4, within an interval of approximately 2
million base pairs. Using yeast artificial chromo-
some cloning, we have generated an overlapping
set of genomic clones from this region and are
utilizing a variety of novel technologies to search
for candidate genes.
A new project in our laboratory this year is an
intense effort to identify the gene responsible for
early onset breast cancer, which has recently
been mapped by Mary-Claire King and her col-
leagues (University of California, Berkeley) to
chromosome 17. Using genetic linkage analysis
in families, as well as the analysis of tumors, we
are narrowing down the responsible interval. Sig-
nificant portions of the candidate interval have
now been cloned in yeast artificial chromosomes.
Utilizing the relatively new technology of physi-
cal microdissection of chromosomes, we have
been able to produce many new markers in this
region of the long arm of chromosome 17; thus
the opportunity to identify a major gene contrib-
uting to this common and devastating disease in
Western society is now quite real.
•■.»•
Colocalization of neurofibromin (the normal protein product of the neurofibromatosis gene)
and microtubules. The same rodent cell has been double-labeled and visualized by fluorescence
microscopy. Red shows staining with an antibody directed against tubulin, the major component
of cytoplasmic microtubules. Green shows staining of neurofibromin that appears to decorate the
microtubules.
Research of Paula Gregory, David Gutmann, and Francis Collins.
82
Development of the Immune System
Max D. Cooper, M.D. — Investigator
Dr. Cooper is also Professor of Medicine, Pediatrics, and Microbiology at the University of Alabama at
Birmingham. He received his M.D. degree and specialty training in pediatrics at Tulane Medical School
and his postdoctoral training in immunology at the University of Minnesota.
INFORMATION obtained from studies of im-
mune system development in a variety of verte-
brate species is used to investigate diseases of the
immune system in humans. We are particularly
interested in the pathogenesis of immunodefi-
ciency diseases and lymphoid malignancies.
Comparative studies in birds and mammals ini-
tially revealed the separate developmental path-
ways of T and B lymphocytes, the two major types
of immunocompetent cells. T cells provide help
for antibody-producing B cells and are primarily
responsible for immunity against viruses and
fungi. We wish to know whether the thymus is
the only source of T cells in vertebrates or
whether they may also arise in other tissues.
The central tissues may vary for production of B
cells, the source of antibodies. In birds, immuno-
globulin (Ig) -bearing B cells are derived from the
hindgut bursa of Fabricius, whereas in mammals
such cells are generated in blood-forming organs,
mainly the fetal liver and adult bone marrow.
Multipotent stem cells in these tissues use an elab-
orate gene program to generate millions of T and
B lymphocyte clones, each expressing a T cell
receptor (TCR) or Ig receptor of different antigen
specificity. These are seeded via the bloodstream
to the peripheral tissues, where they execute im-
mune surveillance of foreign and self antigens.
Inherited or acquired gene defects may specifi-
cally alter growth or maturation of these cell
lines to cause immune system dysfunction or
malignancy.
Comparative Analysis of T Cell Development
We have embarked on a comparative analysis of
T cell development in representative mamma-
lian, avian, and amphibian species. Our studies in
birds reveal remarkable evolutionary conserva-
tion of the pattern of T cell development found in
mammals, including the sequential development
of cells bearing either the yh or a/3 TCR.
Cells expressing the yb TCR are generated first
during ontogeny. They may not undergo clonal
selection during their intrathymic development.
They migrate preferentially to splenic red pulp
regions and the intestinal epithelium. Unlike afi
T cells, the yb T cells cannot recognize class II
molecules of the major histocompatibility
(MHC) gene complex to initiate a graft-versus-
host (GVH) attack. Analysis of their physiological
role is facilitated by their relative abundance in
birds (25-50 percent of T cells), where we are
examining their development and functional
capabilities.
In collaboration with Craig Thompson (HHMI,
University of Michigan) , we have used the TCR2
and TCR3 antibodies to identify two discrete sub-
lineages of a/3 cells that use different variable-
region gene families: V/Sl and V|82. The avian V/Jl
and V/32 genes contain highly conserved se-
quences that distinguish the two major sub-
groups of mammalian V/3 genes. Birds thus pro-
vide a simple model system for study of the
functional significance of these prototypic gene
families.
In birds, as in mammals, TCR/? diversity is cre-
ated largely by nucleotide sequence variations in
the joints between rearranged V/3, D (diversity),
and J (joining) genes. Since V/51 and V|82 genes
both combine with the same D and J genes, the
V|S 1 ^ and V/32^ T cells should be capable of recog-
nizing the same spectrum of antigenic peptides.
However, the V/31 and Vj82 genes have little se-
quence homology (less than 30 percent), so their
protein products may interact differently with
peptide-presenting MHC molecules.
Indeed, although both V/31+ and V/32+ T cells
can cause GVH disease in recipients with the
same MHC class II genes, they do so with differ-
ent efficiency. Selective inhibition of VjSl T cell
development revealed that only the V;3l^ T cells
can help B cells produce protective IgA antibod-
ies along the mucous surfaces of the body.
Current studies explore in more detail the func-
tional capabilities of the VjSl and V(82 subpopula-
tions of avian afi T cells.
To analyze T cell development in an amphib-
ian, we have made monoclonal antibodies that
identify 7^ and a/3 TCR candidates expressed by
separate lymphocyte subpopulations in Xenopus
laevis. Frog T cells bearing the putative yb and
TCR homologues exhibit the same tissue distri-
83
Development of the Immune System
bution patterns shown previously for their avian
and mammalian counterparts. Cloning of the frog
TCR genes is one current focus, and the thymus
dependence and functional capabilities of the
two sublineages are being explored. The goals of
th^se studies are to understand the evolutionary
strategy for generating T cells that can discrimi-
nate between self and nonself and to gain fresh
clues for some of the unresolved mysteries of the
human immune system.
B Cell Development
Bone marrow stromal cells can influence the
growth and development of B cell precursors, ei-
ther by direct cell contact or via soluble products
like interleukin-7 (IL-7). We are examining cell
surface molecules through which pre-B cells
may receive these environmental cues. One such
candidate molecule is an ectoenzyme named
aminopeptidase A (APA), the increased expres-
sion of which is induced by exposure to IL-7. We
are testing the hypothesis that IL-7 is either a li-
gand or a catalytic substrate for APA. We are using
a mouse gene probe to identify the human gene
for APA for sequence analysis, with the goal of
understanding its expression as a function of nor-
mal and leukemic pre-B cell differentiation.
Before pre-B cells rearrange their light-chain
genes to become B cells, they express a complex
receptor that is composed of surrogate light-
chain proteins (VpreB and X5), m heavy chains, and
two small transmembrane proteins called a and
/3. We have produced monoclonal antibodies
against exposed configurations of these protein
receptor units in order to explore the nature of
the receptor ligand(s) and the signals transduced
to the pre-B cell. One of these antibodies, which
is specific for the /3-chain of the Ig receptor unit,
may prove to be a universal B cell suppressant
because it down-modulates the antibody recep-
tors on all B cells.
Immunodeficiency Diseases
The pathogenesis of three primary immunodefi-
ciency diseases is being investigated. X-Iinked
agammaglobulinemia features an arrest in pre-B
cell differentiation, the precise nature of which is
under study in affected boys. Other studies ad-
dress the hypothesis that selective IgA deficiency
(IgA-D) and common variable immunodeficiency
(CVID) may represent polar ends of a clinical
spectrum reflecting a single underlying gene de-
fect. The cellular defect in both deficiencies con-
sists of arrested B cell differentiation. Both dis-
orders frequently occur in members of the same
family, and the same MHC class III haplotypes are
associated with IgA-D and CVID. Current studies
focus on the C4A complement gene.
84
Structure and Function of RNA Polymerase II
Jeffry L. Corden, Ph.D. — Associate Investigator
Dr. Corden is also Associate Professor of Molecular Biology and Genetics at the Johns Hopkins University
School of Medicine. He received his B.S. and Ph.D. degrees in biochemistry and biophysics from Oregon
State University. His postdoctoral work was done with Pierre Chambon in the Laboratoire de Genetique
Moleculaire des Eucaryotes at the Faculte de Medecine, Strasbourg, France.
THE first step in the expression of genetic in-
formation is carried out by RNA polymer-
ase II. How this enzyme determines which genes
are to be expressed in which cells and under
which physiological conditions remains largely
unknown.
RNA polymerase II contains more than 10 dif-
ferent proteins whose precise functions are only
now beginning to be addressed. The goal of my
laboratory is to understand the structure and
function of the subunits of RNA polymerase II.
Several years ago we began to analyze the largest
subunit of the mouse RNA polymerase II com-
plex. This subunit comprises one-half of the mass
of the enzyme and, through the work of many
laboratories, is known to be involved in the enzy-
matic synthesis of RNA. We have cloned and se-
quenced the mouse gene encoding this largest
subunit and have also isolated and characterized
several mutations in this gene. These mutant
genes are being used to study the function of dif-
ferent domains of the largest subunit. The aim of
these experiments is to understand how RNA
polymerase II orchestrates the ordered expres-
sion of 100,000 genes during the vertebrate life
cycle.
The gene encoding the largest subunit of RNA
polymerase II comprises 28 segments (exons)
that cover 30,000 bases of mouse chromosomal
DNA near the center of chromosome 1 1 . The
amino acid sequence deduced from the DNA se-
quence has revealed two interesting properties of
the subunit. The major portion of the protein is
similar in sequence to a bacterial RNA polymer-
ase subunit (from Escherichia coif) that carries
out an equivalent function. This evolutionary
conservation is much stronger than had been ex-
pected and has allowed us to predict that the
mouse subunit is involved in the transcription
elongation process.
Although the major part of the largest subunit is
related to the bacterial enzyme, our DNA se-
quence analysis has also revealed a domain that is
unique to RNA polymerase II. This domain is lo-
cated at one end of the molecule and constitutes a
52-fold repeat of a seven-amino acid sequence.
This unusual sequence, although absent in bacte-
ria, is found in other large subunits of virtually
every RNA polymerase II, including those of ani-
mals, plants, insects, and protists. We are
currently focusing our efforts on understanding
the role of this domain in the process of
transcription.
Our genetic approach to the function of this
carboxyl-terminal domain (CTD) has grown from
analysis of mutations in the largest subunit gene.
We first isolated mutant mouse tissue-culture cell
lines that are resistant to the mushroom toxin a-
amanitin. The largest subunit genes from several
of these cell lines have been cloned and, by re-
introduction into amanitin-sensitive cells, have
been shown to confer resistance to a-amanitin.
We have used this gene transfer technique to
map the mutations responsible for amanitin
resistance.
The availability of a well-defined, selectable ge-
netic marker in the largest subunit gene has
proved useful in the analysis of the role of the
CTD. Deletion, insertion, and substitution muta-
tions have been created in the CTD of an amani-
tin-resistance gene. The effect of these mutations
has been tested by introducing the altered resis-
tance genes into cells and scoring for amanitin
resistance. Removing the entire CTD eliminates
the ability to transfer amanitin resistance, demon-
strating that the CTD plays an essential role in
transcription. We have also shown that up to 20
of the 52 repeats are dispensable for growth in
tissue culture, indicating either that the CTD is
functionally redundant or that dispensable re-
peats are only necessary in some cell types. Para-
doxically, we have recently shown that the CTD
is highly conserved among all mammals. We
are now designing experiments to assess the role
of dispensable heptapeptide repeats during
development.
We have also been examining postsynthetic
modifications of the CTD. This domain is rich
in amino acids (such as serine, threonine, and
tyrosine) that can be modified by addition of
phosphate groups. RNA polymerase II is a phos-
phorylated enzyme, but no function for phos-
phorylation is known. We have used synthetic
85
Structure and Function of RNA Polymerase II
CTD peptide repeats as substrates to identify en-
zymes (CTD kinases) that carry out this phos-
phorylation. Two CTD kinases have been iso-
lated, each containing protein kinase catalytic
subunits previously identified in yeast as cell-
division control proteins. Further cloning studies
have revealed several more related mouse kinases
whose role in CTD phosphorylation is currently
under investigation. This investigation is sup-
ported by a grant from the National Institutes of
Health.
Using the purified CTD kinases, we have
mapped the sites of phosphorylation on the CTD.
These sites, serines that precede proline residues,
correspond to sites in other proteins that are mod-
ified in a cell cycle-dependent fashion. Current
studies are directed toward examining the level
of CTD phosphorylation at different times in the
cell cycle. One consequence of CTD phosphory-
lation is that the CTD becomes greatly extended.
The basis of this conformational change is also an
area of investigation in our laboratory.
To define more precisely the amino acid se-
quences required for CTD function, we have de-
vised a strategy to clone synthetic CTD repeats.
This work is being done in the yeast Saccharo-
myces cerevesiae, where as few as 1 0 of the re-
peats are sufficient for viability. We have con-
structed mutant CTDs in which the residues
identified as phosphorylation sites have been
changed. In most cases these mutations are lethal,
indicating that phosphorylation is essential for
RNA polymerase II function. We are currently us-
ing yeast genetics to isolate suppressors of these
mutations in an effort to identify proteins that
interact with the CTD. Through characterization
of such proteins we hope to determine the func-
tion of the CTD in RNA polymerase II function.
Hair cells from the bullfrog inner ear. A tuft of cilia, which sense mechanical
stimuli, protrude from the apical surface of each cell. Deflections of the bundle
toward the longer cilia are excitatory.
Research of David Corey and fohn Assad.
86
Mechanically Activated Ion Channels
David p. Corey, Ph.D. — Associate Investigator
Dr. Corey is also Associate Professor of Neuroscience at Harvard Medical School and Assistant Physiologist
at Massachusetts General Hospital, Boston. He studied physics as an undergraduate at Amherst College,
conducted research for a year at Harvard Medical School, and then entered the neurobiology program at
the California Institute of Technology. His thesis work, with James Hudspeth, focused on mechanical
transduction in auditory receptor cells. His postdoctoral work with Charles Stevens at Yale Medical School
was on voltage- sensitive ion channels.
OUR laboratory is interested in how protein
channels in cell membranes mediate the
electrical activity of the brain. Such channels,
which open and close to regulate the flow of min-
ute amounts of electrical current into a cell, are
intimately involved in the brain's information
processing. They are important in detecting sen-
sory signals such as light and sound, in the trans-
mission of this information from the sense organ
to the brain, and in communication from one
brain cell to another. We are focusing primarily
on ion channels in the sensory receptor cells of
the inner ear, especially on the mechanism of ac-
tivation of those channels that detect sound.
The sound-activated channels occur in a spe-
cial type of inner-ear cell called a hair cell. These
cells are named for a bundle of cilia that extends
from their top surface and that tilts back and forth
with each cycle of a sound wave. Moving the cilia
opens and closes the channels. Over 10 years ago,
it was found that the ion channels could open
extremely quickly — in just a few millionths of a
second — when the bundle was deflected. That
speed rules out the kind of biochemical chain
used by receptor cells of the eye and nose. We
proposed instead that mechanical forces on the
cilia could pull the channels open directly.
Others have shown that this was true, and it was
possible to measure the opening movement of
channels when force was applied to them. But
where are the channels, and what pulls on them?
Other workers, having explored around the
bundle with a fine electrode, indicated that the
channels were near the tips of the cilia. This was a
surprising result, since a lever action of the cilia
might focus forces near the bases. We have re-
cently confirmed this, however, by putting a fluo-
rescent dye inside the cell that reports the entry
of calcium through the channels. When calcium
is allowed to enter the cell, the fluorescent signal
appears first at the tips, showing that the channels
are there.
With attention focused on the tips of the cilia, a
group in England discovered that the tips are
linked by extremely tiny filaments, just 1 50 nano-
meters long, which they called tip links. To-
gether these clues led to a beautifully simple
theory for the operation of hair cells. Deflecting
the bundle in one direction would stretch the tip
links, and they would pull directly on the ion
channels to open them and let ions into the cell.
Moving in the other direction would relax the
links and allow channels to close. However, no
way to test the theory could be found.
This past year, we have been able to confirm it.
We found that removing calcium from the fluid
around the cilia, for just a few seconds, com-
pletely eliminated the tip links, as observed with
either transmission electron microscopy or scan-
ning electron microscopy. When we tested the
mechanical sensitivity of the hair cells, by mov-
ing a single bundle and measuring the electrical
response of the cells, we found that the same
treatment eliminated the electrical response in a
few tenths of a second. A further test indicated
that the electrical response was lost specifically
because the mechanical tension on the channels
was gone. Thus tip links do convey the stimulus
tension to the ion channels.
In addition to this transduction mechanism in
these cells, there is an adaptation mechanism that
enables them to be sensitive to extremely small
displacements while retaining an ability to re-
spond over a large range of stimuli. Hair cells
have a very narrow range of sensitivity. A deflec-
tion of about a third of a micrometer — the diame-
ter of one cilium — is sufficient to open all chan-
nels. We found some years ago, however, that a
steady deflection that opens all channels is rap-
idly followed by the spontaneous closure of
channels. To reconcile this with the context of
the tip-links model, one must suppose that clo-
sure is a consequence of relaxing the tension on
the channels. A further deflection to stretch the
tip links does, in fact, reopen channels. Thus
there seems to be a continuous adjustment sys-
tem, acting to set the tension on the ion channels.
Indeed, another group found that the stiffness of
the bundle relaxed with the same time course as
adaptation, in keeping with a tension adjustment
scheme.
In contrast, experiments from a different group
87
Mechanically Activated Ion Channels
suggested that adaptation is not a mechanical ad-
justment, but comes about through calcium entry
through the transduction channels. Calcium in-
side the cell would close channels, to create a
similar kind of negative feedback. It was clearly
important to distinguish these models.
In the past two years, we have obtained good
evidence for the mechanical adjustment hypoth-
esis. A key feature of this hypothesis is that the
putative adjustment of tension in the tip links
would also be felt by the cilia themselves and
would act to move the bundle. We found, first,
that we could change the rate of adaptation by
changing the voltage across the cell membrane.
Because voltage affected the relaxing more than
the tightening, we could then predict that posi-
tive voltage would increase the tension, and how
much and how quickly. Judging from the activa-
tion of transduction channels, these predictions
were borne out. Third, we could then predict
how much the tension change should move the
bundles. The predicted change was only about
100 nanometers — a quarter the wavelength of
light — so we worked out a video microscopy sys-
tem to measure the motion with a resolution of 5
nanometers. When the voltage was changed, the
bundles did move, with the predicted amplitude
and time course. Thus a quantitative theory that
describes the adaptation also describes the move-
ment, strongly suggesting that adaptation is
mechanical.
How could tension on the tip links be adjusted?
Others have speculated that the adaptation comes
about by a movement of the points where the tip
links are attached to the structure of the cilia.
When the tip links are stretched to open chan-
nels, the attachment points might slip to allow
the links to shorten. Conversely, if the tip links
are relaxed, the attachment points move up to
stretch the links to restore the resting tension.
This past year, we began to test this structural
theory, by measuring the position of the attach-
ment points before and after deflections that
cause adaptation. Bundles were given calibrated
deflections and then fixed with glutaraldehyde in
the deflected position. Measurements from trans-
mission electron micrographs suggest that the at-
tachment points do move in response to deflec-
tion, although we still need to test whether the
movement is quantitatively consistent with the
adaptation.
What could move the attachment point? Some
circumstantial evidence suggests that the "mo-
tor" molecule is like the myosin protein that
causes the contraction of muscle: The stiff cores
of the cilia are composed largely of actin, on
which myosin moves in muscle; the motor mole-
cule in these cilia moves in the same direction
and at the same rate as myosin; and glass beads
coated with muscle myosin do move on the actin
cores of these cilia. We are now seeking further
evidence as to whether the motor is a form of
myosin.
Our ultimate aim is to describe each link in the
mechanical chain from cilia to channels, in terms
of the protein identity of the links, their biophysi-
cal properties, and their relationship to each
other. Answers to these specific questions will
contribute to the long-range goal of a compre-
hensive theory of mechanically activated chan-
nels, not only in the ear but in the many other cell
types that display a mechanical sensitivity.
In the past two years, we have started to use
some of what has been learned about ion chan-
nels to understand human disease. Hyperkalemic
periodic paralysis is a dominantly inherited mus-
cle disease that causes sporadic weakness or paral-
ysis. Exercise or certain foods that raise the level
of potassium in the blood can bring on a paralytic
attack. Colleagues at Massachusetts General Hos-
pital had found that the voltage-sensitive sodium
channel in muscle was genetically linked to this
disease, suggesting that a defect in the channel
might be the cause. Last year we were able to
show that the channels are in fact defective. In
the presence of excess potassium, they fail to in-
activate normally, which allows sodium to enter
the muscle and (indirectly) causes the paralysis.
A puzzle, however, is that paralytic episodes
are often preceded in patients by myotonia — an
excessive contraction of the muscle. Could the
same defect cause both myotonia and paralysis?
This year we have used a toxin derived from sea
anenomes, which mimics the genetic defect in
sodium channels when applied to normal mus-
cle. Toxin-treated muscle did, in fact, show clas-
sic myotonia, with both increased electrical activ-
ity and increased tension. Thus a single genetic
defect can have a graded effect. Induced only
slightly, it causes myotonia; induced to a greater
extent by potassium, it causes paralysis. A com-
puter model of electrical activity in muscle
shows the same result with a single defect. Ex-
plaining the pathology at the molecular level
gives hints for effective drug treatment of the
disorder.
88
Genetic Regulatory Mechanisms
in Cellular Differentiation
Gerald R. Crahtree, M.D. — Associate Investigator
Dr. Crabtree is also Associate Professor of Pathology at Stanford University School of Medicine. He received
his B.S. degree from West Liberty State College, West Virginia, and his M.D. degree from Temple University
School of Medicine, Philadelphia. He was a senior investigator at NIH before coming to Stanford University.
CELLS acquire their final differentiated func-
tion by a complex interplay between primary
genetic regulatory events in the nucleus and in-
teractions at the cell membrane. Building on con-
cepts largely provided from studies on lower
animals, our laboratory has been exploring regu-
latory mechanisms that help determine how
mammalian cells differentiate to assume their
normal functions.
Isolation of a Trans-acting Regulator
of Homeodomain Protein Function
Several years ago, we identified a tissue-
specific transcription factor, HNF-la, that inter-
acts with essential regions of the promoters of a
large family of genes expressed in endodermally
derived tissues. After purifying the protein and
cloning its gene, we found that HNF-1 contains a
homeodomain similar to that found in genes de-
termining body form in insects. Curiously, the
protein dimerizes through an amino acid se-
quence unlike that found in other homeodomain-
containing proteins. This led us to look for a pro-
tein that might heterodimerize with it and hence
diversify its regulatory capabilities. We found
such a protein by screening a hepatocyte cDNA
library at low stringency. This protein, HNF-1/8,
which is expressed in a partially overlapping
group of tissues with HNF-la, contains a dimer-
ization and homeodomain similar to those of
HNF-la, but a different transcriptional activation
domain.
Surprisingly, HNF-la was only able to activate
transcription in certain tissues, suggesting the ex-
istence of a second tissue-specific protein that
regulated its function. Dirk Mendel began a
search for such a protein by purifying HNF- 1 a by
means that do not disturb hydrophobic interac-
tions. HNF-la copurified with an 1 1-kDa protein
that participated in the formation of a tetrameric
complex. The 1 1 -kDa protein, which was cloned,
expressed, and found to enhance the affinity of
dimerization between HNF molecules, was
named DCoH (dimerization cofactor for HNF- 1 ) .
In both cases DCoH required the amino-terminal
32 amino acids that constitute the dimerization
domain for binding to the complex consisting of
two molecules of HNF-la and two molecules of
DCoH. Developmentally, the DCoH protein is ex-
pressed early (about day 9 or 10 after fertiliza-
tion) in mice and is expressed in a group of tis-
sues that do not always express HNF-la or -1/3,
suggesting that other tissues such as the brain
contain proteins that can bind to DCoH. When
co-expressed with HNF-la, DCoH enhances tran-
scriptional activation by several hundredfold; we
are presently investigating the mechanism by
which this occurs.
T Lymphocyte Activation
and Differentiation
T lymphocytes undergo two biologically and
medically important types of differentiation. The
first occurs in the thymus and generates cells ca-
pable of directing an immune response to nearly
any antigen. However, the cells produced by the
thymus that circulate in our blood are immuno-
logically nearly inert. They acquire immunologic
function as a result of a second process of cellular
differentiation that takes about 10-14 days and
produces T cells that coordinate the actions of
other cells involved in the immune response by
production of cytokines and cell-cell interac-
tions. This differentiation pathway is initiated by
a complex interaction between the T cell and an
antigen-presenting cell. The essential require-
ment for a commitment to specialized function is
a highly specific interaction between histocom-
patibility molecules, antigen, and the antigen re-
ceptor. This critical interaction is only effective
when stabilized by transient nonspecific interac-
tions based on intracellular adhesive molecules.
Finally, lymphokines such as interleukin-1 and -6
that are the secreted products of the antigen-
presenting cell are accessory signals necessary to
initiate differentiation. These three requirements
for the initiation of differentiation — a highly spe-
cific cell-cell interaction, a nonspecific adhesive
interaction, and cytokines — are similar to the re-
quirements for the cellular commitment to dif-
ferentiate in other systems.
Because the interaction between the antigen-
89
Genetic Regulatory Mechanisms in Cellular Differentiation
presenting cell and a T lymphocyte is transient,
all of the molecular events required for the deci-
sion to proceed down this 1 0- to 1 4-day pathway
of cellular differentiation must occur in the short
period during which the T cell and the antigen-
presenting cell interact. Our laboratory is seeking
an understanding of the molecular basis of this
cellular decision.
Our approach to understanding the events that
initiate this pathway was to begin by defining the
molecules essential for the activation of genes
that are known to be essential for T cell activation
and difl'erentiation. This approach led us to de-
fine two transcription factors that specifically re-
ceive signals from the antigen receptor and are
responsible for activation of early genes such as
interleukin-2. One of these proteins, nuclear fac-
tor of activated T cells (NFAT), is expressed al-
most exclusively in T lymphocytes, and its tran-
scriptional activity is under rigorous control by
the antigen receptor. Furthermore, its transcrip-
tional activity is induced by the antigen receptor
immediately before the activation of most early
genes. We thus focused our attention on this tran-
scription factor rather than others that are acti-
vated in many cell types and under many biologic
circumstances.
Immunosuppressants Cyclosporin A
and FK-506 Block Nuclear Translocation
of the Cytosolic Component of NFAT
Cyclosporin A and FK-506 specifically inhibit
T cell activation, a characteristic that underlies
their clinical use as immunosuppressants to pre-
vent transplant rejection. Although the mecha-
nism of action of these drugs is unknown, they
appear to work early during the commitment pe-
riod for T cells. By acting early, they block the
late functions of T cells and many of the functions
of B lymphocytes and other hematopoietic cells
that are directed by T cells. Other groups have
found that these drugs bind and inhibit the func-
tion of cis-trans prolyl isomerases. These en-
zymes accelerate the folding of newly synthe-
sized proteins. In studies with Stuart Schreiber
(Harvard University) , we have shown that the iso-
merase activity is not involved in the action of the
cyclosporin and FK-506, but rather that an inhibi-
tory complex formed between either cyclosporin
A and cyclophilin or between FK-506 and FKBP
blocks signal transduction. This inhibitory com-
plex binds and blocks the activity of calcineurin,
a heteromeric protein phosphatase also known as
PP2B. This inhibition is likely to be related to the
mechanism of action of the FK-506 and cyclo-
sporin A, since immunosuppressive drugs form
complexes that block the activity of calcineurin
but inactive ones do not. We have found that the
specific transcriptional activity of NFAT, but not
its DNA-binding activity, is affected by cyclo-
sporin A and FK-506. Mike Flanagan, Peter Kao,
and Blaise Corthesy found that NFAT is a complex
heterodimeric protein, one subunit of which is
constitutive, T cell specific, and located in the
cytoplasm of resting cells; the other subunit is
located in the nucleus and is induced. The tran-
scriptionally active protein forms when the cyto-
plasmic component translocates to the nucleus in
response to stimulation through the antigen re-
ceptor of T cells. Cyclosporin A and FK-506 ap-
pear to function by inhibiting the translocation
but do not interfere with the induction of the
nuclear component. The inhibition of nuclear
import is specific to the extent that the drugs do
not block nuclear import of NF-kB. Recently,
Peter Kao and Blaise Corthesy have purified the
proteins and isolated the genes for the two sub-
units of NFAT. This should allow us to determine
directly if calcineurin is involved in the signaling
pathway initiating T cell activation.
90
The Mechanism of a Bacterial Transposition
Reaction
Nancy L. Craig, Ph.D. — Associate Investigator
Dr. Craig is also Associate Professor of Molecular Biology and Genetics at the Johns Hopkins University
School of Medicine. After receiving a bachelor's degree in biology and chemistry from Bryn Mawr College,
she did graduate work on bacterial responses to DMA damage with Jeffrey Roberts at Cornell University,
leading to her Ph.D. degree in biochemistry. She did postdoctoral research on the integration/excision
cycle of the bacteriophage X with Howard Nash at the National Institute of Mental Health. Before joining
HHMI, she was Associate Professor of Microbiology and Immunology and of Biochemistry and Biophysics
at the University of California, San Francisco.
DESPITE DNA's essential role in maintaining
accurate genetic information, this molecule
displays a surprising degree of plasticity. We now
know that DNA rearrangements — i.e., the reorga-
nization of DNA sequences through breakage,
translocation, and rejoining reactions — mediate
a wide variety of fundamental cellular processes.
DNA rearrangements play an important role in the
control of gene expression during development,
the acquisition of new genetic elements such as
viruses, the repair of damaged DNA, and the cre-
ation of genetic diversity. We are interested in
understanding at the molecular level how DNA
rearrangements occur and are controlled.
We are particularly interested in the type of
recombination called transposition. In this reac-
tion, a discrete DNA segment moves from one po-
sition in a genome into another, nonhomologous
target position. This translocation may occur be-
tween different positions on the same chromo-
some, between chromosomes, or between chro-
mosomes and extrachromosomal elements such
as plasmids. Transposable elements have been
identified in a wide variety of organisms. A nota-
ble consequence of transposition is the promo-
tion of rapid and extensive genetic change. Trans-
poson insertion results in the stable linkage of
information encoded by the transposon with the
target DNA. Transposon insertion into a gene
will likely inactivate the gene, and insertion into
DNA sequences that control the expression of
nearby genes may inactivate or activate those
genes. Probably because of its potential for
profound influence, transposition is highly
regulated.
Our research is focused on understanding the
transposition of Tn7, a bacterial transposon with
several unusual properties. Of particular note is
Tn7's unusual target selectivity. Most transpos-
able elements display little insertion-site selectiv-
ity, inserting into many different targets. By con-
trast, Tn7 inserts at high frequency into a single,
specific site in the chromosomes of many bacte-
ria. In Escherichia colt, the organism in which
we study Tn7, this special target is called an at-
tachment site and designated attTnl. When
attTn 7 is unavailable, Tn7 resembles most other
transposable elements, inserting at low fre-
quency into many target sites.
Like most mobile DNA segments, Tn7 encodes
the machinery that mediates its transposition
from place to place. We have established that Tn7
encodes a surprisingly complex array of transposi-
tion proteins, and we have also identified the par-
ticular DNA sequences at the ends of Tn7 and at
its insertion sites that are the actual substrates for
transposition. The high-frequency insertion of
Tn7 into attTnl is mediated by four Tn7-
encoded genes, tnsABC + tnsD\ Tn7 insertion
into other random target sites, a low-frequency
reaction, is mediated by a distinct set of Tn7-en-
coded genes, tnsABC + tnsE. Tn7 also encodes
resistance to the antibiotics trimethoprim and
streptomycin/spectinomycin. The ability of Tn7
and other mobile DNA segments that encode anti-
biotic resistance to insert into and thus be joined
to plasmids that can, in turn, move among a vari-
ety of bacterial species underlies the rapid dis-
semination of antibiotic resistance among bacte-
rial populations.
Our overall goals are to understand in molecu-
lar detail how Tn7 moves from place to place
and how the frequency of this movement is
modulated. We expect that understanding the
macromolecular interactions that underlie Tn7
transposition will contribute not only to the
understanding of DNA recombination but also
to the understanding of other complex protein-
nucleic acid transactions such as DNA replica-
tion, transcription, and RNA processing. We are
using a variety of biochemical and genetic ap-
proaches to dissect Tn7 transposition.
The fundamental steps in transposition are the
DNA cleavages that separate the transposon from
its flanking donor DNA and the subsequent break-
age of the target DNA and the joining of the trans-
poson to the target DNA. Little is known in molec-
ular terms about how such reactions occur. Our
91
The Mechanism of a Bacterial Transposition Reaction
HHMI-supported effort is focused on dissecting
the mechanism of the DNA strand breakage and
joining reactions during Tn7 transposition.
We previously developed a cell-free system for
Tn7 transposition to attTnl. Tn7 inserts effi-
ciently and specifically into attTnl in vitro in
the presence of four purified TnV-encoded pro-
teins: TnsA + TnsB + TnsC + TnsD. Thus these
proteins participate directly in transposition. We
have also established that Tn7 strand breakage
and joining can be carried out by a subset of these
proteins: TnsA + TnsB + TnsC. We now want to
know which transposition protein carries out
which particular step(s) in transposition and the
chemical basis of these reactions. We are probing
these questions by examining the ability of each
transposition protein to individually execute a
subset of the chemical steps that underlie the
complete transposition reaction. Our strategies
to reveal what may be usually suppressed activi-
ties include manipulating the reaction conditions
and using DNA substrates derivatized in particu-
lar ways that we suspect may provoke strand
cleavage and joining.
Another biochemical method we are using to
identify the domains of the recombination pro-
teins that are most intimately involved in DNA
strand breakage and joining is to determine,
through protein-DNA crosslinking studies, the
segments of the recombination proteins that most
closely appose the positions of DNA strand break-
age and joining during transposition.
In a complementary genetic approach, we are
seeking to identify "active sites" in the recombi-
nation proteins that promote DNA strand break-
age and joining, by isolating and characterizing
mutant transposition proteins altered in their abil-
ity to promote these reactions. We suspect that
these active sites actually lie in TnsB, which binds
specifically to the ends of Tn7, i.e., the sites of
strand breakage and joining during transposition.
We are also particularly interested in TnsB be-
cause there is some sequence similarity between
TnsB and the recombinases of retrotransposons,
including the integrases of retroviruses. Our
long-term goals are to describe in detail the chem-
ical steps that underlie DNA breakage and joining
and to understand how the recombination pro-
teins promote these reactions.
We are also interested in understanding how
transposition is controlled and, in particular, in
understanding the interplay between Tn7 and its
bacterial host.
Much of our biochemical characterization of
Tn7 transposition has focused on dissecting the
high-frequency insertion of Tn7 into its spe-
cific insertion site attTn 7. We are now working
to extend the biochemical analysis of Tn7 trans-
position to its low-frequency insertion into ran-
dom target sites. We are developing a cell-free
system for Tn7 insertion into random target
sites. We will then be poised to dissect the dif-
ferences between high- and low-frequency Tn7
transposition in molecular detail. We also sus-
pect that characterization of Tn7 insertion into
random target sites will reveal host proteins
that likely participate in this recombination re-
action. We are also examining how the struc-
ture of the E. coli chromosome may influence
Tn7 transposition. Another strategy we are us-
ing to probe the control of Tn7 transposition is
to isolate and characterize mutant Tns proteins
that display altered transposition properties.
The above work on the control of Tn7 transposi-
tion is supported by a grant from the National
Institutes of Health.
92
Mechanisms of Antigen Processing
Peter Cresswell, Ph.D. — Investigator
Dr. Cresswell is also Professor of Immunobiology at Yale University School of Medicine. He was born
and educated in the United Kingdom. He received undergraduate degrees in chemistry and microbiology
from the University of Newcastle upon Tyne and his Ph.D. degree in biochemistry and immunology
from London University. Dr. Cresswell took postdoctoral training at Harvard University with
Jack Strominger. Before assuming his position at Yale, he was Chief of the Division of Immunology
at Duke University Medical Center.
FOREIGN protein antigens must be proteo-
lyzed into peptides and must associate with
specialized membrane glycoproteins before they
can be recognized by T lymphocytes. These
membrane glycoproteins are known as major his-
tocompatibility complex (MHC) molecules, be-
cause of their original definition as the critical
antigens responsible for organ graft rejection be-
tween members of the same species. Two types of
MHC molecules have evolved, apparently to deal
with two different categories of antigens. Class I
MHC molecules bind peptides derived from cyto-
solic proteins synthesized by the cell that bears
them, and are critical for T cell recognition of
virally infected cells. Class H MHC molecules
bind peptides derived from proteins internalized
by a number of class Il-positive cell types, such
as B cells or macrophages, collectively known as
antigen-presenting cells. The major current inter-
est of my laboratory is in the molecular mecha-
nisms involved in generating these MHC-peptide
complexes.
Class I MHC-Peptide Association
The association of cytosolic protein-derived
peptides with class I MHC molecules presents a
topological problem, in that the peptides or their
precursor proteins must cross an intracellular
membrane for binding to occur. Evidence from
the literature suggests that peptide binding oc-
curs early in the intracellular transport of class I
MHC molecules, perhaps in the endoplasmic re-
ticulum (ER).
A mutant cell line, T2, derived in our labora-
tory from a similar cell line, .174 (produced by
Robert DeMars at the University of Wisconsin-
Madison) , is defective in one or more of the steps
involved in generating class I MHC-peptide com-
plexes. Somatic cell genetic evidence suggested
that the gene (or genes) involved is localized in a
region of the MHC that is deleted in T2. Two of
these genes are homologous to a group of pro-
teins with multiple membrane-spanning domains
known as the ATP-binding cassette (ABC) family
of membrane transporters.
These molecules are generally involved in the
active transport of small molecules, or occasion-
ally larger proteins, across membranes. It has
been proposed that the members of the family
encoded in the MHC are responsible for the trans-
location of cytosolic peptides into the ER, where
they bind to class I MHC molecules. Data from my
laboratory have shown that most class I alleles
expressed in T2 indeed lack peptides when they
are affinity purified and associated peptides are
analyzed chromatographically.
Human class I MHC (HLA) molecules generally
fail to be transported to the cell surface when
expressed in T2, a probable consequence of the
lack of associated peptides. An exceptional al-
lele, HLA-A2, is significantly surface-expressed at
20-50 percent of wild-type levels. HLA-A2 is as-
sociated with a limited set of peptides in T2,
three of which have been isolated and se-
quenced. Two proved to be derived from the sig-
nal sequence of an interferon-inducible protein
known as IP- 30, and the third corresponded to no
known protein.
The two identified peptides, a nonamer and
1 1 -mer respectively, are presumably transpor-
ted into the ER by the conventional signal-
recognition mechanism involved in the translo-
cation of secretory and transmembrane proteins.
This would explain their association with HIA-A2
in a cell line defective in the normal mechanisms
of peptide transport. In wild-rype cells, class I-
associated peptides are generally nonameric. The
existence of the 1 1-mer peptide, and the fact that
the third peptide is a 1 3-mer, argues that the gen-
eration of nonamers in wild-type cells may be a
normal facet of peptide generation and transport.
Curiously, mouse class I MHC (H-2) molecules
in general are well surface-expressed in the T2
cell line, even though they are devoid of asso-
ciated peptides. This has led us to hypothesize
that the human class I molecules are subject to a
specific retention mechanism that prevents them
from leaving the ER unless they are associated
with a peptide. Understanding the difi'erential
transport properties of class I H-2 and HLA mole-
cules in both normal and mutant cells is a major
area of emphasis in our laboratory and should un-
93
Mechanisms of Antigen Processing
cover additional details of the mechanisms in-
volved in the in vivo generation of class I MHC-
peptide complexes.
Class II MHC-Peptide Association
Class II MHC molecules associate late in trans-
port with peptides derived from internalized pro-
tein antigens. It is believed that in the early stages
of transport, they follow the same pathway as
class I molecules, namely from the ER through
the Golgi complex. This leads to the question.
Why do class II molecules nothind peptides from
the same set of cytosolically derived peptides that
associate with class I MHC molecules?
At least a partial answer lies in the association
of class II molecules with an additional glycopro-
tein, the invariant chain. This molecule is not a
product of the MHC and is not structurally related
to class I or class II glycoproteins. It associates
with class II molecules immediately upon synthe-
sis in the ER, forming a large nine-chain structure
that we have found to consist of a core trimer of
invariant chains associated with three class II mol-
ecules, each of which is a two-subunit hetero-
dimer. Class II molecules in this structure cannot
bind antigenic peptides, though they acquire the
capacity to do so if released by mild denaturation
or partial proteolysis of the invariant chain. Pre-
sumably their association with the invariant chain
in vivo prevents class II MHC molecules
from binding inappropriate peptides early in
transport.
After assembly, the nine-chain class II MHC-
invariant chain complex is transported from the
ER and through the Golgi apparatus. We have
found that the invariant chain then targets the
complex to endosomal structures, where it en-
counters internalized foreign antigens, shown
strikingly when influenza virus is used as a test
antigen . Evidence obtained with proteinase inhib-
itors, such as leupeptin, suggests that the invari-
ant chain is proteolytically degraded in the endo-
some, releasing class II molecules capable of
binding peptides.
The simplest model for the generation of class
II MHC-peptide complexes suggests that internal-
ized proteins are endosomally degraded into pep-
tides that bind to the newly released class II
molecules. Recent evidence, however, suggests
that this model is too simple. We have found
that the T2 cell line, referred to above as a class I
MHC antigen-processing mutant, is also a class
II MHC antigen-processing mutant. T2 itself lacks
class II molecules as a result of the homozygous
deletion of the MHC region encoding class II mol-
ecules. However, if mouse class II molecules (I-
A'') are expressed in T2 by transfection, the cell
line is unable to process internalized protein an-
tigens into class II MHC-associated peptides rec-
ognizable by T cells. To date, it appears that both
mouse and human class II molecules expressed in
T2 associate normally with the invariant chain
and are transported to an endosomal compart-
ment where the invariant chain is proteolytically
cleaved. The missing steps in generating func-
tional class II MHC-peptide complexes remain to
be determined, but again appear to be the proper-
ties of MHC-linked genes.
94
Regulation of Human Retroviral
Gene Expression
Bryan R. Cullen, Ph.D. — Associate Investigator
Dr. Cullen is also Associate Professor in the Section of Genetics and the Department of Microbiology and
Immunology and Associate Medical Research Professor in the Department of Medicine at Duke University
Medical Center. He received his master's degree in virology from the University of Birmingham, England.
After emigrating to the United States, Dr. Cullen worked as a research technician for several years before
reentering graduate school at the University of Medicine and Dentistry of New Jersey, where he received
his Ph.D. degree in microbiology. Before accepting his current position at Duke, Dr. Cullen studied gene
regulation in higher eukaryotes, as a laboratory head in the Department of Molecular Genetics
at Hoffmann- la Roche.
RETROVIRUSES derive their name from their
ability to reverse the normal flow of genetic
information from DNA to RNA. They possess the
remarkable ability to synthesize a double-
stranded DNA copy of their single-stranded RNA
genome and then to integrate this DNA copy into
the genome of the infected host cell. Once the
genome of a retrovirus is integrated into a host
chromosome, it is indistinguishable from a host
gene and may be actively transcribed by the host's
transcriptional machinery.
The infection of an animal by retroviruses can
result in a number of disease states, of which the
most common is leukemia. Indeed, the avian leu-
kemia virus (ALV) was the first oncogenic virus to
be defined experimentally. ALV and the some-
what similar murine leukemia viruses continue to
be studied extensively as models for this virus
group. This research has not only helped to delin-
eate the retroviral replication cycle but has also
greatly advanced our understanding of retroviral
oncogenesis. Most importantly, this research
has allowed the definition of a number of cellular
genes, the oncogenes, whose inappropriate ex-
pression can contribute to cellular transformation.
Although animal retroviruses have been the
subject of scientific research for some time, the
discovery of human retroviruses occurred only
within the last decade. Two major groups of
pathogenic human retroviruses have been identi-
fied thus far. Human T cell leukemia viruses
(HTLV-I and HTLV-II) are known to be causative
agents of human leukemias, including adult T
cell leukemia, and are significant disease agents
in several parts of the world, including Japan and
the Caribbean basin. Of even more concern are
the human immunodeficiency viruses (HIV-1 and
HIV-2), which are a leading cause of disease and
death in parts of Africa and in the United States. A
third group of retroviruses, the human foamy vi-
ruses (HFV), has recently been detected in hu-
man populations in both Africa and Oceania but
has not yet been clearly associated with any
disease.
A striking feature of all three classes of human
retroviruses is that they each encode regulatory
proteins that control both the quantity and qual-
ity of viral gene expression. This regulatory com-
plexity is not observed in many animal retrovi-
ruses, including the avian and murine leukemia
viruses, and may strongly influence the patho-
genic potential of these "complex retroviruses."
The major focus of this laboratory has been the
determination of the role and mechanism of ac-
tion of these viral trans-activators, with a concen-
tration particularly on the Tat and Rev regulatory
proteins of HIV- 1 .
In the past we demonstrated that the Tat pro-
tein of HIV-1 acts on sequences located within
the HIV-1 promoter element (the long terminal
repeat or LTR) to increase the level of expression
of linked genes. This increased viral gene expres-
sion occurs via a bimodal mechanism that in-
volves an increase in the rate of transcription of
HIV-1 mRNAs and in the efficiency of transla-
tional utilization of those RNAs. The target se-
quence for Tat is a 59-nucleotide RNA stem-loop
structure located at the very 5' end of all viral
mRNA molecules. The direct interaction of Tat
with this RNA structure leads to an enhancement
of viral transcription initiation and elongation.
This mechanism, which may be unique to Tat,
remains poorly understood but is likely to in-
volve the interaction of Tat with cellular proteins
that are currently poorly defined. The identifica-
tion and characterization of these cellular pro-
teins is a major research aim of this laboratory.
A second HIV- 1 protein, Rev, is required for the
expression of viral structural proteins but is dis-
pensable for the expression of viral regulatory
proteins. Our research has demonstrated that Rev
acts post-transcriptionally to induce the cytoplas-
mic expression of the unspliced or incompletely
spliced RNAs that encode the viral structural pro-
teins Gag and Env, while simultaneously repress-
ing the expression of the fully spliced RNAs that
encode the viral regulatory proteins, including
95
Regulation of Human Retroviral Gene Expression
Rev itself. The Rev protein therefore regulates its
own expression via a negative feedback mecha-
nism. It has been proposed that Rev achieves this
effect by specifically regulating the export of
viral RNAs from the cell nucleus to the cyto-
plasm. This specificity is conferred by a cis-acting
viral RNA target sequence, the Rev response ele-
ment (RRE), which has been shown to form a
complex RNA secondary structure. Recent data
demonstrate that Rev specifically recognizes, and
binds to, a short, approximately 13-nucleotide
primary sequence within the context of the larger
RRE structure. Rev function also appears to re-
quire the subsequent binding of additional Rev
protein monomers to secondary target sites
within the RRE.
Mutational analysis of the Rev protein has
demonstrated the existence of two functional
domains. The first is a sequence-specific RNA-
binding domain required for binding to, and mul-
timerization on, the RRE, while the second is be-
lieved to interact with a currently unidentified
cellular protein that may form part of the cellular
RNA transport machinery. Mutations of this latter
domain, the Rev activation domain, give rise to
Rev proteins that act as competitive inhibitors of
the wild-type Rev trans-activator. Mutant HIV-1
proteins of this type (dominant negative mu-
tants) may have future application in the gene
therapy of HIV-1 -infected individuals. A major
focus of this laboratory is the development of
these trans-dominant Rev mutants and, in particu-
lar, the further investigation of the role of cellu-
lar proteins in the Rev response.
Finally, we have begun to expand our research
to other human retroviruses, including HIV-2 and
the apparently nonpathogenic HFV, as well as to
related animal retroviruses, such as visna virus.
The elucidation of similarities and differences in
the regulation of gene expression among these
retroviruses should facilitate the identification
and understanding of the cis- and trans-acting ele-
ments required for their replication and
pathology.
96
Mechanism of Retrovirus Infection
James M. Cunningham, M.D. — Assistant Investigator
Dr. Cunningham is also Assistant Professor of Medicine at Brigham and Women's Hospital and Harvard
Medical School. He received a B.S. degree in chemistry from the University of Michigan and an M.D. degree
from Stanford University School of Medicine. After clinical training in internal medicine (Peter Bent
Brigham Hospital) and oncology (Dana Farber Cancer Institute), he was a postdoctoral fellow
in the laboratory of Robert Weinberg at the Massachusetts Institute of Technology. Dr. Cunningham was
an HHMI Associate at Brigham and Women 's Hospital and Harvard Medical School
before assuming his current appointment.
VIRUSES are parasites. They cannot produce
progeny on their own, but must rely on the
machinery provided by the host cell to replicate
the viral genome and assemble new virus parti-
cles. Infection is initiated by attachment of the
virus to the host cell — the first step in a complex
reaction that results in transfer of the viral ge-
nome through the cell membrane and into the
cytoplasm. This attachment, or binding, is a con-
sequence of the interaction between proteins ex-
posed on the surface of the virus and the host cell
plasma membrane.
Cells that do not express a suitable virus-
binding protein, called a receptor, are not suscep-
tible to infection by a particular virus. Indeed,
the ability of many pathogenic viruses, such as
human immunodeficiency virus, poliovirus, and
certain herpesviruses, to infect specific host tis-
sues has been closely correlated with the expres-
sion of specific receptors.
Our laboratory has been interested in the mech-
anism of infection utilized by Moloney murine
leukemia virus (MoMuLV) , a member of a group
of related leukemogenic retroviruses found in vir-
tually all vertebrates. We have isolated a molecu-
lar clone, MCAT, which confers MoMuLV infectiv-
ity upon introduction into mammalian cells that
are not normally susceptible to infection. Subse-
quent experiments have demonstrated that MCAT
encodes for a membrane protein that serves as the
MoMuLV receptor. Our current research is ad-
dressed toward dissecting the molecular details
of the virus-receptor interaction that mediates in-
fection and understanding the function of the re-
ceptor in normal cell metabolism.
The MoMuLV receptor is not present in mice
for the convenience of the virus, but rather pro-
vides a portal for entry of lysine, arginine, and
ornithine, amino acids that carry a net positive
charge. There is a similarity between MCAT and
two membrane proteins in yeast that are also
amino acid transporters. This suggests conserva-
tion of a single mechanism for transport of these
amino acids over evolutionary time and predicts
that proteins similar to the MoMuLV receptor are
used by all animals. Inherited disorders of cat-
ionic amino acid transport have been described
in patients that may be explained by mutations in
MCAT genes, a hypothesis we are now examining.
A protein that is closely related to the MoMuLV
receptor has been identified in liver tissue. We
have demonstrated that this protein is also an
amino acid transporter with properties that may
help explain the specific requirements for argi-
nine metabolism by the liver: the capacity of he-
patocytes to clear the portal vein of the high con-
centration of amino acids present after a big meal
and the regulation of the urea cycle, the meta-
bolic pathway used to eliminate nitrogen waste.
Related forms of this protein are expressed in T
and B lymphocytes and macrophages that are ac-
tivated as part of the host response to bacterial
and parasitic infection. Within the past few years,
arginine has been identified as the substrate for
nitric oxide, an important mediator of the host
defense against these pathogens. Currently we
are investigating how the family of related amino
acid transporters can influence nitric oxide pro-
duction by regulating arginine availability. As
part of these studies, we have identified a molecu-
lar clone that encodes nitric oxide synthase, the
enzyme that is responsible for the synthesis of
nitric oxide from arginine. Expression of the gene
encoding this enzyme is stimulated in macro-
phages that are activated in response to infection.
Our laboratory remains interested in how retro-
viruses interact with MCAT to gain entry into the
cell. Recently we have prepared an antibody that
can recognize the MCAT protein and thereby per-
mits examination of its synthesis in infected and
uninfected cells. These studies have demon-
strated that MCAT is normally modified by the
addition of carbohydrate during its transit from
the ribosome to the plasma membrane. In cells
that have been infected with MoMuLV, we have
identified an interaction between MCAT and the
virus envelope protein that impairs MCAT matura-
tion and decreases amino acid uptake. This de-
crease may result from failure of the mature
MCAT protein to arrive at the plasma membrane.
This finding has important implications for un-
97
Mechanism of Retrovirus Infection
derstanding the host response to retrovirus infec-
tion and also for understanding virus interference
— the production by cells of retroviruses that are
resistant to additional infection by the same virus.
To investigate this problem, we are now examin-
ing the behavior of MCAT proteins that contain
mutations at the site of virus attachment. Our
long-term goal is to understand the chemical
basis of the virus-transporter interaction in suffi-
cient molecular detail to design small molecules
that can block virus binding and prevent
infection.
98
The Nuclear Pore Complex
Laura I. Davis, Ph.D. — Assistant Investigator
Dr. Davis is also Assistant Professor in the Section of Genetics and the Department of Cell Biology at Duke
University Medical Center. She received her undergraduate degree from the University of California, San
Diego, and her Ph.D. degree from the Rockefeller University, where she studied with Giinter Blobel.
Her postdoctoral work was done with Gerald Fink at the Whitehead Institute.
ONE of the hallmarks of eukaryotic cells is the
presence of membranous barriers that di-
vide the cell into functional compartments called
organelles. How these compartments are gener-
ated and how traffic through them is controlled
are matters of intense interest in cell biology. Our
focus is on the nuclear envelope, which separates
the genomic material from the rest of the cell.
Both RNAs and proteins move across the enve-
lope, between the nucleus and cytoplasm, and
the populations found in the two compartments
are very different.
For example, newly synthesized mRNA must be
extensively edited or processed before leaving
the nucleus. The mechanism that restricts its ex-
port until processing is complete remains un-
clear. Similarly, only certain proteins are im-
ported into the nucleus. These contain specific
signals that must be recognized by some compo-
nent of the import apparatus. Recently it has be-
come clear that the cell can regulate the availabil-
ity of these signals by changing the context in
which they are found. For example, transcrip-
tional activators that control the growth state of
the cell by turning on gene expression can be
held in an inactive form in the cytosol if their
signals are hidden from the import apparatus.
When the cell is stimulated to begin growth, the
signals are uncovered, allowing the transcrip-
tional activators to enter the nucleus and exert
their function. Thus nucleocytoplasmic transport
has an important regulatory role in controlling
gene expression.
It is thought that all nucleocytoplasmic trans-
port proceeds through large proteinaceous chan-
nels that perforate the nuclear envelope. Called
nuclear pore complexes (NPCs), these structures
are about 30 times as large as an ion channel and
are probably composed of over 200 different
polypeptides. Among the few known compo-
nents of the NPC are members of a family of re-
lated proteins called nucleoporins. These pro-
teins have been localized to an iris-like structure
in the middle of the NPC, called the central trans-
porter, and are thought to play an essential role in
mediating protein (and perhaps RNA) transport.
They are required for binding of proteins to the
NPC prior to nuclear import, and antibodies that
bind to the nucleoporins can block the energy-
dependent movement of proteins through the
NPC. While it is possible that the nucleoporins
themselves recognize the signals on proteins des-
tined for the nucleus, most of the available evi-
dence suggests that they are more likely to act as
docking points for cytosolic receptors that actu-
ally recognize the signals.
Our goal has been to understand more about
the function of the nucleoporins and to identify
other components of the transport apparatus. We
use the budding yeast Saccharomyces cerevi-
siae, since it is amenable to both genetic and bio-
chemical analysis. Using antibodies that recog-
nize the mammalian nucleoporins, I previously
identified homologues of these proteins in yeast
and cloned the gene encoding one of them
{NUPl). The genes encoding two others have
now been isolated (NUP2 and NSPl) . The pro-
tein encoded by NUPl is essential for viability,
and we have isolated nupl mutants that confer
growth at one temperature but not another. These
conditional mutants have been used to assay the
phenotype resulting from loss of NUPl function.
Using immunofluorescence, we have found that
protein import into the nucleus is severely inhib-
ited at the nonpermissive temperature, providing
in vivo evidence that NUPlp is required for pro-
tein transport.
Conditional mutants will also be useful for
identifying proteins that functionally interact
with NUPlp. To do this, we are using genetic
screens to find mutations in new genes that either
enhance or suppress the conditional phenotype
of the nupl mutants. The assumption is that mu-
tations in proteins that functionally interact with
NUPlp will either exacerbate the defect in nupl
mutants and lead to death at all temperatures, or
will overcome the defect and allow grouth at the
restrictive temperature. Using this approach, we
hope to identify new NPC components, as well as
cytoplasmic and nuclear proteins that may inter-
act transiently with the nucleoporins in a func-
tional manner.
We have also taken a biochemical approach to
99
The Nuclear Pore Complex
identifying proteins that interact with NUPlp.
Using a specific antibody, we isolated the NUPl
protein from cell extracts and have found it to be
stably associated with several other proteins.
Some of these appear to be nucleoporins, while
others are as yet unidentified. We are now purify-
ing the individual members of the complex, in
order to obtain protein sequences that we can use
to isolate the genes encoding them. We expect
some of these proteins to correspond with those
we identify through genetic interactions with
NUPl.
100
Molecular Approaches to T Lymphocyte
Recognition and Differentiation
Mark M. Davis, Ph.D. — Investigator
Dr. Davis is also Professor of Microbiology and Immunology at Stanford University School of Medicine. He
received his B.A. degree from the Johns Hopkins University and his Ph.D. degree in molecular biology from
the California Institute of Technology. He held positions at NIH as a postdoctoral fellow and Staff Fellow
before joining the staff at Stanford. He is the recipient of the Eli Lilly award in Microbiology
and Immunology and a Gairdner Foundation award.
WE have focused on several major areas in
immunology that generally resolve into
two questions: How do T cells recognize foreign
entities? How is lymphocyte differentiation con-
trolled, both in the thymus and in the periphery?
An additional goal is to refine and better integrate
recombinant DNA technology with other power-
ful techniques in immunology, as an approach to
defining the function of unknown genes or
poorly understood genes and their products.
Topology of T Cell Recognition
The work of many investigators over the years
has shown that T cells, through their antigen re-
ceptor molecules, recognize fragments of foreign
proteins (peptides) embedded in major histo-
compatibility complex (MHC) molecules. In
contrast, antibodies, although closely related to T
cell antigen receptors (TCRs) , bind intact foreign
particles directly. This suggests fundamental dif-
ferences in the rules governing T cell recognition
versus antibody-mediated recognition by B cells.
Because of the consistently high concentration of
sequence diversity in the V-J junctional region of
TCRs (equivalent to the third complementation-
determining region [CDR3] of immunoglobulins)
as well as structural considerations, we have pro-
posed that this is the important region for peptide
recognition and that other V region-encoded resi-
dues might contact the surface of the MHC
molecule.
To test this hypothesis, we recently developed
an immunological version of the classical genetic
technique of second-site suppression. In these
experiments we change residues that are impor-
tant for T cell recognition (and not MHC interac-
tion) on a peptide, immunize mice, and then ana-
lyze the responding T cells that emerge with
respect to their TCR sequences. To hold part of
the original receptor constant, we use mice that
are transgenic for either chain of the original
TCR. With this approach, we have shown that one
of the two residues on the peptide that are impor-
tant in the T cell response is governed by the
CDR3 of the Va polypeptide, and that the other
(three amino acids downstream on the peptide
sequence) is specified by the CDR3 of the VjS
polypeptide. Thus we have generated significant
support for the original hypothesis. The results
indicate that T cell recognition is a much more
stylized event than antibody-antigen interactions.
Kinetics of T Cell Recognition
and Activation
To learn more about the dynamics of TCR-
peptide-MHC interactions, we have developed
expression systems that allow us to produce ei-
ther TCR or MHC class II heterodimers in a solu-
ble form. This involves replacing the normal
membrane-spanning sequences of these polypep-
tides with a signal sequence for lipid linkage,
such as employed normally by a number of cell
surface proteins. Molecules expressed in this fash-
ion can then be conveniently cleaved from the
surface of expressing cells with the enzyme phos-
phatidylinositol-specific phospholipase C. By
utilizing high-density mammalian cell culture
machines, we are able to make milligram quanti-
ties of a soluble TCR and its cognate MHC mole-
cule. This has provided us with the raw material
to initiate structural studies such as x-ray crystal-
lography and nuclear magnetic resonance analy-
sis. We have also used the soluble MHC protein to
show greatly enhanced uptake of antigenic pep-
tides at low pH. This is important both in making
significant quantities of a pure antigen-MHC
complex and in understanding the biology of this
type of MHC molecule (class II) that recycles
through low-pH endosomal corripartments. Our
current data suggest that the low pH triggers a
specific conformational change in the MHC mole-
cule, which allows it to bind new peptides more
easily.
Recently we have studied the kinetics of TCR
peptide-MHC interactions, using soluble MHC-
peptide complexes that competitively inhibit
binding of labeled anti-TCR antibody fragments
to TCRs. In several cases studied, we derive
values for this interaction of 5 X 10~^ M, 1,000-
to 10,000-fold weaker than antibodies to protein
ligands of comparable size. These values are con-
sistent with the scanning nature of T cell recogni-
101
Molecular Approaches to T Lymphocyte Recognition and Differentiation
tion and indicate that TCR binding to peptide-
MHC ligands is so weak energetically that other,
antigen-independent receptor-Iigand systems must
govern the initial stages of T cell contact with
antigen-presenting or target cells. The best candi-
dates for such regulators of T cell interaction are
adhesion molecules, such as LFA-1 (lymphocyte
function antigen 1) or CD2. This would give ad-
hesion molecule-ligand interactions a major role
in orchestrating which T cells interact with
which antigen-presenting cells. This has impor-
tant implications for autoimmunity: normally T
cells would be expected to focus on appropriate
antigen-presenting cells (such as B cells or macro-
phages) and would be hindered in surveying
most other cells or tissues (which would lack the
appropriate ligand expression) .
Generation of Memory T Cells in Vitro
A hallmark of the vertebrate immune system is
its ability to respond much more strongly and
quickly to a second encounter with an antigen.
Although some of this effect is due to the clonal
expansion of B and T cells, a significant, and per-
haps the major, contribution is thought to involve
the induction of antigen-specific memory lym-
phocytes. Evidence suggests that this differen-
tiation step is crucial to mounting a successful
immune response, both in appropriate and inap-
propriate (e.g., autoimmune) circumstances.
Studies of B or T lymphocyte memory have been
hampered by the lack of well-defined systems in
which uniform populations of primary and mem-
ory cells can be studied. We have approached this
problem in T lymphocytes by making use of TCR
a/3 transgenic mice that have an essentially mono-
clonal T cell immune system, thereby eliminating
the clonal expansion component of T cell mem-
ory and allowing us to focus on the physiological
changes that characterize this transition.
We find that primary T cells can only produce
interleukin-2 (IL-2) in response to TCR stimula-
tion with immobilized peptide-MHC complexes
if PMA (phorbol 12-myristate 12-acetate) is in-
cluded in the culture. In contrast, secondary, or
memory, T cells, derived from the identical start-
ing population, readily produce lymphokine on
this substrate, irrespective of PMA addition. Both
T cell populations are inhibited by protein kinase
C (PKC) inhibitors H7 and staurosporine. Thus
secondary T cells are functionally distinct from
primary cells; the establishment of T cell mem-
ory, which may include alterations in PKC signal-
ing, followed by the addition of exogenous IL-2,
is sufficient to convert primary T cells to second-
ary cells in vitro. This provides an excellent op-
portunity to characterize and manipulate this im-
portant juncture in T cell differentiation.
102
Signal Transduction by the Epidermal Growth
Factor Receptor
Roger J. Davis, Ph.D. — Assistant Investigator
Dr. Davis is also Associate Professor in the Program in Molecular Medicine and the Department
of Biochemistry and Molecular Biology at the University of Massachusetts Medical School. He received
his undergraduate and graduate education at Cambridge University and was a postdoctoral fellow
with Michael Czech at the University of Massachusetts.
CELLULAR proliferation is a highly regulated
process. During embryonic development,
rapid cell growth is required to form the tissues
of the body. In contrast, cellular proliferation in
adults is slow, primarily serving to replace senes-
cent cells. Adults, however, retain a limited ca-
pacity for rapid growth — for example, during
wound healing. Regulation of this proliferative
capacity is critically important. Errors in growth
control can result in a variety of diseases, includ-
ing cancer.
The local production of protein growth factors
is an important mechanism that can account for
the control of cellular proliferation. Our research
group is investigating the action of a family of
peptides that includes epidermal growth factor
(EGF) and transforming growth factor-a (TGF-
a). These agents are synthesized as cell surface
glycoproteins that are split to release small solu-
ble peptides. Both the membrane-bound precur-
sor and the diffusible peptides are biologically
active and bind to specific receptor molecules
located at the surface of responsive cells. Secre-
tion of these peptide growth factors contributes
to the rapid growth of some tumors.
The long-term goal of this laboratory is to un-
derstand the molecular basis for the control of
cellular proliferation by the EGF receptor. It is
known that the binding of growth factors to this
receptor at the cell surface triggers a complex
series of chemical reactions that lead to DNA syn-
thesis within the nucleus and to cell division.
However, the molecular details of the signaling
pathways utilized by the receptor are poorly
understood.
Regulation of EGF Receptor Function
The EGF receptor is a glycoprotein consisting
of an extracellular domain that binds growth fac-
tor, a membrane-spanning domain, and a cyto-
plasmic domain. The cytoplasmic domain is an
enzyme, tyrosine kinase, that causes the covalent
attachment of phosphate to tyrosine components
of substrate proteins (phosphorylation). The
binding of EGF to the receptor's extracellular do-
main causes an increase in the tyrosine kinase ac-
tivity of the cytoplasmic domain. EGF also causes
the receptor to aggregate and to associate tran-
siently with intracellular regulatory molecules to
form a signal transduction complex. We are study-
ing these interactions and investigating the con-
sequences of the phosphorylation process.
The ability of EGF to increase the tyrosine ki-
nase activity of its receptor is blocked when cells
are incubated with a tumor promoter or with
other growth factors. Under these conditions, the
EGF receptor is itself phosphorylated at multiple
serine and threonine residues. We are investigat-
ing the significance of this phosphorylation. Our
approach is to construct receptors with point
mutations at the sites of phosphorylation, using
recombinant DNA technology. These studies have
demonstrated that the phosphorylation of a single
threonine residue blocks the ability of EGF to
stimulate the receptor's tyrosine kinase activity.
Phosphorylation also alters the internalization of
the receptor. We are investigating the structural
basis for the effects of phosphorylation on the
regulation of EGF receptor function. These stud-
ies are supported by a grant from the National
Institutes of Health.
Signaling by the EGF Receptor
A principal question that we must answer to
understand the mechanism of signal transduction
by the EGF receptor is how a signal that is initi-
ated at the cell surface can be transmitted to the
nucleus to cause DNA replication. One class of
regulatory molecules that could account for this
process is the protein kinases. We are focusing
our research on one subclass of these enzymes
that cause the phosphorylation of proline-rich
target sequences in substrate proteins. Examples
are represented by the mitogen-activated protein
kinases and the cyclin-dependent protein ki-
nases. Each of these types of protein kinases exists
in multiple forms as part of an extended family of
enzymes that are regulated by growth factors and
by the cell cycle. We are investigating the struc-
ture of additional members of this family by mo-
lecular cloning.
Substrates for these growth factor-regulated
103
Signal Transduction by the Epidermal Growth Factor Receptor
protein kinases that we have identified include
the EGF receptor and the nuclear transcription
factors expressed by the proto-oncogenes c-myc
and c-jun. We are investigating the mechanism
and significance of the phosphorylation of these
proteins in EGF-treated cells. The overall goal of
these studies is to establish the molecular details
of a signal transduction pathway that is initiated
at the EGF receptor and leads to the regulation of
nuclear function.
Tissue Specificity of Tumor Induction
The gene for the EGF receptor is a frequent site
of integration by avian leukosis viruses. Insertion
of the virus causes the expression of a truncated
EGF receptor. The formation of a virus containing
a copy of the truncated receptor gene can also
occur. This truncated gene is the dominantly ac-
tive oncogene erbB. The primary disease asso-
ciated with erbB is erythroblastosis. However,
mutations in the carboxyl terminus of erbB that
occur during viral replication cause additional
tumors — fibrosarcomas and angiosarcomas. We
are studying the molecular basis for the change in
tissue specificity of the erbB oncogene.
The approach we are taking is to construct re-
combinant viruses containing erbB. The advan-
tage of this procedure is that we can undertake a
systematic analysis of the effects of erbB muta-
tions on the tissue specificity of tumor formation.
The results should help explain why erbB causes
tumors in one tissue but not in others. This infor-
mation will be useful in designing strategies for
clinical intervention in tumor development.
A hippocampal neuron in primary culture, showing axosomatic and axo-
dendritic synapses at the surface. The perikaryon and the dendrites of the
neuron are visualized by immunostain (green fluorescence) for MAP2, a
protein associated with microtubules of perikarya and dendrites. Presynap-
tic nerve terminals originating from neurons not visible in the field are
visualized by immunostain (orange-yellow fluorescence) for the synaptic
vesicle protein synaptotagmin.
From the work of Michela Malleoli, Gary Banker, Thomas Siidhof and
Pietro De Camilli.
104
Traffic of Synaptic Vesicle Proteins in Neurons
and Endocrine Cells
Pietro De Camilli, M.D. — Investigator
Dr. De Camilli is also Professor of Cell Biology at Yale University School of Medicine. He received his M.D.
degree from the University of Milano, Italy, where he worked with Jacopo Meldolesi. He did his
postdoctoral studies with Paul Greengard at Yale University. Prior to his current position, he held
appointments both at Yale and the University of Milano. Dr. De Camilli is a recipient of a McKnight
Scholars Award and a member of the European Molecular Biology Organization (EMBO ).
WE are interested in the mechanisms of sig-
naling between nerve cells, with emphasis
on neuronal secretion of neurotransmitter mole-
cules. Most neurotransmitters are stored in vesi-
cles within nerve terminals and are secreted in
response to depolarization of the terminal by fu-
sion of the vesicle with the plasmalemma, a pro-
cess called exocytosis. Most, and possibly all,
nerve terminals contain two classes of secretory
vesicles: synaptic vesicles (SVs) and the so-called
large dense-core vesicles (LDCVs), which differ
in a variety of properties, including neurotrans-
mitter content.
SVs are small, morphologically homogeneous
vesicles (50-nm diameter), containing nonpep-
tide neurotransmitters only. They are very abun-
dant in nerve terminals, and clusters of SVs ap-
posed to the plasmalemma represent a typical
structural feature of axon endings. Their exocyto-
sis takes place selectively at the presynaptic plas-
malemma, is faithfully linked to nerve terminal
depolarization, and plays the dominant role in
the fast, point-to-point intercellular signaling typ-
ical of the nervous system (the so-called synap-
tic transmission). SV membranes are rapidly re-
internalized after exocytosis and are used to
re-form SVs, which are loaded locally with
neurotransmitters .
LDCVs are larger vesicles with a dense protein-
rich core. Their contents are predominantly pep-
tide neurotransmitters, though they may also
contain amines. Their exocytosis is preferentially
triggered by trains of action potentials and is in-
volved primarily in a modulatory type of signal-
ing. They have long been recognized as the neuro-
nal organelles equivalent to secretory granules of
endocrine cells, which secrete peptide hormones
and amines. In contrast, many of the properties of
SVs set them apart from other secretory organ-
elles. The elucidation of these properties repre-
sents the main focus of our laboratory. We are
following three major lines of research. (Several
aspects of this research have previously been sup-
ported by the National Institutes of Health.)
Biogenesis and Traffic of Synaptic Vesicles
First, we are investigating the biogenesis and
traffic of SVs, using in vivo and in vitro systems.
The large body of information recently accumu-
lated on the molecular structure of SVs has made
available numerous molecular probes that can be
used in these studies (some of which are carried
out in collaboration with Reinhard Jahn [HHMI,
Yale University] and Thomas Siidhof [HHMI, Uni-
versity of Texas Southwestern Medical Center at
Dallas]). We have developed an immunocyto-
chemical assay to detect SV exocytosis. Using this
assay on hippocampal neurons in primary cul-
ture, we have shown that SVs undergo a high rate
of constitutive exocytosis and recycling in devel-
oping neurons before synaptic contacts are
formed.
These findings suggest that synapse formation
coincides with a reorganization of the exocytotic
machinery (rather than with its de novo assem-
bly) and that signaling via SV exocytosis may play
an important role in nervous system development
prior to synapse formation. We are working at de-
veloping assays, some cell free, to define specific
steps of the exo-endocytotic pathway of SVs. Such
assays aid greatly in elucidating the underlying
molecular mechanisms.
Synapse-like Microvesicles
in Endocrine Cells
Second, we are exploring the relationship of
SVs to organelles found in nonneuronal cells. Do
SVs represent the adaptation of recycling vesicles
to organelles found in other cells? How is the
exo-endocytotic recycling of SVs related to the
plasmalemma-endosome recycling that operates
in all cells and occurs, for example, in the recy-
cling of receptors?
We and others have found that a variety of SV
proteins are expressed at significant levels by
peptide-secreting endocrine cells. Investigating
the subcellular localization of these proteins, we
have found that at least some of them (synapto-
physin, protein p29) are concentrated on a popu-
lation of microvesicles (synapse-like microvesi-
cles, SLMVs) distinct from the secretory granules
that store and secrete peptide hormones.
A first biochemical characterization of these
105
Traffic of Synaptic Vesicle Proteins in Neurons and Endocrine Cells
microvesicles has demonstrated that they are simi-
lar in composition to bona fide neuronal SVs.
Consistent with these findings, SLMVs, like SVs,
undergo exocytosis, endocytosis, and recycling.
We have therefore explored the possibility that
SLMVs may share with SV the property to store
and secrete nonpeptide neurotransmitters. The
model endocrine cell we have used for these stud-
ies is the insulin-secreting cell of the endocrine
pancreas, also called pancreatic |8-cell.
Previous work had suggested that the nonpep-
tide neurotransmitter GABA, the major inhibitory
neurotransmitter in the brain, may play a para-
crine role in the physiology of the endocrine pan-
creas. GABA was reported to be concentrated in
pancreatic jS-cells and to act on surrounding pan-
creatic endocrine cells. Work we have carried out
so far has suggested that SLMVs of /3-cells may in-
deed store GABA. These findings support the hy-
pothesis that the neuronal secretory pathway in-
volving SVs represents an amplification and an
adaptation of a secretory pathway that is ex-
pressed in a more rudimentary form in peptide-
secreting endocrine cells.
Such results, in addition to having important
implications for the field of endocrine physiol-
ogy, raise the possibility of using endocrine cells
to answer questions on the life cycle of SVs that
would not be feasible to address in neurons. Us-
ing endocrine cell lines as a model system, we
have obtained evidence indicating that the bio-
genesis of SLMVs (and presumably of SVs) is ef-
fected by protein sorting from endosomal
membranes.
Molecular Mechanisms of Exocytosis
Third, we are interested in elucidating the mo-
lecular mechanisms of SV exocytosis. The ap-
proach we are currently taking is based on the
assumption that at least some basic feature of the
exocytotic process may have been conserved in
evolution. We are exploring the potential of yeast
genetics to identify some of the proteins in-
volved. A variety of temperature-sensitive yeast
mutants defective in vesicle exocytosis {sec mu-
tants), and therefore in growth, have been iso-
lated in recent years. We are currently searching
for gene products that rescue sec mutations in a
functional assay, by transfecting yeast sec mutants
with mammalian brain cDNAs.
Stiflf-Man Syndrome and Diabetes
The presence of SV-like organelles in endocrine
cells suggests that pathological processes affecting
SV proteins (mutations, autoimmunity) might af-
fect both the nervous and endocrine systems.
These considerations led us to identify a possible
link between stiff-man syndrome and insulin-
dependent diabetes mellitus (IDDM). Stiff-man
syndrome is a rare and severe disease of the central
nervous system characterized by a progressive ri-
gidity of the body musculature. In a series of stud-
ies that involved a collaboration with the Univer-
sity of Milano, we have obtained evidence for an
autoimmune pathogenesis of stiff-man syndrome
and identified the GABA-synthesizing enzyme,
GAD, as the major autoantigen.
GAD is concentrated at the cytoplasmic surface
of SVs of GABAergic neurons and of SLMVs of pan-
creatic jS-cells. IDDM, which results from an au-
toimmune destruction of (S-cells, was found to
occur at high frequency in stiff-man syndrome
patients. This observation then led us to hypothe-
size an identity between GAD and the 64-kDa
autoantigen of IDDM previously identified by
Steinunn Baekkeskov, Ake Lemmark, and their co-
workers (Hagedorn Research Laboratory, Den-
mark). A collaboration between our group and
that of Dr. Baekkeskov (now at the University of
California, San Francisco) confirmed that the 64-
kDa antigen is GAD.
These findings may lead to a diagnostic assay
for IDDM and for preclinical stages of the disease.
In addition, it opens the possibility of investigat-
ing molecular mechanisms of autoimmunity in
IDDM.
During the next year the work on SV exocytosis
and on GAD autoimmunity in stiff-man syndrome
will be partially supported by the National Insti-
tutes of Health.
106
Three-Dimensional Structures of Biological
Macromolecules
Johann Deisenhofer, Ph.D. — Investigator
Dr. Deisenhofer is also Regental Professor and Professor of Biochemistry, and he holds the Virginia and
Edward Linthicum Distinguished Chair in Biomolecular Science at the University of Texas Southwestern
Medical Center at Dallas. He was born and educated in Germany. His Ph.D. research in protein
crystallography was done at the Max Planck Institute for Biochemistry, Martinsried, and at the Technical
University of Munich. As a postdoctoral fellow and as a staff scientist at the Max Planck Institute, he
continued his structural analysis of biological macromolecules by x-ray crystallography. He has received
several honors for his structure analysis of a photosynthetic reaction center, including the 1988 Nobel
Prize in chemistry, which he shared with Hartmut Michel and Robert Huber.
MY colleagues and I study the three-
dimensional structures of proteins to un-
derstand their folding, structural stability, and
function. We are particularly interested in
protein-pigment complexes catalyzing photo-
chemical energy conversion, energy transfer, and
electron transfer, and in membrane-spanning and
membrane-associated proteins.
Cytochrome b/ Cj complex
The b/Cj complexes (also called ubiquinol-
cytochrome c oxidoreductases) are integral
membrane proteins that play crucial roles in pho-
tosynthesis and cell respiration. Their function in
these fundamental processes is to oxidize quinols
to quinones and to transfer electrons and protons
through the membrane. The electrons go to cy-
tochrome c, and the protons build up an electro-
chemical gradient across the membrane, which,
for example, drives the synthesis of ATP.
Photosynthetic purple bacteria have the sim-
plest b/c, complexes, consisting of only three or
four different protein subunits with three heme
groups and an iron-sulfur cluster as cofactors.
Photosynthetic reaction centers and b/c, com-
plexes cooperate in the bacterial inner mem-
brane; b/Ci complexes occur at a concentration
significantly lower than that of reaction centers
and are therefore more difficult to isolate. David
Knafif's research group (Texas Technical Univer-
sity, Lubbock) succeeded in purifying the b/Cj
complex from Rhodospirillum ruhrum. In col-
laboration with this group we have prepared this
complex in milligram quantities with high pu-
rity. This preparation serves as the starting mate-
rial for crystallization experiments.
Mitochondria of higher organisms have in their
inner membranes b/Cj complexes consisting of at
least 10 different polypeptide chains, 3 of which
are similar to those in the b/Cj complex of purple
bacteria. We collaborate with Chang-An Yu and
his colleagues (Oklahoma State University, Still-
water), who produced large crystals of the b/Cj
complex from beef heart mitochondria. Prelimi-
nary diffraction experiments at the Cornell High
Energy Synchrotron Source (CHESS) were encour-
aging, and we hope to begin a structure analysis at
medium resolution (approximately 5 A) in the
near future.
DNA Photolyase
Light energy plays an important role in reac-
tions other than photosynthesis. The molecular
machinery that enables cells to repair DNA dam-
aged by ultraviolet (UV) light includes an en-
zyme that uses light energy to drive the repair of
one frequent type of damage caused by UV irra-
diation: the crosslinking of two neighboring thy-
mine bases in a strand of DNA. Most of these
crosslinks are in the form of cyclobutane rings
connecting four carbon atoms, two from each
thymine base. The enzyme DNA photolyase can
locate and bind to such defects, and upon input
of light of suitable wavelength (300-500 nm)
cleave the carbon-carbon bonds between the
bases, thus repairing the damage.
DNA photolyase has been found in prokaryotes,
eukaryotes, and archaebacteria. Aziz Sancar and
his colleagues (University of North Carolina at
Chapel Hill) sequenced, overexpressed, and puri-
fied the enzyme from Escherichia coll. It consists
of a single polypeptide chain of 471 amino acids
and two cofactors — a flavin-adenine dinucleo-
tide (FAD) and 5, 10-methenyltetrahydrofolate.
The FAD cofactor fully reduced to FADH2 is es-
sential for the enzyme's function; the folate acts
as a light-harvesting antenna.
Despite significant problems with the en-
zyme's tendency to denature, we were able to
crystallize DNA photolyase from E. coli in two
crystal forms; both forms difi'ract x-rays to at least
2.8-A resolution. We collected x-ray diffraction
intensity data from one of these crystal forms; to
solve the phase problem, experiments are under
way to bind heavy-atom compounds to the pro-
tein in the crystal. We are also trying to crystallize
the enzyme in complex with a substrate, a five-
nucleotide piece of single-stranded DNA contain-
107
Three-Dimensional Structures of Biological Macromolecules
ing a crosslinked pair of thymine bases. These ex-
periments have to be done in the dark or under
yellow or red light, to prevent the enzyme from
repairing and releasing the substrate.
Because light is an essential ingredient of the
enzymatic reaction of DNA photolyase, the en-
zyme-substrate complex will be a suitable system
to study the time course of the reaction, using the
Laue technique. This technique employs the
broad spectrum of x-ray wavelengths in the
synchrotron's powerful beam to record within a
very short time a sufficient fraction of a crystal's
dilfraction pattern for structural interpretation.
In such an experiment the crystal is irradiated
with a light pulse suitable to trigger the enzy-
matic reaction, and diffraction patterns are re-
corded at different times after the pulse. These
data can provide snapshots of the structural rear-
rangements during the reaction and thus contrib-
ute to an understanding of the enzyme's
mechanism.
Cytochrome P-450 Enzymes
P-450 enzymes are members of a superfamily
of b-type heme proteins that catalyze hydroxyla-
tion of organic substrates using electrons from
reduced nicotinamide adenine dinucleotide
phosphate (NADPH) and molecular oxygen.
They consist of a polypeptide chain of about 420
amino acids and a heme group as part of their
active site. P-450s occur in organisms as different
as bacteria and humans. Some of them play spe-
cific roles in biosynthetic pathways, such as for-
mation of glucocorticoids, androgens, and estro-
gens from cholesterol; others are involved in
detoxification of foreign substances by making
them more soluble and so facilitating their excre-
tion. However, in some cases hydroxylation of
foreign substances creates highly efficient carcin-
ogens; thus the effects of P-450s are not always
beneficial.
P-450s occur in two classes, distinguished by
the way in which electrons from NADPH are
transferred to the heme group. Class I P-450s re-
ceive the electrons via a flavoprotein and an iron-
sulfur protein; class H enzymes are reduced di-
rectly by a protein containing the cofactors flavin
mononucleotide and FAD. The three-dimensional
structure of only one class I P-450, the camphor
hydroxylase P-450cam' is known at atomic
resolution.
To learn more about the structural variation
within the cytochrome P-450 family, about dif-
ferences between class I and class II enzymes, and
about the spectrum of reactions catalyzed by
P-450s, we collaborate with Julian Peterson and
his colleagues (University of Texas Southwestern
Medical Center at Dallas) to analyze the three-
dimensional structures of a class I and a class II
enzyme. The class I P-450 was isolated from
Pseudomonas; it hydroxylates the hydrocarbon
a-terpineol and was therefore named P-450,p^.
The class II P-450 is the cytochrome part of an
enzyme from Bacillus megaterium that is
unique in combining a P-450 domain and a P-450
reductase domain in a single polypeptide chain
of 1,066 amino acids. The cytochrome domain,
P-450bm-3. consists of the 471 amino-terminal
amino acids; it was cloned and expressed at high
levels in E. coli. Its functions include hydroxyla-
tion and epoxidation of long-chain fatty acids.
Well-ordered crystals of both P-450s have been
grown. At CHESS, crystals of P-450t^rp diffracted
to about 2.0 -A resolution, and crystals of
P-450bm-3 diffracted to better than 1.5 -A resolu-
tion. Molecular replacement calculations, using
the P-450cam model, produced a plausible solu-
tion for the crystal structure of P-450terp- Multiple
isomorphous replacement with three heavy-atom
derivatives allowed the calculation of an electron
density map at 3-A resolution for P-450bm.3; this
map is currently being interpreted. Model build-
ing and crystallographic refinement will result in
accurate models of both proteins in the near fu-
ture. We also plan to study substrate binding and
structures of site-specific mutants.
Other Projects
In addition to the projects described above, we
also work on the determination of the three-
dimensional structures of several other proteins,
including the catalytic domain of human HMG-
CoA reductase, a key enzyme in the synthesis of
cholesterol and a likely target for drugs; human
synapsin I, a protein binding to synaptic vesicles
and mediating their release; the small GDP-
binding protein smgp25A from bovine brain; the
DNA-binding protein myogenin; the SecA protein
from E. coli, one of the key parts of the protein
export system; mammalian phosphofructokinase;
and ribonuclease inhibitor from pig liver.
The work on cytochrome b/Cj complexes and
on DNA photolyase is also supported by a grant
from the Welch Foundation.
108
Molecular Mechanisms of Lymphocyte
Differentiation
Stephen V. Desiderio, M.D., Ph.D. — Associate Investigator
Dr. Desiderio is also Associate Professor of Molecular Biology and Genetics at the Johns Hopkins University
School of Medicine. He received his undergraduate degree in Biology and in Russian from Haverford
College and his M.D. and Ph.D. degrees from the Johns Hopkins University School of Medicine in
biochemistry, and cellular and molecular biology. After a postdoctoral fellowship at the Massachusetts
Institute of Technology with David Baltimore, he returned to Johns Hopkins.
ONE remarkable feature of the immune sys-
tem is its ability to recognize and respond to
an extraordinary variety of foreign molecules, or
antigens. This exquisite specificity is achieved
through protein receptors, which bind tightly to
specific antigens. Such receptors are found on
the surfaces of two types of immune cells: B and T
cells.
One type of antigen receptor is the antibody
molecule, or immunoglobulin. The site on the
antibody molecule that binds to a specific anti-
gen is genetically encoded by multiple short seg-
ments of DNA. At the onset of development of the
immune system, these DNA segments are located
at separate places in the genome, but during the
maturation of antibody-producing cells (B cells),
segments are joined to form intact immunoglobu-
lin genes.
In addition to antibody, another class of anti-
gen receptor is found on the surfaces of cells that
mediate cellular immunity (T cells). The T cell
receptor's antigen-binding site, like that of the
antibody molecule, is encoded by multiple DNA
segments that are brought together during T cell
maturation.
After their genes are assembled, the antigen re-
ceptor molecules are expressed at the surfaces of
B and T cells. Here, several interactions combine
to trigger cell division and further maturation.
These include the binding of antigens and special-
ized hormones called lymphokines with their re-
spective receptors.
Lymphocyte Activation
A major goal of our laboratory is to find out how
specific antigens and growth factors trigger the
activation of B and T cells. Biochemical evidence
has long suggested that enzymes called tyrosine
kinases might be intimately involved. These en-
zymes regulate the activities of other proteins,
and of each other, by adding regulatory chemical
groups, phosphates, to specific sites. The kinases
involved in B and T cell activation have, for the
most part, proved elusive. We therefore set out to
find new tyrosine kinase genes that are expressed
in B or T cells, and we identified seven. Two are
expressed only in immune cells, and we have
concentrated on those.
One of the genes resembles a growth-
regulating gene called c-src. Unlike c-src, how-
ever, which is turned on in many different kinds
of cells, this new gene is only turned on in B cells
and their developmental precursors. Accord-
ingly, we call the gene blk, for B lymphoid ki-
nase. The blk gene is activated early in B cell de-
velopment and is expressed along with several
proteins that are known to transmit signals across
the B cell membrane, including the antibody
heavy chain.
When B cells develop into antibody-secreting
or plasma cells, they no longer express these sig-
nal transmission proteins or blk. This pattern sug-
gests that the product of the blk gene interacts
with a receptor that spans the B cell membrane
and that senses the presence of antigen or a spe-
cific growth signal. In the spleen, the protein en-
coded by blk is found specifically in those loca-
tions where resting B cells encounter antigen,
reinforcing the idea that this kinase functions in
the triggering of the B cell's immune response.
How might extracellular signals activate the
blk kinase? The kinase is inactivated by addition
of a phosphate group to a specific site, and is
activated when the phosphate is removed. If the
target site on the blk kinase is mutated so that it
cannot accept the inactivating chemical group,
the enzyme is locked into the "on" state. Cells
that contain this "on" version of the blk kinase
grow in an unregulated way. Having an activated
version of the blk kinase has made it easier to
identify its targets. We have found one target to
be an enzyme called phospholipase C-7 (PLC-7),
a key component of the B cell activation pathway.
Evidence from other systems suggests that PLC-7
is activated by tyrosine kinases. We speculate that
activation of the blk kinase in B cells activates
PLC-7, initiating subsequent signaling events.
The hormone interleukin-2 (IL-2) plays a cen-
tral role in the immune response by inducing the
multiplication of T cells. The binding of IL-2 to
its receptor turns on a tyrosine kinase of unknown
identity. We have recently identified six new tyro-
sine kinase genes that are expressed in cells de-
109
Molecular Mechanisms of Lymphocyte Differentiation
pendent on IL-2 for growth. One of these genes,
which we call itk (for IL-2-inducible T cell ki-
nase), is expressed specifically in T cells. The itk
gene encodes a tyrosine kinase that resembles src
kinases, but from its sequence we predict that it
will differ from those kinases with respect to its
location in the cell and the regulation of its activ-
ity. IL-2 is known to regulate expression of a num-
ber of genes, including a gene for its own recep-
tor. In cells that have been withdrawn from IL-2,
itk is expressed at low levels. Soon after addition
of IL-2, itk is strongly turned on in parallel with
the IL-2 receptor gene, suggesting that itk func-
tions in the IL-2 response.
Antibody Gene Rearrangement
The total number of immunoglobulin or T cell
receptor gene segments is large, but when any
particular immunoglobulin or T cell receptor
gene is assembled, only a handful of segments are
selected and joined. As a result, many different
combinations of segments are possible. It is this
shuffling of small bits of DNA that generates
much of the diversity of the immune response.
By using artificial DNA molecules that rear-
range after they are introduced into immature B
cells, we have been able to outline the general
features of antibody gene assembly, but the mech-
anism at the level of interactions between mole-
cules is not known. The products of at least three
genes are known to play an intimate role in rear-
rangement. Two of these genes, RAG-1 and RAG-
2, have been shown by others to activate some-
how antibody gene rearrangement. A third gene
encodes a protein that all cells use to help repair
DNA damage and that also functions at a late step
in antibody gene assembly. In addition to the
products of these genes, our laboratory has found
a protein — NBP — that binds specifically to a
DNA sequence element required for rearrange-
ment, suggesting that it may be a part of the re-
combinational machinery. Members of our labora-
tory are studying the products of the RAG genes
and NBP, with the aim of understanding their
functions and interrelationships.
Since NBP was isolated from calf thymus, we
cloned the RAG-1 and RAG-2 genes of the cow
and identified regions of identity among the
genes from cow, mouse, and human. Based on the
predicted protein sequences of these conserved
regions, antibodies were raised that specifically
bind to the products of the RAG-1 and RAG-2
genes. Using these antibodies, RAG-1 and RAG-2
proteins were identified in extracts of calf thy-
mus and isolated. We found these proteins are
distinct from NBP. Interestingly, the sizes of the
RAG-1 or RAG-2 proteins are slightly different
from those predicted by their gene sequences,
suggesting that they are modified after they are
made. We have gone on to find that both RAG-1
and RAG-2 are modified by addition of phosphate
groups. As was discussed earlier, this type of mod-
ification often serves to regulate protein func-
tion. We are intrigued by the possibility that such
is the case for RAG-1 and RAG-2.
Molecular Mimicry by Antibody Molecules
As a result of antibody gene rearrangement, the
structures of potential antigen-binding sites vary
almost endlessly. It has been a long-standing no-
tion in immunology that the structure of an anti-
gen could be mimicked by a special type of anti-
body— an antibody raised against another
antibody that was originally elicited by the anti-
gen of interest. We can draw an analogy for this
idea from the art of casting. A sculptor creates a
positive image, which is used to produce a nega-
tive image, the mold, which in turn is used to
reproduce the final work. An essential difference
between this metaphor and reality is that in cast-
ing, one object is shaped by another, while in the
antibody response, the binding of antigen selects
a complementary shape from a universe of preex-
isting ones.
We have tested this idea in collaboration with
research groups in Baltimore and Paris (led by
Mario Amzel of the Johns Hopkins School of Medi-
cine and Pierre Ronco, INSERM, Hopital Tenon).
Antibodies (Abl) were produced against a hor-
mone, angiotensin II. Then a second antibody
(Ab2) was raised against the anti-angiotensin II
antibody. A third antibody (Ab3) was in turn
raised against Ab2. The Ab3 antibodies bound to
angiotensin II just as well as the original Abl anti-
bodies, suggesting that Ab2 could mimic the hor-
mone's structure.
This was proved by analyzing the structures of
Abl and Ab3. Remarkably, we found that the se-
quences of their antigen-binding sites were
nearly identical. We then determined the three-
dimensional structure of the antigen-binding site
of Ab3 in complex with angiotensin II. We discov-
ered that the atoms in critical contact with angio-
tensin II in this complex are also found in the
sequence of Abl. Thus both Abl and Ab3 were
elicited by similar structures, one represented by
the original antigen, angiotensin II, and the other
by a surface feature of the Ab2 antibody that mim-
ics angiotensin II.
110
1
Expression of the gene BK27 in a Drosophila embryo with an extended germ band. This gene,
which encodes a homeodomain and a Paired domain, exhibits prominent expression in the
ventral labial lobe, marking the progenitor cells for the salivary gland placode. The staining seen
in the head is localized in the subantennal region.
Research and photograph by Susie Jun in the laboratory of Claude Desplan.
112
Transcription Control During Early
Drosophila Development
Claude Desplan, Ph.D. — Associate Investigator
Dr. Desplan is also Associate Professor and Head of the Laboratory of Molecular Genetics at the Rockefeller
University. He graduated from the Ecole Normale Superieure de Saint Cloud in France. He received
his D.Sc. degree from the University of Paris, working with Baty Moukhtar and Monique Thomasset
at the INSERM on calcium regulation. He was a Maitre de Conference at the Ecole Normale Superieure
de Fontenay until he moved to the University of California, San Francisco, to work as a postdoctoral
fellow with Pat O'Farrell on the functions of homeodomain proteins during early development.
ALL information required for the develop-
ment of a fertilized egg into a complete or-
ganism is contained in its own genetic material,
contributed by both parents, and in products pro-
vided by the mother as the egg is formed. Genetic
studies on the fruit fly Drosophila have identi-
fied most of the genes involved in the process of
pattern formation. Many of these genes appear to
encode transcription factors and to share some
common protein motifs, such as homeodomains
or zinc finger domains. The zygotic genome re-
sponds to maternal organizing factors through a
network of transcriptional regulators to set up the
body pattern of the embryo. Many of the Drosoph-
ila developmental genes have mammalian homo-
logues expressed during embryogenesis. Thus it
is likely that the mechanisms uncovered in flies
are of general significance for the development of
multicellular organisms.
The goal of our laboratory is to understand the
molecular mechanisms involved in the regula-
tory interactions among developmental genes. In
light of the complexity of the system, a produc-
tive approach is, first, to characterize the molecu-
lar interactions in vitro and, second, to design in
vivo systems to test models consistent with prop-
erties uncovered through the first approach. We
are investigating the events leading to patterning
in two major systems. One involves the establish-
ment of the early anterior pattern, while the other
is a structure-function analysis of a gene that con-
trols later steps of development. Independent
support will be sought for the research described
in the first part of this report.
Interactions Between Maternal
and Zygotic Genes
The maternal homeodomain-encoding gene bi-
coid is one of the few examples of a morphogen.
A concentration gradient of its protein product
appears to control the establishment of the ex-
pression pattern of the first set of zygotic genes,
in particular the anterior gap gene hunchback
{hb^. Using transgenic flies that carry artificial
genes containing combinations of binding sites
for the products of bicoid and hb, we have shown
that the establishment of the simple hb pattern
requires not only bicoid but also the synergistic
participation of hb itself. This hb autoregulation
may help explain the sharp, precise boundaries
of its expression domain. It also has evolutionary
implications for the development of other organ-
isms that do not possess bicoid.
The role of bicoid is not only to control hb
expression but also to direct the expression of a
set of newly discovered genes required for head
formation. Using the same approach as described
above, we have demonstrated that bicoid
action is modulated by the maternal genes from
the terminal torso group. These genes, which
encode a cascade of kinases, appear to act post-
transcriptionally on the bicoid protein (Bed) to
repress its action in the terminal region. This
leads to the expression of the head gap genes as
an anterior stripe, responsible for development
of a specific head region. While bicoid is setting
up their posterior margin, the torso group, pre-
venting activation by bicoid, sets up their ante-
rior margin.
We recently identified a gene that has such an
expression pattern and encodes a homeodomain
with a DNA-binding specificity related to that of
Bed. We are investigating the control of the new
gene's pattern of expression, and are analyzing
phenotypes in the head region for genetic dele-
tions of the locus.
The demonstration of a direct interaction be-
tween bicoid and the maternal torso group is the
first clue for understanding the formation of head
structure. It also presents us with a unique oppor-
tunity to use a combination of biochemical and
genetic approaches to address the nature and reg-
ulation of the post-transcriptional modifications
of a transcription factor in the context of a devel-
oping organism.
Regulation of the Segmentation Gene hb
Molecular dissection of the hb promoter by sev-
eral groups has allowed definition of the sites of
action of bed. Our in vivo studies have uncov-
ered a more complex regulation of hb. As de-
scribed above, hb appears to be involved in the
establishment of its own early expression, in syn-
ergy with bed. We have identified in vitro, in the
113
Transcription Control During Early Drosophila Development
region of the hb promoter, an Hb protein-bind-
ing site potentially required for full anterior acti-
vation. Mutations of this site are being tested in
vivo for their ability to respond to hb function. In
another region of the hb promoter, we have iden-
tified a cluster of Hb-binding sites that may medi-
ate the other instance of hb autoregulation at a
later stage in development.
Interestingly, these two examples of positive
hb autoregulation contrast with the general re-
pressive nature of hb. We have obtained in vivo
evidence that, depending on the promoter con-
text, hb can act as a positive or a negative regula-
tor of gene expression. Finally, the hb promoter
is also regulated by the neighboring gap gene
Kriippel, for which two putative binding sites
have been identified. This description and the ex-
periments presented above may allow us to re-
construct a functional hb promoter made exclu-
sively of minimal response elements identified
for the regulators of hb.
DNA-binding Specificity
of the Homeodomain
Most developmental processes involve genes
that encode a homeodomain (HD). The HD in-
cludes a domain similar to the helix-turn-helix
motif present in many prokaryotic DNA-binding
transcriptional regulators. Our analysis of the
function of the HD has led us to propose that the
specificity changes among classes of HDs are due
to the nature of a single amino acid at position 9
of the recognition helix. This position is not criti-
cal in the prokaryotic helix-turn-helix proteins,
and indeed structural analysis of HD-DNA com-
plexes recently showed that the HD-DNA interac-
tion involves a new mode of recognition. In this
mode, amino acid 9 fits in the major groove of
DNA, in close contact with the base pairs being
recognized.
We have generalized the role of amino acid 9 to
several classes of HDs. Using a powerful selection
procedure from a library of random sequences,
we have isolated specific sequences recognized
by a set of mutant HD proteins carrying different
residues at position 9. These sequences confirm
our previous observations, but indicate that the
different proteins interact with DNA in different
modes.
The paired Gene Encodes a Multifunctional
Transcription Factor
In addition to an HD, the paired gene product,
the Prd protein, contains regions that are also
found in other developmental genes. We have
shown that the so-called Paired domain is a sec-
ond DNA-binding domain in the Prd protein,
which makes Prd a bifunctional transcription fac-
tor. Although both the Paired domain and the HD
can bind to DNA independently, they can also
bind cooperatively to adjacent sites, but only
when both domains are present in the same mole-
cule. The cooperation between the two motifs
may refine the functional specificity of genes
containing highly related domains.
We have undertaken an in vivo structure-
function analysis of the product of prd. In this we
are attempting to correlate the multiple DNA-
binding functions of Prd defined in vitro with the
multiple genetic functions of the prd gene, as a
regulator of segmentation genes and of genes in
the nervous system. Transgenic flies that carry the
prd coding sequence driven by its own promoter
rescue the lethal phenotype of prd" flies. Other
lines carrying versions of the prd gene modified
in regions encoding each of the subdomains of
the Prd protein are now being tested for their
ability to rescue some of the differential molecu-
lar phenotypes of prd~ .
From Segmentation to Organogenesis
We recently discovered a gene that appears to
be involved in the early steps of organogenesis of
the salivary gland, but not in segmentation. This
gene, highly homologous to prd, is first ex-
pressed in a group of cells that represent the pro-
genitor of the salivary gland placode, at a stage
when no tissue differentiation has occurred. This
expression is controlled by positional cues from
the dorsoventral and anteroposterior determi-
nants. Later, when the placode invaginates, other
transcription factor-encoding genes are ex-
pressed in the same tissue. Since the expression
of this pr^Z-related gene precedes morphological
events, it may provide an excellent model of tran-
scriptional commitment to a particular differen-
tiation pathway. We are now generating mutants
to analyze the early expression of the putative
target genes and the resulting morphological
phenotype.
Because the protein encoded by this gene has a
Prd-like organization but a Prd domain from a di-
vergent class, this molecule will also be very use-
ful in dissecting the residues that determine the
specificity of the Prd domain. In collaboration
with John Kuriyan's group (HHMI, Rockefeller
University) , we have undertaken a structural anal-
ysis of the Prd domain and the HD present in Prd.
114
Immune Evasion by Parasites Causing
Tropical Diseases
John E. Donelson, Ph.D. — Investigator
Dr. Donelson is also Distinguished Professor in the Department of Biochemistry at the University of Iowa,
Iowa City. He obtained his bachelor's degree in biophysics from Iowa State University, served as a Peace
Corps volunteer for two years in Ghana, West Africa, and then completed a Ph.D. degree in biochemistry
at Cornell University. His postdoctoral research was conducted at the MRC laboratory of Molecular
Biology in Cambridge, England, and at Stanford University. He has received the Iowa Governor's
Science Medal and the Burroughs Wellcome Award in Molecular Parasitology.
THE risk of acquiring parasitic infections is a
part of the daily lives of more than 3 billion
people living in the developing countries of the
tropics. The various protozoan and helminthic
parasites that are responsible for these infectious
diseases possess a variety of mechanisms for evad-
ing the immune response of their hosts. Our labo-
ratory studies the interactions between several of
these parasites and the immune system, in antici-
pation that a more detailed understanding of
these immune evasion mechanisms will suggest
better ways to combat or prevent the infections.
The parasites that cause three of these tropical
diseases are described here.
Trypanosomiasis
African trypanosomes are protozoan parasites
that cause sleeping sickness or trypanosomiasis
throughout equatorial Africa. They are transmit-
ted from tsetse flies to the mammalian blood-
stream, where they continually confront the hu-
moral and cellular immune systems. Each
trypanosome in the bloodstream is covered by
about 10 million copies of a single protein, the
variant surface glycoprotein (VSG). The trypano-
some population survives the continuous
immune assault because individual parasites
occasionally switch spontaneously from the ex-
pression of one VSG to another — a process called
antigenic variation. A new immune response
must be mounted against the VSG of the switched
parasite and its descendants, enabling the trypan-
osome population as a whole to stay "one step
ahead" of the host immune defenses. We are ex-
amining the events at the DNA and RNA level that
are responsible for this antigenic variation.
We know that the trypanosome genome con-
tains about 1,000 different genes encoding anti-
genically distinct VSGs. Usually one, and only
one, of these VSG genes is expressed at a time.
Rearrangements and duplications of these genes
are partly responsible for the selection of which
VSG gene is to be transcribed and for the switch
event itself.
The rearrangements maneuver specific VSG
genes into and out of special chromosomal loca-
tions, called expression sites, where transcrip-
tion occurs. These expression sites are always lo-
cated near the ends of the chromosomes, i.e.,
near the chromosomal telomeres, for reasons that
are not well understood. The expression process
is complicated by the fact that several, and per-
haps many, potential telomere-linked expression
sites exist in the genome, yet only one is normally
activated at any time. Our goal is to understand
this activation mechanism at the molecular level.
In one project we have identified a protein that
specifically binds to a region upstream of a VSG
gene in an expression site activated during the
final developmental stage of the parasite in the
tsetse fly, i.e., the metacyclic stage. We think that
this protein may contribute to the developmental
regulation of that VSG gene's transcription. In an-
other project we have detected a high rate of mu-
tation in the telomere-linked VSG genes, which
may increase the effectiveness of antigenic varia-
tion still further. A third project involves a charac-
terization of a group of very small chromosomes
that are unique to African trypanosomes and con-
tain many of the VSG genes destined for sequen-
tial expression. In still another project we have
introduced into trypanosomes several plasmids
containing combinations of sequences upstream
of an expressed VSG gene and an easily detected
luciferase gene in an effort to identify the se-
quences that regulate the transcription of the
VSG gene.
Leishmaniasis
Leishmania parasites are protozoan pathogens
that cause a spectrum of diseases, including cuta-
neous, mucocutaneous, and visceral leishmania-
sis, in many tropical countries. During their life
cycle these parasites reside in both the sandfly
vector and a mammalian host. Within sandflies
they exist as uniflagellar promastigotes that de-
velop from a less-infectious form to the final
highly infectious form. This developmental pro-
cess can be mimicked during cultivation of pro-
mastigotes in liquid medium. After transmission
from the sandfly to the mammalian host blood-
stream, the promastigotes penetrate host macro-
115
Immune Evasion by Parasites Causing Tropical Diseases
phages, where they reside as spherical amasti-
gotes within acidic phagolysosomes. Thus the
parasites evade the immune response by "hid-
ing" inside macrophages — one of several cell
types of the immune system that normally help to
destroy foreign pathogens and substances. When
a sandfly ingests amastigote-laden macrophages
during a blood meal, the life cycle is completed.
With a grant from the National Institutes of
Health, we are studying how these organisms can
survive in hostile environments as diverse as a
sandfly midgut and an acidic macrophage
phagolysosome.
A major glycoprotein on the surface of both
promastigotes and amastigotes is a metal lopro-
tease of about 63 kDa (gp63) . This enzyme partic-
ipates in the promastigotes' uptake by macro-
phages and contributes to the amastigotes'
survival within them. The amount of gp63 on the
surface of promastigotes increases about 1 0-fold
as the parasites develop into the highly in-
fectious, virulent form during growth in culture.
We have discovered that three different RNA spe-
cies encoding gp63 occur during cultured pro-
mastigote growth.
One RNA species occurs in the promastigotes
only during their early, logarithmic growth when
they are less infectious and have a small amount
of gp63 on their surface. This gp63 RNA is re-
placed by another gp63 RNA species as the pro-
mastigotes enter stationary phase and become
more infectious. The presence of this second RNA
correlates with the increased amount of gp63
protein. The third gp63 RNA species is continu-
ously present at a low level throughout cultured
growth and encodes an altered, transmembrane
form of the protein.
The three RNA species are derived from a fam-
ily of about 1 5 different genes, some of which are
identical and all of which are continuously tran-
scribed independent of the stage of cultured
growth. The 3'-untranslated regions of the three
RNA species have completely different se-
quences, and we have shown that these se-
quences contribute to the different steady-state
amounts of the RNAs at the different growth
stages. In addition, we are introducing recombi-
nant DNAs into leishmania to amplify, alter, or
delete individual gp63 genes so that we may de-
termine the functions of each of the three differ-
ent gp63 proteases during the promastigote and
amastigote stages. One of our long-term goals is
to alter genetically, or attenuate, the parasite so
that it can be readily maintained under laboratory
culture conditions but cannot survive in human
macrophages. Such parasites may be useful in the
development of a vaccine against leishmaniasis.
Onchocerciasis
Onchocerca volvulus is a filarial nematode
that causes onchocerciasis, or river blindness, in
Africa and Latin America. Female O. volvulus
worms grow to 50 cm in length and reside
throughout the body within nodules. In the nod-
ules they produce thousands of microfilariae
each day that migrate throughout the body and
enter the eyes, where they produce lesions that
can lead to blindness. It is not known how any of
the developmental stages of the parasite evade
the immune response of an infected person.
The parasites are difficult to study in the labora-
tory, because there is no good experimental ani-
mal model; they infect only humans and
chimpanzees.
We have constructed cDNA expression librar-
ies of the mRNAs from the infective L3 stage lar-
vae of the parasite, and we are using specific
cDNA clones to overproduce parasite antigens
from this larval stage that are recognized by anti-
sera from onchocerciasis patients. This approach
has revealed several proteins that are unique to
the parasite and may be valuable for improved
diagnosis, treatment, and prevention of the
disease.
116
Post-transcriptional Regulation
of Gene Expression, RNA-Protein Complexes,
and Nuclear Structures
Gideon Dreyfuss, Ph.D. — Investigator
Dr. Dreyfuss is also Professor of Biochemistry and Biophysics at the University of Pennsylvania School
of Medicine. He received his Ph.D. degree in biological chemistry from Harvard University and his
postdoctoral research training as a Helen Hay Whitney fellow at the Massachusetts Institute of Technology.
Prior to his present appointment, he was Professor and Established Investigator of the American Heart
Association at Northwestern University.
MESSENGER RNAs (mRNAs) are the func-
tional translatable intermediates in the
pathway of gene expression from DNA to pro-
teins. They are formed in the nuclei of eukaryotic
cells by extensive and tightly regulated post-
transcriptional processing of primary RNA poly-
merase II transcripts. These transcripts are
termed heterogeneous nuclear RNAs (hnRNAs) , a
term that describes their size heterogeneity and
cellular localization. The terms hnRNA and pre-
mRNA are often used interchangeably, though it
is possible that only a subset of hnRNAs are actu-
ally precursors to mRNA and that the rest turn
over in the nucleus. From the time hnRNAs
emerge from the transcription complex and as
long as they remain in the nucleus, they are asso-
ciated with proteins.
The collective term for the proteins that bind
hnRNAs (but are not stable components of other
classes of ribonucleoprotein complexes, such as
snRNPs [small nuclear ribonucleoproteins]) is
hnRNPs. The significance of hnRNPs is that they
are bound to the hnRNAs and thus influence their
structure and therefore their fate and processing
into mRNAs. They are also abundant in the nu-
cleus and in hnRNA-hnRNP complexes (hnRNP
complexes) and of interest as major nuclear
structures.
Once formed, the mRNAs are transported to the
cytoplasm via nuclear pores — a process that ap-
pears to be one of the most important regulatory
steps in the post-transcriptional pathway of gene
expression and about which very little is pres-
ently known. In the cytoplasm, mRNAs are asso-
ciated with proteins — the mRNPs — and these are
likely to be involved in the regulation of the
translation and stability of mRNAs and in their
cellular localization. Our goal is to understand,
in molecular detail and cellular architecture,
how the post-transcriptional portion of the path-
way of gene expression operates in the cell. To
that end, we investigate the structure, function,
and localization of the hnRNPs and mRNPs and
the RNP complexes.
We have used photochemical RNA-protein
crosslinking in intact cells and affinity chromato-
graphic methods to identify and purify the
hnRNPs and mRNPs and have produced monoclo-
nal antibodies to many of them. The monoclonal
antibodies were used for immunopurification of
hnRNP complexes from vertebrate and Drosoph-
ila melanogaster cells and for the characteriza-
tion of the hnRNPs. Immunopurified hnRNP
complexes contain large RNA of up to 10,000
nucleotides and at least 20 major proteins, desig-
nated A-U, in the range of 34- 1 20 kDa. There are
also many less abundant hnRNPs, and these ap-
pear to bind only to specific subsets of hnRNAs.
The specific arrangement of the proteins on spe-
cific hnRNAs is probably important in determin-
ing the structure of the hnRNA and is one of the
areas on which we concentrate our investigation.
Related to this issue, we found that several of
the hnRNPs have RNA-binding specificities. Inter-
estingly, some of the specificities of the major
hnRNPs are for sequences important in pre-
mRNA processing and polyadenylation, and it is
likely that this binding specificity is directly re-
lated to a role for these proteins in mRNA
formation.
The molecular cloning and sequencing of
cDNAs for several RNPs made possible the discov-
ery of a conserved RNA-binding domain (RBD)
and a ribonucleoprotein consensus sequence
(RNP-CS). The RNP-CS, Lys/Arg-Gly-Phe/Tyr-
Gly/Ala-Phe-Val-X-Phe/Tyr, is the most highly
conserved segment in a generally conserved do-
main of about 90 amino acids found in many
RNA-binding proteins of the nucleus and cyto-
plasm in all eukaryotes examined. Overall, RNP-
CS proteins have a modular structure reminiscent
of DNA-binding transcription factors. They fre-
quently contain several similar but nonidentical
RBDs, and all contain at least one auxiliary
domain that is unique to each type of protein
(e.g., glycine-rich, glutamine-rich, or acidic) and
that most likely functions in protein-protein
interactions.
The hnRNP C1/C2 proteins are abundant, avid
hnRNA binders. Deletional analysis showed that a
93-amino acid segment of the human hnRNP C
protein that contains the RNP-CS is, as predicted,
117
Post-transcriptional Regulation of Gene Expression, RNA-Protein Complexes,
and Nuclear Structures
sufficient for RNA binding. This RBD was pro-
duced in bacteria and purified to homogeneity in
active form. Nuclear magnetic resonance (NMR)
methods (in collaboration with Luciano Mueller
and Michael Wittekind, Bristol-Myers Squibb),
including '^C- and ^^N-edited three-dimensional
NMR, were used to determine the structure of the
RBD in solution. The compact folded structure
exhibits a four-stranded antiparallel |8-sheet and
two well defined a-helices. The structure of this
RBD complexed with an RNA substrate is being
determined.
Experiments on mitotic cells unexpectedly
provided important insights into the assembly
and general nature of hnRNP complexes and into
the transport of proteins to the nucleus. In mito-
sis, as the nuclear envelope breaks down, hnRNPs
disperse throughout the cell but remain asso-
ciated in complexes with RNA. After mitosis,
once the nuclear envelope re-forms, preexisting
hnRNPs return to the nucleus. We observed, how-
ever, using double-label immunofluorescence
microscopy with monoclonal antibodies to
various hnRNPs on postmitotic cells, that at the
end of mitosis the hnRNP complexes dissociate in
the cytoplasm and the different proteins are trans-
ported to the nucleus separately. Some, includ-
ing CI, C2, and U, like snRNPs and lamins, are
transported immediately (early group), while
others, including Al, A2, Bl, B2, E, G, H, and L,
are transported into the nucleus later (late
group). Thus, immediately following reassembly
of the nuclear envelope at the end of mitosis,
pairs of cells are detected in which some hnRNPs
are in the nucleus, and others are in the cyto-
plasm. These observations show that hnRNP com-
plexes are dynamic structures, in that hnRNPs
can dissociate from the complexes and return to
the nucleus separately.
Surprisingly, the transport of the late group re-
quires transcription by RNA polymerase II: inhibi-
tors of this polymerase cause the late proteins to
remain in the cytoplasm. Thus there are two path-
ways for nuclear transport localization of pro-
teins: a transcription-independent pathway and a
novel transcription-dependent pathway. The dif-
ferent hnRNPs utilize one of the two pathways,
and both pathways operate throughout the cell
cycle. The signals in the proteins that spec-
ify the pathway to be used, the mechanism of
transcription-dependent transport localization,
and the relevance to mRNA transport are being
investigated.
Immunofluorescence microscopy indicated
that the hnRNP proteins A, B, C, E, I, K, L, M, and
U are nucleoplasmic — that is, localized to the
nucleus but excluded from nucleoli. In contrast,
the mRNPs that have been characterized so far [in
particular the poly (A) -binding protein] are con-
fined to the cytoplasm. It was therefore con-
cluded that the mRNA must exchange most if not
all of the proteins with which it is associated in
the nucleus as it is transported to the cytoplasm.
The dissociation of these proteins from the mRNA
and the subsequent binding of mRNPs must be an
important aspect of nuclear-cytoplasmic trans-
port of mRNA.
Recently we found that several of the abundant
hnRNPs, including Al, are not confined to the
nucleus but rather shuttle continuously between
the nucleus and the cytoplasm. Thus hnRNPs may
have cytoplasmic functions in addition to their
nuclear roles in the processing of pre-mRNA to
mRNA. Furthermore, Al is bound to mRNA in the
cytoplasm and its return to the nucleus requires
RNA polymerase II transcription. It is therefore
possible that the novel cytoplasmic ribonucleo-
protein complex of mRNA with hnRNPs is the
substrate of RNA nuclear-cytoplasmic transport.
Thus it is likely that hnRNPs play an active role
in mRNA export and that the mRNA is transported
to the cytoplasm as a result of its association with
the shuttling hnRNPs. Understanding the traffick-
ing of hnRNPs in the cell and the mechanisms
that regulate the assembly and disassembly of
RNPs with hnRNAs and mRNAs should be impor-
tant for elucidating the post-transcriptional path-
way of gene expression, including the nuclear
transport process of mRNA.
118
Genetic Basis of Hearing Loss
Geoffrey M. Duyk, M.D., Ph.D. — Assistant Investigator
Dr. Duyk is also Assistant Professor of Genetics at Harvard Medical School and a member of the Eaton-
Peabody Laboratory at the Massachusetts Eye and Ear Infirmary. He received his undergraduate degree
at Wesleyan University and his M.D. degree and a Ph.D. degree in biochemistry at Case Western Reserve
University. After internship and residency at the University of California, San Francisco, he held a
fellowship in medical genetics under the direction of Charles Epstein. His postdoctoral research work was
also at UCSF, with David Cox and Richard Myers. While at UCSF, Dr. Duyk was awarded an HHMI
Physician Research Fellowship and was a Lucille P. Markey Scholar in Biomedical Science.
HEARING loss is the commonest form of sen-
sory impairment. In the United States the in-
cidence of congenital deafness is approximately
1 per 1,000, and at least half of these cases are
likely to be determined genetically. It is esti-
mated that by age 65 one in six of us will have a
clinically significant hearing loss. While environ-
mental factors play a role in hearing impairment,
genetic factors often determine predisposition.
Deafness is a major aspect of over 100 defined
genetic disorders.
Our laboratory is interested in sensorineural
hearing loss in which the abnormality should lie
along the pathway between the sensory receptors
of the inner ear and the auditory centers of the
brain. Our strategy is to combine genetic map-
ping and positional cloning with candidate gene
approaches to identify genes responsible for
hearing loss and to further our understanding of
the molecular mechanisms of hearing.
Candidate Genes
Within the inner ear reside the sensory organs
for hearing and balance. Both of these systems
represent mechanical transduction. In each, a
specialized "hair cell" serves as a sensor of sound
or motion. Hair cells are highly polarized, with
the basal surface forming synapses with afferent
and efferent nerve fibers. Finger-like projections
referred to collectively as the hair cell bundle are
distinctive features of the hair cell's apical sur-
face. Deflection of the hair cell bundle by the
sensory input activates mechanically sensitive
ion channels, initiating signal transduction. This
process is distinct from other sensory systems
such as taste, olfaction, and vision, where a ligand
binds to a specific receptor, which in turn acti-
vates a second messenger system to initiate sen-
sory transduction. The molecular components of
this highly specialized mechanical sensory sys-
tem are logical targets for specific mutations re-
sulting in hearing loss.
In collaboration with David Corey (HHMI,
Massachusetts General Hospital), we have begun
the process of identifying the molecular compo-
nents of the hearing apparatus by combining mo-
lecular biological and biophysical approaches.
Although extensive microanatomical and bio-
physical characterization of this apparatus has
been undertaken, little biochemical or molecular
information is available. As a starting point, we
are constructing cDNA libraries from microdis-
sected inner ear material highly enriched for hair
cells. In addition, we are exploring strategies for
analyzing mRNA populations of single cells with
the polymerase chain reaction (PGR) technique
to augment the traditional approaches for re-
covering and studying genes from specialized tis-
sue sources. As we identify components of the
sensory transduction pathway, human and mu-
rine homologues will be recovered and mapped
to specific chromosomes, and closely linked DNA
polymorphic markers will be identified. The de-
rived information will be an important resource
as we begin our search for the mutations responsi-
ble for this group of inherited disorders.
"Reverse Genetics"
The genetic analysis of hearing loss in human
populations is complicated by the difficulty of
diff^erentiating many of these syndromes (genetic
heterogeneity) and the fact that deaf individuals
often intermarry (nonassortive mating). As a con-
sequence, most of the families suitable for study
are small, increasing the problem of placing the
disease locus with the degree of precision re-
quired to clone the gene. The availability of
mapped candidate genes helps to bridge the gap
between genetic and physical mapping. Analysis
of these small families will be enhanced by the
availability of dense genetic maps. Toward that
end, we have developed an efficient technique
for the construction of genomic libraries aug-
mented for selected classes of simple sequence
repeats (SSRs). SSRs correspond to runs of di-,
tri-, or tetranucleotide sequences (e.g., [CA]„ or
[GGC]„) that often demonstrate length polymor-
phisms detectable by PGR assay. These libraries
will aid in the production of high-resolution ge-
119
Genetic Basis of Hearing Loss
netic maps composed of very polymorphic, ho-
mogeneously distributed PCR-based markers.
Genetic mapping can localize a disease gene to
a 1- to 2-Mb region of a particular chromosome.
The next challenge is to identify the genes in this
region. We have developed a strategy, termed
exon trapping, that utilizes splicing signals as
primary identifiers of coding sequence in cloned
genomic DNA. This approach is also of great util-
ity for cloning the human homologue of a gene of
interest when the temporal or spatial pattern
of gene expression limits the availability of
mRNA for analysis. We are developing second-
generation exon-trapping strategies based on
large-capacity cloning vectors (e.g., cosmids,
PI ) that should extend the range for gene search-
ing and help to integrate the recovery of genes
with physical mapping.
120
Molecular Genetics of Intracellular
Protein Sorting
Scott D. Emr, Ph.D. — Associate Investigator
Dr. Emr is also Professor of Cellular and Molecular Medicine at the University of California, San Diego,
School of Medicine. He did his graduate work in microbiology and molecular genetics with Thomas
Silhavy and Jonathan Beckwith at Harvard Medical School, where he identified and characterized the first
secretion- defective signal sequence mutants as well as the first component of the bacterial protein export
apparatus. Dr. Emr did postdoctoral work on protein secretion in yeast with Randy Schekman at the
University of California, Berkeley. He was a faculty member at the California Institute of Technology
before moving to UCSD. Dr. Emr counts among his honors a Searle Scholars Award and an NSF
Presidential Young Investigator Award.
AN essential feature of all eukaryotic cells is
their highly compartmentalized organiza-
tion. Different, often competing biochemical
processes are segregated into distinct compart-
ments. The identity, stability, and function of
each of these compartments, or organelles, is
conferred in large part by the unique set of pro-
teins that reside within it. These proteins must be
routed from their common site of synthesis in the
cytoplasm to their unique site of function in the
appropriate intracellular organelle. Toward a de-
tailed understanding of the molecular mecha-
nisms that direct the delivery of one class of these
proteins, we have focused our attention on the
transport and sorting of proteins through the
Golgi complex to the lysosome.
Lysosomal Hydrolase Sorting
Our laboratory is using a simple unicellular eu-
karyote, the yeast Saccharomyces cerevisiae, as a
model genetic system to study protein trafficking
through the secretory pathway to the lysosome-
like vacuole. The fundamental similarities be-
tween yeast and other eukaryotic cells in their
pathways and mechanisms for protein delivery
have clearly established yeast as an important
model system for the study of these problems.
The importance of the lysosomal protein sort-
ing pathway in humans is revealed when the ef-
fects of mislocalizing such enzymes are exam-
ined. A number of lysosomal hydrolases are
secreted from naturally occurring tumor cells
(e.g., cathepsin D by breast cancer cells). It has
been proposed that the mislocalization of lyso-
somal enzymes enhances cell growth and in-
creases the metastatic potential of tumor cells by
contributing to the hydrolysis of extracellular
matrix components of target tissues. The impor-
tance of this sorting pathway is further exempli-
fied by the serious inherited disorders (e.g., I cell
disease) that result in mislocalization of lyso-
somal hydrolases.
Proteins destined for the vacuole/lysosome in
yeast and mammalian cells transit through early
stages of the secretory system. The transport and
processing characteristics of several vacuolar hy-
drolases in yeast, including the soluble protease
carboxypeptidase Y (CPY), have been well char-
acterized. CPY is translocated into the endoplas-
mic reticulum (ER), where it is modified with
four core oligosaccharides to generate the ER
precursor form of CPY (pi CPY). Subsequent
transport events are mediated by vesicular car-
riers. Delivery from the ER to the Golgi complex
and passage through the Golgi are accompanied
by elongation of the core oligosaccharides on
CPY, resulting in formation of the Golgi-
modified form of CPY (p2CPY). Sorting of
p2CPY from proteins destined for secretion ap-
pears to take place in the late Golgi. Upon arrival
in the vacuole, an amino-terminal propeptide
segment on p2CPY is proteolytically removed,
generating the active mature form of the protease
(mCPY).
Sorting-Defective Mutants
We have focused our efforts on identifying and
characterizing the cellular machinery that directs
the sorting and transport of vacuolar hydrolases.
A gene fusion-based selection scheme we de-
signed enabled us to isolate more than 600 yeast
mutants that exhibit severe defects in vacuolar
protein sorting. Each of the mutants missorts and
secretes CPY and other vacuolar enzymes. The
recessive mutations define more than 33 comple-
mentation groups referred to as vps (vacuolar
protein sorting defective) . Extensive genetic, bio-
chemical, and morphological characterization of
the vps mutants has allowed us to organize them
into phenotypically related groups. Each group
appears to function at a common stage of the pro-
tein sorting pathway.
The apparent genetic complexity of the vps
mutant collection presumably reflects the bio-
chemical complexity of the protein sorting reac-
tion. Specific cellular components must recog-
nize vacuolar proteins, segregate them from
other secretory proteins, and package them into
121
Molecular Genetics of Intracellular Protein Sorting
specific transport vesicles that ultimately must
recognize and fuse with the appropriate target
organelle, the vacuole. Therefore the list of po-
tential activities and structures required for the
sorting and transport of lysosomal/vacuolar pro-
teins can easily accommodate the large number
of gene products presently implicated by the ge-
netic studies.
Role for a Protein Kinase Complex
An understanding of this protein sorting path-
way and the individual activities of different VPS
gene products is being facilitated by the molecu-
lar isolation and characterization of these genes.
Thus far, we have cloned and sequenced eight
VPS genes. Comparison of the yeast VPS gene se-
quences with other known mammalian genes has
revealed several informative structural and func-
tional similarities. Two of the genes, VPS\ 5 and
VPS54, are of particular interest. The sequence of
the VPSl 5 gene predicts a protein with the fol-
lowing features: a consensus site for amino-
terminal myristoylation, a region that shares sig-
nificant sequence similarity with the family of
Ser/Thr protein kinases, and a region of homol-
ogy with the regulatory subunit of protein phos-
phatase 2A (PP2A). The sequence similarity
Vpsl 5 protein (Vpsl 5p) shares with Ser/Thr pro-
tein kinases and the PP2A regulatory subunit
raises the interesting possibility that protein
phosphorylation/dephosphorylation may play an
important role in the regulation of protein sort-
ing events.
To assess the functional significance of these
sequence similarities, we used site-directed mu-
tagenesis to change several highly conserved
amino acid residues in the putative kinase do-
main of the Vpsl 5p. Each of the mutations inacti-
vates the complementing activity of the VPS15
gene. The mutant cells exhibit extreme vacuolar
protein sorting defects; amino acid changes in
less highly conserved positions gave only weak
protein sorting defects.
These data suggest that the Vpsl 5p is an active
kinase and that this kinase activity is required
during some step in vacuolar protein sorting.
Protein phosphorylation may act as a "molecular
switch" in this protein sorting pathway by ac-
tively diverting vacuolar hydrolases away from
the default secretion path and toward the vacu-
ole. One can imagine several points in the path-
way that may need to be regulated precisely, such
as the budding and transport of carrier vesicles or
the recognition and fusion of these with the
correct target organelle.
Biochemical and genetic evidence indicate
that the Vpsl 5 protein forms a complex (on the
cytoplasmic face of the membrane) with another
Vps protein, Vps34p. Overexpression of Vps34p
suppresses the vacuolar protein sorting and
growth defects caused by mutations within the
kinase domain of Vpsl 5p, but will not suppress a
null allele of VPSl 5. Therefore, Vps34p cannot
bypass the cells' requirement for Vpsl5p. This
genetic interaction between VPSl 5 and VPS54 is
consistent with the observation that mutations in
both the VPSl 5 and VT534 genes result in a com-
mon set of phenotypes.
The Vps34 protein shares sequence similarity
with a mammalian gene recently identified in
Mike Waterfield's laboratory that codes for the
catalytic subunit of the phosphatidylinositol
3-kinase (PI3-kinase). P13-kinase catalyzes the
formation of PI3-phosphate, a rare membrane
lipid that has been proposed to act as a second
messenger in cell signaling. The enzyme appears
to play an important role in cell proliferation and
transformation. PI3-kinase activity has been
shown to be associated with several cell surface
protein-tyrosine kinase receptors (e.g., the PDGF
[platelet-derived growth factor], insulin, and
CSF-1 [colony-stimulating factor 1] receptors).
The role of this lipid-modifying enzyme in pro-
tein trafficking events is not yet clear. However,
we are in the process of mutating several con-
served sequence motifs to analyze the signifi-
cance of this intriguing sequence similarity in
Vps34 protein function.
Among its functions, PI3-phosphate formation
at localized sites in the membrane (adjacent to
membrane receptor proteins) may facilitate vesi-
cle formation/targeting required during protein
transport to the lysosome from both the Golgi
complex (delivery of newly synthesized lyso-
somal hydrolases) and the cell surface (endocytic
uptake and down-regulation of cell surface re-
ceptors). The Vps 15 kinase may regulate the ac-
tivity of Vps34p/PI3-kinase by phosphorylation
of Vps34p.
We recently developed an in vitro assay that
reconstitutes intercompartmental protein trans-
port to the yeast vacuole. We are now using this
assay to assign the Vpsl 5 and Vps34 proteins to a
specific step(s) in the reaction.
122
Mechanisms Involved in Preventing Unwanted
Blood Clots
Charles T. Esmon, Ph.D. — Investigator
Dr. Esmon is also a member of the Oklahoma Medical Research Foundation and Professor of Pathology
and Associate Professor of Biochemistry at the University of Oklahoma Health Sciences Center, Oklahoma
City. He received his B.S. degree in chemistry from the University of Illinois and his Ph.D. degree
in biochemistry from Washington University. He conducted his postdoctoral research at the University
of Wisconsin before joining the faculty at the University of Oklahoma Health Sciences Center.
later he joined the Oklahoma Medical Research Foundation.
PROTEIN C, protein S, and thrombomodulin
constitute one of the natural anticoagulant
complexes that prevents unwanted blood clots.
We have focused our attention on this system be-
cause we have been able to identify patients with
a history of unwanted blood clots who have ab-
normal protein C, protein S, or thrombomodulin.
To understand how the system functions, it is
useful to review the function of the components.
Thrombomodulin is found primarily on the sur-
face of endothelial cells, the cells that line the
blood vessels. Thrombin, the enzyme that causes
blood to clot, can bind to thrombomodulin;
when thrombin is bound, it no longer clots the
blood but instead converts protein C into the ac-
tive blood clotting inhibitor, activated protein C.
Activated protein C then binds to protein S on the
surface of platelets (small cells in the blood) or
endothelial cells, where it functions as an antico-
agulant. The activated protein C-protein S com-
plex works as an anticoagulant by cutting up and
inactivating two of the clotting proteins, factor
VIII (the protein missing in hemophilia) and fac-
tor V.
This broad outline of how the system functions
fails to tell us much about where, when, or how
the system might function in human disease pro-
cesses. This knowledge is important both in
terms of understanding the basic properties of the
system and in the design of new therapeutic ap-
proaches to diagnosis and prevention of blood
clots. Of particular interest, and still unex-
plained, is the observation that administration of
activated protein C at levels that can prevent un-
wanted blood clots does not increase blood loss
at surgical sites. This contrasts with available anti-
coagulants, such as heparin, which block un-
wanted clot formation but also dramatically in-
crease blood loss at surgical sites. One of our
goals is to understand how this natural anticoagu-
lant can accomplish this remarkable specificity.
New therapeutic agents with these properties
could greatly decrease morbidity and mortality
associated with thrombotic complications.
Most healthy individuals have adequate
amounts of protein C and the other components
of the system that control blood clot formation
under normal circumstances. When people be-
come sick, unwanted clotting is often a problem.
Studies from other laboratories have shown that
protein S circulates in humans both free and
bound to an inhibitor of the complement system
(the system that helps protect from infection),
called C4b-binding protein (C4bBP).
We found that only the free form of protein S
could work to form the anticoagulant. Patients
with clinical conditions known to cause an in-
creased risk of blood clots also had reduced lev-
els of free protein S and more C4bBP-protein S
complex. Families with inherited thrombotic
complications were identified in which the fam-
ily members who developed blood clots had high
levels of the complex. These observations sug-
gested that alteration in the levels of free protein
S might contribute to the clotting complications
observed in these patients. To test this hypothe-
sis, we developed some animal models that in-
volved blood clotting as a complication.
Clearly thrombosis is caused by more than sim-
ply an alteration of C4bBP levels. Infection and
agents that cause inflammation are known to trig-
ger coagulation by a variety of mechanisms.
Three very different responses occur. The first is
related to bacterial infection, which can result in
septic shock, a process characterized by small
blood clots in the circulation, damage to organs,
and death. Once this process begins, treatment is
very difficult. A second response is formation of
solid blood clots in the small vessels. This results
in death of the affected organs. A third response is
occlusion of the large vessels. Why inflammation
causes these three dramatically different re-
sponses is unknown, complicating rational ap-
proaches for both early diagnosis and effective
therapy.
Animal models for these three processes were
developed in collaboration with Fletcher Taylor.
These models indicate that shifts in the balance
between free protein S and protein S bound to
C4bBP contribute to the dift'crent disease pro-
cesses and can be involved in shifting the ob-
served type of thrombosis. Increased levels of
123
Mechanisms Involved in Preventing Unwanted Blood Clots
C4bBP-protein S complex cause the inflamma-
tory response to low levels of bacteria to convert
from a mild response to a severe response with all
of the characteristics of toxic shock. Moreover,
under certain conditions the shift in amounts of
G4bBP-protein S complex can change the re-
sponse from that of circulating small clots to oc-
clusion of small vessels. Most data suggest that
bacteria cause clotting by causing the formation
of inflammatory mediators, called cytokines. In-
flammatory cytokines do not lead to the small
clots characteristic of bacterial infection but
rather to blood clots in the major vessels. This
response occurs only when the C4bBP-protein S
levels are high. In every case tested to date, the
deleterious effects of the C4bBP can be reversed
by adding protein S.
Why then do bacterial infections lead to the
small clots in the circulation? Possible insights
were gained from the observation that small cell
fragment-like particles can convert this throm-
botic response back into the small-clot syndrome
seen in septic shock. Complement is known to
cause such cell fragments to be formed. These
studies may provide insights into how different
inflammatory and coagulation components work
in concert to generate such distinct — but
related — thrombotic complications.
These results suggest that the balance between
free protein S and that bound to C4bBP may be
critical to thrombotic complications as a result of
inflammation. The data imply that increasing
protein S would prevent these thrombotic com-
plications, suggesting new therapeutic ap-
proaches to thrombotic diseases. If we could
block the decrease in free protein S, the risk of
thrombosis might be reduced. This unique ap-
proach could potentially return the patient to
normal status without significantly increasing the
risk of bleeding.
A major interest in our laboratory has been to
understand how thrombomodulin causes throm-
bin to change its function from a clotting to a
clot-inhibiting enzyme. To understand this, we
have examined the binding of thrombin to throm-
bomodulin and to its other targets. Of particular
interest is the ability of thrombin to activate cells
and platelets, leading to platelet plugs and clots,
especially in the arteries. Thrombomodulin can
block platelet activation by thrombin, but how it
does so had been a mystery. In collaboration with
Shaun Coughlin's group, we demonstrated that
the thrombin receptor on platelets and thrombo-
modulin bind thrombin at the same place,
thereby explaining the earlier observations. Both
protein C and the thrombin receptor share com-
mon sequences near the cleavage sites, and both
sequences are inherently poor sequences for
cleavage by thrombin. Binding of thrombomodu-
lin fragments or the fragments from the thrombin
receptor results in changing how thrombin
cleaves its substrates. Thus both receptors have a
built-in switch that allows thrombin to function.
124
Molecular Genetics of Steroid and Thyroid
Hormone Receptors
Ronald M. Evans, Ph.D. — Investigator
Dr. Evans is also Professor at the Gene Expression Laboratory of the Salk Institute for Biological Studies
and Adjunct Professor in the Departments of Biology and of Biomedical Sciences at the University of
California, San Diego. He received his Ph.D. degree in microbiology and immunology from the University
of California, Los Angeles, School of Medicine. After postdoctoral training with James Darnell at the
Rockefeller University, he joined the faculty of the Salk Institute. Dr. Evans is a member of the
National Academy of Sciences. His research interests are in developmental and inducible regulation
of gene expression.
AN understanding of the mechanisms by
which apparently distinct regulatory sys-
tems integrate to modulate body function and be-
havior poses one of the most important chal-
lenges of modern biology. Hence we have
focused our attention on the action of steroid,
retinoid, and thyroid hormones in regulatory cel-
lular and organ physiology. This field has under-
gone an extraordinary development in the last
several years as a consequence of the cloning and
sequencing of the genes encoding the receptors
for these hormones in target cells.
It has been demonstrated that these receptors
are all structurally related and constitute a super-
family of nuclear regulatory proteins that are ca-
pable of modulating gene expression in a ligand-
dependent fashion. One challenge is to define
each receptor's molecular properties that deter-
mine its interactions with the transcription ma-
chinery regulating gene expression. Another
challenge is to elucidate the contributions of indi-
vidual regulatory systems to the integrated and
complex processes associated with cell growth,
differentiation, and organ function.
A Novel Retinoic Acid Response Pathway
The retinoids, a group of compounds that in-
clude retinoic acid, retinol (vitamin A), and a se-
ries of natural and synthetic derivatives, exert
profound effects on development and differentia-
tion in a wide variety of systems. (The retinoic
acid receptors are collectively designated RARs.)
Retinoic acid has also been shown to induce the
transcription of several genes, suggesting a role
analogous to those of steroid and thyroid hor-
mones. In previous studies we described the
cloning and characterization of a retinoic acid-
dependent transcription factor referred to as
RARa. Additional RAR-related genes have been
isolated, and at least three different RAR subtypes
(a, /3, and 7) are now known in mice and humans.
Retinoic acid receptors share homology with
the superfamily of steroid and thyroid hormone
receptors and have been shown to regulate
specific gene expression by a similar ligand-
dependent mechanism. Complicating these ob-
servations is our recent identification of a group
of receptors termed RXRs (retinoid X receptors) ,
which are only distantly related to the RARs. The
discovery of this second retinoid transduction
pathway led us to investigate its functional prop-
erties and determine its relationship to the RARs.
We now know that there are at least three RXR-
related genes (termed a, j3, and 7) located at ge-
netically distinct loci. Northern blot analyses of
the RXRs indicate that each isoform has a unique
pattern of expression in adult tissue and is tempo-
rally and spatially expressed in the embryo.
These studies suggest a role for RXRs in adult
physiology and embryonic development.
Binding experiments demonstrate that the RXR
protein has low affinity for retinoic acid (RA)
and, taken together with the transactivation stud-
ies, indicate that the RXR ligand may be a metabo-
lite of RA. Based on these assumptions, we de-
vised a strategy to identify the putative
metabolite referred to as retinoid X. The implicit
concept of the strategy was that all-trans RA may
be converted to retinoid X by a natural cellular
process. Accordingly, high doses of all-trans RA
were fed to recipient tissue culture cells. After
allowing for metabolic conversion, material ex-
tracted from these cells was fractionated to re-
solve the various retinoid peaks, and each peak
was assayed for its ability to activate the RXRs in a
transfection assay. A specific peak, termed reti-
noid X, was identified and characterized by mass
spectrometry.
Based on this process, we have now demon-
strated that 9-cis RA is the high-affinity ligand for
the RXR. While not previously seen in living or-
ganisms, 9-cts RA is apparently a new and widely
used vertebrate hormone. It transactivates RXRa
up to 40 times more efficiently than all-trans RA
and binds to RXR with high affinity. We also con-
firmed that all-trans RA shows no detectable
binding affinity for the RXR. Furthermore, each
of the RXR subtypes (a, /3, and 7) is activated by
9-c« RA with increased potency and efficacy rela-
tive to all-trans RA.
125
Molecular Genetics of Steroid and Thyroid Hormone Receptors
A point of potential physiological significance
is that 9-cis RA also binds to and transactivates
both RXRs and RARs and may thus serve as a com-
mon or "bifunctional" ligand. Conversion of the
all-trans to the 9-cis isomer could provide a
novel means for differential cell-speciftc regula-
tion of the activity of these retinoid pathways.
The hypothesis that 9-cis RA may be functionally
distinct from its all-trans precursor raises the in-
triguing possibility that the regulation of its iso-
merization could be a key step in retinoid physiol-
ogy. It is unknown whether this reaction is
catalyzed by an enzyme.
The 3-4-5 Rule
Members of the receptor superfamily modulate
target gene expression by binding as either homo-
or heterodimers to hormone response elements
(HREs). We recently described the properties of
direct repeats of the consensus half-site sequence
AGGTCA as HREs for nuclear receptors. Receptor
specificity for binding and activation was shown
to be conferred through the number of nucleo-
tides separating the two half-sites. Spacers of 3, 4,
or 5 nucleotides were originally shown to serve
as optimal response elements for the vitamin D
receptor (VDR), thyroid hormone receptor (TR),
and RAR, respectively. We now refer to this physi-
ological code built into HREs as the "3-4-5 rule."
More recently, we have characterized in the up-
stream regulatory region of the cellular retinol-
binding protein type II (CRBPII) gene an HRE
that confers selective responsiveness to the RXRs.
This response element consists of tandem repeats
of the AGGTCA sites separated by a single
nucleotide.
The ability of the VDR, TR, and RAR to recog-
nize their cognate response elements is depen-
dent upon their ability to form a heterodimeric
complex with an unknown nuclear factor. Re-
markably, the search for this factor has recently
identified RXR as a common heterodimeric
partner for the VDR, TR, and RAR. These results
emerged from our initial finding of a functional
interaction between the RAR and RXR, followed
by the demonstration that these two receptors
form stable heterodimers in solution. Further-
more, the heterodimer binds target DNA with
more than 100-fold increased efficiency over ei-
ther partner alone.
Similarly, RXR heterodimers with the VDR and
TR have greatly increased affinity for the respec-
tive cognate response elements. Apparently RXR
is serving as a type of master receptor, gating the
activities of vitamin D, thyroid hormone, and reti-
noic acid. As noted above, however, RXR can act
independently of all these receptors when re-
sponding to its novel hormone 9-cis RA. Despite
the apparent level of complexity revealed by
these interactions, it appears that the physiologi-
cal response is built upon a series of simple prin-
ciples in which heterodimers, each consisting of
an RXR component, interact with common DNA
sequences, each varying by a single nucleotide. It
is through this simplicity that such exquisite
specificity can be maintained and thus permit co-
ordinate control of a vast gene network in a
highly selective and orderly fashion.
Opposite: Expression pattern of the genes for RXRa, ^, and 7 in a mouse embryo. RXRs are
receptor proteins that mediate the actions of the newly described hormone, 9-cys retinoic acid.
Parasagittal sections from the embryo taken at gestation day 16.5 were hybridized in situ with
radioactive probes that mark the specific areas where RXR mRNA is expressed. After exposure to
x-ray film, the resulting image is digitally scanned into a computer and colorized. Dark red to
yellow colors represent low to high levels of RXR expression. Thus the figure demonstrates that
RXRa is strongly expressed in skin and metabolic organs such as intestine; RXR^ is expressed
ubiquitously; RXRy shows marked expression in the pituitary and corpus striatum, the brain
center that regulates muscle coordination and is affected in parkinsonian disorders.
Reprinted with permission from Mangelsdorf D.J., Borgmeyer, U., Heyman, R.A., Zhou,J.Y.,
Ong, E.S., Oro, A.E., Kakizuka, A., and Evans, R.M. 1992. Genes Dev 6:329-344. Computer
photograph by Jamie Simon, the Salk Institute for Biological Studies.
126
1 mtji I fji mry
DEVELOPMENT
alpha
gamma
Spring Harbor Laboratory Press
in association with
The Genetical Sticiety of Great Britain
127
Molecular Mechanisms Involved in the Actions
of Calcium-mediated Hormones
John H. Exton, M.D., Ph.D. — Investigator
Dr. Exton is also Professor of Molecular Physiology and Biophysics and of Pharmacology at the Vanderbilt
University School of Medicine. He received his medical degree from the University of New Zealand and his
Ph.D. degree in biochemistry from the University of Otago, New Zealand. His postdoctoral research was
done in the Department of Physiology at Vanderbilt University School of Medicine with Charles Park,
where he has remained as a faculty member. His honors include the Lilly Award of the American Diabetes
Association and the M.D. degree with distinction from the University of Otago.
THE major objective of my laboratory is to elu-
cidate the mechanisms of action of hor-
mones, neurotransmitters, and other agents that
transmit information in the nervous system and
other organ systems by altering membrane lipids
and increasing the concentration of calcium ions
in their target cells. Agents that act this way in-
clude regulators of heart function and blood
flow, such as epinephrine, norepinephrine, ace-
tylcholine, angiotensin, and vasopressin; other
neurotransmitters, such as serotonin, neuroten-
sin, and substance P; and agents that control cer-
tain pituitary and pancreatic secretions, food di-
gestion, bladder and uterine contraction, platelet
aggregation, and certain responses to trauma and
infection.
We initially established that many hormones
and neurotransmitters act by increasing the intra-
cellular concentration of calcium ions. The next
phase of our work involved the demonstration
that the increase is due to both mobilization of
calcium ions from internal stores and stimulation
of their inflow across the cell membrane. We also
demonstrated that the receptors for calcium-me-
diated agents are located on the outer surface
of their target cells. Thus these cells must
have some means of signaling from the receptors
to the internal calcium stores, and our eff^orts
were directed toward elucidating the signaling
mechanism.
Initially we tested the hypothesis that the sig-
nal is generated by the breakdown of phosphati-
dylinositol, a phospholipid in the cell mem-
brane. However, this was found to be too slow to
account for the changes in calcium. Attention was
then focused on a related phospholipid, phos-
phatidylinositol 4,5-bisphosphate (PIP2), which
breaks down more rapidly in response to hor-
mones. The situation became clearer when inosi-
tol 1,4,5-trisphosphate (IP3) was identified as
the signaling molecule for intracellular calcium
release. This compound is generated when PIP2 is
broken down by the enzyme phospholipase C. The
other compound produced is 1 ,2-diacylglycerol
(DAG), which is also a signaling molecule, since
it activates a specific protein-phosphorylating
enzyme, protein kinase C.
The present activities of the laboratory encom-
pass two major research areas. The first involves
elucidating how calcium-mediated agents stimu-
late the breakdown of PIP2. A major discovery has
been the finding that a G protein (a regulatory
protein that binds the nucleotide GTP) is in-
volved in coupling the receptors for these agents
to the phospholipase C enzyme that breaks down
PIP2. Our work has involved purifying and char-
acterizing the relevant G proteins from liver cell
membranes and reconstituting them with other
components of the signaling system. Reconstitu-
tion of the G proteins with PIP2 phospholipase C
has been achieved, and the system has been used
to purify the G proteins to homogeneity in both
the complete form {a^y heterotrimers) and in
the form of free a-subunits. Two a-subunits (42
and 43 kDa) have been identified, and both have
been shown immunologically to be members of
the Gq family of G proteins. Partial sequencing of
tryptic peptides of both proteins has confirmed
that they are Gaq and Gq:,i.
Purification of the G protein activators of the
phospholipase C in the heterotrimeric form has
likewise yielded two G proteins corresponding to
Gq and Gn. Using Gofq and Gofn activated by a
GTP analogue, we have shown that the specific
isozyme form of the PIP2 phospholipase C con-
trolled by these G proteins in both liver and brain
is the 148-kDa -isozyme. Evidence that Gq
and G,] are coupled to receptors for calcium-
mobilizing agents has also been obtained. Three
calcium-mediated agents (vasopressin, epineph-
rine, and angiotensin) specifically stimulate the
labeling of Gaq and Ga,, by a radioactive, light-
reactive analogue of GTP in liver cell mem-
branes. Binding of a GTP analogue is enhanced by
an acetylcholine analogue when Gq and G,j are
reconstituted with the M, (calcium-mobilizing)
muscarinic receptor but not with the M2 recep-
tor. Co-reconstitution of the purified Mj recep-
tor, Gq/Gi,, and phospholipase C-/3i permits
agonist-stimulated hydrolysis of PIP2 in a GTP ana-
logue-dependent manner. The activation of the
phospholipase is paralleled by activation of the G
proteins, as measured by their binding of the
129
Molecular Mechanisms Involved in the Actions
of Calcium-mediated Hormones
GTP analogue. These findings demonstrate the
complete reconstruction of a signaling pathway
for a calcium-mediated agonist from purified
components.
The second major research area arose from the
observation that many hormones, neurotransmit-
ters, and growth factors break down another cell
membrane phospholipid, phosphatidylcholine
(PC), in a wide variety of cells. This breakdown
yields DAG and phosphatidic acid (PA) and in-
volves two other phospholipases (C and D),
which are being purified and characterized. Fur-
ther work has shown that PC breakdown is regu-
lated by mechanisms involving G proteins, cal-
cium ions, protein kinase C, and tyrosine kinases.
The latter two mechanisms are currently being
explored in several cell types.
In a related area, the regulation of different iso-
zymes of protein kinase C by certain molecular
species of DAG and PA is being studied. This in-
volves the purification of a protein kinase C iso-
zyme that is stimulated by PA but not DAG. These
studies should elucidate the functions and mecha-
nisms of this novel signaling system.
130
Tumor-Suppressor Genes
Andrew p. Feinberg, M.D., M.P.H. — Associate Investigator
Dr. Feinberg is also Associate Professor of Internal Medicine and Human Genetics at the University
of Michigan Medical School. He received his B.A., M.D., and M.P.H. degrees from the Johns Hopkins
University. He received clinical training at the University of Pennsylvania and Johns Hopkins and did
postdoctoral research at the University of California, San Diego, and Johns Hopkins. Before moving to the
University of Michigan, Dr. Feinberg was Assistant Professor of Oncology and Medicine at Johns Hopkins.
ONE of the most important areas of cancer ge-
netics is the identification and characteriza-
tion of tumor-suppressor genes, whose inactiva-
tion contributes to cancer. Since almost all genes
are present in two copies in the cell, cancer
would develop from deletion or inactivation of
both copies of these suppressor genes. Inactiva-
tion of one copy could be transmitted in families
from parent to child. Thus individuals inheriting
one nonfunctional copy of the gene would be at
increased cancer risk.
Wilms' tumor (WT), a childhood kidney
cancer, was one of the earliest models of suppres-
sor gene action. Strong and Knudson showed 20
years ago that WT apparently resulted from two
mutations, based on two peaks in the age of onset.
Furthermore, it was discovered in the 1970s that
some children with WT lack a large portion of
chromosome 1 1 (in band llpl3), which is visi-
ble microscopically. Before joining HHMI, I
observed (with Eric Fearon and Bert Vogelstein)
that more subtle gene deletions can be detected
indirectly on chromosome 1 1 in WTs, through
use of restriction fragment length polymor-
phisms (RFLPs), which can distinguish the ma-
ternal and paternal copies of a given gene.
In collaboration with David Schlessinger
(Washington University, St. Louis) and Bryan Wil-
liams (Hospital for Sick Children, Toronto), our
laboratory has now cloned the 1 1 p 1 3 WT gene
region in yeast artificial chromosomes (YACs) . In-
terestingly, we found that this region contains
multiple genes turned on specifically in develop-
ing kidney. At least two of these genes showed
reduced or absent expression in approximately
half of sporadically occurring WTs, notably those
of the same histologic type that occur in children
with 1 lpl3 deletions. One of these genes codes
for a DNA-binding protein that is mutated in
some WTs.
These mutations, however, occur infrequently,
and the laboratory is investigating whether re-
duced expression of the gene may be a more
common mechanism for tumorigenesis and
whether other genes from this complex may play
a role. We are also "retrofitting" WT YACs with a
gene that allows growth in mammalian cells, in
order to demonstrate directly a tumor-suppressor
phenotype of the 1 lpl3 WT gene(s) and to de-
termine by deletion experiments the functional
role of the multiple genes from this region.
In addition to the known gene on chromosome
1 1 that predisposes to WT, we have discovered a
second predisposing gene at a different location
on the chromosome (band 11 p 1 5 ) . This gene ap-
pears to be involved in bladder, breast, and lung
cancer, as well as WT. Thus WT causation is more
complex than investigators had previously be-
lieved, and this second WT gene may turn out to
be important in common cancers.
Consistent with this gene being a tumor sup-
pressor, the laboratory has applied genetic
linkage analysis to map to the same region
of 11 pi 5 a cancer-predisposing disorder, Beckwith-
Wiedemann syndrome (BWS). Using YACs, the
laboratory has now isolated several DNA break-
points from BWS patients with germline chromo-
somal rearrangements. This should enable the lab-
oratory to determine whether the BWS gene is
indeed the second WT-suppressor gene on the
llpl5 band.
Adding to the complexity of WT is the fact that
some non-BWS families with hereditary predispo-
sition to WT do not show linkage of this trait to
chromosome 1 1 . Recently the laboratory found,
in collaboration with former sabbatical member
Anthony Reeve (University of Otago, New Zea-
land), involvement of chromosome 16.
The laboratory has recently developed a novel
approach to isolating tumor-suppressor genes di-
rectly. Human chromosomes are fragmented into
2-5 million base pair "superfragments" in a way
that allows their transfer into any mammalian
cell. The advantage of this technique is that it
enables one to transfer the giant DNA pieces into
a recipient cell and screen directly for a func-
tional gene. It may thus have immediate ap-
plication to cloning other chromosome 1 1
tumor-related genes, or more general application
to cloning a variety of genes on other chromo-
somes— genes for which one can currently
screen but not select, such as those for cellular
aging.
For example, one should be able to exploit the
131
Tumor-Suppressor Genes
cancer-suppressing property of tumor-suppressor
genes to identify the key chromosomal region
containing the gene. In a test of this method, 8 of
nearly 100 chromosome 11 superfragments that
the laboratory isolated, when introduced into WT
cells, suppressed their neoplastic growth. Inter-
estingly, the tumor-suppressing hybrids contain a
small portion of 1 lpl5, demonstrating the exis-
tence of a suppressor gene on this band. The labo-
ratory is now testing whether these superfrag-
ments also suppress important common cancers
that involve lip, such as lung cancer. This novel
technique may allow investigators to bridge a gap
in cloning methods between chromosome-size
pieces (averaging 100 million nucleotides) and
YACs (averaging 300,000 nucleotides), and it
may have general application to cloning a wide
variety of genes.
Another long-standing interest of the labora-
tory is the role of epigenetic changes (not involv-
ing DNA sequence, and thus potentially re-
versible) in cancer. The laboratory recently
identified several CpG islands (DNA sequences
rich in cytosine-guanine dinucleotides, often
found near actively expressed genes) within the
1 Ipl 3 WT gene region. Some of these sequences
were methylated, a reversible modification of the
nucleotide cytosine. Previously CpG islands
were only known to be methylated on the inac-
tive X chromosome. The surprising finding of
methylated autosomal CpG islands suggests that
epigenetic changes may play a role in Wilms' tu-
morigenesis. We are now testing this hypothesis
by examining the islands directly.
We also observed that balanced BWS chromo-
somal translocations on llpl5 always involve
the maternal chromosome, while unbalanced
BWS duplications always involve the paternal
chromosome, suggesting an epigenetic differ-
ence (or imprint) between the two alleles. If
such epigenetic changes are found to contribute
to inactivation of 1 Ip suppressor genes, this will
represent an exciting convergence of our two
major interests.
132
' ' ' ' w
Genetics, Structure, and Function
of Histocompatibility Antigens
Kirsten Fischer Lindahl, Ph.D. — Investigator
Dr. Fischer Lindahl is also Professor of Microbiology and Biochemistry at the University of Texas
Southwestern Medical Center at Dallas. She began the study of histocompatibility with Morten Simonsen
in Copenhagen, Denmark, and received her Ph.D. degree in immunobiology from the University of
Wisconsin-Madison. She was a postdoctoral fellow with Darcy Wilson at the University of Pennsylvania,
Philadelphia, and with Klaus Rajewsky at the Institute for Genetics in Cologne, West Germany. Before
accepting her current position, she was a member of the Basel Institute of Immunology in Switzerland.
HISTOCOMPATIBILITY (H) antigens are cell
surface molecules that, when foreign, lead
to the rejection of grafted tissues and organs by
the vertebrate immune system. Because they form
a major obstacle to clinical transplantation, H an-
tigens have been studied for over 50 years. They
are complexes of a small peptide ligand and an
MHC molecule (encoded by genes of the major
histocompatibility complex). A given individual
has MHC molecules of a few different kinds, each
of which can present to the immune system a
large variety of peptides on the surface of cells.
These peptides might be derived from proteins
produced by intracellular parasites, bacteria, or
viruses or by the body's own cells, such as tumor-
specific antigens or minor H antigens. The amino
acid side chains that line the peptide-binding
groove of an MHC molecule determine which
peptides that molecule can bind and therefore
what antigens can be presented to induce an im-
mune response in the individual with this MHC.
The immune system is capable of recognizing a
difference in either of the H antigens' two parts. A
difference in the MHC molecule itself will alter
many complexes and induce a strong immune re-
sponse, hence the term "major" H antigen. By
contrast, a difference in a peptide alters only one
of many kinds of complexes and induces a weaker
response, hence the term "minor" H antigen. Un-
like the major H antigens, human minor H anti-
gens remain ill defined. In the mouse, however,
more than 50 genes that encode minor H antigens
have been mapped. Almost every chromosome,
including the mitochondrial genome, carries at
least one.
The Maternally Transmitted Antigen
In the mouse mitochondrial protein NDl, the
sixth amino acid is polymorphic. When cells with
one form are transplanted to a mouse with an-
other form, the amino-terminal peptide of NDl
will act as a transplantation antigen, called Mta.
This peptide is presented on the cell surface by
an MHC class I molecule called M3. M3 binds the
NDl peptide only when the methionine at the
end carries a formyl group, and M3 can also bind
other peptides with a formyl-methionine. This
characteristic of the amino terminus of mito-
chondrial and bacterial proteins distinguishes
them from proteins made in the cytoplasm of
mammalian cells. We are collaborating with
Michael Bevan (HHMI, University of Washington,
Seattle) , who is studying the immune response of
mice to the bacterium Listeria. His group reports
that the M3 molecules present a Listeria antigen
on the cell surface to cytotoxic T cells, which kill
the infected cells.
It is of great interest to understand how the
formyl-methionine peptides are bound by M3 and
which amino acids in M3 are important for this
specificity. We cloned the gene for M3 from mice
and rats. The rat and mouse M3 genes are more
similar to each other than they are to the other
MHC class I genes of their own species. This is
particularly striking in the rat, where the class I
genes are otherwise very similar. These results
suggest that the specialized function of M3
evolved long ago in a species from which both
rats and mice are descended, and that it has been
conserved in both species during their separate
evolution, presumably because it is useful in the
immune response against bacterial peptides.
We can express M3 in mouse fibroblasts, where
Mta can be detected by killer T lymphocytes. This
system allows us to change single amino acids in
the protein by introducing mutations at specific
sites in the gene. We are currently looking at
amino acids 34 and 171, which differ in M3 from
the consensus of MHC class 1 molecules and are
near the site where the formyl-methionine of the
peptide is expected to lodge. In sequencing natu-
rally occurring variants of the M3 gene from wild
mice, we are learning more about which amino
acids are essential for the ability to present the
NDl peptide. Amino acid 95 points straight up in
the peptide-binding site, probably right under
the variable sixth residue of the peptide. The sin-
gle change of 95 from leucine to glutamine com-
pletely abolishes T cell recognition of Mta. We
are collaborating with the laboratory of Johann
Deisenhofer (HHMI, University of Texas South-
133
Genetics, Structure, and Function of Histocompatibility Antigens
western Medical Center at Dallas) to produce M3
in amounts sufficient for a structural analysis.
RMA-S Mutant Cells
The heavy chains of MHC class I molecules are
not stable in the properly folded conformation at
body temperature unless they bind a peptide as
well as the 182-microglobulin light chain. RMA-S
mutant cells make MHC class I heavy chains and
j82-microglobulin, but do not display them on
their surface. The defect can be circumvented by
adding suitable synthetic peptides to the mutant
cells, which will then display those MHC mole-
cules that bind the added peptide. Not only clas-
sical MHC class I molecules and M3, which were
already known to present peptides, are reduced
or missing from these cells, but also other MHC
class I molecules, such as Qa-1 . These molecules
are recognized by antibodies or killer T lympho-
cytes, but their physiological function is not yet
clear. The RMA-S cell line shows that they all bind
peptides.
The RMA-S mutation affects one of two MHC
genes whose protein products together make a
channel to transport peptides from the cytoplasm
into the endoplasmic reticulum, where MHC
class I molecules fold. Surface display of all such
molecules, including Mta and Qa-1, is restored
when a normal version of this gene is expressed
in the RMA-S cell line. This tells us that the mito-
chondrial peptide is transported from the cyto-
plasm into the endoplasmic reticulum by the
same mechanism that other peptide antigens use.
We have found that the drug oligomycin, which
stimulates protein degradation in mitochondria,
increases the surface display of both Mta (as ex-
pected) and Qa-1 on RMA-S cells, suggesting that
Qa-1 can also bind a mitochondrial peptide. We
believe that the increased supply of these pep-
tides floods the inefi'ective transport system and
overcomes the defect.
/?2-Microglobulin Polymorphism
Mice are the only mammals in which allelic
forms of /32-microglobulin have been described.
Because of its intimate interaction with the pep-
tide-binding parts of MHC class I heavy chains, an
allelic difi'erence of 182-microglobulin that alters
amino acid 85 from aspartic acid to alanine can
subtly aff'ect the structure of MHC class I mole-
cules and their ability to bind particular pep-
tides. We have found four new alleles of (82-
microglobulin among wild mice of the species
Mus musculus. These differ by only one or two
amino acids relative to the |82microglobulin of
inbred mice. Three of these wild alleles have va-
line in position 85 and may therefore alter pep-
tide binding.
We also looked at the |82-niicroglobulin gene
from the species Mus spretus, which has been
separated from Mws by about 1 million
years of evolution. To our surprise, this (82-
microglobulin differed from that of inbred mice
by 12 amino acids and had only a single, silent
mutation, which afi'ected the DNA sequence, but
not the protein. For comparison, rat and mouse
/32-microglobulins differ by 14 amino acids. We
would have expected at least an equal number of
silent mutations and amino acid changes for the
spretus form.
Such drastic divergence contrasts with the mod-
est changes seen in other members of the immu-
noglobulin superfamily and suggests that the
molecule has been selected for its differences.
The diversification may be part of the speciation
process, but it may be limited to molecules
that can tolerate change. All changes in the
microglobulin molecule are in loops or on the
face away from the MHC class I heavy chain, spar-
ing the face that interacts with the conserved do-
main of the heavy chains.
134
Borrelia burgdorferi, the Lyme disease agent in the gut of deer ticks. Ticks that feed on immunized
mice show elimination of the spirochetes ( right), whereas ticks that feed on control mice contain
numerous spirochetes (left).
From Fikrig, E., Telford, S., Barthold, S.W., Kantor, F.S., Spielman, A., and Flavell, R.A. 1992.
Proc Natl Acad Sci USA 89:5418-5421.
136
Genetic Approaches to Immune Function
and Tolerance
Richard A. Flavell, Ph.D. — Investigator
Dr. Flavell is also Professor of Immunobiology at Yale University School of Medicine. He received k>is B.Sc.
and Ph.D. degrees in biochemistry from the University of Hull, England, and performed postdoctoral work
in Amsterdam and Zurich. Before accepting his current position, Dr. Flavell was first Assistant Professor
at the University of Amsterdam, then Head of the Laboratory of Gene Structure and Expression at the
National Institute for Medical Research, Mill Hill, London, and subsequently President and Chief Scientific
Officer of Bio gen Research Corporation, Cambridge, Massachusetts. Dr. Flavell is a Fellow of the Royal
Society and a member of other distinguished societies.
MY laboratory has concerned itself for many
years with the regulation and function in
the immune system of the genes of the murine
MHC (major histocompatibility complex). In the
mouse these genes are encoded on chromosome
17, and prior work has shown that there are a
large number of linked class 1-related genes and
a handful of class II genes. Class I genes encode a
protein of approximately 45,000 molecular
weight that is found in association with a small
subunit, i82"r"icroglobulin. Together this com-
plex forms a symmetrical molecule consisting of
four extracellular globular domains anchored
through the cell membrane with the transmem-
brane segment and having a short stretch of
amino acids that extend into the cytoplasm. Class
II molecules achieve a similar symmetry, but
with two polypeptide chains, a and /?, each of
which has two extracellular domains and a trans-
membrane and cytoplasmic segment.
Both class I and class II gene products serve as
recognition elements, which bind antigenic pro-
tein fragments and present them to T cells. In the
case of class I genes, the presentation is to T cells
carrying the CDS co-receptor molecule. These
cells are usually cytotoxic T cells, whose role is
to destroy cells that are virally infected. In the
case of class II molecules, it is commonly intra-
cellular pathogens that are presented, this time to
helper or inflammatory T cells that carry the CD4
co-receptor. Both types of T cells secrete hor-
mone-like molecules called lymphokines, which
in turn act on other cell types — for example, on B
cells, which are stimulated to multiply and to
make antibody.
Class I and class II genes are both regulated in
vivo by various lymphokines. Of these, the inter-
ferons are important and are currently utilized in
the therapy of several human diseases. For exam-
ple, interferon-7 secreted by activated T cells
stimulates the synthesis of MHC class 1 and II mol-
ecules and, as a result, presumably renders a cell
better able to present antigen and thus to poten-
tiate an immune response. We have taken a ge-
netic approach to attempt to understand how in-
terferon-7 activates the synthesis of these MHC
molecules. A new strategy was used to isolate a
series of mutant cell lines that are not capable of
responding to interferon-7. These mutants ap-
pear to have a series of different defects. In some
of these cell lines, the mutations result in a com-
plete loss of the ability of all genes to respond to
interferon-7 and even, surprisingly, interferon-a
and which were previously believed to use
distinct mechanisms. Other cell lines are defec-
tive only in the response of the class II and related
genes. This genetic approach should help us un-
derstand how these important molecules regulate
gene expression by elucidating the molecular
steps that the cell utilizes to activate genes
through interferons.
One important issue in the functioning of the
immune system is how the body discriminates its
own tissues (self) from foreign components such
as pathogens. The process that protects the indi-
vidual against the destruction of self tissues is
known as immune tolerance. Tolerance is gener-
ally believed to be established during the produc-
tion of new T cells in the thymus by a process
called negative selection, which is mediated by
clonal deletion; that is, self-reactive T cells are
destroyed at the site of synthesis. The failure of
self-tolerance leads to the autoimmunity that
characterizes human diseases such as autoim-
mune diabetes (insulin-dependent diabetes mel-
litus [IDDM]) and rheumatoid arthritis.
In the past few years we have been interested in
determining the mechanisms of tolerance to
those components of the body that are never
found in the thymus and therefore pose a prob-
lem for tolerance mechanisms operating in the
thymus. Transgenic mice can be used to study
this process, since the expression of a given gene
— and hence the protein encoded by that gene —
can be directed to the tissue of choice by linking
the gene for the desired protein to the regulatory
signals that function in that specific tissue. We
have previously performed such experiments by
directing the synthesis of MHC class II proteins to
the pancreatic jS-cells of transgenic mice. In these
137
Genetic Approaches to Immune Function and Tolerance
experiments the mice were indeed found to be
tolerant to the MHC antigens. T cells that would
normally react with this MHC product are not
eliminated, which is what would be found if the
class II antigen is expressed in thymic tissues. In-
stead, these T cells are present but have been in-
activated in some way, such that they are no
longer able to respond to the MHC antigen, either
in the animal or in test-tube experiments. As a
result, no destruction of the pancreatic tissue
occurs.
To determine mechanisms of peripheral toler-
ance to protein antigens, rather than MHC, we
made transgenic mice that express the T cell re-
ceptor (TCR) reactive with a fragment of the T
antigen of SV40 (simian virus 40). We have ob-
tained transgenic mice that express SV40 T anti-
gen in various peripheral tissues in the body, in-
cluding the |8-cells of the pancreas and the
secretory exocrine cells of the pancreas that pro-
duce digestive enzymes. By crossing the TCR
transgenics with mice expressing the antigen, we
can determine the immune status to SV40 T anti-
gen. Tolerance is found to the SV40 T antigen
provided that expression has occurred prior to
the release of T cells from the thymus in the new-
born animal. This tolerance results in partial elim-
ination of antigen-specific T cells and clonal
inactivation of the remainder. If, however, ex-
pression of antigen is delayed, the T cells
are not tolerant but instead become activated,
with resultant autoimmune destruction of the
pancreas.
Delayed expression of antigen therefore seems
to place an individual at risk for autoimmunity.
However, even in individuals genetically predis-
posed to become autoimmune, nongenetic fac-
tors play an important role. One model suggests
that a precipitating event for autoimmunity can
be the onset of a local inflammatory response in
the tissue, for example, as a consequence of in-
fection. To test this we made transgenic mice that
express the inflammatory cytokine TNF (tumor
necrosis factor) on the pancreatic islets, which
simulates local inflammation. This cytokine ex-
pression causes a massive infiltration of T and B
lymphocytes, which mimics the infiltrate seen in
mouse autoimmune diabetes and causes partial
destruction of the insulin-producing j8-cells of
the pancreas. We are currently determining
whether the inflammatory cells present in the
TNF transgenic mice are specific for pancreatic
antigens, as is the case in true IDDM.
We also study the response of the immune sys-
tem to the spirochete Borrelia burgdorferi,
which causes the notorious inflammatory dis-
ease, Lyme disease. We showed last year that mice
vaccinated against an outer surface protein of
Borrelia (OspA) are protected against infection
by Borrelia burgdorferi. We have now shown
that protection can also be mediated by a second
protein (OspB) but not by the flagellar antigen,
even though a potent antibody response is ob-
tained to this protein.
For such a vaccine to work it must be eff'ective
against most, if not all, strains of the infectious
agent and be effective against a natural challenge.
Since the natural vector is the deer tick, we also
infected laboratory mice with B. burgdorferi by
placing infected ticks on these mice. Encourag-
ingly, mice vaccinated with either OspA or OspB
were protected; furthermore, spirochetes in the
gut of ticks feeding upon vaccinated mice were
eliminated, whereas they were unaffected in non-
vaccinated mice. This shows an additional poten-
tial benefit of the vaccine, which is to eliminate
spirochetes from the vector. We are now con-
sidering strategies for the eradication of the spiro-
chetes in the wild population of ticks by a related
approach.
138
Biophysical Genetics of Protein Structure
and Folding
Robert O. Fox, Ph.D. — Associate Investigator
Dr. Fox is also Associate Professor of Molecular Biophysics and Biochemistry at the Yale University School
of Medicine. He received his B.S. degree in biochemistry from the University of Pittsburgh and his M.Phil,
and Ph.D. degrees in molecular biophysics and biochemistry from Yale University while working with
Frederic Richards in the area of x-ray crystallography. He carried out postdoctoral studies at Yale University
in protein engineering with Nigel Grindley and studied protein folding using NMR spectroscopy
with Christopher Dobson at Oxford University as a Fellow of the Jane Coffin Childs Memorial Fund
for Medical Research. Before moving to Yale, Dr. Fox was Assistant Professor in the Department
of Cell Biology at Stanford University Medical School.
ALTHOUGH the information that directs the
folding of a protein molecule into a defined
three-dimensional structure is genetically en-
coded, the mechanisms and pathways of the fold-
ing process are poorly understood. One approach
to this problem is an analysis of partially struc-
tured folding intermediates, combined with a
mutational analysis. We use nuclear magnetic res-
onance (NMR) spectroscopy and chemical meth-
ods to probe for structural and kinetic interme-
diates in the folding process.
Many polypeptide sequences adopt a common
folded motif, but they frequently differ in the de-
tailed arrangement or conformation of structural
elements in ways that are functionally significant.
Certain loops of the immunoglobulins (antibod-
ies) are examples. We are using staphylococcal
nuclease as a model protein system to understand
how the amino acid sequence of a secondary
structural element dictates its detailed conforma-
tion in the context of a folded protein molecule.
We combine a number of methodologies in these
studies, including x-ray crystallography, NMR
spectroscopy, and molecular biology.
Mapping Structure in the Unfolded State
of Proteins
Protein molecules in the unfolded and molten
globule states are often more compact than
would be expected for a true random-coil confor-
mation. If this conformational bias is toward that
of the folded structure, it may explain the rapid
rate at which proteins fold. We have developed a
chemical approach to map close contacts be-
tween a variable-reporter residue site and all
other residues of a protein chain in these states.
This approach is being used to investigate staphy-
lococcal nuclease variants, nuclease fragments,
and the molten globule state of myoglobin. A po-
lar chelator has been designed and synthesized
that can be specifically attached to a cysteine resi-
due engineered into the protein chain. When this
chelator is loaded with iron and the reaction is
initiated with a reducing agent, hydroxyl radicals
and other reactive oxygen species are generated;
these in turn cleave peptide bonds at positions in
the protein chain in proximity to the chelator.
The cleavage sites can be determined by peptide
mapping and protein sequencing. The reagent
cleaves native proteins at a number of solvent-
accessible sites close to the site of attachment.
The reagent has also been used to map the prox-
imity of several residues of 76 resolvase to its
DNA-binding sites.
Analysis of Protein Folding Using NMR
Spectroscopy
Protein molecules are generally thought to
adopt a final tertiary structure where all back-
bone and side chain conformations and tertiary
contacts are within local energy minima. Several
well-refined protein molecules display excep-
tions to this view, where significant strain is in-
duced at a particular point in the structure,
usually involving a residue in the enzyme ac-
tive site. Examples include c/s-peptide bonds,
eclipsed side chain rotamers, and energetically
unfavorable van der Waals contacts. How a pro-
tein imposes the stress needed to favor locally
strained conformations remains unclear.
Staphylococcal nuclease provides an excellent
system to examine the relationship between
stress and strain in a globular protein. Residues
115-118 adopt a type Via /3-turn, containing a
c/5-peptide bond, forming one wall of the nu-
cleotide-binding pocket of the active site. Al-
though the conformation of the Lysl l6-Prol 17
peptide bond of this jS-turn is predominantly in
the c/5-configuration, the equilibrium between
cis- and ^ra«5-conformers can be monitored by
NMR spectroscopy. Site-directed mutants within
this type Via ;8-turn have been examined by NMR
spectroscopy and x-ray crystallography. Limiting
the backbone conformational space available to
the residue preceding the cw-proline contributes
to the stress favoring the strained c/5-peptide
bond conformation. We have begun a collabora-
tion with Axel Brunger (HHMl, Yale University)
139
Biophysical Genetics of Protein Structure and Folding
to simulate this system using molecular dynamics
techniques.
A structural analysis of early protein-folding in-
termediates has been difficult because of their
transient nature. Amide hydrogen exchange has
been used by others to examine the appearance of
hydrogen-bonded protein-folding intermediates
for a number of proteins. We have begun to use
this technique to examine early folding interme-
diates of a staphylococcal nuclease variant. Par-
tial protection of amides occurs early in the fold-
ing process, despite the long overall folding time
for nuclease.
Genetic Analysis of a /S-Turn
A sharp change in the trajectory of a polypep-
tide chain between secondary structure elements
in a globular protein has been defined as a reverse
turn or j8-turn. These structures occur in a num-
ber of defined geometric types and frequently
contribute side chains to the active site of the
enzymes, such as staphylococcal nuclease, or the
combining site of binding proteins, such as the
immunoglobulins. We wish to determine the se-
quence requirements for the formation of differ-
ent /3-turn types to better understand the detailed
structure of globular proteins and to define de-
sign principles for protein engineering.
We have developed a genetic approach to de-
termine which amino acid sequences are consis-
tent with a particular |S-turn structure in staphylo-
coccal nuclease. Each member of our gene
library contains a unique sequence at this jS-turn.
Only a small fraction of the sequences examined
are consistent with an enzymatically active and
stable protein in Escherichia coli. The |8-turn
under consideration is well removed from the ac-
tive site, suggesting that the modulation in the
observed enzyme activity is due to changes in the
stability of the protein. There are strong biases in
the amino acids occurring at each position in the
i8-turn. Recently, a statistical analysis of these data
has led to a predictive model for this /3-turn type
in all globular proteins. This approach may be
useful in defining other sequence-secondary
structure relationships. Results of this and related
experiments should provide insight into the rela-
tionship between amino acid sequence and struc-
ture required for the rational design and engineer-
ing of protein molecules.
Structural Studies of Trypanosome
Calmodulin
Calmodulin serves as a calcium-dependent reg-
ulatory subunit of a variety of cytoskeletal pro-
teins and cytoplasmic enzymes, including a num-
ber of protein kinases. The novel structure of rat
calmodulin was determined in the laboratory of
Charles Bugg. The protein is composed of two
largely helical domains, each with two calcium-
binding sites, separated by an extended solvent-
exposed a-helix. We have obtained crystals of
calmodulin from Trypanosoma brucei rhode-
siense, in collaboration with Curtis Patton (Yale
University). The crystal structure in progress
should serve as the basis for rational drug design
against this and related organisms.
140
Molecular Basis of Genetic Diseases
and Chromosome Mapping
Uta Francke, M.D. — Investigator
Dr. Francke is also Professor of Genetics and Pediatrics at Stanford University School of Medicine. She
received her M.D. degree from the University of Munich, Germany, trained in pediatrics at Los Angeles
Children 's Hospital, and carried out postdoctoral research and clinical training in medical genetics
at the University of California (Los Angeles and San Diego). Before moving to Stanford, Dr. Francke
was Professor of Human Genetics and Pediatrics at Yale University School of Medicine.
THE genetic maps of humans and the mouse
are undergoing rapid growth and develop-
ment. Through the efforts of many laboratories,
including ours, several hundred homologous
genes have been mapped in both species, and
over 60 chromosome regions have been delin-
eated that contain conserved groups of genes.
Thus it has become possible, after mapping a
gene in one species, to predict the location of its
homologue in the other. Comparative mapping
information is used to evaluate the possibility of a
mouse mutation being a true model of a human
genetic disorder.
Our laboratory is employing in situ hybridiza-
tion, with multicolor nonradioactive detection of
chromosomal signals, as well as somatic cell ge-
netic approaches, to locate cloned genes of
known function on human and mouse chromo-
somes. We are using this information to define
candidate genes for human inherited disorders or
for phenotypic mutations in mice and to further
delineate regions that contain homologous genes
in both species. Our goal is to identify genes in-
volved in producing phenotypic abnormalities
in chromosomal imbalance syndromes and in in-
herited disorders, to understand their function,
and to devise precise diagnostic tests and rational
treatment strategies.
Search for Genes Involved in Diseases
In collaboration with the laboratory of Eric
Shooter, Stanford University, we have mapped a
peripheral myelin protein gene, identified by
others as having a growth arrest function, to
mouse chromosome 1 1 and human chromosome
17. It thus became a candidate gene for involve-
ment in the mouse mutation Trembler and the
inherited human neuropathy Charcot-Marie-
Tooth disease, type lA. Point mutations in this
gene were indeed demonstrated in Trembler
mice, and the work on the human disorder is in
progress.
In collaboration with Stuart Lefif and Tim Don-
Ion, also at Stanford, we have assigned a gene
whose product is involved in mRNA-processing
(splicing) events to the region of human chromo-
some 15 that is commonly deleted in patients
with the Prader-Willi syndrome. In this deletion
syndrome, hypotonia, hypogonadism, mental re-
tardation, and obesity due to lack of appetite con-
trol are associated with often submicroscopic
(micro-) deletions of region 15qll.2-ql3. Since
no gene of known function has yet been mapped
to the smallest deletion overlap region, our as-
signment of a gene to this region makes it a candi-
date for contributing to the microdeletion pheno-
type. It is well established that in Prader-Willi
syndrome the chromosome with the deletion is
paternally derived, while the maternally derived
homologous genes do not appear to be expressed
(imprinting) . If we can show that our candidate
gene is also imprinted on the maternally derived
chromosome, it would support the hypothesis
that this gene contributes to the deletion
phenotype.
While the classical forms of X-linked-recessive
progressive muscular dystrophy (Duchenne and
Becker types) are due to deletions or mutations in
the dystrophin gene on Xp21, there is a distinct
autosomal recessive form of muscular dystrophy,
affecting both sexes and clinically resembling
Duchenne muscular dystrophy, for which the de-
fect is unknown. We have collected and studied
several families with more than one affected indi-
vidual. The dystrophin-like gene on chromosome
6 has been excluded as a candidate, and linkage
to other chromosomal sites is being tested. We
are also collaborating with Kevin Campbell
(HHMI, University of Iowa, Iowa City), whose
laboratory has characterized and isolated a dys-
trophin-associated complex of glycoproteins lo-
cated at the sarcolemma that interacts with dys-
trophin. We have begun to map genes for these
proteins to chromosomal sites as a prerequisite
for testing them as possible candidates for in-
volvement in autosomal childhood-onset progres-
sive muscular dystrophy.
Search for Mutations in Candidate Genes
In order to find mutations in a candidate gene,
we are employing screening methods in which
the gene is amplified in small portions of DNA of
141
Molecular Basis of Genetic Diseases and Chromosome Mapping
an obligate heterozygote. The amplified DNA du-
plexes are screened for the presence of base dif-
ferences by denaturing gradient gel electrophore-
sis (DGGE) or single-strand conformational
polymorphism (SSCP) studies. Amplified exons
that appear to have base differences, identified by
altered melting behavior or conformational prop-
erties, are sequenced to reveal the precise base
differences.
In a rare form of inherited dv^^arfism, the Laron
syndrome, individuals are unresponsive to
growth hormone, lack the activity of a specific
growth hormone-binding protein in serum, and
have low levels of insulin-like growth factor I.
The growth hormone receptor is a likely candi-
date gene for mutations leading to this disorder.
The receptor gene consists of 10 exons that are
amplified individually from flanking primers. In
a study of 38 Laron syndrome individuals from a
highly inbred population in the mountains of
southern Ecuador, we identified a base substitu-
tion in the extracellular domain of the receptor
that does not alter the encoded amino acid but
creates a new donor splice site. The resulting mu-
tant mRNA is deleted for 24 bases, which predicts
a protein lacking eight amino acids in a highly
conserved region of the molecule. The location
of the deleted eight amino acids does not involve
the hormone-binding or receptor dimerization
sites. The most likely effect of this mutation is a
poorly folded, unstable protein that might de-
grade rapidly. Further studies are under way to
test this prediction.
The same strategy is now being pursued to find
mutations in the growth hormone receptor gene
in other patients with Laron syndrome who are
from different ethnic and racial populations. The
Ecuadorean growth hormone receptor mutation
represents the first instance of a disease-causing
base substitution that acts by creating a new do-
nor splice site and leads to its exclusive use while
the original site is unaltered.
The gonadotropin-releasing hormone (GnRH)
is produced in specialized cells in the hypothala-
mus, travels to the pituitary gland, and causes the
release of the gonadotropic follicle-stimulating
hormone (FSH) and luteinizing hormone (LH).
We previously mapped the gene to the short arm
of human chromosome 8. In the mouse, partial
deletion of this gene causes a recessive pheno-
type, termed hypogonadal, leading to lack of sex-
ual development. In human families with in-
herited incomplete sexual development due to
low levels of gonadotropin hormones, the GnRH
gene is a candidate for involvement, as in the
mouse model. We have studied a family with
three affected siblings that show a sequence
polymorphism in the signal peptide of the GnRH
gene, but have found that this polymorphism
does not cosegregate with the disease phenotype,
thus excluding a GnRH gene defect in this partic-
ular family.
The Marfan syndrome is an autosomal domi-
nant condition that involves abnormalities in
elastic tissue of the aorta and the fibrils that sus-
pend the optic lens and in connective tissue of
many organs, including bones, tendons, and lung.
The estimated incidence is 1 in 10,000, and clin-
ical severity is highly variable. Dissection and
rupture of the aorta is a life-threatening compli-
cation. During the past year, linkage studies in
other laboratories have pinpointed the site of the
Marfan gene on human chromosome 15, and co-
incidentally the gene encoding fibrillin, the ma-
jor microfibrillar protein, was also mapped to the
same location. Subsequently, a mutation in the
fibrillin gene was documented in two unrelated
Marfan patients.
Stanford University is a center for Marfan diag-
nosis, treatment, and research. Over the last three
years, we have established a broad-based research
program that involves a detailed clinical and ge-
nealogical database and the collection of blood
and tissue samples from affected persons and fam-
ily members. We hypothesize that a mutation in
the fibrillin gene causes the Marfan phenotype by
producing an abnormal protein that when in-
corporated into the microfibrils causes their
instability.
We are now screening the mRNA produced by
skin or aortic fibroblasts from patients with clas-
sical Marfan syndrome and related connective tis-
sue disorders. After reverse transcription of the
10,000-bp fibrillin mRNA, small overlapping
sections are amplified and screened for point mu-
tations by DGGE and SSCP studies. Our goal is not
only to develop a molecular diagnostic test for a
disorder that is clinically often difficult to diag-
nose, but also to gain insight into the relation-
ships between particular types of mutations and
the phenotypes they produce.
Since there are fibrillin-related genes else-
where in the genome and many connective tissue
disorders that overlap the Marfan phenotype to
some degree, this will be a large fruitful field of
investigation.
142
Molecular Biology of Obesity and Diabetes
Jeffrey M. Friedman, M.D., Ph.D. — Assistant Investigator
Dr. Friedman is also Associate Professor and Head of Laboratory at the Rockefeller University. He received
his B.S. and M.D. degrees from the Renssalaer Polytechnic Institute-Albany Medical College. After
completing a residency in internal medicine at Albany Medical College and a gastroenterology fellowship
at Cornell University Medical College, he enrolled in the graduate program at Rockefeller, where he
received his Ph.D. degree in molecular biology.
EXTENSIVE studies of humans and other organ-
isms have suggested that body weight, body
composition (percent body fat) , and food intake
are under strict physiological control. The "set
point" hypothesis holds that both the intake and
expenditure of energy are physiologically regu-
lated in the individual to maintain a predeter-
mined body weight. Implicit in this hypothesis is
the notion that signal molecules reflecting the
nutritional state are synthesized in the periphery
and sensed by brain centers whose appropriate
response is set to stabilize body weight.
Studies in which the rodent brain has been se-
lectively lesioned suggest that this feeding con-
trol center resides, at least in part, in the hypothal-
amus. However, the site of synthesis of the
molecules that signal nutritional state is un-
known, though fat cells or cells of the gastrointes-
tinal tract have been proposed. Knowledge of the
site of synthesis and the molecular nature of sig-
nals aff'ecting the control of appetite and body
composition could have important implications
for our understanding of nutritional disorders.
To learn more about these signaling mecha-
nisms, we have been taking a variety of ap-
proaches, both genetic and molecular, to study
the role of specific gene products in the control
systems.
Molecular Basis of Obesity in oh/ ob
and db/db Mice*
If one wishes to understand the basis for differ-
ences in the complicated system of energy ho-
meostasis, there are a number of experimental
advantages to studying mutant mice, including
the ability to control for environment and to set
up genetic crosses. For these reasons, we have
begun a study of mice carrying recessive muta-
tions that result in profound obesity. At least four
obesity-causing mutations are available: obese
(ob), diabetes (db),fat (fat), and tubby (tub).
In each case, a mouse becomes obese because of a
single-gene defect. We have focused on the ob
and db mutations for several reasons. The mutant
mice become very obese, often three times nor-
mal weight; and the obese phenotype in mutant
animals, as in humans, appears to result from
both increased food intake and diminished en-
ergy expenditure. Furthermore, Douglas Cole-
man at the Jackson Laboratory has suggested that
ob mice lack a circulating factor that suppresses
appetite and that db mice, which are unable to
respond to this factor, may lack its receptor.
Current techniques in molecular genetics,
such as Southern blots and chromosome walking,
make it possible to clone genes such as ob and
db whose function is known on the basis of a mu-
tant phenotype but whose gene product is un-
known. This approach, called positional cloning,
utilizes restriction fragment length polymor-
phisms (RFLPs) — genetic markers defined by spe-
cific cloned pieces of DNA. These markers define
genetic differences in inbred mouse strains. The
first step in attempts to clone a mutant gene uti-
lizes genetic crosses between normal and obese
(or diabetic) mice. Inheritance of the obese phe-
notype is compared in individual animals with
that of individual RFLPs, which if inherited along
with the obese phenotype, are said to be linked
genetically to the obesity gene.
By performing this analysis on several thousand
mice with several dozen different RFLPs, we have
been able to identify a series of DNA probes that
are very tightly linked to the ob and db genes.
RFLPs linked genetically to these mutations are in
physical proximity to ob and db and can be used
as starting points to characterize the adjacent
DNA and clone the mutant genes.
In the case of ob mice, we first used three dif-
ferent RFLPs to generate a detailed genetic map
around the ob locus. The genes represented were
the met oncogene carboxypeptidase A, which
codes for a pancreatic enzyme; the trp gene; and
the cystic fibrosis gene. Similarly, the mouse db
gene has been mapped relative to RFLPs for inter-
feron-a and a complement gene. We have also
found that the rat obesity mutation fatty (fa) is
also flanked by the genes for interferon-a and
* Studies aimed at cloning the mutant obesity genes ob and
db from mice have been supported by the National Institute
of Diabetes and Digestive and Kidney Diseases.
143
Molecular Biology of Obesity and Diabetes
complement. These data suggest that mutations
in the same gene can cause obesity in two differ-
ent rodent species and raise the possibility that
similar mutations can cause obesity in humans.
We are endeavoring to test this possibility, in col-
laboration with Rudolph Leibel, by using the in-
terferon-a and complement genes to characterize
human families with a high incidence of obesity.
To identify other RFLPs that are more tightly
linked than the probes mentioned above, we have
used the technique of chromosomal microdissec-
tion, in which small slices of individual chromo-
somes are dissected and cloned. Separate libraries
have been made from proximal chromosome 6,
where ob maps, and from mid chromosome 4,
where db maps. Two probes have been isolated
from the chromosome 4 library that flank db and
are about 1 cM apart. Similarly, two probes that
flank ob and are about 0.2 cM apart were isolated
from the chromosome 6 library. A genetic dis-
tance of 0.2 cM corresponds to about 400,000
base pairs of DNA.
We are currently attempting to clone the DNA
near these probes by using techniques such as
pulsed-field gel electrophoresis and cloning of
large fragments in yeast artificial chromosomes
(YACs). YACs of 300,000 to 1 million base pairs
have already been isolated for each of the probes
that flank ob, and the search for the gene in these
artificial chromosomes should begin shortly. The
cloning of these genes should further our under-
standing of the mechanisms that control food in-
take and body weight.
Polygenic Inheritance of Type II Diabetes
The mice carrying the ob and db obesity muta-
tions develop diabetes of adult onset that is quite
similar to type II diabetes in humans. However,
differences in the type II phenotype are seen in
mutant mice, depending on the strain carrying
the mutation. C57BL/6J ob/ob mice develop a
mild insulin-resistant diabetes with high levels of
plasma concentration of glucose and insulin. In
contrast, DBA/2J mice develop a severe diabetes
characterized by very high plasma levels of glu-
cose and relatively low plasma insulin concentra-
tion. These data suggest that genetic differences
in insulin sensitivity and output from pancreatic
/3-ceIls can influence the severity of the diabetes
in genetically obese mice of different inbred
strains.
Further analyses of these data have suggested
that the genetic differences are a result of poly-
genic inheritance — genetic variation in several
genes. Advances in molecular genetics now make
it possible to dissect polygenic traits (such as dia-
betes in mice) into single gene components. This
requires that a collection of genetic markers,
such as RFLPs, be available for genotyping the
diabetic animals. To facilitate the analysis of poly-
genic traits, we have helped Eric Lander and Wil-
liam Dietrich (Massachusetts Institute of Technol-
ogy) generate a linkage map of the mouse
genome, using a new type of genetic marker
known as the SSR (simple sequence repeat).
These markers greatly facilitate genetic mapping
experiments. We are now analyzing obese prog-
eny of various genetic crosses with these SSR
markers to identify novel genes that predispose
ob animals to diabetes.
Regulation, Function, and Expression
of Cholecystokinin in Human Tumors
The hormone cholecystokinin (CCK) was origi-
nally found in the small intestine by virtue of its
ability to stimulate gallbladder contraction and
pancreatic secretion in response to feeding. High
levels of CCK have also been found in neurons of
the mammalian brain, where it functions as a neu-
rotransmitter. The first demonstration that CCK
could affect behavior was reported by Gerry Smith
and Dick Gibbs, who showed that peripherally
administered CCK had an appetite-suppressing
effect on rats. It has also been demonstrated that
CCK antagonists increase feeding behavior in ro-
dents. These observations suggest that the regula-
tion and function of this gene is important in the
control of appetite.
At present, we are using a variety of techniques
to explore CCK's function(s). One of our first
objectives was to determine whether any human
tumors associated with weight loss overexpress
this peptide, and we therefore began screening
tumor cell lines for CCK production. Several pe-
diatric tumors were found to synthesize the appe-
tite depressant, including peripheral neuroepi-
thelioma (a rare nerve tumor that usually
develops in the chest wall) , Ewing's sarcoma of
bone, and rhabdomyosarcoma (a malignant mus-
cle tumor) . These data suggest that measurement
of the CCK levels in the blood may be of diagnos-
tic and prognostic value in the management of
these tumors. Studies are being initiated to ascer-
tain whether overproduction of CCK in patients
with these tumors is associated with excessive
weight loss.
144
Test-tube filaments made from genetically engineered human epidermal keratins. On the
left are filaments from normal keratins; on the right, filaments from keratins with a
single amino acid substitution (arginine to cysteine). This point mutation was discov-
ered in a patient with epidermolysis bullosa simplex, a disease involving abnormalities in
the keratin filament network.
From Coulombe, P. A., Hutton, M.E., Letai, A., Hebert, A., Paller, A.S., and Fuchs, E.
1991. Cell 66:1301-131 1. Copyright© 1991 by Cell Press.
146
Regulation of Keratin Expression During
Differentiation and Development in Human Skin
Elaine Fuchs, Ph.D. — Investigator
Dr. Fuchs is also Professor in the Departments of Molecular Genetics and Cell Biology and of Biochemistry
and Molecular Biology at the University of Chicago. She received her B.S. degree in chemistry from the
University of Illinois and her Ph.D. degree in biochemistry from Princeton University, where she studied
with Charles Gilvarg. Her postdoctoral research was done with Howard Green at the Massachusetts
Institute of Technology. Dr. Fuchs counts among her honors the R.R. Bensely Award from the American
Association of Anatomists.
THE long-range objective of our research is to
understand the biochemical mechanisms that
operate and regulate the expression of human
genes during development and differentiation in
skin. Present knowledge of the biochemistry of
human skin and its diseases is limited. Although
dermatologists have always directed their interest
toward human skin biology and skin diseases, the
field of molecular biology has only recently ap-
proached a level of understanding that permits
the complex biochemistry of human skin to be
explored. A major factor facilitating such studies
is the ability to grow human skin cells in tissue
culture, including epidermal cells, dermal fibro-
blasts, melanocytes, and dermal papillae cells.
The recent development of technology for target-
ing foreign genes to transgenic animals' skin has
provided a valuable in vivo model system for the
study of genetic skin diseases and skin cancers.
Much of our research on human skin has fo-
cused on the epidermis, which comprises about
20 cell layers whose outermost is the skin sur-
face. Only the inner or basal layer of the epider-
mis is truly living and undergoes DNA synthesis
and cell division. Under an influence as yet un-
identified, a basal cell ceases to divide and makes
a commitment to differentiate terminally. As the
cell moves outward to the skin surface, it under-
goes a variety of morphological and biochemical
changes. The most pronounced of these is the pro-
duction of a dense network of keratin filaments,
which are tough, resilient protein fibers. Many
skin diseases of the epidermis, including psoria-
sis and basal and squamous cell carcinomas, in-
volve a malfunctioning of the differentiative pro-
cess that is frequently associated with some
abnormality in the production or organization of
these filaments. Our investigation is focused on
the regulation of the expression of keratin pro-
teins and their genes in human epidermis and in
epidermal cells differentiating in tissue culture.
The keratins are a group of 1 0-20 related pro-
teins (40-70 kDa) that form the 10-nm keratin
filaments in the cytoplasm of epidermal cells.
Only a subset (typically 2-6) of keratins are ever
expressed at one time. As a normal epidermal cell
differentiates, it changes the subset of keratins it
makes. In addition, the cell increases its keratin
synthesis, leaving the fully differentiated epider-
mal cell with 85 percent of its total protein as
keratins. In diseases of the skin involving epider-
mal hyperproliferation, including psoriasis and
squamous cell carcinomas, a new subset of kera-
tins not normally made in the epidermis is pro-
duced, and can be diagnostic.
A coordinated genetic and biochemical ap-
proach is necessary to determine the regulation
of the multiple keratins and to decipher their
structural and functional roles in the differentiat-
ing epidermal cell. A number of years ago, we
showed that expression of different subsets of
epidermal keratins is due to changes in the syn-
thesis of different mRNAs. We used DNA recombi-
nant technology to show that these mRNAs are
encoded by about 20 different genes of two dis-
tinct types. Type I encodes small keratins (40-53
kDa); type II, larger keratins (53-67 kDa).
Keratins are expressed as specific pairs of type
I and II proteins. The basic subunit of keratin fila-
ments is a heterodimer, composed of the two
types. Approximately 20,000 heterodimers form
a single 10-nm filament; the assembly process is
energy independent and does not appear to re-
quire auxiliary proteins or factors. Using DNA
sequencing, we determined the amino acid
sequences for several keratin pairs. The cytoskele-
tal architecture of keratin filaments may be specif-
ically tailored to suit the particular structural
needs of each epidermal cell at various stages of
differentiation and development.
To determine the details of the filament assem-
bly process and to investigate the interactions of
keratin filaments with other proteins and organ-
elles, we used deletion and site-directed muta-
genesis to alter the coding sequences of K5 and
Kl4, the pair expressed in the living cells of the
epidermis. We generated substantial quantities of
keratins for filament assembly studies, using ge-
netic engineering to overexpress wild type and
mutant human keratins in bacteria, which do not
have keratin. We purified these keratins and iso-
lated and examined the consequences of muta-
tions and deletions on keratin filament assembly
in vitro.
147
Regulation of Keratin Expression During Differentiation and Development
in Human Skin
In the past few years, we identified those se-
quences involved in filament elongation and
those that are more important for lateral associa-
tions. In addition, we used gene transfection of
human tissue culture cells to examine the dy-
namics of filament assembly in vivo and to deter-
mine how abnormal expression of keratins in ma-
lignant and hyperproliferating epidermal cells
might change their biology and cytoskeletal ar-
chitecture. The results of these studies have be-
gun to yield valuable insights into the complex
assembly process of keratin filaments and the
function they perform in providing a tough skin
surface.
We have also been interested in determining
whether natural mutations in human epidermal
keratin genes might lead to genetic skin diseases.
We had previously shown that elevated expres-
sion of epidermal keratin genes is a relatively late
event in development and that certain keratin
mutants have a deleterious effect in cultured
cells, disrupting the endogenous keratin filament
network. In the past two years, we have made
transgenic mice and used epidermal keratin pro-
moters to target expression of some of these domi-
nant mutant keratins to the epidermis.
Unexpectedly, the transgenic mice exhibit a
phenotype resembling that of humans with epi-
dermolysis bullosa simplex (EBS) , a class of blis-
tering skin diseases that are dominant, sometimes
life-threatening, and of previously unknown etiol-
ogy. Mice expressing Kl4 mutants that severely
disrupt filament assembly exhibit severe blister-
ing over body trunk regions — lesions resembling
those of Dowling Meara EBS, the severest form;
while mice expressing Kl4 mutants that mildly
perturb the network exhibit blistering only over
paws, as in the mildest form of EBS, or Weber-
Cockayne, where patients show blistering pre-
dominantly on their hands and feet. Thus we
were able to demonstrate that multiple mutations
in a single gene, namely K14, can give rise to
most if not all forms of EBS, thus strongly suggest-
ing that these diseases may be genetically linked.
Further studies enabled us to demonstrate that
skin blistering in EBS is due to the fragility of
basal epidermal cells caused by an abnormal ker-
atin filament network, thereby compromising the
mechanical strength of this cell layer.
The similarities between the phenotypes in
mice and humans prompted us to focus on this
disease. In the past year we characterized the Kl 4
and K5 mRNAs, genes and proteins from two pa-
tients with Dowling Meara EBS. Both patients
have point mutations in an amino acid of the Kl4
protein that we had previously shown to be criti-
cal for filament assembly. When we engineered
and tested the two mutations in our wild-type
K14 gene, we verified that these defects are re-
sponsible for the phenotype of the EBS patients. It
will take much work to develop better diagnostic
and therapeutic tools for EBS, but elucidating the
genetic basis is an important first step.
While EBS is a disease of defects in the coding
portions of keratin genes, there are other skin dis-
eases that appear to arise from abnormalities in
the control of epidermal genes or genes that influ-
ence growth and/or differentiation. As a prerequi-
site to investigating the bases for these types of
diseases, we are 1 ) analyzing the molecular mech-
anisms underlying the differential expression of
epidermal keratin genes and 2) utilizing our Kl4
promoter to make transgenic mice that overex-
press various regulatory factors — e.g., transform-
ing growth factor-a (TGFa), epidermal growth
factor (EOF), cytokines, hormone receptors, and
proto-oncogenes in the epidermis.
In the past few years, we have identified proxi-
mal and distal domains that act synergistically to
regulate expression of the human Kl4 gene. Our
goal is to identify the sequences and transcription
factors involved. Analysis of the factors control-
ling the cell's major structural genes should lead
us to the factors determining keratinocyte fate.
Unraveling the nature of promoter and enhancer
sequences involved in regulating epidermal
genes will be important not only for understand-
ing epidermal development but also for targeting
products to the epidermis. In the last year, we
used the Kl4 promoter/enhancer to engineer
transgenic mice that overexpress TGFa, the major
autocrine growth factor of the epidermis. Our re-
sults have shown that TGFa plays an important
role in controlling the thickness of the epidermis
and that certain features of TGFa overexpression
are similar to those of psoriasis, a hyperprolifera-
tive skin disease.
The Kl4 promoter/enhancer, so effective in
transgenic technology, should also be useful for
drug therapy. Because epidermal cells can be re-
moved from a patient, cultured in vitro, and
grafted back, it should be possible to introduce
foreign genes, driven by the Kl4 promoter/en-
hancer, into the cultured epidermal cells prior to
grafting. Such techniques have potentially power-
ful applications for future medical research.
148
The Molecular Basis of Viral Replication
and Pathogenesis
Donald E. Ganem, M.D. — Associate Investigator
Dr. Ganem is also Professor of Microbiology and Immunology and of Medicine at the University of
California, San Francisco. He received an A.B. degree in biochemistry from Harvard College and the M.D.
degree from Harvard Medical School. Following his clinical training in infectious diseases, he did
postdoctoral research training in the laboratory of Harold Varmus at UCSF.
OUR laboratory studies the molecular mecha-
nisms by which pathogenic human viruses
infect the host and cause disease. We are espe-
cially interested in those viruses that produce
persistent infections and engender chronic pa-
thology. Several fundamental questions underlie
our work: How are persistent infections estab-
lished and maintained? What factors regulate
viral replication and spread? How does the persis-
tent presence of virus evoke disease?
Most of our work centers around the human
hepatitis B virus (HBV) and its animal homo-
logues. HBV is a small DNA virus that replicates
principally in liver cells (hepatocytes) and pro-
duces acute and chronic type B hepatitis. Most
initial infections are transient: following a rela-
tively brief period of liver injury (hepatitis), the
immune system eliminates virus from the liver
and invokes lasting immunity. However, 5-10
percent of infections are not successfully elimi-
nated. In these cases, viral replication persists in
the liver for the life of the host, evoking various
degrees of chronic liver injury that can lead to
premature death from liver failure. Most strik-
ingly, persistence of viral replication for several
decades enormously increases the risk of liver
cancer.
There are many reasons to be interested in this
remarkable infection. First is its great public
health significance. Worldwide, there are over
250 million chronic HBV carriers. In large re-
gions of Asia and Africa, 10-15 percent of all hu-
man inhabitants are persistently infected by the
virus. Second, the virus replication cycle is dis-
tinctly unusual: replication of the viral DNA ge-
nome is accomplished via reverse transcription of
an RNA intermediate. The HBV life cycle is thus a
permuted version of the retroviral life cycle, and
a fuller understanding of its details allows in-
structive comparisons to be made with cognate
steps in retroviral replication. More importantly,
unraveling the mechanism of HBV replication
should identify new potential targets for antiviral
therapy. Finally, the nature of the link between
chronic HBV replication and hepatocellular carci-
noma represents one of the great unsolved prob-
lems in human cancer biology, and a deeper un-
derstanding of HBV persistence is expected to
provide new clues to its solution.
Our studies of HBV replication have examined
many different steps in the viral life cycle. The
initial step in all viral infections is binding of the
virus to the cell surface receptor and its entry into
the cell. Little is known about this step in HBV
infection, but we have begun to explore it in a
convenient animal model, the duck hepatitis B
virus (DHBV) . By using recombinant DHBV sur-
face proteins to look for cellular proteins that
will bind them, we have identified a single host
glycoprotein that will interact with DHBV parti-
cles with high affinity and specificity. Such mole-
cules are good candidates for the receptor, and
we are actively attempting to clone the gene for
this host protein to allow its more-detailed
characterization.
Next, internalized virus particles must be deliv-
ered to the nucleus, where their DNA genome is
transcribed. This is perhaps the most poorly un-
derstood of all reactions in virology. For no virus
has the cellular machinery involved in this essen-
tial step been identified. Recently, we discovered
that drugs affecting the integrity of cellular mi-
crotubules block this step, implying that these
structures are involved in the intracellular trans-
port of subviral particles. This raises the exciting
possibility that such particles might also be used
to identify host proteins that recognize them and
perhaps bind not only virus particles but normal
cellular components (e.g., organelles) to cyto-
skeletal motors for transport within the cell.
Once in the nucleus, viral genes are tran-
scribed into RNA to be transferred to the cyto-
plasm for translation. Following translation, prog-
eny particles assemble in the cytoplasm. In this
remarkable reaction, viral RNA is packaged into
subviral particles along with the viral reverse
transcriptase. Surprisingly, this RNA packaging
process requires the participation of the reverse
transcriptase itself. To understand this key step
better, we are currently attempting to reproduce
this RNA recognition in vitro, using recombinant
reverse transcriptase and cloned viral RNA. Sim-
149
The Molecular Basis of Viral Replication and Pathogenesis
ilar in vitro reconstruction reactions are being
attempted for the reverse transcription reaction
itself. This enormously complex reaction gener-
ates the duplex DNA genome from its RNA pre-
cursor. By fully characterizing the molecules that
catalyze the reaction and the intermediates
through which it proceeds, we hope to identify
novel targets for drugs that would be useful in
interrupting virus growth.
The development of liver cancer during persis-
tent infection is another major focus of interest.
For this we study two relatives of HBV: the wood-
chuck hepatitis virus (WHV), which is potently
oncogenic in its normal host, and the ground
squirrel hepatitis virus (GSHV), a less-efficient
carcinogenic stimulus in woodchucks. Our stud-
ies thus far suggest fundamental differences in
the oncogenic pathways employed by these
closely related viruses. WHV-induced tumors fre-
quently (30-45 percent) contain integrated
WHV genomes adjacent to the N-myc oncogene,
whose expression is thereby activated. Such in-
sertional activation is not apparent in GSHV-
induced hepatomas. We are currently looking for
other genes that are regularly disrupted by WHV
integration, with the view that such loci may de-
fine other proteins centrally involved in the con-
trol of normal hepatocyte growth.
150
Second Messengers and Cell Regulation
David L. Carters, Ph.D. — Investigator
Dr. Garters is also Professor of Pharmacology at the University of Texas Southwestern Medical Center at
Dallas. He received his B.S. degree in agriculture and his Ph.D. degree in biochemistry at the University
of Wisconsin. His postdoctoral research was done at Vanderbilt University. Before assuming his present
position, Dr. Garbers was Professor of Pharmacology and of Molecular Physiology and Biophysics
at Vanderbilt University School of Medicine.
THE focus of the research in this laboratory
centers on the mechanisms by which ceils
communicate with each other — specifically the
mechanisms by which sea urchin or mammalian
spermatozoa detect signals from the egg. These
studies have turned out to be applicable to so-
matic cells and have led to the identification of a
new cell surface receptor family in humans and
other mammals. Members of this family serve as
receptors for molecules that regulate blood pres-
sure, as well as a large number of other physiolog-
ical processes.
Around 1981 our laboratory reported the puri-
fication of a peptide that can stimulate sperm mo-
tility. This peptide was derived from media in
which sea urchin eggs had been allowed to stand.
Subsequently it was demonstrated that different
species of sea urchins contain different peptides
and that the molecules from one species do not
necessarily stimulate the sperm cells from an-
other. In later research it became clear that sperm
cells detect higher concentrations of peptide and
swim toward them. Since the highest concentra-
tions are around the egg, the sperm cell swims
directly toward the egg under normal conditions.
This laboratory set out to determine how the
cell detects the egg peptide. We found that a par-
ticular protein on the sperm plasma membrane
specifically bound the egg peptide. This protein
appeared to serve as the detector or receptor mol-
ecule, but how it signaled to spermatozoa that a
specific egg peptide had been bound was not
clear.
To help resolve the question, we purified the
receptor protein on the plasma membrane. It was
identified as the enzyme guanylyl cyclase, which
catalyzes the formation of cyclic GMP, a small
molecule that causes a change in the behavior of
many different cells. It seemed possible that the
receptor is situated with part of it outside the
cell, where it could bind the egg peptide, and the
other part inside, where it forms cGMP. The
cGMP thus formed would then serve as a signal
that egg peptide is being detected, lending
greater speed and direction to the spermatozoon.
To provide evidence that the membrane recep-
tor protein is in fact guanylyl cyclase, we isolated
complementary DNA clones for this enzyme.
Such clones allow one not only to predict the
primary structure of a protein, the receptor in
this case, but to direct protein synthesis in quan-
tity. Unfortunately, the sea urchin sperm receptor
was not formed in the proper manner, and re-
search continues on its expression.
Under appropriate conditions, DNA will bind
(hybridize) to closely related DNA. Therefore sea
urchin DNA was used to determine whether
mammals contain a related protein that might
serve as a receptor. Clones containing comple-
mentary DNA were isolated from rat brain. The
DNA sequence revealed that a rat brain guanylyl
cyclase is a component of the plasma membrane,
with approximately one-half of the protein out-
side and one-half inside the cell. The intracellu-
lar region of the sea urchin sperm enzyme is very
similar to the part of the rat brain enzyme inside
the cell. In the regions outside the cell, however,
the two proteins show little similarity. This
would be expected if guanylyl cyclase serves as a
cell surface receptor for peptides and the peptide
of mammals is different from that of the sea ur-
chin. That is, the detector part of the molecule
would need to change to recognize a different
peptide, but intracellular regions of the receptor
could remain unaltered.
Subsequent binding studies demonstrated that
the cloned rat brain guanylyl cyclase could specif-
ically bind and be activated by certain peptides
synthesized in the heart and brain. These atrial
natriuretic peptides (ANPs) regulate blood pres-
sure, as well as various other physiological
events.
Since sea urchin sperm cells respond to pep-
tides that do not resemble ANP, other animals
may be expected to contain yet other peptides
that interact with guanylyl cyclase. In addition,
multiple membrane forms of guanylyl cyclase
may exist within mammals. We have used the
DNA that encodes the ANP receptor to determine
whether other guanylyl cyclase receptors exist.
Another receptor with properties similar to the
one described above has been identified. This
151
Second Messengers and Cell Regulation
guanylyl cyclase receptor is similar to the first
ANP receptor within intracellular regions but is
only 43 percent identical in the extracellular,
ligand-binding region. It now appears that a pep-
tide other than ANP regulates this receptor.
We subsequently discovered the existence of a
third guanylyl cyclase receptor. This receptor
binds small peptides released from various bacte-
ria that cause acute diarrhea. This form of diar-
rhea (often referred to as traveler's diarrhea) is
prevalent in infant humans and young domestic
animals. The receptor has the same general fea-
tures as the two receptors described above but
has a markedly different amino acid sequence
within the toxin-binding region. Whether a regu-
latory molecule not of bacterial origin normally
exists that binds to this receptor is not yet known.
During the past year, we have obtained evi-
dence to suggest that at least two other guanylyl
cyclase receptors exist. The hormones that nor-
mally regulate these receptors have not yet been
identified. It seems, therefore, that different cells
contain unique guanylyl cyclase receptors,
which allow them to respond to specific hor-
monal signals with an increased production of
the messenger molecule cGMP. The cells then
change their behavior. For example, one receptor
may control smooth muscle relaxation and be
principally involved in the regulation of blood
pressure, another may be involved in neural
functions, and yet another may regulate epithe-
lial cell secretion. These results also suggest that
drugs specific to different forms of the guanylyl
cyclase receptors may prove clinically relevant.
152
Molecular Genetics of the Major
Histocompatibility Complex
Jan Geliebter, Ph.D. — Assistant Investigator
Dr. Geliebter is also Assistant Professor and Head of the Mammalian Molecular Genetics Laboratory at the
Rockefeller University. He received his Ph.D. degree in microbiology and immunology from the State
University of New York, Downstate Medical Center. He was a postdoctoral fellow and research associate
in the laboratory of Stanley Nathenson at the Albert Einstein College of Medicine, Bronx, New York.
THE immune system functions to rid the body
of foreign objects such as bacteria, viruses,
tumors, and transplants. The portion of such mat-
ter that is recognized as foreign by the immune
system is called an antigen. Antigens that are
found on cells are "presented" to the immune
system by cell surface molecules called histocom-
patibility molecules (also called HLA molecules
in humans and H-2 molecules in the mouse). His-
tocompatibility molecules are able to bind anti-
genic fragments of, for example, viruses, and
stimulate white blood cells (lymphocytes) to
attack the virus-infected cell, thereby limiting
the spread of infection. Without these antigen-
presenting molecules the host would be unable
to mount an immune response against pathogens
and would not survive.
Different H-2 molecules can bind and present
different types of antigens. Because inbred mice
have about three different types of H-2 molecules
on their cells, they can bind and present a large,
but limited, number of antigens to the immune
system.
To ensure the survival of the species, it is bene-
ficial that many varieties of H-2 molecules be
present in the population. In this way there will
always be some portion of the population that
will mount an immune response to a given anti-
gen. Indeed, an extraordinary number of differ-
ent histocompatibility molecules have been
found in almost all species investigated. In hu-
mans, the large variety of HLA molecules ensures
our survival but is the major obstacle confound-
ing tissue transplantation. Our interest lies in the
genetic mechanism that generates the different
histocompatibility genes in mice and other
species.
The H-2 genes of the mouse are part of the
larger major histocompatibility complex class I
multigene family. This gene family also contains
genes that are structurally similar to H-2 genes
and have unknown functions. The genetic mecha-
nism that generates variety in H 2 genes is the
microrecombination process, which reassorts
DNA among H-2 genes and other related class I
genes. By substituting small segments of class I
gene sequences into //-2 genes, the microrecom-
bination process can create new H-2 molecules
that have different antigen-presenting capabili-
ties, thereby expanding the immune responsive-
ness of the population.
Our interest is to understand better the mecha-
nism underlying the microrecombination pro-
cess. This process has previously been studied by
identifying microrecombinant mice that differed
from their otherwise identical siblings by altered
H-2 genes. Since microrecombinant H-2 mole-
cules elicit skin graft rejection, these studies
were accomplished by skin graft testing thou-
sands of mice. The rejection of a skin graft by a
sibling mouse signaled an alteration in H-2 mole-
cules. These labor-intensive studies found that,
on the average, one microrecombinant mouse
was detected for every 5,000 skin grafts
performed.
To gain further insight into the microrecom-
bination process, we are using in vitro-engi-
neered constructs to detect microrecombinant
H-2 genes. We have constructed a fusion gene in
which /3-galactosidase sequences replace two cy-
toplasmic exons for the gene. The fusion pro-
tein can be detected by staining for |8-galactosi-
dase activity, which is manifested as blue-colored
cells. We have also site-directed two in-frame ter-
mination codons in the gene at positions that
undergo frequent microrecombinations. This
prevents the expression of |S-galactosidase.
/3-Galactosidase expression can be rescued by a
microrecombination with a linked class I gene,
Q4, that recombines away the termination co-
dons. Thus microrecombinations can be scored as
blue cells.
This microrecombination construct, once in-
troduced into a variety of cell types, will help
determine the microrecombination frequencies
of different cells. Its introduction into transgenic
mice will help determine frequencies in germ
cells. Data from previous studies indicate that mi-
crorecombinations occur in female germ cells.
We also hope to determine if they occur in sperm
cells as well and at what frequency. Some strains
of mice may undergo microrecombinations more
153
Molecular Genetics of the Major Histocompatibility Complex
frequently than others. Placing this construct
onto different genetic backgrounds may help to
determine microrecombination frequencies in
different mouse strains and perhaps identify criti-
cal parameters in the process. These studies will
contribute to our understanding of the genetic
processes that control the evolution and ulti-
mately the function of the mammalian immune
system.
Although sequence diversity and polymor-
phism are the hallmark of genes, region genes are
characterized by sequence conservation among
alleles and limited polymorphism. The lack of
polymorphism among these genes has been sug-
gested to preclude an immunological function
for the gene products. We have identified a re-
gion gene w^hose sequence differs greatly be-
tween alleles of the C57BL/6 and C3H mice. The
sequence differences between the two alleles are
manifested in both scattered and clustered nu-
cleotide substitutions. The clustered substitu-
tions are similar to those observed in the micro-
recombinations that diversify genes and may
reflect past microrecombination events with H-2
and other Qa region genes. These data may pro-
vide the first evidence that Qa genes can be recip-
ients in the microrecombination process.
Polymerase chain reaction (PGR) analysis has
indicated that this Qa gene is transcribed in some
strains of mice. We are currently engaged in an
in-depth analysis of the transcription, translation,
and cell surface expression of this gene and its
product to ascertain its function. This gene is
polymorphic in at least three strains of mice, and
the analysis of other strains is under way. The di-
versity and polymorphism of this gene suggest an
immunological function for its product, perhaps
the first Qa gene to be ascribed such a function.
154
The Decoding Code in mRNA
Raymond F. Gesteland, Ph.D. — Investigator
Dr. Gesteland is also Professor of Human Genetics at the University of Utah School of Medicine and
Professor of Biology and Adjunct Professor of Bioengineering at the University of Utah. He received his B.S.
degree in chemistry and his M.S. degree in biochemistry from the University of Wisconsin. He earned his
Ph.D. degree in biochemistry from Harvard University, where he studied with J. D. Watson. Dr. Gesteland
was an NIH postdoctoral fellow at the Institute de Biologie Moleculaire, Geneva, Switzerland. He served
as Assistant Director and Investigator at Cold Spring Harbor Laboratory, New York, before assuming his
present responsibilities.
THE genetic code precisely specifies the iden-
tification of each of the 20 amino acids from
the set of 6 1 triplets in mRNA, with three triplets
dedicated to termination. As a ribosome moves
progressively along the mRNA deciphering it, the
contiguous triplets contain all the information
necessary to specify amino acid sequence. How-
ever, there is another code, also encrypted by the
linear order of bases in mRNA, that carries infor-
mation about the mechanism of translation.
This code specifies how the common genetic
code should be implemented for each mRNA. In
some cases the additional information may be
quite simple, consisting of instructions that tell
the ribosome where to start and where to stop
decoding. In other cases this secondary code radi-
cally changes the decoding process so that indi-
vidual mRNAs use a different version of the ge-
netic code, altered either in the meaning of
certain codons or in the linear readout mecha-
nism. This set of instructions could be called the
"decoding code," in that it operates on the ge-
netic code. Different decoding codes individual-
ize mRNAs.
A decoding code operating during the deci-
phering of a specific mRNA often results in a vio-
lation of the conventional genetic code. This can
occur with a set frequency, so that the same
mRNA produces two different protein products;
but the frequency can also depend on metabolic
state, thus providing another point at which gene
expression is controlled.
For example, the ribosomal machinery can be
instructed to read a nontriplet number of bases at
one site so that an alternate reading frame will be
used from that point on. Or a proportion of the
ribosomes can be instructed to jump from one
site to another on the same mRNA so that noncon-
tiguous codons will be read out. Or the meaning
of specific codons can be altered so that a ribo-
some reads a stop codon as an amino acid, or even
as a new amino acid that is not a member of the
conventional set of 20. Each of these unusual
events is specified by the decoding code
information.
The diversity of schemes for encrypting infor-
mation for the decoding code in individual
mRNAs is just beginning to be appreciated. In
some cases, rather simple sequences are in-
volved; in others, sequence dictates complex
RNA structures.
The signals in mRNA that carry the instructions
for programmed ribosomal frameshifting usually
include a shift site where the frame changes and a
stimulator sequence that greatly increases the
shift site's efficiency. The shift site consists of
4-7 nucleotides that allow a ribosomal-bound
tRNA decoding in the first frame to slip forward
or backward by one nucleotide to get into the
new frame. The stimulator information can be
upstream of the shift site, as in the case of the RF2
gene in Escherichia coli, where 5-6 nucleotides
need to pair directly with a specific sequence in
ribosomal RNA in order to stimulate the shift. Or
it can be downstream, as in the case of many retro-
viruses and retroviral-like genes where the spe-
cific interactions are less clear and the mRNA
contains secondary and tertiary structures that
somehow stimulate frameshifting at the upstream
shift site. The combination of shift site and stimu-
lator sequences constitutes the decoding code
that tells how the standard genetic code should
be corrupted for translation of a specific message.
There is only one compelling case of ribosomal
jumping, and here the decoding code is complex.
The mRNA for gene 60 of bacteriophage T4 has a
gap of 50 nucleotides that is very efficiently by-
passed by ribosomes. The mRNA sequence ele-
ments involved include both structural informa-
tion in the gap and upstream information in the
form of the amino acid sequence of the growing
peptide chain that must interact with the same
part of the complex.
Other versions of the decoding code reprogram
stop codons to specify amino acids. Some retrovi-
ruses (e.g. , murine leukemia virus, MuLV) make a
fusion protein from two in-frame genes separated
by a UAG stop codon, which in this case is de-
coded as glutamine. The decoding code that spe-
cifies this includes a downstream element whose
155
The Decoding Code in mRNA
secondary and tertiary structures are crucial. In
the handful of known cases of proteins having the
unusual amino acid selenocysteine, a UGA stop
codon is reprogrammed to be decoded as this
21st amino acid. The decoding code clearly in-
cludes structural elements that are downstream,
in one case 150 nucleotides away in the non-
translated part of the message.
An emerging theme of the decoding code is
mRNA secondary and tertiary structural elements.
Stem-loop structures can often be identified by
inspection of sequence, and evidence for their
participation can be obtained by making disrup-
tive and restorative mutations. The functional
stems can be short in some contexts, long in
others, and can be near the site of action or well
downstream.
In an increasing number of cases, the crucial
structure is more complicated, involving a pseu-
doknot where a sequence downstream of a stem-
loop folds back to pair with the loop of the
stem, resulting in two stems and two loops that
are intertwined. Physical-chemical properties of
model pseudoknots show that the two stems are
coaxial, with considerable stability coming from
the extended base stacking that is achieved.
To test the biological importance of each of the
bases of these complicated structures requires
construction of a very large number of mutants,
and analysis of the resulting structural changes is
so far minimal. In some cases the actual sequence
of the proposed stems is unimportant so long as
base-pairing is maintained and the sequence of
the loops is not crucial. In fact, these criteria are
used to define the importance of a stem or pscu-
doknot. In at least one case (MuLV stop codon
read-through) , a few bases in the loop region of a
pseudoknot are crucially important.
Structures in ribosomal RNA are beginning to
be delineated clearly, largely from evolutionary
comparisons of sequence. We are still quite igno-
rant, however, about the complexity of structures
in mRNA. At least for the elements involved in the
decoding code, there is a biological assay for
function.
We are even less clear about how these struc-
tures in mRNA alter the decoding process. The
simplest view — that mRNA structures merely
cause a ribosomal pause allowing alternative reac-
tions— is almost certainly naive. But whether the
structures bind protein factors, interact with ribo-
somal proteins, pair with ribosomal RNA, or even
interact with other parts of mRNA is completely
untested. Investigation of these possibilities in
the context of the decoding code will be
revealing.
Ribosome Interaction Sequence
GGA UAG ^ J- GGA UUA-
■ -STOP ■ \
Coding Gap
GLY LEU
u
UCA AAA AAC UUG
Decoding of the genetic code. Top: Model of a
ribosome traversing the 50-nucleotide coding
gap in mRNA of bacteriophage T4 's gene 60,
illustrating the involvement of upstream pep-
tide sequences and sequences in the gap.
Bottom: Model of a ribosome reading a stop
codon in mRNA of a murine leukemia virus,
stimulated by a downstream pseudoknot
structure.
Research of Norma Wills, Robert Weiss,
Dianne Dunn, and fohn Atkins in the labora-
tory of Raymond Gesteland.
UC AAA AAA CUU C
156
Molecular Analysis of Proteins Involved
in Human Disease
Mary-Jane H. Gething, Ph.D. — Investigator
Dr. Gething is also Professor of Biochemistry at the University of Texas Southwestern Medical Center at
Dallas. She received her bachelor 's and Ph.D. degrees from the University of Melbourne. After holding
research fellowships in Cambridge, England, with Brian Hartley and in London with Michael Waterfield,
she joined the scientific staff of the Imperial Cancer Research Fund, London. After that she was a senior
staff investigator at the Cold Spring Harbor Laboratory, New York, before going to the Southwestern
Medical Center.
OUR studies on experimental models of hu-
man disease involve three systems: 1) hu-
man tissue-type plasminogen activator (t-PA), a
serine protease involved in fibrinolysis, tissue re-
modeling, and metastasis; 2) the hemagglutinin
of influenza virus, which is being used to develop
models of autoimmune disease in transgenic
mice; and 3) the tumor-suppressor protein p53
and its interaction with cytosolic stress-70
proteins.
Role of t-PA in Thrombolysis and Metastasis
Many normal and abnormal biological pro-
cesses that require extracellular proteolysis, in-
cluding thrombolysis, tissue remodeling, and
metastasis, are mediated by plasminogen activa-
tors that cleave plasminogen to the active pro-
tease plasmin. One such activator, t-PA, is the
principal thrombolytic agent in the circulation,
and its elevated expression is thought to be
linked to increases in the metastatic potential of
some types of tumor cells, including malignant
melanomas.
The t-PA protein is composed of a number of
independent structural domains. The finger do-
main and an epidermal growth factor (EGF)-like
domain are involved in the initial, high-affinity
binding of t-PA to fibrin, while stimulation of t-PA
activity requires secondary, lower-affinity inter-
actions of fibrin with either of two kringle do-
mains of the molecule. The binding of t-PA to
specific clearance receptors on hepatic cells also
involves sequences within the finger and/or EGF-
like domains. Finally, a specific inhibitor, plas-
minogen activator inhibitor ! (PAI-1), interacts
with the active site in the carboxyl-terminal cata-
lytic domain.
Although the three-dimensional structure of
t-PA has not been solved, we have been able to
model all the domains through use of known
structures of homologous domains in other pro-
teins. Site-directed mutants using these proposed
structures have provided information about the
role of individual amino acid sequences of the
protein, and variant enzymes have been gener-
ated that are efficient, fibrin-stimulated plasmin-
ogen activators but are resistant to inhibition by a
variety of serpins, including PAI- 1 , or do not bind
to the t-PA receptor(s) involved in clearance of
the enzyme in the liver. Because these mutant
enzymes should have an extended effective life in
the circulation, they may have significant poten-
tial for use in thrombolytic therapy of patients
with myocardial infarction.
The variant enzymes are also being utilized to
test the role of t-PA in metastasis of malignant
melanoma cells. Transgenic mice expressing the
T antigen oncogene from simian virus 40 (SV40)
under the control of the mouse tyrosinase pro-
moter develop ocular melanoma with high fre-
quency. These tumors are transplantable to non-
transgenic animals but are not metastatic. We are
currently developing additional lines of trans-
genic mice that express wild-type or inhibitor-
resistant forms of t-PA from the same tyrosinase
promoter. Crossing of the Ty-Tag and Ty-tPA
transgenic mice will reveal whether an increased
level of t-PA production in melanoma cells would
result in increased metastatic potential.
Transgenic Models of Immune Tolerance
and Autoimmune Disease
We are using transgenic mice to study the de-
velopment of immunological responses to the
hemagglutinin (HA) of influenza virus. RIPHA
mice, which express HA from the rat insulin II
promoter/enhancer only in the /3-cells of the pan-
creas, promise to provide a valuable model for
the study of immune tolerance and autoimmune
diabetes. These mice display no physiological
problems until 4-5 months of age. After that, in-
creases in the blood glucose levels begin to ap-
pear, and shortly thereafter these mice develop a
severe hyperglycemia that is responsive to
insulin.
In the RIPHA-33 line, both male and female
animals can develop an immune response to HA
and other antigens of pancreatic |8-cells. How-
ever, male mice become hyperglycemic with
much higher frequency (45 percent ) than female
mice ( 5 percent ) . Recent studies have shown that
the amount of fat in the diet is an important regu-
157
Molecular Analysis of Proteins Involved in Human Disease
lator of diabetes in RIPHA-33 mice and that the
sex bias may reflect a decreased ability of male
mice to control their blood glucose levels.
Current studies focus on a comparison of the
RIPHA model with mice of the non-obese dia-
betic (NOD) line and on the role played by cyto-
kines, such as interleukin-1 , tumor necrosis fac-
tor, interferon-7, and interleukin-6, during the
progression of the disease.
Folding and Assembly
of the Tumor-Suppressor Protein p53
Until recently it was widely assumed that the
folding and oligomerization of newly synthesized
polypeptides and their subsequent molecular re-
arrangements are spontaneous processes that do
not require the intervention of other cellular pro-
teins. However, it is now apparent that members
of the stress-70 protein family are "chaperones"
intimately involved in facilitating protein folding
and assembly within prokaryotic and eukaryotic
cells.
Our previous studies of the interaction of BiP,
the endoplasmic reticulum stress-70 protein,
with newly synthesized membrane and secretory
proteins has led to an understanding of the role of
stress-70 proteins in stabilizing unfolded or
partly folded polypeptides in a form competent
for further folding and oligomeric assembly.
Current studies focus on the role of hsc70, the
cytosolic stress-70 protein, in modulating the
structure or activity of mutant forms of the
tumor-suppressor protein p53.
Other workers have shown that p53 plays a
role in regulation of normal cell growth and that
it binds to a number of viral-transforming pro-
teins, including SV40 T antigen. Mutations in p53
convert the protein to an oncogenic form capable
of co-operating with another oncogene, activated
ras, to transform cells. These mutant forms of
p53 appear to be altered conformationally, as in-
dicated by loss of T antigen recognition, altered
reactivity with monoclonal antibodies, and co-
translational binding to hsc70. We are analyzing
the interaction between p53 and hsc70, using the
fd bacteriophage expression system. By screening
p53-peptide-expressing bacteriophage for bind-
ing to hsc70, with subsequent enrichment and
amplification in bacteria of those expressing
hsc70-binding epitopes, the region(s) in p53 rec-
ognized by hsc70 will be identified.
158
Signal Transduction Pathways
in B Lymphocytes
Sankar Ghosh, Ph.D. — Assistant Investigator
Dr. Ghosh is also Assistant Professor of Immunobiology and of Molecular Biophysics and Biochemistry
at Yale University School of Medicine. After receiving his Ph.D. degree from the Albert Einstein College
of Medicine, Bronx, New York, he conducted postdoctoral work at the Whitehead Institute
in Cambridge, Massachusetts, under the supervision of David Baltimore.
THE focus of research in our laboratory is to
understand the signals that regulate B cell de-
velopment and the pathways through which they
are transduced. We propose to address these ques-
tions by studying the regulation of developmen-
tal-stage-specific expression of the immunoglob-
ulin K light-chain gene. Primarily, we wish to find
the signal that initiates k gene expression during
B cell development and determine whether ex-
pression of the gene is necessary for rearrange-
ment of the K locus.
The expression of the k gene in mature B cells
is driven by an enhancer located in the intron
between the J/c and Ck genes. The activity of this
enhancer depends on the binding of a transcrip-
tion factor, called NF-/cB, to a specific site in the
element. In keeping with its role as a regulator for
developmental-stage-specific expression of the k
gene, NF-/cB can only be detected in an active
form in the nucleus of mature B cells. However, it
is present in the cytoplasm of pre-B cells and
other cell types as an inactive precursor, bound
to an inhibitory protein called I/cB.
NF-kB activity can be induced by treatment of
cells with agents such as phorbol esters or bacte-
rial lipopolysaccharide. These agents activate sig-
nal transduction pathways that result in the modi-
fication of I/cB and the subsequent dissociation of
the NF-/cB:IkB complex. The free NF-zcB then
enters the nucleus and activates gene expression.
Therefore the inducible system of NF-/cB behaves
like a second messenger, transducing cell surface
signals to the nucleus.
To understand the change in IkB that results in
the dissociation of the NF-kB:IkB complex, we es-
tablished an in vitro system to look at the involve-
ment of protein kinases in the activation process.
The results strongly suggested phosphorylation
of I/cB as a key event in the activation of NF-kB.
However, further analysis of the activation pro-
cess was hampered by the lack of reagents, such
as antibodies, or clones for the genes encoding
NF-kB. Therefore we undertook to clone the
genes; the information we obtained has resulted
in significant new insights into the function of
these proteins.
NF-kB is composed of two subunits of 50 kDa
(p50) and 65 kDa (p65) and exists as a hetero-
dimer. Cloning the cDNA encoding p50 revealed
that it was actually synthesized as a larger protein
of 105 kDa, which was processed to the mature
50-kDa size. The full-length protein did not bind
to DNA, but removal of the carboxyl-terminal re-
gion revealed the DNA-binding activity. How-
ever, the most interesting feature of the clone was
the homology for over 300 amino acids in the
amino-terminal half (i.e., the region encoding
p50) to the oncogene \-rel, its cellular counter-
part c-rel, and the Drosophila morphogen
dorsal.
Subsequent cloning of the p65 subunit re-
vealed a similar re/-homology region, which con-
tains both DNA-binding and dimerization do-
mains. In fact, with the recent cloning of other
re/-homologous proteins, a picture has begun to
emerge in which the different members associate
with one another to form transcription factors
that can activate different genes.
The homology between NF-kB and dorsal also
has interesting implications for understanding
the biology of both proteins. The developmental
morphogen dorsal is responsible for the estab-
lishment of the dorsoventral polarity in Drosoph-
ila embryos. The basis for its action is the estab-
lishment of a gradient of dorsal protein that is
progressively localized to the nucleus. The cyto-
plasmic localization of dorsal was genetically de-
termined to be due to the action of a gene called
cactus, which behaves like IkB, and recent clon-
ing of the gene has shown that it is a homologue
of mammalian IkB.
The signal for the nuclear localization of dor-
sal goes through a pathway involving eight up-
stream maternal-effect genes. However, for those
of us studying NF-kB, what has been of great inter-
est is the finding that one of these upstream
genes, toll, is a membrane receptor with a cyto-
plasmic domain that is homologous to one of the
mammalian interleukin-1 (IL-1) receptors. This
is significant because, among the different inter-
leukins, only IL-1 can activate NF-kB. Therefore
the two genes that lie between toll and dorsal
159
Signal Transduction Pathways in B Lymphocytes
(i.e., pelle and tube) are probably components of
the signal transduction pathway that resuh in the
modification of cactus. The challenge for the fu-
ture is to find the analogues of pelle and tube in
mammalian cells, and we are exploring difl'erent
approaches in our attempt to elucidate the signal
transduction pathway that results in the constitu-
tive activation of NF-/cB in mature B cells.
The next major advance was the cloning of I/cB,
and once again the cloning revealed a sequence
homology that is important for I/cB function. In
this case the sequence elements are the ankyrin
repeats that are found in a diverse group of pro-
teins, including the pi 05 precursor of NF-/cB
p50, Drosophila Notch, yeast cdclO, and the hu-
man erythrocyte ankyrin protein. Because all of
these proteins are either cytosolic or membrane
associated, the initial idea was that ankyrin se-
quences mediate cytosolic retention of proteins
by binding to cytosolic or membrane structures.
However, a nuclear transcription factor was re-
cently found to contain these sequences, necessi-
tating a change in our idea of the role of ankyrin
sequences. A more comprehensive model envi-
sions them as domains responsible for protein-
protein interaction. We also found that pp40, a
protein associated with v-Rel and c-Rel in
chicken lymphoid cells, is a homologue of mam-
malian IkB. This observation, along with the find-
ing that cactus is also homologous to I/cB, sug-
gests that the NF-KB/rel/dorsal family has
evolved over millions of years and used the same
general principles for regulating the activity of
these positive-acting transcription factors.
One of the most interesting questions about
NF-/(B is its regulation during B cell development.
NF-kB is an inducible cytosolic protein in pre-B
cells but changes into a constitutively active nu-
clear protein in mature B cells. There can be at
least two simple explanations for this: either a
transcriptional shutofl" of I/cB synthesis or a con-
stitutive modification of I/cB in mature B cells.
Preliminary results indicate that transcription of
the major form of I/cB is not shut off in mature B
cells, but the UB that is synthesized is somehow
inactivated. We are trying to determine how IkB
activity is regulated in mature B cells. We also
want to determine the signal responsible for the
developmental transition of a pre-B cell to a ma-
ture B cell.
The other major focus of our research is to un-
derstand the regulation of k gene expression in
plasma cells. Unlike mature B cells, plasma cells
express large amounts of the immunoglobulin
chains, including the k chains, which form com-
plete immunoglobulin molecules that are se-
creted. The high-level expression of the k gene is
not driven by the intronic enhancer; instead, an
enhancer located 9 kb downstream of the k gene
appears to be the dominant element in plasma
cells. Therefore k expression appears to be con-
trolled by a unique dual enhancer system during
development: the intronic enhancer drives low-
level expression in mature B cells, while the 3'
enhancer is responsible for high-level expression
in plasma cells. The questions we wish to address
in this system are, What is the protein(s) responsi-
ble for the activity of the 3' enhancer in plasma
cells? WTiat is the signal that activates the 3' en-
hancer? and finally. What role does crosslinking
of the surface immunoglobulins play in this pro-
cess? We would also like to study the surface im-
munoglobulin complex on mature B cells that in-
cludes proteins such as mb-1 and B-29 and
elucidate the signal transduction pathway that
leads to the activation of the 3' enhancer.
160
Molecular Genetics of Blood Coagulation
David Ginsburg, M.D. — Associate Investigator
Dr. Ginsburg is also Associate Professor in the Departments of Internal Medicine and Human Genetics
at the University of Michigan Medical School. He received his B.A. degree in molecular biophysics
and biochemistry from Yale University and his M.D. degree from Duke University School of Medicine.
His postdoctoral research training was done in the laboratory of Stuart Orkin at the Children's
Hospital, Harvard Medical School. While in Boston, Dr. Ginsburg was also Instructor in Medicine
at Brigham and Women 's Hospital, Harvard Medical School.
THE major research activities of my laboratory
concern two important blood clotting pro-
teins, von Willebrand factor and plasminogen
activator inhibitor- 1, and their associated hu-
man diseases. In addition, we are applying mo-
lecular tools to the study of bone marrow
transplantation.
von Willebrand Factor
One major function of von Willebrand factor
(vWF) , which is an important part of the body's
blood clotting system, is to serve as a bridge be-
tween blood platelets and the wall of an injured
blood vessel, thereby helping to control bleed-
ing. vWF is also the carrier for factor VIII, the
missing substance in patients with hemophilia.
Abnormalities in vWF result in von Willebrand
disease (vWD), the most common inherited
bleeding disorder, occurring in 1-3 percent of
the general population. Over 20 different types
of vWD have been described.
Our aim is to understand how the various parts
of the vWF protein work in the body and how
they interact with other factors in the blood clot-
ting system. Recently we have made considerable
progress in identifying the defects within the
vWF gene that are responsible for vWD. In addi-
tion to aiding in the diagnosis and management of
vWD, this information has provided important in-
sights into the function of vWF.
Over 90 percent of patients with type IIB vWD
have one of four specific defects, all within a
small region of the vWF gene critical for its inter-
action with blood platelets. By introducing one
of these defects into the DNA of tissue culture
cells, we have shown that this single change is
responsible for the type IIB variant. In similar
studies of type IIA vWD, we have identified an-
other set of defects clustered in a different region
of the vWF gene. We have introduced these de-
fects into cultured cells and have shown that type
IIA vWD may be due to abnormalities in the pro-
cess whereby vWF is manufactured inside the
cell. In studies of a patient whose vWF is unable
to bind factor VIII, we identified a single change
in the gene that has helped to pinpoint the region
of vWF responsible for factor VIII transport.
We have also identified in some patients with
type III vWD (the most severe form) a defect in
the ability to copy the vWF gene into normal mes-
senger RNA.
Despite considerable progress, type I vWD, the
most common variant, remains a mystery. We
have recently begun a new project to study the
molecular basis for a disease in mice that closely
resembles type I vWD of humans. Surprisingly, it
appears that the defect in the mouse may be in a
gene other than vWF. If we are successful in iden-
tifying the mutant gene in the mouse, our find-
ings should be directly transferable to the study
of human type I vWD. Through these studies, we
hope to expand our understanding of vWF, to ad-
vance our ability to diagnose and classify vWD,
and eventually to improve the medical treatment
for this common human disorder. The work on
von Willebrand factor has been funded in part by
a grant from the National Institutes of Health.
Plasminogen Activator Inhibitor-1
The fibrinolytic system is the body's mecha-
nism for breaking down blood clots. This system
must be precisely balanced with the clot-forming
system, since an imbalance can result in un-
wanted blood clotting or uncontrolled bleeding.
The protein that turns on the fibrinolytic system
is plasminogen activator. Its activity is controlled
by a regulator protein, plasminogen activator in-
hibitor! (PAI-1).
Synthetic plasminogen activators, such as re-
combinant tissue-type plasminogen activator
(t-PA) and urokinase, are now used in patients to
dissolve blood clots, particularly in the early
stages of a heart attack when a major blood vessel
to the heart muscle has become blocked. There is
also increasing evidence that patients with abnor-
mally high blood levels of PAI-1 (disrupting the
normal clot-dissolving activity of natural plas-
minogen activator) are at particularly high risk
for heart attacks and other diseases due to in-
creased blood clot formation. Thus an under-
standing of the structure and function of PAI-1
and its interaction with plasminogen activators is
161
Molecular Genetics of Blood Coagulation
of great significance in the study of the fibrino-
lytic system and its role in a number of important
diseases.
We have established a system for directing bac-
teria to make synthetic PAI-1 in the test tube. This
recombinant PAI-1 has all of the important prop-
erties of the natural protein. We have made over
170 variants of PAI-1 in which a small part of the
molecule has been changed. Studying the effects
of these small changes, which can vary consider-
ably in their ability to regulate the different types
of plasminogen activators, has advanced our un-
derstanding of how PAI-1 functions and how it
interacts with other parts of the blood clotting
system. These observations might eventually aid
in the design of new drugs for the treatment of
bleeding and clotting diseases. This work has
been funded in part by a grant from the National
Institutes of Health.
We have recently identified the molecular de-
fect in our first patient with complete deficiency
of PAI-1 . This nine-year-old girl has a moderately
severe bleeding disorder. Studies of her DNA dem-
onstrated complete inactivation of the gene, ac-
counting for the total absence of PAI-1 in her
blood. These observations have provided impor-
tant insights into the true function of PAI- 1 and
the fibrinolytic system in the body. Further stud-
ies will provide valuable information for the fu-
ture treatment of this and other patients with
PAI- 1 deficiency and for studying how PAI- 1 actu-
ally works inside the body. We are currently at-
tempting to disrupt the PAI-1 gene in mice to
create an animal model for this human disease.
Bone Marrow Transplantation
Bone marrow transplantation is being used
with increasing frequency to treat a variety of dis-
eases, including several types of leukemia and a
number of other cancers. In this procedure, doses
of radiation and chemotherapy are given that are
designed to destroy the patient's diseased bone
marrow. As this would ordinarily prove fatal, the
patient is then "rescued" by the transplantation
of marrow from a healthy donor. Our laboratory
has had a long-standing interest in bone marrow
transplantation and has developed techniques to
monitor what happens to the blood — to the pa-
tient's own blood cells and those received from
the donor — following the procedure.
Currently the major obstacle to the more wide-
spread use of bone marrow transplantation is an
often-fatal complication called graft-versus-host
disease (GVHD). This is caused by small differ-
ences between the genes of patient and donor
that cannot yet be detected. We have begun a new
research effort to identify the gene or genes re-
sponsible for GVHD. If this effort is successful,
the process of matching patients and donors will
be much improved and bone marrow transplanta-
tion may become considerably safer and more
widely applied.
162
Uncovering the Molecular Basis
of X-linked Disorders
Jane M. Gitschier, Ph.D. — Assistant Investigator
Dr. Gitschier is also Associate Professor of Medicine ( Genetics) at the University of California, San
Francisco. She received a B.S. degree in engineering science from Pennsylvania State University, an M.S.
in applied physics from Harvard University, and a Ph.D. in biology from the Massachusetts Institute
of Technology. She did postdoctoral research with Richard lawn at Genentech, before joining the faculty
at the University of California.
X-LINKED disorders result from mutations in
genes on the X cliromosome, one of the two
sex chromosomes in humans. These common in-
herited conditions almost exclusively affect
males because they lack the protection of a sec-
ond, normal X chromosome. Females, though
they have such protection, can be silent carriers
of X-linked diseases and genetically transmit
them to their sons. Consequently, X-linked dis-
eases are often referred to as "sex-linked."
Our laboratory is interested in uncovering the
molecular biology of some of these disorders. We
are seeking to identify both the genes responsible
for several diseases of unknown biochemical
basis and the underlying mutations.
Much of our effort is concentrated on a particu-
lar region of the X chromosome called Xq28, the
terminal portion of the chromosome's long arm.
To date, 24 inherited disorders have been
mapped to this region, which consists of approxi-
mately 9 million base pairs of DNA. As this density
of disease loci is unusually high compared with
that found in other regions of the chromosome, it
suggests that Xq28 may be very rich in genes.
The incidence of Xq28-linked disorders is vari-
able. Color blindness is extremely common, af-
fecting about 1 of every 10 males. However, the
majority of diseases occur infrequently, making
accurate genetic mapping difficult. These rare
diseases are clinically diverse and of unknown
biochemical basis. They include Emery-Dreifuss
muscular dystrophy, nephrogenic diabetes insi-
pidus, adrenoleukodystrophy, incontinentia pig-
menti, and dyskeratosis congenita.
We are attempting to isolate genes in Xq28 and
determine whether they are mutated in individ-
uals affected by Xq28-linked disorders. To date
we have isolated six candidate genes and have
collected 5 1 patient samples, including at least
one example of most of the rare diseases. In some
cases the sequence of the cloned gene has pro-
vided a clue to its function and suggested possi-
ble involvement in a particular disease. On the
other hand, one of the isolated genes does not
appear to produce a protein. Until a match is
made between candidate gene and inherited dis-
ease, we are screening all patient samples for mu-
tations in all genes, searching for gross rearrange-
ments in genes as well as single-base alterations.
One Xq28-linked disorder we have studied in
depth is classic hemophilia, or hemophilia A. It
results from mutations in the gene coding for a
blood coagulation protein called factor VIII. The
gene was identified previously because part of
the amino acid sequence was known. Our labora-
tory is interested in understanding what types of
mutations lead to hemophilia A and how these are
generated.
Hemophilia is well suited to studies on muta-
genesis because it is clinically heterogeneous,
implicating a wide variety of mutations in the fac-
tor VIII gene. Correlating the types of mutations
with particular clinical findings may be very
helpful in understanding the role of factor VIII in
coagulation. To that end we recently completed a
study of mutations that lead to amino acid
changes in the protein. These mutations, which
affect a single base pair in DNA, were revealed by
a sensitive technique called denaturing gradient
gel electrophoresis. By comparing the amino acid
sequence of factor VIII with sequences from evo-
lutionarily related proteins, we were able to infer
that some mutations are likely to destroy activity
by disrupting the protein's structure, while
others alter amino acids that play a role unique to
factor VIII function.
Hemophilia can occur in families lacking a ge-
netic history of hemophilia, reflecting newly aris-
ing mutations. It is possible to uncover the origin
of factor VIII gene mutations in these cases. In the
past, mutations were assumed to occur as isolated
events in either eggs or sperm. However, we have
found evidence in several sporadic cases for "mo-
saicism"— i.e., a mixture of cells with and with-
out mutation within the individual. These results
suggest that mutations can occur during embryo-
logical development. In addition, these findings
demonstrate that the unaffected mother of a child
with sporadic disease may be at substantial risk of
having a second child with the disease.
Another aspect of the hemophilia-related re-
search has been a continued investigation of the
factor VIII gene. We discovered that it may en-
Uncovering the Molecular Basis of X-linked Disorders
code two additional proteins. The two extra mes-
senger RNAs, termed factor Vlll-associated tran-
scripts A and B, emanate from an intron, one of
the noncoding regions of the factor VIII gene.
The function of the gene is unknown, but the
RNAs are produced at high levels in many differ-
ent cell types, suggesting a "housekeeping" role.
It is interesting to speculate that the A and B tran-
seripts influence the expression of factor VIII and
that mutations in them may in turn be responsible
for some instances of hemophilia.
Finally, we are attempting to create a hemophi-
lic mouse. Genes on the X chromosome in hu-
mans are located on the X in other mammals as
well, and hemophilia has been recognized in
male sheep, dogs, and cats. It would be prefera-
ble, however, to use a mouse model for rapid
testing of factor VIII products and for future gene
therapy experiments. In progress are experi-
ments to disrupt the normal mouse factor VIII
gene, in hopes that male mice bearing the disrup-
tion will have hemophilia. If successful, this ap-
proach will be extended to create mouse models
for other Xq28-linked disorders.
164
Membrane Lipids and Cell Regulation
John A. Glomset, M.D. — Investigator
Dr. Glomset is also Professor of Medicine and Biochemistry at the University of Washington School of
Medicine. He received his M.D. and M.D. /Ph.D. degrees from the University of Uppsala, Sweden. He then
joined the Department of Medicine at the University of Washington. He received an honorary M.D. degree
from the University of Oslo for his discovery of a plasma enzyme, lecithin:cholesterol acyltransferase
(LCAT). Dr. Glomset is a member of the National Academy of Sciences.
ONE of the major research areas in my labora-
tory concerns the phosphorus-containing
lipids (phospholipids) of animal cell mem-
branes. In particular, we have been focusing at-
tention on the biosynthesis and function of phos-
pholipids that contain a polyunsaturated fatty
acid, arachidonic acid. This fatty acid or its pre-
cursor, linoleic acid, must be present in the diet
if animals are to reproduce, develop, grow, and
maintain their health. The basis for this require-
ment is not fully understood, but most of the ara-
chidonic acid that becomes available to animal
cells is rapidly incorporated into membrane phos-
pholipids. Furthermore, in response to various
stimuli, a small amount of the arachidonic acid is
released from the phospholipids and converted
into products that can trigger many different cell
functions.
The mechanisms that promote the incorpora-
tion of arachidonic acid into cell membrane
phospholipids remain to be characterized, but it
is clear that the metabolic pathways involved are
not the classical ones described in most text-
books. For example, phosphatidylinositol (PI), a
phospholipid that plays an important role in cell
signaling, typically contains high amounts of ara-
chidonic acid. As much as 80 percent of the PI
present in animal cell membranes consists of a
molecular species that contains arachidonic acid
and a saturated fatty acid, called stearic acid. Ex-
periments with radioactive precursors have
shown that cells form this exceptional species of
PI, 5n-l-stearoyl-2-arachidonoyl (SA) PI, several
hours after forming Pis that contain other fatty
acids.
To investigate the special metabolic pathway
that forms SA PI, we recently examined the pos-
sibility that animal cells in culture might be
able to incorporate a radioactive precursor, sn-2-
arachidonoyl monoacylglycerol, into this phos-
pholipid. Our experiments showed that to be the
case. Furthermore, follow-up incubation experi-
ments with cell fractions identified two new en-
zyme activities that seemed to be involved. A
monoacylglycerol kinase activity could catalyze
the phosphorylation of sn-2-arachidonoyl mono-
acylglycerol to form the phospholipid sn-2-
arachidonoyl lysophosphatidic acid, and a stear-
oyl transacylase could convert the lysophospha-
tidic acid into SA phosphatidic acid. In the
presence of appropriate cofactors, other enzymes
could convert the phosphatidic acid into the
corresponding species of PI. Thus it appeared
that we had discovered a new pathway of PI
biosynthesis.
More recent experiments have suggested that
additional steps may contribute to the pathway.
We have been able to solubilize the stearoyl trans-
acylase from membranes and examine its specific-
ity. Surprisingly, the enzyme can use SA PI as a
stearoyl donor in the transacylase reaction. When
it does, a major product of the reaction is sn-2-
arachidonoyl lysophosphatidylinositol. If the
enzyme forms this product in intact cells, the ly-
sophosphatidylinositol might well have some spe-
cial function or metabolic fate. To investigate this
possibility we are currently conducting a search
for enzymes that catalyze the conversion of lyso-
phosphatidylinositol into PI. If we find some, we
will characterize their activities in order to de-
fine the pathway more completely. Then we will
attempt to localize the various enzymes within
cells in order to explore the pathway's intracellu-
lar role.
One potential role of pathways that form ara-
chidonic acid-containing phospholipids may
relate to the fine structure of animal cell mem-
branes. Thus the arachidonic acid in phospho-
lipids might conceivably affect their ability to
pack tightly with one another in membranes. We
began to investigate this possibility some time
ago in parallel with our studies of phospholipid
biosynthesis. We used a computer-based ap-
proach to examine the effects of arachidonic acid
on the conformation and packing properties of
model phospholipids in simulated monolayers.
The results of this molecular modeling approach
have suggested that arachidonic acid-containing
phospholipids may be able to form straight con-
formations and pack together in regular, tight
arrays.
We are currently attempting to test this possi-
bility, in collaboration with Howard Brockman of
165
Membrane Lipids and Cell Regulation
the Hormel Institute. If we find that semisyn-
thetic arachidonic acid-containing phospho-
lipids form tightly packed arrays in experimen-
tally produced phospholipid monolayers and
bilayers, we will expand our experiments to in-
clude animal cell membranes. A demonstration of
such tightly packed arrays in cell membranes
would raise fundamental questions about the role
of phospholipid domains in programming mem-
brane-associated events.
Fruiting bodies of the slime mold Dictyostelium, each a mass of spore cells supported by a thin
column of stalk cells.
Research of Richard Gomer.
166
Determination and Maintenance of Cell Type
Richard H. Gomer, Ph.D. — Assistant Investigator
Dr. Gomer is also Assistant Professor of Biochemistry and Cell Biology at Rice University and Adjunct
Assistant Professor of Cell Biology at Baylor College of Medicine. He received his B.A. degree in physics
from Pomona College and his Ph.D. degree in biology from the California Institute of Technology. He was
a postdoctoral fellow in the laboratory of Richard Firtel at the University of California, San Diego.
OUR laboratory is interested in the general
problem of differentiation and morphogen-
esis. We are trying to understand at a molecular
level some of the factors that determine the cell
type into which an embryonic cell differentiates
and how the ratios of the different cell types are
then maintained in an organism. As a model sys-
tem, we are using the slime mold Dictyostelium
discoideum.
Dictyostelium normally exists as undifferen-
tiated single cells called amoebae that eat bacte-
ria in soil and decaying leaves and proliferate by
cell division. When the amoebae eventually over-
grow their food supply and starve, they aggregate
together in groups of about 100,000. Roughly 80
percent of these cells become spores. (A spore is
a cell with a tough outer coat that forms an
"escape capsule.") The remaining 20 percent of
the cells form a stalk about 2 mm high that holds
the spore cells off the ground. A spore, dispersed
by the wind, will crack open to release an
amoeba, which may luckily find itself in the
midst of a new supply of bacteria. The advantage
of this organism is its simplicity: cells differen-
tiate into just two main types.
Determining Cell Fate
In the presence of a food source, Dictyoste-
lium cells grow to a certain size and then separate
their chromosomes to opposite sides of the cell
and divide in two. The cycle of growth and divi-
sion then repeats. In a field of cells, there will be
cells at all different phases of this cycle.
Dictyostelium uses a simple and elegant mech-
anism based on this cycle to determine whether a
cell will become stalk or spore. When the cells
starve, those cells that have recently separated
their chromosomes and divided will become
stalk cells; the remaining cells will become
spores. As long as the cells are randomly distrib-
uted with respect to the phase of their cell cycles,
there will always be the proper percentage of
cells in the "stalk" quadrant. We refer to this as a
musical chairs mechanism, since the decision of
any given cell to become either stalk or spore is
made by the phase of the cycle that the cell hap-
pens to be in at the time of the differentiation
signal (starvation).
Cell-type choice determination mechanisms of
this sort may operate in humans, and aberrations
could thus lead to birth defects. We are develop-
ing a variety of techniques that use DNA inserted
into Dictyostelium cells to identify genes in-
volved in cell-type differentiation. In the past
year we have engineered DNA constructs to mu-
tate Dictyostelium cells three ways: by having
too many copies of a sequence of DNA, by insen-
ing foreign DNA into the chromosomal DNA, and
by making RNA that binds to and renders useless
the RNA encoding a specific protein. In prelimi-
nary experiments with these constructs, we have
identified mutants that may have abnormal ratios
of the two cell types. In addition, to examine the
extent of the linkage between the cell cycle and
cell-type choice, we have treated cells with drugs
that disrupt the cell cycle. These experiments
confirmed that the cell-type choice mechanism is
linked to the cell cycle and does not use a sepa-
rate timer.
Sensing Cell Density
During Dictyostelium development, cells turn
on the stalk- or spore-specific genes only when
above a certain cell density. Being able to sense
whether other cells are nearby represents a para-
digm for possible mechanisms that would allow,
for instance, liver cells or others to sense how
much of the body is composed of liver cells. At
present, little is known about the molecular
mechanisms whereby the size and density of a
tissue are sensed by individual cells. Such mecha-
nisms would be centrally involved in the regula-
tion of growth during development, wound heal-
ing, and regeneration. In addition, an important
and relevant aspect of tumor cells is that they
have lost their ability to regulate the size and/or
density of the tissue and, as a result, proliferate.
One way this could happen would be if the tumor
cells could no longer properly sense the total
mass of the tissue.
167
Determination and Maintenance of Cell Type
In experiments funded by the National Insti-
tutes of Health, we have found that Dictyoste-
lium cells sense whether they are near other
starving cells by, upon starvation, secreting small
quantities of a protein that we have named den-
sity-sensing factor (DSF). Cells are sensitive to
DSF: above a threshold concentration they will
express cell-type-specific genes. We have done
jheoretical diffusion calculations and have
found, in agreement with our observations, that
DSF secreted from cells that are quite far from
other DSF-secreting cells diffuses away so quickly
that it never reaches the threshold concentration.
However, at a sufficiently high density of starving
cells, the DSF concentration will be above the
threshold value. We have purified DSF protein
and have found that DSF eventually breaks down
to small protein fragments that are much more
effective in activating differentiation. This might
be a safety mechanism to allow cells that cannot
find other starving cells to stimulate themselves
and perhaps form an isolated spore. Interestingly,
DSF is made and stored in growing cells and is
only secreted upon starvation.
We have cloned and sequenced the DNA en-
coding DSF. This has been used in turn to obtain
the sequence of amino acids for the DSF protein.
Computer comparison with data banks of other
protein sequences indicates that DSF is not
closely related to any known protein. We have
used the DSF DNA to generate Dictyostelium
cells that do not make DSF; these cells do not
aggregate unless DSF protein is added to their me-
dium. Dictyostelium cells, which eat bacteria
that are almost their size, starve asynchronously.
The behavior of the cells lacking DSF suggests
that its function is to coordinate development by
triggering aggregation when most of the cells in a
given area have starved, as signaled by DSF secre-
tion, and are ready to form one large fruiting
body. Finally, we have made antibodies that will
allow us to examine where DSF is stored in Dic-
tyostelium cells and to search for similar pro-
teins in other organisms.
These two photographs show the expression offasciclin II (brown axon
bundles) in the developing central nervous system of the grasshopper
(left) and T)Tosop\\i\2L embryos ( right). Although separated by well over
300 million years, these two insects continue to express the protein on
a specific subset of longitudinal axon pathways. The brown staining
results from HRP immunocytochemistry using anti-fasciclin II anti-
bodies in both species.
left photograph from Allan Harrelson and Corey Goodman. Right
photograph from Gabriele Grenningloh and Corey Goodman.
168
Growth Cone Guidance and Neuronal
Recognition in Drosophila
Corey S. Goodman, Ph.D. — Investigator
Dr. Goodman is also Class of '33 Professor of Neurobiology and Genetics in the Department of Molecular
and Cell Biology at the University of California, Berkeley, and Adjunct Professor in the Department of
Physiology at the University of California, San Francisco. He received his B.S. degree in biology from
Stanford University and his Ph.D. degree in developmental neurobiology from the University of California,
Berkeley. His postdoctoral work in developmental neurobiology was done at the University of California,
San Diego. Prior to his present position, Dr. Goodman was a faculty member at Stanford University. His
honors include the Alan T. Waterman Award from the National Science Board.
WE are interested in understanding the mo-
lecular mechanisms by which neuronal
growth cones find their way toward, and ulti-
mately recognize, their correct targets during de-
velopment. Growth cones navigate over long dis-
tances and often through a series of complex
choice points, appearing to follow signals on the
surfaces of cells and in the extracellular environ-
ment. Our studies are aimed at uncovering the
molecules and mechanisms that impart specific-
ity to the developing nervous system and thus al-
low growth cones to recognize their correct path-
ways and targets. To address these issues, we use
molecular genetic approaches in the fruit fly
Drosophila.
Guidance of Pioneers
Within and just outside of the developing cen-
tral nervous system (CNS) , certain glial cells and
other special midline cells provide instructive in-
formation for the differential guidance of the ini-
tial, "pioneering" growth cones as they choose
which cells to extend toward or along. For exam-
ple, a specific subset of cells at the midline ap-
pears to provide an attractive cue for the growth
cones that extend toward the midline and pio-
neer the commissural axon pathways that con-
nect the two sides of the developing CNS. Simi-
larly, a specific pattern of longitudinal glia
appears to provide an important substrate for the
formation of the longitudinal axon tracts that
connect adjoining segments of the CNS. We have
conducted a large-scale screen for mutations that
perturb the guidance of pioneering growth
cones.
Of the hundreds of new mutations identified in
this screen, mutations in three genes are of partic-
ular current interest. Mutations in longitudinals
lacking (fold) have a dramatic phenotype. al-
though both commissural and peripheral path-
ways are normal, as are most other aspects of em-
bryogenesis, the CNS of these mutants lack all
longitudinal axon tracts. In lola mutant embryos,
the longitudinal glia are born, initially migrate,
and divide as normal; the earliest defect is seen
about the time that the first growth cones contact
these glia and fail to extend along them.
Mutations in the second gene, commissure-
less, have an equally dramatic phenotype: al-
though all other axon pathways are normal, the
CNS of these mutants lack all commissural path-
ways. They also have normal peripheral nervous
system (PNS) axon pathways, muscles, sensory
organs, and body organization. In these mutant
embryos, the growth cones of CNS neurons do not
extend across the midline and commissures never
form, although other aspects of embryonic pat-
tern formation and neuronal development appear
normal. The commissureless gene product is a
good candidate to be either the signal or the re-
ceptor for the guidance of growth cones toward
the midline.
Mutations in a third gene, roundabout (robo) ,
lead to a dramatic misrouting of the growth cones
that normally pioneer the MPl pathway. For ex-
ample, the MPl growth cone extends across the
anterior commissure where it contacts its homo-
logue from the other side, rather than proceeding
posteriorly. In contrast, many other longitudinal
pathways develop as normal in robo mutant
embryos.
Pathway Recognition
Once the initial axon pathways are established,
the predominant guide for "follower" growth
cones is the surface of the earlier axons in these
pathways. Growth cones are able to distinguish
one particular axon bundle, or fascicle, out of an
array of many. The experimental analysis of these
phenomena led to the labeled pathways hypothe-
sis, which holds that axon pathways are diff'eren-
tially labeled by recognition molecules that en-
able growth cones to navigate through complex
choice points. To identify such recognition mole-
cules, we used an immunological approach to
identify and subsequently clone the genes encod-
ing five diff'erent surface glycoproteins: fasciclin
I, fasciclin II, fasciclin III, fasciclin IV, and neuro-
glian (a sixth, connectin, has recently been
cloned using a different method, as described in
169
Growth Cone Guidance and Neuronal Recognition in Drosophila
the following section) . Five of the six glycopro-
teins are dynamically expressed on overlapping
subsets of growth cones, axon fascicles, and glia
during embryonic development; the sixth, neuro-
glian, is more broadly expressed on the surface of
most axons and glia. Fasciclin II, neuroglian, and
fasciclin III are members of the immunoglobulin
superfamily; connectin is a member of the leu-
,cine-rich repeat family; fasciclin I and fasciclin
rv are neither related to each other nor thus far to
anything else in the data bank. Five of the six
proteins (except for fasciclin IV) can function as
homophilic cell adhesion molecules.
Genetic analysis has shown that fasciclin II is
indeed a neuronal recognition molecule that
plays an important role in specific growth cone
guidance. The fasciclin II protein is normally ex-
pressed on a subset of growth cones and axons
that pioneer and selectively fasciculate in the
MPl axon fascicle; later it is expressed on several
other axon pathways. In embryos that are mutant
for the fas II gene, although these specific
growth cones extend, they do not properly recog-
nize one another and fail to fasciculate. These
studies are the first molecular confirmation of the
existence of functional labels on specific axon
pathways in the developing nervous system.
A major current goal is to determine the molec-
ular machinery whereby the activation of fasci-
clin II, by either another fasciclin II molecule or
a different ligand on the surface of neighboring
neurons, instructs a fasciclin Il-expressing
growth cone to extend toward and along these
specific cells. We have established a highly sensi-
tive assay that utilizes a mutation in the fas II
gene that reduces the amount of fasciclin II pro-
tein to the minimal amount required for fas II
function. Using this assay we are looking for other
genes in which a 50 percent reduction in the
level of their protein product results in a failure
of fas II function. In this way we are in the pro-
cess of identifying genes that appear to encode
products that interact with the fas II gene and
thus function in the events of neuronal
recognition.
Our previous studies have shown that the ex-
pression of surface recognition molecules is dy-
namic and regional on the surface of individual
neurons; i.e., parts of the cell are differentially
labeled in accordance with the processes around
it for which it has a selective affinity. We wish to
uncover the signals that instruct a neuron, as it
navigates from one pathway to another, to change
the expression of surface recognition molecules
on its growth cone. For example, the growth
cones of many commissural interneurons extend
right past longitudinal axon pathways on their
own side; once they cross the midline, they dis-
play a high affinity for one of these axon pathways
on the other side of the CNS. This change in the
behavior of the growth cone is accompanied by
an equally dramatic change in its expression of
surface recognition molecules; neurons express
different molecules on the surface of their longi-
tudinal and commissural processes. We are inves-
tigating whether extension across the midline is
required for growth cones to make the appro-
priate changes in surface expression and subse-
quent pathway choices, by examining the behav-
ior of specific growth cones in commissureless
mutant embryos in which guidance toward the
midline does not occur. Thus far it appears that
some aspects of longitudinal guidance are per-
turbed in these mutant embryos.
Target Recognition
Having navigated along a series of pathways,
growth cones are ultimately capable of recogniz-
ing their correct target cells. In the Drosophila
embryo, the specificity of neuronal growth cones
for target cells is most clearly studied in the abil-
ity of motoneuron growth cones to recognize spe-
cific muscle fibers. A large-scale molecular ge-
netic screen has led to the identification of
several genes that are expressed by different sub-
sets of undiff^erentiated muscle fibers prior to in-
nervation. A new gene, connectin, and a previ-
ously identified gene. Toll, encode membrane
proteins that are members of the leucine-rich re-
peat family of cell adhesion and signaling mole-
cules. The connectin gene encodes a protein that
is expressed on the surface of eight (of the 31)
muscle fibers in each abdominal segment and on
the grovvT:h cones and peripheral axons of the
very motoneurons that innervate these eight mus-
cle fibers. The growth cones of this subset of mo-
toneurons, and the undifferentiated muscle
fibers they innervate, express the connectin pro-
tein prior to contact and synapse formation. In
vitro studies show that the connectin protein is
capable of functioning as a homophilic cell adhe-
sion molecule. Thus this protein is a good candi-
date to be a bona fide target recognition molecule
involved in the formation of specific synaptic
connections. A genetic analysis of connectin
function is presently in progress. In addition, a
large-scale genetic screen is under way to identify
mutations that perturb the guidance of motoneu-
ron growth cones.
170
Mechanisms of Immunological Self-Tolerance
and Autoimmunity
Christopher C. Goodnow, B.V.Sc, Ph.D. — Assistant Investigator
Dr. Goodnow is also Assistant Professor of Microbiology and Immunology at Stanford University School of
Medicine. He was educated in the United States and Australia and received B.S. and veterinary degrees
from the University of Sydney. After training in molecular immunology with Mark Davis at Stanford
University, he returned to the University of Sydney to complete doctoral and postdoctoral studies
on immunological tolerance in the laboratory of Antony Basten.
EACH individual B and T lymphocyte in the
immune system expresses antigen receptors
of one type on its surface, which confer on the
cell an ability to recognize one of the millions of
different antigens; and there are millions of dif-
ferent lymphocytes in the immune system. Given
the system's annihilative powers, it is remarkable
that tissue components of our own bodies are
spared during immunological attacks on invading
foreign organisms. Normally the immune system
can recognize one's tissue components as "self"
and tolerate them. Self-tolerance is lost, however,
in a variety of "autoimmune" diseases, such as
systemic lupus erythematosus, type I diabetes
mellitus, and rheumatoid arthritis, resulting in
inexorable destruction of particular organs and
tissues. The mechanisms that maintain self-
tolerance in healthy individuals, and the factors
that lead to its breakdown in autoimmune dis-
ease, are the main focus of our laboratory.
It has been theorized for many decades that
self-tolerance might somehow result from the si-
lencing or elimination of lymphocytes bearing
antigen receptors that happen to recognize self
antigens. To determine whether this idea was
correct, however, has been difficult, since it is
almost impossible to track the life of any one cell
among millions.
Advances in biotechnology, in particular the
advent of transgenic mice, have opened doors to
the development of ways to follow the life of par-
ticular immune cells in vivo. Transgenic mice
are genetically altered at the outset of embryonic
development by microinjecting carefully de-
signed cassettes (transgenes) into fertilized oo-
cytes. With colleagues in the laboratory of Antony
Basten at the University of Sydney, we produced
transgenic mice in which most of the B lympho-
cytes expressed identical, rather than widely dif-
fering, antigen receptors.
This was done by introducing transgenes that
coded for a single antibody molecule (since anti-
body molecules serve as antigen receptors on B
lymphocytes) . The particular molecule was one
that recognized and bound a foreign protein, hen
egg-white lysozyme, and because the transgene
was expressed in essentially all the B lympho-
cytes, they all now recognized lysozyme in an
identical fashion.
The extraordinary abundance of lysozyme-
binding B cells in the transgenic mice has made it
possible to track the development and fate of
these cells in the body. To determine the fate of B
cells that might happen to recognize a self anti-
gen rather than a foreign one, we prepared addi-
tional transgenic mice in which transgene cas-
settes led to the synthesis and production of
lysozyme by the mouse itself. When the two types
of mice are mated, a fraction of their offspring
inherit both types of transgene and thus contain
large numbers of lysozyme-binding B cells that
encounter lysozyme expressed as if it were a nor-
mal self constituent.
The self-reactive B cells that develop in these
mice undergo one of three distinct fates. First, if
lysozyme is present at too low a concentration or
it binds to the B cell's receptors with too low an
affinity, no signal is registered by the cell. As a
result, the B cells are neither activated nor ren-
dered tolerant but remain fully capable of mount-
ing an antibody response if they receive a
stronger stimulus subsequently. Second, when
higher concentrations of soluble lysozyme are
bound with sufficient affinity, the B cells con-
tinue to follow their normal maturation program
but store some sort of negative signal, such that
they are unable to mount an antibody response
when stimulated to do so. Third, if lysozyme is
encountered in a form where it is displayed on
cell surfaces, and thus bound by the B cell with
extremely high avidity, developing B cells in the
bone marrow are triggered to abort their matura-
tion program and die.
These distinct cellular events help to explain
the long-standing observation that self-tolerance
is an actively acquired but incomplete process
and suggest a number of possible avenues along
which autoimmunity may ultimately develop.
For example, our laboratory has found that toler-
ant B cells that have stored a negative signal can
nevertheless recover the ability to mount an effi-
cient antibody response. Recovery of function
171
Mechanisms of Immunological Self-Tolerance and Autoimmunity
and breakage of tolerance depends, however, on
a balance between stimuli that promote antibody
formation and stimuli that reinforce tolerance.
One line of research in our laboratory involves
developing transgenic mouse models for study-
ing how the interplay between these factors
shapes the development and remission of autoim-
mune disease in vivo. This research is funded by
the National Institutes of Health.
Identifying the molecular events that prevent
tolerant cells from mounting an antibody re-
sponse, or lead to the death of the autoreactive
cells, will be crucial for fully understanding how
self-tolerance is induced and maintained. Our ap-
proach to this problem is severalfold. First, we
are continuing to delineate the precise condi-
tions under which tolerance rather than activa-
tion is induced in B cells in vivo. Second, we
have developed tissue culture systems that repro-
duce the cellular responses observed in the
whole animal, allowing much closer scrutiny of
the sequence of events accompanying tolerance
induction. In autoreactive B cells that are unable
to mount an antibody response, this is due to in-
terference with the cell's capacity to replicate
and to differentiate into an antibody-secreting
cell. For B cells that are triggered to die, the pri-
mary event appears to be an arrest of cell develop-
ment at an immature, short-lived stage of develop-
ment. Third, we are using a combination of
biochemical, molecular, and genetic strategies to
search for the key differences in biochemical sig-
naling cascades or changes in gene expression
that underlie these events.
Self-reactive B lymphocytes sorted from the bone marrow and visual-
ized by electron microscopy. The antigen receptors on these cells have
already been triggered, and as a result the cells are destined to die
within a few days in vivo. At this stage, however, the process is revers-
ible and the cells show none of the ultrastructural changes that precede
programmed cell death.
Research of Suzanne Hartley and Christopher Goodnow.
172
Developmental Control of Gene Expression
Rudolf Grosschedl, Ph.D. — Assistant Investigator
Dr. Grosschedl is also Associate Professor of Microbiology and Immunology and of Biochemistry and
Biophysics at the University of California, San Francisco. He completed his undergraduate studies on the
replication of lambdoid bacteriophages in the laboratory of Gerd Hobom in Freiburg, West Germany. His
graduate studies, on the regulation of histone gene expression, were carried out in the laboratory of Max
Birnstiel in Zurich, Switzerland. Dr. Grosschedl spent his postdoctoral years with David Baltimore at the
Massachusetts Institute of Technology and the Whitehead Institute.
THE process of terminal differentiation turns a
multipotential cell into a cell that carries out
a particular function or synthesizes a specific
product. The lymphoid B cell lineage ultimately
generates a cell that secretes antibody. During
cell differentiation, genes that encode the anti-
body or associated proteins are expressed in a de-
fined cell type-specific and temporally ordered
pattern.
Transcription of the m immunoglobulin (Ig)
gene encoding the heavy chain of the antibody
can be detected in virtually all lymphocytes. By
contrast, the k immunoglobulin light-chain gene
is transcribed only in late-stage B cells, and the
mb l gene encoding an antibody-associated pro-
tein is expressed only in early-stage B cells. The
goal of our research is to gain some insight into
the molecular mechanisms that mediate the de-
velopmental control of lymphoid-specific gene
expression.
Tissue-Specific Regulation of Ig Gene
Expression in a Transgenic Model
We are attempting to understand how multiple
cis-acting regulatory information of a gene is in-
tegrated to govern the correct developmental
pattern of expression. Toward this goal, we trans-
ferred rearranged wild-type and mutated Ig
genes into the mouse germline and analyzed ex-
pression in various tissues and developmental
stages. Our first set of experiments suggested that
lymphoid-specific expression is dependent upon
multiple tiers of regulation.
We found that the intragenic enhancer directs
high-level expression in lymphoid tissues and ba-
sal expression in most but not all nonlymphoid
tissues. Basal expression appears to be governed
by negative regulation, since mutations of the
juE5- and ME2-binding sites in the enhancer in-
creased At gene expression in many nonlymphoid
tissues to levels similar to those found in lym-
phoid tissues. However, the "off-state" of ex-
pression observed in a few other nonlymphoid
tissues was not affected by these mutations, sug-
gesting that another tier of negative regulation
may be involved in further decreasing basal
expression.
An additional set of experiments suggested that
the Oct-binding site in the promoter contributes
to the positive regulation of ^l gene expression
during B cell differentiation. A mutation in the
Oct-binding site decreased ti gene expression in
fetal pre-B cells (which represent early B cells)
by a factor of 5, but decreased ^L gene expression
in stimulated splenocytes (representing late-
stage B cells) by a factor of 100 to 200. Together
these experiments indicate that the multiple
modes of regulation are necessary to maximize
the difference between the on-state and off-state
of gene expression in vivo.
In a second set of experiments, we addressed
the regulation of the accessibility of binding sites
for nuclear factors in native chromatin. With the
aim of uncoupling changes in chromatin struc-
ture from the process of transcription, we linked
DNA fragments comprising the m enhancer region
to the promoter for bacteriophage T7 RNA poly-
merase and incorporated the gene construct into
the mouse germline. Subsequently, we examined
the accessibility of the T7 promoter in isolated
pre-B cell nuclei by adding exogenous T7 RNA
polymerase and measuring the synthesis of
TT-specific RNA.
The T7 promoter was accessible in 7/7 lines
when linked to the ti enhancer. By contrast, the
T7 promoter alone was only weakly accessible in
two lines and not accessible in six. Because the
enhancer fragment used in this experiment
lacked any known promoter activity, we are
currently using this assay to examine which se-
quences confer accessibility in native chromatin.
Novel Lymphoid-Specific Regulators
of Gene Expression
To identify and clone novel lineage-specific
transcriptional regulators, we adopted two ap-
proaches. First, we included in our analysis of
lymphoid-specific gene expression the mb-1
promoter, which has a cell type-specific pattern
of activity distinct from that of the Ig promoters.
We found that the mb- 1 promoter is functional in
173
Developmental Control of Gene Expression
pre-B and surface Ig-positive B cells but not in
later-stage B cells or T cells. In our study of the
mb- 1 promoter, we identified a novel nuclear
factor, termed early B cell factor (EBF) .
EBF was shown to bind a functionally impor-
tant sequence in the distal promoter region. The
factor was found to be present specifically in pre-
B and surface Ig-positive B cells, thus paralleling
the pattern of activity of the mb- 1 promoter. Re-
cently we purified EBF to homogeneity and ob-
tained partial amino acid sequences. Based on
this information, we isolated cDNA clones that
correspond to EBF. Currently we are characteriz-
ing its DNA-binding and regulatory properties.
The second approach to identify lineage-
specific regulators of gene expression consisted
in differential screening of a pre-B cell (minus
erythroid) cDNA library. One of the cDNA clones
was found to encode a DNA-binding protein with
homology to the chromosomal nonhistone high-
mobility group protein HMG- 1 and to regulators
of cell specialization. This pre-B and T cell-spe-
cific DNA-binding protein, termed LEF-1 (lym-
phoid enhancer-binding factor 1), was found to
bind a functionally important site in the T cell
receptor (TCR) a gene enhancer.
LEF-1 was shown to participate in the regula-
tion of the TCRa enhancer. First, co-transfection
of cells lacking endogenous LEF-1 with TCRa en-
hancer-containing reporter genes together with
an LEF-1 cDNA expression plasmid increased en-
hancer function. Second, in the context of the
intact minimal TCRa enhancer, the LEF-1 -bind-
ing site is required for full enhancer function.
We showed that the HMG domain of LEF- 1 en-
compasses a sequence-specific DNA-binding do-
main that interacts with the minor groove of DNA.
Moreover, we defined amino acids that are con-
served among various members of the family of
HMG-domain proteins as residues important for
DNA binding.
In previous experiments, we found that LEF-1
by itself is unable to augment basal promoter ac-
tivity. This observation raised the possibility that
LEF-1 functions by aiding the binding and/or ac-
tion of other nuclear factors. Recently we demon-
strated that DNA binding by the HMG domain in-
duces a sharp bend in the DNA helix. Therefore,
we examined whether LEF- 1 -induced DNA bends
can facilitate communication between proteins
bound at distant sites.
We replaced one binding site for the bacterial
integration host factor with one for LEF- 1 in the
att? locus of bacteriophage X and found that
LEF- 1 is capable of precisely aligning widely sepa-
rated integrase (Int) protein-binding sites and
stimulating Int-dependent recombination. These
data suggest that LEF-1 can serve as an "architec-
tural" element in the assembly of higher-order
nucleoprotein structures. Currently we are exam-
ining the role of DNA bending for the regulation
of TCRa enhancer function and are attempting
to gain insight into the importance of LEF- 1 for
lymphoid differentiation by targeted gene
inactivation.
174
Polypeptide Hormone Gene Regulation
Joel F. Habener, M.D. — Investigator
Dr. Habener is also Professor of Medicine at Harvard Medical School and Associate Physician and Chief
of the Laboratory of Molecular Endocrinology in the Department of Medicine at Massachusetts General
Hospital, Boston. He obtained his B.S. degree in chemistry at the University of Redlands and his M.D.
degree at the University of California, Los Angeles. After medical residency training at the Johns Hopkins
Hospital, he spent two years in research at the National Cancer Institute. Dr. Habener completed his
medical training in endocrinology and metabolism at Massachusetts General Hospital.
OUR laboratory seeks an understanding of the
molecular processes involved in the regula-
tion of gene expression. The general hypothesis
being tested is that peptide hormones are impor-
tant regulatory molecules in conveying informa-
tion among cells via ligand-receptor interactions
and corresponding signal transduction, resulting
in the expression of specific genes. These pro-
cesses are important in determining cellular meta-
bolic responses such as secretory activity, cellu-
lar differentiation, and growth.
Peptides in Cellular Metabolism
Peptides activate metabolic responses in cells
by way of interactions with specific receptors on
distant (endocrine), adjacent (paracrine), or the
same (autocrine) cells. These ligand-receptor in-
teractions lead to the activation of signal trans-
duction pathways involving postulated cascades
of protein phosphorylation enzymatically cata-
lyzed by protein kinases, eventuating in the as-
sembly of active transcriptional complexes.
Under intensive investigation are two such signal-
ing pathways mediated by cAMP-dependent pro-
tein kinase A and by phospholipid/diacylglyc-
erol-stimulated protein kinase C. A major focus of
the laboratory is to understand how specific
phosphorylated DNA-binding proteins interact
with cognate DNA sequences and thereby induce
gene expression.
Genes Encoding Polypeptide Hormones
We have determined the structures, organiza-
tion, and regulation of the expression of genes
encoding several of the polypeptide hormones.
Our work has centered on the genes encoding the
glucagon and glucagon-related peptides, somato-
statin, insulin, angiotensinogen, and the gonado-
tropins. These studies have progressed through
several stages: 1) cloning of the genes and eluci-
dation of their structures, 2) determinations of
the DNA enhancer and suppressor sequences re-
sponsible for the regulation of the transcriptional
expression of the genes, and 3) isolation and char-
acterization of DNA-binding proteins involved in
the regulation of expression. These studies have
led to the identification of cell-specific enhancer
sequences within the "promoter" regions of the
genes — sequences that determine in which cel-
lular phenorype the genes are expressed and how
gene transcription responds to activator sub-
stances such as cAMP and phorbol esters. They
have also led to the identification of complex,
cell-specific post-translational processing of pro-
tein precursors (prohormones) encoding the
peptide hormones.
Recently we determined that specific nuclear
proteins bind to these important DNA enhancer
sequences and that the binding specificities, as
well as the transactivation activities, of these pro-
teins are regulated by their phosphorylation.
Regulation of Glucagon and Somatostatin
Gene Expression
Our analyses of the regulation of the expres-
sion of the glucagon gene in pancreatic islet cell
lines reveal that islet cell-specific expression re-
sides in at least two enhancer-like sequences and
that A cell expression in the islets is determined
by an additional enhancer/promoter combina-
tion. The expression of the somatostatin gene
is restricted to the D cells because suppressor
elements prevent expression in the glucagon-
producing A cells and insulin-producing B cells.
Transcriptional activation of the glucagon gene is
mediated through both protein kinase C and pro-
tein kinase A pathways, whereas activation of the
somatostatin gene is regulated by protein kinase A
and a calcium-regulated pathway.
Cloning and Structure of a cAMP-Dependent
DNA-binding Protein
In studies of the somatostatin and gonadotro-
pin genes, we have determined that their expres-
sion is stimulated via a cAMP-dependent signal
transduction pathway. We have discovered that
DNA-binding proteins interact with specific, short
DNA sequences to generate cAMP-responsive
complexes. These DNA-protein complexes that
mediate either cAMP or phorbol ester control of
gene transcription share certain related struc-
tures. Cooperative interactions among several
175
Polypeptide Hormone Gene Regulation
DNA-binding proteins and with adjacent target
DNA sequences appear to determine cellular
specificity of gene expression as well as metaboli-
cally regulated expression. Recently we have
cloned several members of a family of cAMP-
responsive enhancer-binding proteins (CREBs)
and have discovered a domain on one of the
CREBs that is phosphorylated by the cAMP-
dependent protein kinase A, as well as by addi-
tional protein kinases. The phosphorylation of
CREB has a profound effect of increasing the tran-
sactivation of gene transcription.
Discovery of an Insulinotropic Peptide
We have discovered a new glucagon-like pep-
tide related in its structure to pancreatic gluca-
gon and co-encoded with glucagon in the precur-
sor protein of the two hormones (proglucagon) .
These peptides are differentially cleaved from
the proglucagon, the initial translation product,
in the pancreas and the intestines. One of the
peptides, glucagon-like peptide ! (7-37) pro-
duced in the intestine and released in response to
oral nutrients has potent insulinotropic activi-
ties. Concentrations as low as 10""- 10"'^ M
stimulate insulin secretion in the perfused rat
pancreas and stimulate both cAMP formation and
proinsulin gene transcription and mRNA levels in
insulinoma cell lines. We have determined that
the glucagon-like peptide regulates insulin se-
cretion in humans and propose that it may be in-
volved in the pathogenesis of certain types of dia-
betes mellitus due to a faulty regulation.
Particular emphasis is on analyses of the |8-cell
receptor for the glucagon-like peptide and the
mechanisms operative in the stimulation of insu-
lin gene transcription in response to the actions
of the peptide.
Our goals are 1) to characterize further the
genes encoding the regulatory (DNA-binding)
proteins that interact with tissue-specific en-
hancers and to determine how cAMP-mediated
expression of the peptide hormone-encoding
genes is regulated in the specific cellular pheno-
types and 2) to test further the new glucagon-like
peptides for their potential regulatory actions
within the pancreatic islets and the intestinal
tract and to explore the possible role of the pep-
tides in diabetes mellitus.
Some of these studies were also supported by
funds from the National Institutes of Health.
176
Structural Studies of Macromolecular Assemblies
Stephen C. Harrison, Ph.D. — Investigator
Dr. Harrison is also Professor of Biochemistry and Molecular Biology at Harvard University and Research
Associate in Medicine at the Children's Hospital, Boston. He received his A.B. degree in chemistry
and physics from Harvard College and his Ph.D. degree in biophysics from Harvard University.
Dr. Harrison is a member of the National Academy of Sciences.
HOW do regulatory proteins activate or in-
hibit transcription of particular genes? How
do viruses leave and enter cells? How do recep-
tors and their ligands cycle from cell surface to
ceil interior and back? These questions deal with
molecular recognition in the determination of
cell organization. They represent groups of proj-
ects in our laboratory, all of which involve eluci-
dation of detailed atomic structures as a prerequi-
site for tackling functional problems.
Transcriptional Regulatory Complexes
A common characteristic of eukaryotic tran-
scriptional regulatory elements is the presence of
sites in multiple copies varying slightly in se-
quence, often with two or more related proteins
that can bind to them. The best-understood pro-
karyotic paradigm is in the immunity region of
temperate bacteriophages, where two proteins,
repressor and Cro, bind two sets of three sites,
with appropriately graded affinities. We have
made an effort to understand the mechanism of
this regulatory switch, by determining the struc-
tures of a series of specific protein/DNA com-
plexes containing the Cro protein of phage 434
or the DNA-binding domain of its repressor. We
are using computational approaches to link ob-
served structural differences among these various
complexes with the corresponding free energies
of binding.
We are studying structural aspects of eukaryo-
tic transcriptional regulation, initially by examin-
ing the DNA-binding domains of regulatory pro-
teins in complex with synthetic binding sites.
GAL4, the prototype of a class of such proteins, is
a regulator of galactose metabolism in yeast. It
binds as a dimer to the upstream activity se-
quences of several genes involved in galactose
and melibiose catabolism. The DNA-binding re-
gion is at the amino terminus of the 881 -residue
polypeptide chain. We have determined the
structure of a fragment containing residues 1-65,
in a specific complex with DNA. A small domain
containing zinc ions (residues 8-40) recognizes
a conserved CCG triplet at each end of the 17-
base pair binding site, through direct, major-
groove contacts. A short a-helical coiled-coil ele-
ment imposes twofold symmetry. A segment of
extended polypeptide chain, linking the metal-
binding module to the dimerization element,
specifies the length of the site. The complex has a
relatively open structure, which would allow an-
other protein to bind coordinately with GAL4 . Co-
ordinate binding of two or more proteins has
been shown to be an important feature of many
eukaryotic control elements.
GCN4, also a yeast transcriptional regulator,
represents yet another class of DNA-binding pro-
teins. It contains a dimerization element, gener-
ally called a leucine zipper, which forms an
a-helical coiled coil about 30 residues in length.
This segment is preceded in the protein sequence
by a positively charged region, which has little
ordered structure in the free protein but which
also acquires a-helical structure when it binds
to DNA. We have prepared crystals of the basic
region/leucine zipper fragment of GCN4, in
complex with a synthetic binding site (having a
so-called AP- 1 sequence) . The structure determi-
nation is nearly complete. Preliminary analysis
shows that each chain is a continuous a-helix.
The part corresponding to the basic region lies
along the major groove, and contacts can be made
by suitable side chains to four base pairs on either
side of a central GC.
TflllA, which controls 5S RNA transcription in
Xenopus, represents the so-called zinc finger
class of regulatory proteins. The finger is a small,
30-residue domain, stabilized by a tightly bound
zinc ion. A recombinant fragment comprising
seven of the nine fingers from TflllA binds to a
30-base pair DNA containing an appropriate part
of the total binding site, and we have crystallized
this complex. The fragment also binds specifi-
cally to 5S RNA.
Understanding how these various structures
recognize their DNA sites is only a beginning. The
specificities of interactions between other do-
mains of these proteins and additional compo-
177
Structural Studies of Macromolecular Assemblies
nents of a transcriptional initiation complex pres-
ent even more challenging puzzles for the future.
Viruses
The small, double-stranded DNA viruses SV40
(simian virus 40) and polyoma have given us a
first glimpse of virus particles that package a mini-
chromosome in one cell and deliver it to the nu-
cleus in another. The shells of these viruses are
composed of 72 pentamers of the major struc-
tural protein VPl and 30-60 copies each of two
internal proteins, VP2 and VP3. These compo-
nents package the viral DNA. The VPl polypep-
tide chain folds in such a w^ay that two large
j8-sheets with radially directed strands form a
framework, with very tight interactions between
adjacent subunits in a pentamer. The carboxyl
terminus of VPl forms an extended arm that in-
teracts with subunits of another pentamer, gener-
ating three kinds of interpentamer contact in the
virus particle. This tying together of standard
building blocks allows for the required variabil-
ity in packing geometry without sacrificing speci-
ficity. Flexibly extended arms, which form or-
dered structures only when the units assemble
into a particle, appear to be an important feature
of complex assemblies.
A recently determined structure for the murine
polyoma virus shows that an important surface
loop is larger in the polyoma subunit than in
SV40. Mutational evidence suggests that this loop
creates a shallow pocket for binding sialic acid,
required for cell entry by polyoma but not by
SV40. A number of viruses of various structural
types use cell-surface sialic acid for attachment,
and it is of broad interest to understand how spe-
cific viruses accomplish the interaction. Compar-
ison of SV40 and polyoma shows that this func-
tion can readily be added or lost by small changes
at the surface of a viral coat protein.
We have recently begun to study the double-
stranded RNA viruses. Crystals of rotavirus single-
shelled particles and reovirus cores diflfract to
at least 7-A resolution. These particles are elab-
orately organized molecular machines, con-
taining the complete transcription and RNA-
modification activities.
Receptors
The receptor for human immunodeficiency
virus (HIV) is the lymphocyte surface antigen
CD4. Its extracellular portion is composed of
four immunoglobulin-like domains. We have de-
termined the atomic structure of a two-domain,
amino-terminal fragment, which binds HIV as
tightly as does the intact receptor. The first two
domains are joined by a continuous (8-strand con-
nector, and they have an extensive hydrophobic
interface. Thus they form a rigid, rod-like seg-
ment. The HIV-binding site appears to be a ridge
along one edge of the first domain. Binding and
mutational studies, carried out collaboratively,
show that a projecting phenylalanyl side chain is
critical for the interaction with HIV gpl20 and
that various positively charged residues sur-
rounding it are also important. The same region
appears to be involved in contacts with class II
MHC (major histocompatibility complex) mole-
cules. We have also crystallized a four-domain
fragment of CD4, corresponding to the entire ex-
tracellular region, and the structure determina-
tion is in progress.
Many important receptors are taken up into
the cell by a process of endocytosis mediated
by clathrin-coated vesicles. The transferrin re-
ceptor is one such molecule. It undergoes a well-
characterized cycle of uptake and return to the
cell surface. We have crystallized the extracellu-
lar domain, which makes up about three-fourths
of the molecule. This domain exhibits reversible
conformational changes at low pH that we be-
lieve to be significant for intracellular sorting
steps. The determination of the structure has re-
cently been facilitated by the observation that
much better diffraction data can be collected by
studying crystals at liquid-nitrogen temperatures.
Such freezing techniques are now being broadly
applied in our laboratory, enhancing our ability
to study radiation-sensitive crystals of very com-
plex structures.
178
Control of Gene Expression During the Cell Cycle
and Development of the Mammalian Cerebellum
Nathaniel Heintz, Ph.D. — Investigator
Dr. Heintz is also Professor at the Rockefeller University. He received his Ph.D. degree at the State
University of New York at Albany, where he studied the genetics and biochemistry of bacteriophage SPOl
gene expression. During postdoctoral studies with Robert Roeder at Washington University, St. Louis,
he initiated his work on histone gene expression during the cell cycle. Continuation of these studies
and examination of the developing mammalian cerebellum are his current research interests.
OUR studies are focused on the identification
of molecular mechanisms controlling gene
expression during the cell cycle and in the devel-
oping cerebellum. The elucidation of these mech-
anisms should provide fundamental insights into
the biological transitions that underlie the con-
trol of cell division and the development of the
mammalian central nervous system.
Control of Gene Expression During
the Cell Cycle
Transcription of histone genes during the S
phase of the eukaryotic cell cycle is achieved
through the agency of subtype-specific consensus
elements within histone gene promoters and
their cognate distinct transcription factors. Co-
ordinate transcription of this gene family is ac-
complished through biochemically distinct tran-
scription factors; this suggests pleiotropic
regulatory mechanisms that control the activities
of these factors during the cell cycle. We wish to
identify these mechanisms at the molecular level
and elucidate their importance for cell-cycle pro-
gression. Identification of such mechanisms may
provide highly specific targets for intervention in
cell growth.
Recently we have examined the post-transla-
tional modifications to the transcription factor
Octl, which participates in histone H2b expres-
sion during the cell cycle. Both monoclonal and
polyclonal antisera have been used to analyze
changes in phosphorylation of Octl as cells pro-
ceed to division. We have discovered that multi-
ple forms of Octl exist in the cell, and their dis-
tribution is dramatically regulated during the cell
cycle.
Further analysis of Octl phosphorylation has
established that late in the cell cycle this tran-
scription factor appears to be a substrate for both
CDC2 kinase and protein kinase A (PKA). Phos-
phorylation of Octl by PKA occurs within its
DNA-binding domain and results in loss of DNA
binding in vitro. This correlates with a signifi-
cant decrease in Octl DNA binding during mito-
sis in vivo. These results indicate that modifica-
tion of Octl during the cell cycle results in
modulation of its function. Inactivation of Octl
by phosphorylation during mitosis also provides
a possible explanation of the long-standing ob-
servation that transcription is generally sup-
pressed during this time in the cell cycle. Our
present efforts are to extend this analysis to the
histone HI transcription factor H1TF2, to deter-
mine whether its control during the cell cycle
may occur through the same molecular mecha-
nisms. Demonstration that the timing and nature
of the post-translational modifications on Octl
and H1TF2 are similar in vivo would prove the
existence of the proposed pleiotropic molecular
mechanism for regulation of transcription during
the S phase.
A question that has arisen from these studies is
whether S-phase-specific transcription and DNA
replication are mechanistically coupled. To ad-
dress this issue we have focused on two specific
questions: Are the regulatory proteins for S-phase
histone gene transcription directly involved in
DNA synthesis? Might proteins that regulate DNA
synthesis at specific chromosomal origins of repli-
cation be activated by the same mechanisms that
modulate those transcription factors?
In collaboration with Nicholas Heintz (Univer-
sity of Vermont) and Lisa Dailey (Rockefeller Uni-
versity) we have identified a cellular protein
complex with several properties expected of rep-
lication-initiation factors. We have recently pre-
pared antibodies and obtained primary amino
acid sequences from one of these proteins
(RIP60) and are using these tools to determine
whether this factor participates in cellular DNA
synthesis. Analysis of this factor during the cell
cycle may help answer the two questions posed
above.
Development of the Mammalian Cerebellum
The mammalian cerebellum is a complex and
highly stereotyped structure in which major pat-
tern formation and functional organization occur
postnatally. The precise description of the cellu-
lar events occurring during cerebellar develop-
ment, and the existence of many mutant mouse
strains in which normal development of the cere-
179
Control of Gene Expression During the Cell Cycle and Development
of the Mammalian Cerebellum
bellum is perturbed, recommend it as an amena-
ble system for molecular analysis of central ner-
vous system development. Our initial interests in
this area have been to identify genes that are ei-
ther essential for normal development of the cere-
bellum or that serve as molecular markers for spe-
cific developmental events that occur during its
formation. Our ultimate goal is to utilize these
genes to identify novel proteins that are crucial to
proper development of the cerebellum and to
identify molecular mechanisms that participate
in specific developmental events by analysis of
the pathways that result in their correct spatial
and temporal expression.
To identify genes that are required for normal
development or maintenance of cerebellar struc-
ture and function, we have initiated efforts to
clone the genes responsible for several neurologi-
cal mutants of mice. Our most significant pro-
gress has been in studies concerning the Lurcher
{Lc) and meander tail (mea) loci. Lc is a semi-
dominant mutation that results in death of essen-
tially all cerebellar Purkinje cells, beginning at
about two weeks of age. Secondary loss of cere-
bellar granule cells and olivary neurons is also
observed. We have constructed a detailed genetic
map surrounding the Lc locus on chromosome 6
and have identified an RLFP (restriction fragment
length polymorphism) marker approximately
0.5 cM from the gene. Genomic sequences from
this closely linked marker were used to screen a
yeast artificial chromosome (YAC) library from
Shirley Tilghman (HHMI, Princeton University),
resulting in isolation of a 280-kilobase YAC that
maps to chromosome 6. Using sequences isolated
from this YAC and informative recombinants gen-
erated during genetic mapping of the Lc locus,
we have begun a chromosomal walk toward the
Lc gene.
The gene mea is a recessive mutation resulting
in gross perturbations of cerebellar cytoarchitec-
ture that are confined to the anterior lobes of the
cerebellum. The sharp boundary between the
normal and affected area of the mea/mea cere-
bellum is reminiscent of the discrete boundaries
evident in many Drosophila developmental mu-
tants, suggesting that the mea gene may influ-
ence compartmental cellular organization in
mammalian brain. In this case we have also con-
structed a detailed genetic map surrounding the
mea gene on chromosome 4 and have begun ef-
forts to identify appropriate genomic sequences
to begin isolation of YAC clones containing the
mea locus. The identification of genes responsi-
ble for these and other mouse neurological muta-
tions should provide fundamentally important
insights into cerebellar structure and function.
During the past year we have continued to pur-
sue several different strategies to identify cDNA
clones that are cell specific and developmentally
regulated in the cerebellum. Using both subtrac-
tive hybridization and differential screening
methods, a large number of novel cDNA clones
have been isolated and are presently being
analyzed.
One particularly successful strategy has fo-
cused on the use of antisera to purified granule
neuron precursors to isolate genes that define
stages in granule cell differentiation (in collabo-
ration with Mary Beth Hatten, Columbia Univer-
sity) . A large number of novel granule cell cDNAs
have been isolated that define discrete stages of
granule cell neurogenesis. One biological insight
gained in these studies is the existence of a tran-
sient stage in development of granule cell neu-
rons that occurs just as these cells complete their
migratory journey from the external germinal
layer to the internal granule layer of the develop-
ing mouse cerebellum. Since this step in granule
cell differentiation has not been noted before, it
will be interesting to determine its function, its
relevance to neurogenesis in other areas of the
central nervous system, and the role of the genes
expressed uniquely at this stage in granule cell
development.
180
Structural Biology of CD4 and CDS Involvement
in the Cellular Immune Response
Wayne A. Hendrickson, Ph.D. — Investigator
Dr. Hendrickson is also Professor of Biochemistry and Molecular Biophysics at Columbia University
College of Physicians and Surgeons. He did his doctoral studies in biophysics at the Johns Hopkins
University and remained for a year of postdoctoral research with Warner Love before going to the Naval
Research Laboratory for continued postdoctoral study with Jerome Karle. He stayed on at NRL until
he joined the faculty of Columbia University.
THYMUS-derived lymphocytes, or T cells, dif-
ferentiate upon maturation into two major
cell types. These are principally distinguished by
the exclusive occurrence of either the CD4 or
CDS glycoproteins on their surfaces. The CD4-
bearing cells are known as helper T cells, and the
CD8-bearing cells as cytotoxic or killer T cells. T
cells of both types are stimulated into action
through the interplay of surface receptor mole-
cules with peptide antigens, which must be pre-
sented on target cell surfaces as complexes with
molecules of the major histocompatibility com-
plex (MHC). The helper T cells (CD4+ CDS")
can only interact with class II MHC molecules,
which occur on certain immune system cells
such as macrophages and B cells, whereas the
killer T cells (CD4~ CDS+) interact with the class
I MHC molecules, found on all cells.
The involvement of CD4 or CDS is essential for
the efficient stimulation of the respective re-
sponses. Stimulation of helper T cells by peptides
from an invading pathogen leads to the produc-
tion of cytokines, which in turn stimulate the pro-
liferation of selected lymphocytes for antibody
production or for cellular defense. Subsequent
stimulation of appropriate killer T cells by pep-
tide antigens from infected cells then leads to the
production of lytic factors that can eliminate the
infected cell. Thus the molecules CD4 or CDS
can properly be described as co-receptors in the
cellular immune response.
We have undertaken a series of structural stud-
ies on CD4 and CDS. This work is collaborative
with Richard Axel (HHMI, Columbia University)
and with Ray Sweet and his co-workers at Smith-
Kline Beecham. Results from these studies, when
taken in conjunction with mutational studies by
others, are beginning to provide a picture of the
molecular interactions implied by the specific
stimulatory processes of the immune response.
Moreover, the structural results also relate to the
involvement of CD4 as a receptor for infection by
the human immunodeficiency virus (HIV).
Crystals of Soluble CD4
Both CD4 and CDS are single-pass transmem-
brane proteins. These can be made tractable for
crystallographic study by the expression of solu-
ble extracellular fragments of the protein. A solu-
ble human CD4 fragment containing the entire
extracellular portion of the protein was ex-
pressed in mammalian cells, and Peter Kwong has
crystallized this protein into several different
crystal lattices. Unfortunately, none of these dif-
fracts very well. From a characterization of these
crystals, it is possible, nevertheless, to deduce
some interesting features of the CD4 molecule.
It appears that CD4 oligomerizes at the high
concentrations needed for crystallization, and the
lattice dimensions are consistent with a tetra-
meric model of 1 25 A in length along a diad axis
of the molecule. This propensity to associate may
be relevant to signal transduction during T cell
stimulation. Thus we are pursuing a crystallo-
graphic analysis of one of these crystals, despite
the prospect of resolution limited to approxi-
mately 4 A.
Structure of a CD4 Fragment
The consistently poor diffraction from several
polymorphs of a chemically homogeneous pro-
tein preparation suggested to us that CD4 may be
somewhat flexible. This possibility would be
compatible with the four-domain structure de-
duced from sequence comparisons and intron
positions, and limited proteolysis experiments
produced stable fragments corresponding to the
first two domains (D1D2) and the second two
domains (D3D4). A recombinant construct ex-
pressing the D 1 D2 domain also produced a stable
molecule, and Seong-Eon Ryu obtained diffrac-
tion-quality crystals of this fragment.
The resulting structure analysis revealed a Dl
domain folded as in the variable domains of im-
munoglobulins and a D2 domain in a variation of
the immunoglobulin constant-domain topology.
These two domains are intimately connected,
with the last strand of Dl running directly into
the first strand of D2. Despite the similarity of Dl
to immunoglobulin variable domains, there are
appreciable differences in the loop structures.
181
Structural Biology of CD4 and CDS Involvement in the Cellular
Immune Response
One of these is the major determinant for HIV
binding to CD4.
Structure of a CDS Fragment
Although the function of CDS is quite analogous
to that of CD4, the architecture of the two mole-
cules is different. CDS is a disulfide-bridged dimer,
whereas unactivated CD4 is monomeric, and each
^CDS chain has only one immunoglobulin-like do-
main rather than four. This single domain is Hnked
to the transmembrane segment by a 50-residue
stretch that is highly glycosylated.
Dan Leahy has expressed a soluble fragment of
CDS composed of the immunoglobulin-like do-
main and half of the stalk region. He has recently
solved a crystal structure of this molecule, and
this shows that the CDS dimer is organized very
much as in the variable domain tips of antibodies.
The stalk region from this molecule is disordered
in these crystals, a feature compatible with a role
as flexible tether to the membrane surface.
Interactions with MHC Molecules
Although a direct physical interaction between
isolated CD4 or CDS molecules and their respec-
tive MHC partners has not been demonstrated,
these interactions are expected to be very weak.
Otherwise, in light of the high polyvalency in
cell-cell interactions, unwanted adhesion would
be expected. Less-direct evidence for these inter-
actions, however, is available from suitably con-
structed cell biology experiments.
In the case of CD4, Rafick Sekaly and his col-
leagues at the Clinical Research Institute of Mon-
treal have used point mutations on human CD4 to
identify residues involved in the interaction with
class II molecules. These mutated residues map
to exposed residues on one face of the D1D2
structural model. Further analyses are still in
progress.
In the case of the CDS-class I interaction, we
have been able to model a plausible mode of asso-
ciation between the positively charged CDS mol-
ecule and the largely negative site localized to
the a3 domain of the class I MHC molecule. This
putative interaction is also consistent with results
from a mutational study by Paula Kavathis at Yale
University.
CD4 and AIDS
The hallmark of AIDS (acquired immune defi-
ciency syndrome) is the immunodeficiency
brought on by elimination of CD4"^ T cells. We
continue in our efforts to understand molecular
details of the interaction between CD4 and HIV
that are involved in viral entry. Seong-Eon Ryu is
completing his refinement of the D1D2 frag-
ment, Reza Beigi is analyzing the structures of
mutant proteins with abnormal HIV-binding
properties, and Peter Kwong has characterized an
antibody fragment that permits full HIV binding
to CD4 but inhibits infection. This antibody,
which binds to the D3 domain, indicates an in-
volvement of CD4 flexibility in HIV entry. Peter
Kwong is also attempting to crystallize the viral
coat protein gpl20 and various complexes with
this crucial component of the system.
182
Variegated Position Effects in Drosophila
Steven Henikoff, Ph.D. — Investigator
Dr. Henikoff is also a member of the Basic Sciences Division of the Fred Hutchinson Cancer Research
Center, Seattle. He received a B.S. degree in chemistry at the University of Chicago and a Ph.D. degree
in biochemistry and molecular biology at Harvard University, working in the laboratory of Matthew
Meselson. He did postdoctoral work with Charles Laird at the University of Washington.
EACH individual gene occupies a fixed posi-
tion on a chromosome. By and large, moving
a gene has only a minor effect on expression of
the gene. Thus most studies of gene expression
are able to focus on the gene as an independent
unit, without taking into account larger organiza-
tional features. However, there are exceptional
cases in which the relationship between a gene
and its environment plays a role in expression of
the gene.
The relationship between a gene and its chro-
mosomal environment is especially apparent in
examples of "position effects" associated with
chromosomal rearrangements. In flies a well-
known class of position effects involves inactiva-
tion of genes in the vicinity of rearrangement
breakpoints. Gene inactivation is extremely vari-
able from cell to cell, such that the affected tissue
shows a variegated pattern of expression. In each
case, it is found that the gene has been juxtaposed
to heterochromatin, the deeply staining regions of
chromosomes that flank the centromere. Although
heterochromatin contains a substantial fraction of
DNA in all higher eukaryotes, the repetitive se-
quence structure characteristic of heterochromatin
and the near absence of genes have hampered at-
tempts to understand its role in the genome. Genes
that show variegated expression when placed next
to heterochromatin provide a reporter function,
allowing us to investigate these poorly understood
regions of chromosomes.
Variegated position effects caused by juxtapo-
sition to heterochromatin are seen for a large
number of genes in Drosophila. One well-stud-
ied example is the brown gene, required for full
pigmentation of the eye. Unlike nearly all other
genes, however, such position effects on the
brown gene are dominant over wild type — that
is, placing one copy of brown next to heterochro-
matin can lead to inactivation of the other copy.
We have investigated the genetic basis for this
gene inactivation in trans and have found that a
necessary component is the pairing of homo-
logues in the immediate vicinity of the brown
gene. These findings have led to an explanation
for "trans-inactivation," whereby protein compo-
nents of heterochromatin make direct contact with
the trans copy of the brown gene across paired ho-
mologues. In support of this hypothesis, we have
been able to reproduce trans-inactivation at sites of
transposons carrying tht brown gene, but only for
paired copies of the gene. In addition, we have
found that even very small lesions that disrupt pair-
ing in the immediate vicinity of the gene also re-
duce trans-inactivation.
Mapping of the sequence-specific component
necessary for trans-inactivation to occur has local-
ized it to the brown gene itself, probably to the
region immediately upstream. This supports the
notion that the contact between homologues is
between a DNA-binding protein necessary for
normal brown gene activity and a protein compo-
nent of heterochromatin.
Our current efforts are aimed at identification
of the protein components involved in trans-
inactivation. One approach is to focus on the se-
quences in the immediate upstream region of the
brown gene, where we expect that the sequence-
specific component should bind. Another is to
identify genes that encode proteins involved in
the process by screening for mutations that specif-
ically reduce the degree of trans-inactivation. We
have now isolated several such mutations and are
in the process of precisely mapping them to
clone the corresponding genes.
A new research direction for us came from a
serendipitous finding (during a screen for posi-
tion-effect variegation mutations) of an unstable
chromosome that causes gene markers carried on
it to appear variegated. This chromosome derives
from a fusion between a chromosome arm carry-
ing the markers and a centromere from another
chromosome. The unstable chromosome is inter-
mediate in size among wild-type and rearranged
linear Drosophila chromosomes, all of which are
quite stable in somatic cells. The instability re-
sults from failure of the two products of replica-
tion— called sister chromatids — to come apart
reliably at mitosis, leading to clones and single
cells that have either gained an extra copy of the
chromosome or have lost it entirely.
The appearance of the unstable chromosome
183
Variegated Position Effects in Drosophila
during mitosis indicates that instability is asso-
ciated with premature separation of sister chro-
matids; as a result, the sisters might sometimes
independently attach to the spindle apparatus
that pulls them apart. Current evidence suggests
that the defect results from a position effect on
the centromere, because genetic suppressors of
position-effect variegation also suppress the insta-
bility phenotype. To understand further the basis
for instability, we have initiated physical map-
ping studies on this centromere. Such studies
should help identify sequences necessary for
centromere function in a higher eukaryote.
Phenomena that depend on the position of a
sequence in the chromosome or on somatic pair-
ing of homologues are easily observed in Dro-
sophila, where powerful tools are available for
genetic dissection. Related phenomena are
known to occur in mammals, such as X chromo-
some inactivation, in which one of the female's X
chromosomes becomes heterochromatic. The
many similarities between chromosomes in or-
ganisms as diverse as flies and mammals lead to
the expectation that an understanding of position
effects and centromere function in Drosophila
will have general implications.
184
Biological Roles and Expression
of Complement Receptors
V. Michael Holers, M.D. — Assistant Investigator
Dr. Holers is also Assistant Professor of Medicine and Pathology at the Washington University School
of Medicine and Assistant Physician at Barnes Hospital, St. Louis. He received his undergraduate degree
from Purdue University and his M.D. degree from Washington University. He did postdoctoral research
at the University of Colorado, Denver, and then at Washington University.
THE complement system was initially de-
scribed as an activity found in serum that
could mediate the killing of foreign infectious
organisms such as bacteria or viruses. This sys-
tem, in conjunction with specific immunity gen-
erated by vaccination, was found to be critical to
preventing and fighting infections. Later it was
realized that complement also facilitates the in-
teraction of antigen-antibody complexes with
cells of the immune system, which greatly en-
hances the specific immune response. Therefore
complement not only helps to initiate an immune
response but also plays an important role in the
ability to clear infections and foreign antigens
from the body.
The complement system consists of at least 20
serum proteins that are activated in a cascade fash-
ion. As part of the activation process, protein
fragments are released that attract inflammatory
cells. In addition, antigen-antibody complexes
are coated with specific complement fragments
that covalently attach to this target. One of these
fragments, complement component C3, is able to
be cleaved proteolytically after attachment to a
target. This cleavage reaction results in a number
of different conformations of C3, which allow it
to interact with at least three unique cell surface
receptors. After binding to cells, specific trans-
membrane signals are sent, and the complement
receptors then mediate ingestion and processing
or killing of the targets. Also as part of this pro-
cess, C3 fragments may bind to self tissues, rather
than to the antibody-bound target, thereby attack-
ing at inappropriate sites. Other cell membrane
C3-binding proteins are able to inactivate this C3
and prevent inappropriate damage to self tissues.
We are interested in the interaction of C3 with
its specific receptors and regulatory proteins, par-
ticularly the biological aspects of complement
receptor 2 (CR2). In addition, we are studying
mouse homologues of CR2 and other proteins of
this type. Human CR2 serves as the receptor for
the Epstein-Barr virus (EBV), which is responsi-
ble for most cases of infectious mononucleosis
and is causally associated with a number of hu-
man tumors of B lymphocytes and epithelial
cells. Patients who have forms of congenital or
acquired immunodeficiency (such as AIDS or
after organ transplantation) are particularly sus-
ceptible to tumors associated with EBV.
In the past few years we have cloned and ana-
lyzed the structure and activities of human CR2,
its mouse homologues, and a unique mouse pro-
tein called Crry/p65. We have shown that ex-
pression of recombinant forms of these proteins
in other cells is sufficient to mediate the binding
of ligands or to control inappropriate comple-
ment deposition. By using other recombinant
techniques and creating mutations within these
proteins, we have shown that specific amino
acids in small domains are important for ligand
interactions. In addition, we have synthesized
peptides that have the ability to block binding of
some ligands, in particular, EBV binding to CR2.
These studies should allow us to devise strategies
to alter the function of this receptor in vivo. For
instance, one type of reagent might block EBV
binding to CR2 but not normal binding of C3.
This could be useful in some illnesses associated
with EBV.
We have analyzed the murine homologues of
these proteins to understand further the biologi-
cal role of human CR2, in addition to other com-
plement receptors and regulatory proteins. Once
the activities of these proteins are understood, we
should be able to utilize murine models of the
normal immune response, as well as autoimmune
diseases, to elucidate the in vivo roles of these
proteins.
Another aspect of CR2 expression is also under
analysis. Expression of CR2 varies during human
B lymphocyte development: it is expressed only
on late pre-B cells and mature B cells and not on
very early pre-B lymphocytes or on late immuno-
globulin-secreting cells. The molecular mecha-
nisms that underlie this phenotype, which is also
found among other B cell-specific markers, are
likely fundamental to the overall processes by
which B cells mature and are activated. We are
analyzing these mechanisms. As part of these stud-
ies we have defined a promoter for CR2 and other
sites within the gene that are likely to be impor-
185
Biological Roles and Expression of Complement Receptors
tant in gene regulation. We have determined that
the receptor levels go up and down because the
gene is turned on and off during development,
we have identified functionally important do-
mains and sites of protein interaction within the
CR2 promoter, and we are clarifying the role of
each site in CR2 expression and B cell matura-
tion. These studies will increase our understand-
ing of specific gene expression in B lymphocytes
and further our knowledge of how to alter B cell
phenotypes along pathways that might be more
beneficial during certain disease states.
186
Genetic Control of Nematode Development
H. Robert Horvttz, Ph.D. — Investigator
Dr. Horvitz is also Professor of Biology at the Massachusetts Institute of Technology and Neurobiologist
and Geneticist at Massachusetts General Hospital, Boston. He earned his undergraduate degrees in
mathematics and in economics at the Massachusetts Institute of Technology, followed by the M.A. and
Ph.D. degrees in biology from Harvard University. His postdoctoral research was done at the Medical
Research laboratory of Molecular Biology, Cambridge, England. Dr. Horvitz is a member of the National
Academy of Sciences.
HOW do genes control animal development?
Taking a primarily genetic approach to an-
swer this question, members of our laboratory
have isolated developmental mutants of the
roundworm Caenorhabditis elegans and have
used both genetic and molecular genetic tech-
niques to characterize these mutants. Because the
complete cellular anatomy (including the com-
plete wiring diagram of the nervous system) and
the complete cell lineage of C. elegans are
known, mutant animals can be studied at the
level of single cells and even single synapses.
Genes that play specific roles in cell lineage, cell
signaling, cell death, and cell migration have
been identified and analyzed.
Cell Lineage
The problem of cell lineage — how a single fer-
tilized egg cell undergoes a complex pattern of
cell divisions to generate a multiplicity of dis-
tinct cell types — is one major focus of the re-
search of our laboratory. We have identified
hundreds of genes responsible for controlling
aspects of the C. elegans cell lineage. Many of
these genes function in generating cell diversity
during development. For example, some genes
act to make the two daughter cells generated by a
single cell division different from each other, and
one gene acts to make certain daughter cells dif-
ferent from their mothers. The action of some
cell lineage genes is constrained to a single cell
type, tissue, or organ. For example, one gene acts
only in the nervous system, and another acts only
in the hypodermis. Other genes act in multiple
tissues.
We have analyzed a number of these genes at
the molecular level. These studies have revealed
that many genes that control cell lineage in C.
elegans are strikingly similar to genes found in
other organisms, including humans. Thus the
analysis of developmental control genes in C. ele-
gans should help us to understand aspects of the
development of more-complex organisms.
Cell Signaling
Much of the development of C. elegans, like
that of other organisms, involves intercellular
communication. We have studied cell interac-
tions in nematode development by using a laser
microbeam to kill single cells in living animals: if
destruction of one cell alters the fate of a second
cell, the first cell must normally interact with the
second. We have analyzed in detail the cell inter-
actions involved in inducing the development of
the vulva, which forms the external genitalia,
connects the uterus with the outside environ-
ment, and is used for egg laying and copulation.
We have characterized many genes that function
in the cell interactions of vulval development.
One gene that acts as a switch in the vulval induc-
tive signaling pathway is a member of the ras
gene family. Other ras genes are associated with
many human cancers; the same mutations that
cause extra vulval cell divisions in C. elegans are
oncogenic in mammals. Another gene in the vul-
val signaling pathway has similarities to the src
gene, which is also associated with human
cancers. The study of these and other genes that
function in cell signaling in C. elegans might
provide insights relevant to cancerous growth in
humans.
Cell Death
Naturally occurring or "programmed" cell
death is common during the development of the
nervous system of many animals, including C.
elegans. Why organisms generate cells only to
have them die is an intriguing question. Further-
more, the mechanisms responsible for cell death
might be of medical importance, as the clinical
features of many human disorders (including
trauma, stroke, and a variety of neurodegenera-
tive diseases) are a consequence of nerve cell
deaths.
We have been identifying and characterizing
genes that function in programmed cell death in
C. elegans. Two genes cause cells to die, and
seven other genes are involved in removing the
corpses of dead cells. The two genes that cause
cells to die must be expressed by the dying cells
themselves, indicating that, at least to this extent,
programmed cell deaths are cell suicides. Al-
187
Genetic Control of Nematode Development
though many cells die during the course of
C. elegans development, most cells survive; cell
survival requires the inactivation of the cell death
process, as a tenth gene functions to prevent the
action of the nine cell death genes in surviving
cells. In addition, this regulatory gene is itself
controlled in a cell-specific fashion by other
genes that determine which cells are to live and
which are to die. Molecular analyses of genes in-
volved in programmed cell deaths in C. elegans
have suggested that the biochemical processes re-
sponsible might be similar to those suspected to
cause nerve cells to die in human neurological
disorders. We hope that knowledge of what
makes cells die and of what can block the cell
death process in C. elegans will lead to methods
that will prevent the cell deaths responsible for
human disorders.
Cell Migration
During animal development, cells are often
generated far from their final positions and must
migrate considerable distances before being able
to function. To understand what causes cells both
to migrate and to stop migrating, we are analyzing
two C. elegans cell migrations. The first involves
a pair of muscle precursor cells that are born in
the posterior body region and move to a central
position along the animal's length, near its gonad.
We have discovered that these migrations involve
signaling between the migrating cells and go-
nadal cells located at the termination site of the
migration. We are characterizing genes that func-
tion in this signaling process.
The second migration we have studied involves
a pair of neuronal cells that move from the tail
region to the midbody region of the animal. Thir-
teen genes have been identified that must func-
tion for these neuronal migrations to occur prop-
erly. Some of these genes probably act in the
migration process per se, but some do not. The
actions of some of these latter genes allow these
neurons to acquire their identities; if these genes
do not function, these neurons fail to express
their normal characteristics, including their
long-range cell migration.
188
Protein Folding in Vivo
Arthur L. Harwich, M.D. — Associate Investigator
Dr. Norwich is also Associate Professor of Human Genetics and Pediatrics at Yale University School of
Medicine. He received A.B. and M.D. degrees in biomedical sciences from Brown University. His internship
and residency training in pediatrics were done at Yale. His postdoctoral research training was at the Salk
Institute with Walter Eckhart and at Yale University with Leon Rosenberg.
UNTIL recently it has been assumed that
newly made proteins in the living cell, com-
prising amino acid chains with characteristic se-
quences, are able to fold spontaneously into pre-
cise three-dimensional structures that exhibit
biological activity. Such folding has been ob-
served in test-tube experiments, where many
proteins unfolded by denaturing agents can be
diluted from these agents and observed to refold
into their biologically active forms. Recent stud-
ies, however, suggest that in the cell the process
of folding is assisted by other specialized pro-
teins. We originally identified such a protein
while studying mitochondria, the intracellular
organelles that carry out energy metabolism.
Most of the proteins of mitochondria are first
made outside the organelles, in the cytosol, and
then imported through two membranes to reach
the innermost mitochondrial "matrix" compart-
ment. To traverse the membranes, the newly
made proteins are first unfolded on the cytosolic
side. After import, the proteins refold on the in-
side of the organelles into their biologically ac-
tive conformations. We identified a mutant cell
in which mitochondrial proteins were imported
but failed to fold into biologically active forms.
The mutation was found to affect a protein that
normally resides in the matrix compartment,
called heat-shock protein 60 (hsp60).
This protein was originally identified by the
observation that its abundance is increased about
twofold in response to incubation of cells at high
temperatures. However, it is an abundant protein
even before heat shock, and our genetic analysis
demonstrated that, consistent with a critical base-
line function, hsp60 is required not only at high
temperatures but at all temperatures. The in-
creased level produced in response to heat stress
could represent an effort to try to refold effi-
ciently mitochondrial proteins that heat has
denatured.
In the mitochondrial matrix, hsp60 is found in
a higher order structure, a complex. Fourteen
copies of the protein are arranged in two stacked
rings, a "double donut . ' ' Each ring contains seven
radially arranged copies of hsp60. Our studies
have demonstrated that unfolded mitochondrial
proteins entering the matrix space become asso-
ciated with the surface of the hsp60 complex.
Then, in steps requiring both energy (ATP) and a
second protein, the polypeptides are folded into
their active forms and released from the
complex.
The folding pathway must be dictated by the
amino acid sequence of the "substrate" protein
to be folded, not by the hsp60 complex, because
we have used the complex to fold proteins that
normally reside outside mitochondria. It seems,
however, that hsp60 acts by speeding up, or cata-
lyzing, the folding of proteins. How does it do
this? One possibility is that it simply prevents do-
mains of proteins from wrongfully interacting, ei-
ther with each other or with nearby proteins in
the mitochondrial matrix, a "chaperone" func-
tion. Another possibility is that the complex ac-
tively promotes the progression of an unfolded
protein through a series of folding steps.
Because mitochondria arose from bacteria (one
cell ingested another), it is not surprising that a
structurally related component, the groEL pro-
tein, has been found in bacteria. In the 1970s it
was observed that when Escherichia coli cells
partially defective in this protein are infected
with a virus, the newly made viral coat proteins
are unable to assemble to make new virus
particles.
Additional evidence comes from our reconsti-
tution experiments. When a test protein unfolded
in denaturant was diluted into a mixture contain-
ing purified groEL complex, it became bound to
the complex and was thus prevented from mis-
folding and aggregating. The bound protein was
found as a folding intermediate, called a molten
globule, that has formed its local structures but
lacks the three-dimensional organization of the
active form. When both a cooperating compo-
nent, groES (a small ring structure also with seven
members), and ATP were added, the polypeptide
was observed to reorganize its structure while in
association with the complex during the next
minutes and to be released in its active form. The
cost of folding a single polypeptide was approxi-
mately 100 ATP hydrolyzed, amounting to about
189
Protein Folding in Vivo
one-tenth of the amount of energy needed to syn-
thesize the polypeptide.
How general is the utilization of folding ma-
chinery outside bacteria and mitochondria? We
thought we might find a functionally similar com-
ponent in the cytosol of higher organisms by look-
ing first in the kingdom of archaebacteria. These
organisms are evolutionarily distinct from bacte-
ria and have recently been shown to contain com-
ponents closely related to those in the cytosol of
higher organisms. We transferred a thermophilic
archaebacterium that grows normally at 75 °C to a
near-lethal temperature of 88°C and found that a
single major heat-shock protein was produced.
This protein , already abundant at normal tempera-
ture, is a double-ring complex that looks much
like the complexes of mitochondria and bacteria,
except that each ring contains nine members.
The purified ring complex binds unfolded pro-
teins and exhibits ATPase activity, consistent with
the idea that it might function in the archaebac-
teria similarly to the hsp60 and groEL complexes.
Analysis of the protein's amino acid sequence
demonstrated no significant relationship to ei-
ther hsp60 or groEL, but rather a relationship to a
protein called TCPl , an essential cytosol protein
of higher organisms, whose function has been
largely unknown but which has been implicated
in assembly of the mitotic spindle apparatus. Us-
ing a mutant yeast cell defective in TCPl , we are
now investigating whether the protein has a spe-
cialized function in assembly of the spindle or a
more general function as a "folding machine" for
the cytosol of higher organisms.
MgATP
groEL
Model for folding of polypeptides by groEL. Polypeptide, shown in yellow, is bound by groEL and
stabilized in "molten globule" conformation. When MgATP and groES are present, polypeptide
undergoes conformational changes and is ultimately released, reaching a native, active
conformation.
Derived from Martin, f., Langer, T., Boteva, R., Scbramel, A., Horwich, A.L., and Hartl, F.-U.
1991. Nature 352:36-42.
190
Molecular Mechanisms in the Regulation
of Synaptic Transmission
Richard L. Huganir, Ph.D. — Associate Investigator
Dr. Huganir is also Associate Professor of Neuroscience at the Johns Hopkins University School of
Medicine. He completed his undergraduate work in biochemistry at Vassar College and received his Ph.D.
degree in biochemistry and molecular and cell biology from Cornell University, where he performed his
thesis research in the laboratory of Efraim Packer. After completing a postdoctoral fellowship with Paul
Greengard at Yale University School of Medicine, Dr. Huganir moved to the Rockefeller University.
INFORMATION processing in the brain de-
pends on the transmission of signals between
neurons at specialized areas of contact, called
synapses. At synapses, ion channel proteins in the
neuronal cell membrane generate an electrical
current, which triggers the release of chemical
signals from the first neuron, called the presynap-
tic neuron. These chemical signals, or neurotrans-
mitters, bind to specific receptor proteins in the
membrane of the second neuron, called the post-
synaptic neuron. The neurotransmitter receptors
then generate electrical currents in the postsyn-
aptic neuron and thus complete the process of
synaptic transmission.
The efficiency of synaptic transmission at any
given synapse is constantly changing in response
to a variety of factors; this synaptic plasticity
plays a major role in the function of the nervous
system. Both the amount of neurotransmitter re-
leased from the presynaptic neuron in response
to a given electrical signal and the sensitivity of
the postsynaptic receptor system for a given
amount of neurotransmitter can be modulated.
The molecular mechanisms that underlie the
modulation of synaptic transmission have only
begun to be defined. Recent studies have pro-
vided evidence that protein phosphorylation is
an important mechanism in the regulation of syn-
aptic transmission.
Protein phosphorylation systems consist of
three primary components, a protein kinase, a
substrate protein, and a phosphoprotein phos-
phatase. Protein kinases are enzymes that catalyze
the chemical transfer of phosphate molecules
from ATP to specific substrate proteins. The activ-
ities of many protein kinases are regulated by neu-
rotransmitters and hormones through the actions
of substances called second messengers, such as
cAMP, calcium, and diacylglycerol. Substrate
proteins include many cellular components,
among them enzymes, ion channels, and neuro-
transmitter receptors. The addition of the nega-
tively charged phosphate group alters the struc-
ture of these substrate proteins, thereby
regulating their functional properties. Phospho-
protein phosphatases are enzymes that reverse
the process of protein phosphorylation, remove
the phosphate group from the substrate protein,
and return it to its basal state.
My laboratory is concerned with the structure
and function of neurotransmitter receptors and
the role of protein phosphorylation in the regula-
tion of the properties of the neurotransmitter re-
ceptors. We have used as a model system the best-
characterized neurotransmitter receptor and ion
channel in neurobiology today, the nicotinic ace-
tylcholine receptor. In addition, we have been
studying the major inhibitory neurotransmitter
receptors in the brain, the GABA^ receptors, and
the major excitatory neurotransmitter receptors
in the brain, the glutamate receptors. These re-
ceptors are neurotransmitter-dependent ion
channels that generate electrical currents in the
postsynaptic membrane of the synapse in re-
sponse to their neurotransmitter.
To study the molecular mechanisms involved
in neurotransmitter receptor and ion channel
function, it is essential to identify chemically the
specific proteins required for this activity. We
began by defining the molecular components re-
quired for the functioning of the nicotinic acetyl-
choline receptor ion channel. Using membrane
reconstitution techniques, we solubilized the nic-
otinic receptor and its ion channel from isolated
postsynaptic membranes, purified it, and recon-
stituted it into phospholipid vesicles. These stud-
ies demonstrated that the purified receptor, con-
sisting of four types of protein subunits (a, /?, 7,
5) , contains the ion channel and has all the biolog-
ical properties of the nicotinic receptor in the
intact cell.
We next began to characterize the protein
phosphorylation of these structural components.
Postsynaptic membranes isolated from synapses
highly enriched in the nicotinic acetylcholine re-
ceptor contain at least three different types of
protein kinases that phosphorylate the nicotinic
receptor on six different phosphorylation sites:
cAMP-dependent protein kinase phosphorylates
the 7- and 6-subunits of the receptor; a calcium-
and diacylglycerol-dependent protein kinase
phosphorylates the 5-subunit; and a protein-
191
Molecular Mechanisms in the Regulation of Synaptic Transmission
tyrosine kinase phosphorylates the |8-, 7-, and
5-subunits. These postsynaptic membranes also
contain phosphoprotein phosphatase activity
that dephosphorylates the phosphorylated nico-
tinic acetylcholine receptor.
We are currently using protein purification and
molecular cloning techniques to characterize the
protein-tyrosine kinases that phosphorylate the
receptor and the phosphotyrosine protein phos-
phatases that dephosphorylate the receptor. We
recently identified several cDNA clones for dif-
ferent types of protein-tyrosine kinases that are
expressed in cells enriched in the nicotinic re-
ceptor and are attempting to determine which of
these protein-tyrosine kinases phosphorylate the
receptor. In addition, we recently purified the
phosphotyrosine protein phosphatase that de-
phosphorylates the tyrosine-phosphorylated ace-
tylcholine receptor and are using molecular
cloning techniques to isolate cDNA clones for
this phosphotyrosine protein phosphatase.
What are the functional effects of phosphoryla-
tion of the receptor by these protein kinases? We
have examined this question directly by studying
the properties of the purified and reconstituted
receptor phosphorylated to different degrees by
the various protein kinases. Phosphorylation of
the receptor on the 7- and 6-subunits by cAMP-
dependent protein kinase or phosphorylation of
the receptor on the 0-, 7-, and 5-subunits by the
protein-tyrosine kinase dramatically increases the
rate of desensitization of the receptor. Desensiti-
zation is the process by which the receptor is re-
versibly inactivated in the continued presence of
the neurotransmitter acetylcholine. These stud-
ies provide direct evidence that protein phos-
phorylation of the nicotinic acetylcholine recep-
tor regulates its physiological properties and
plays a role in modulating its sensitivity to
acetylcholine.
To analyze the effect of phosphorylation on the
desensitization of the receptor in more detail, we
have recently used site-specific mutagenesis tech-
niques to mutate the phosphorylation sites on the
receptor subunits. Mutant receptor subunits lack-
ing phosphorylation sites have been expressed in
Xenopus oocytes, and the regulation of desensiti-
zation of these receptors by protein phosphoryla-
tion is being analyzed and compared with normal
receptors.
Using muscle cell cultures that are highly
enriched in the acetylcholine receptor, we have
investigated the regulation of the phosphoryla-
tion of the receptor by neurotransmitters, hor-
mones, and neuropeptides. Calcitonin gene-
related peptide (CGRP), a neuropeptide that is
released from the presynaptic neuron with ace-
tylcholine, increases the intracellular levels of
cAMP and thereby regulates the phosphorylation
of the receptor by the cAMP-dependent protein
kinase. In addition, studies in our laboratory sug-
gest that acetylcholine itself regulates intracellu-
lar levels of calcium and thereby regulates the
phosphorylation of its own receptor by the cal-
cium- and diacylglycerol-dependent protein ki-
nase. We have also demonstrated that tyrosine
phosphorylation of the nicotinic receptor is regu-
lated by the neurons that synapse on muscle.
More recently, in collaboration with Bruce Wal-
lace (University of Colorado Health Sciences
Center), we have found that agrin, an extracellu-
lar matrix protein, may be the factor from neu-
rons that regulates tyrosine phosphorylation of
the receptor. Agrin is a well-characterized pro-
tein that is secreted from neurons and induces
receptor clustering under the nerve during syn-
apse formation. These results suggest that agrin-
induced tyrosine phosphorylation of the receptor
may be involved in the induction of clustering of
the receptor at the synapse.
Our recent studies on GABA^ and glutamate re-
ceptors have supported our hypothesis that pro-
tein phosphorylation of neurotransmitter recep-
tors plays an important role in the modulation of
their function. We have expressed the genes for
these receptors in a variety of cells and have ana-
lyzed the effect of phosphorylation on their func-
tional properties. Using this system, we have
shown that phosphorylation of the /3-subunit by
cAMP-dependent protein kinase decreases the re-
sponse of the GABA^ receptor to its neurotransmit-
ter and alters the desensitization of the receptor.
In addition, recent studies have demonstrated
that glutamate receptors are phosphorylated by a
protein-tyrosine kinase. The functional effect of
tyrosine phosphorylation of the glutamate recep-
tor is currently being examined. This work, com-
bined with our studies of the nicotinic acetylcho-
line receptor, provides strong evidence that
protein phosphorylation of neurotransmitter re-
ceptors is a primary mechanism in the regulation
of synaptic transmission.
192
Molecular Aspects of Signal Transduction
in the Visual System
James B. Hurley, Ph.D. — Associate Investigator
Dr. Hurley is also Associate Professor of Biochemistry at the University of Washington School of Medicine.
He received his undergraduate degree in chemistry from the State University of New York College of
Environmental Science and Forestry, Syracuse, and his Ph.D. degree in physiology and biophysics from
the University of Illinois, Urbana, where he worked with Thomas Ebrey. His postdoctoral research included
studies with Melvin Simon at both the University of California, San Diego, and the California Institute of
Technology, and with Lubert Stryer at Stanford University.
OUR laboratory studies molecular mecha-
nisms responsible for visual transduction in
vertebrate and invertebrate photoreceptors. De-
spite the fact that these two types of cells respond
to light via quite diverse mechanisms, they have
many general features in common. We are inves-
tigating mechanisms that determine such photo-
receptor characteristics as sensitivity, rates of ac-
tivation and deactivation, and ability to adapt to
constant light.
Light hyperpolarizes vertebrate photorecep-
tors via a G protein-mediated cascade that culmi-
nates in cyclic GMP hydrolysis. Depletion of
cGMP reduces the activity of cGMP-gated cation
channels in the photoreceptor plasma mem-
brane. In darkness Ca^"^ enters the cell through
these channels. Light blocks this entry, and the
resulting depletion of cytosolic Ca^"^ promotes re-
covery by stimulating guanylate cyclase to re-
synthesize cGMP.
Invertebrate photoreceptors respond to light
very differently. In these cells light activates
phospholipase C, which produces inositol tri-
phosphate and diacylglycerol as second messen-
gers. Few biochemical details of invertebrate
phototransduction are well understood.
Vertebrate Phototransduction
Our laboratory recently identified a novel
Ca^"^-binding protein that imparts Ca^"^ sensitivity
to photoreceptor guanylate cyclase. This protein,
named recoverin, promotes recovery by stimulat-
ing guanylate cyclase when free Ca^"^ concentra-
tions fall below 300 nM. The amino acid se-
quence of recoverin reveals three Ca^"^-binding
sites. Ca^"^ influences a variety of physical proper-
ties of recoverin, including fluorescence and mo-
bility on electrophoresis gels. We cloned re-
coverin cDNA and expressed recombinant
recoverin in Escherichia coli.
The effects of Ca^"^ on recombinant recoverin
and retinal recoverin are quite different. To ac-
count for these differences, we compared the
masses of recombinant and retinal recoverin di-
rectly by ion-spray mass spectrometry. To our
surprise, the modification turned out to be a
novel type of heterogeneous amino-terminal acy-
lation. Each recoverin is acylated with either a
CI4:0, C14:1, C14:2, or C12:0 fatty acid
residue.
Following stimulation by light, transducin, the
photoreceptor G protein, hydro lyzes its bound
GTP and loses its ability to activate phosphodies-
terase. Photoreceptor cells recover from a light
flash within a couple of seconds, but the steady-
state hydrolysis of GTP is slower. To clarify the
role that GTP hydrolysis plays in recovery from a
photoresponse, we produced transgenic mice
that express a mutant transducin that hydrolyzes
GTP more slowly than their normal counterparts.
Preliminary results suggest that photoreceptors
expressing this form of transducin are abnormally
desensitized. This efi'ect may reflect an attempt
by the cells to compensate for the persistent
phosphodiesterase activity of the mutant
transducin.
Drosophila Vision
Biochemical and physiological evidence sug-
gests that a G protein mediates invertebrate pho-
totransduction by stimulating phospholipase C.
We characterized several Drosophila G proteins
with the aim of understanding their role in inver-
tebrate phototransduction. In addition to several
G protein a-subunits, two G protein /3-subunits
were identified. The first one was detected
throughout the nervous system but not in the
eyes. This prompted us to search for a photore-
ceptor G protein /3-subunit.
Through use of a specific type of cDNA screen-
ing method, we were able to identify a novel type
of G protein /S-subunit that is expressed specifi-
cally in the Drosophila compound eye. Recently,
in collaboration with Charles Zuker and his col-
leagues (HHMI, University of California, San
Diego) , two mutant Drosophila strains have been
identified with reduced expression of this eye-
specific |S-subunit. Biochemical analyses of eyes
from normal Drosophila and from these mutants
are being used to study the role of G proteins in
phototransduction .
193
Molecular Aspects of Signal Transduction in the Visual System
A toxin from the microorganism Bordetella
pertussis specifically inactivates certain G pro-
teins. To study the physiological importance of G
proteins in Drosophila, we produced transgenic
flies that conditionally express endogenous per-
tussis toxin. Pertussis toxin induction in adult
Drosophila alters their visual response and eat-
ing behavior. We are investigating the molecular
mechanisms responsible for these phenotypes.
The long-term objective addressed in these
projects is to identify biochemical mechanisms
by which photoreceptors respond to stimuli.
194
The Molecular Basis of Cell Adhesion in Normal
and Pathological Situations
Richard O. Hynes, Ph.D. — Investigator
Dr. Hynes is also Professor of Biology and Director of the Center for Cancer Research at the Massachusetts
Institute of Technology. He received his undergraduate degree in biochemistry from the University of
Cambridge and his Ph.D. degree in biology from the Massachusetts Institute of Technology. After several
years of postdoctoral work at the Imperial Cancer Research Fund laboratories in London, where he
initiated his early work on fibronectins, he returned to MIT as a faculty member. Dr. Hynes has been the
recipient of a Guggenheim Fellowship and is a Fellow of the Royal Society of London and the American
Association for the Advancement of Science.
MOST cells in the body adhere to their neigh-
bors and to the extracellular matrix, a com-
plex array of proteins that comprise a fibrillar
meshwork throughout the body. Cell adhesion
plays important roles in the normal functions of
cells, contributing to cellular organization and
structure, proliferation, and metabolism. During
embryological development, cell adhesion is im-
portant for the movements of cells that contrib-
ute to modeling of the embryo. In the adult, ap-
propriate cell adhesion is necessary for numerous
physiological processes.
For example, in the blood, cells known as plate-
lets adhere to the walls of blood vessels that are
damaged and help to prevent bleeding. This ad-
hesion process is essential to protect against hem-
orrhage. On the other hand, it is equally impor-
tant that platelets should not adhere at
inappropriate times. If they do, the result is
thrombosis. Thus the control of platelet adhesion
is a matter of life and death. Other blood cells
involved in defense mechanisms during infection
or inflammation need to adhere to the walls of
blood vessels at the sites of infection to emigrate
into the affected tissues.
Another process involving cell adhesion and
migration is wound healing. When skin is dam-
aged, the skin cells migrate in over the wound to
cover it. The processes of cell migration involved
in wound healing have much in common with
those occurring during development.
A final example is that of cancer. Tumor cells
exhibit altered adhesion both to one another and
to their surroundings. This altered adhesion is
thought to be involved in the invasion and metas-
tasis of tumor cells.
These examples illustrate the importance of
appropriate adhesion of cells to their surround-
ings. Our laboratory is involved in molecular anal-
yses of these processes. We seek to understand
the proteins involved in cell adhesion and how
they control adhesion and migration of cells in
both normal and pathological processes.
Two main classes of proteins interest us. The
first comprises the large proteins that make up
the extracellular matrix. These proteins cooper-
ate to build a fibrillar meshwork to which the
cells attach and on and through which they mi-
grate. We have investigated several of these pro-
teins, which we refer to as "nectins" to denote
their role in binding to cells. Fibronectins, a
closely related group of proteins all encoded by a
single gene, are the best understood of these nec-
tins. The different forms of fibronectin are gener-
ated by alternative RNA splicing. We and others
have analyzed in detail the functions and the
structure of these proteins. This work is leading
to a deeper understanding of their roles in cell
behavior. For example, it is now known that fibro-
nectins have several sites in each molecule that
bind cells. The detailed structure of these bind-
ing sites is being elucidated. One intriguing ob-
servation is that fibronectins share with many
other nectins a common recognition site made up
of only three amino acids. This site (designated
RGD in the single-letter amino acid code) is rec-
ognized by receptor molecules on cell surfaces.
This interaction can be blocked by antibodies to
the nectins or to the receptors, which are known
as integrins, or by competitor peptides contain-
ing the RGD sequence. Such blockades interfere
with the cell-adhesive interactions involved in
the physiological processes discussed above. Re-
cent work has identified other cell-binding sites
within fibronectins, which are recognized by dif-
ferent receptors. Cells interact with these mole-
cules in a complex fashion, which is as expected,
given the participation of cell adhesion in many
diverse cellular functions.
Our second major focus of interest is the family
of integrin receptors. These comprise a family of
related cell surface receptors, each composed of
two subunits. Each integrin receptor has a partic-
ular specificity for certain nectins and mediates
the interactions of cells with the extracellular
matrix. In addition, the integrins connect to the
inside of the cell, where they mediate interac-
tions with the internal structures or cytoskeleton
of the cell that are involved in the shape, organiza-
tion, and migration of cells. This integration of
the organization of the extracellular matrix with
195
The Molecular Basis of Cell Adhesion in Normal
and Pathological Situations
the cytoskeleton inside the cells is one of the ori-
gins of the name "integrins."
Using the methods of cell and molecular biol-
ogy, we are studying the structure and function of
fibronectins and integrins, their interactions, and
their roles in various physiological processes, in-
cluding development, blood clotting, inflamma-
jjon, wound healing, and cancer. We observe reg-
ulated expression of these molecules during
these processes, and it is clear that these mol-
ecules are crucial for the appropriate behavior of
cells. For instance, altered expression of both fi-
bronectins and integrins in tumor cells contrib-
utes to their wayward behavior, and expression of
these proteins is altered during wound healing.
Using recombinant DNA methods, we can pro-
duce specific and modified forms of fibronectins
and integrins and thus investigate the ways in
which they alfect the behavior of individual cell
types.
We have recently made progress in analyses of
the role of the intracellular portions of various
integrin receptors in interactions with the cyto-
skeleton. We have also obtained detailed struc-
tural information about the cytoskeletal protein
talin, which is a primary candidate for interac-
tions with integrins. We are now investigating
possible interactions between normal and mutant
integrins and talin. Further progress along these
lines should help explain the effects of cell adhe-
sion on cell structure and behavior.
Work in the past year has uncovered evidence
that integrins do more than provide a physical
connection between the extracellular matrix and
the cytoskeleton. Engagement of integrins trig-
gers tyrosine phosphorylation inside the cells,
strongly suggesting signaling via integrins. Inte-
grin function can also be regulated from inside
cells. Cells need to detach as well as attach. How
is this regulated? We find that cell detachment
can apparently be triggered by phosphorylation
of the cytoplasmic domains of integrins. Thus our
current picture of integrins is that they can medi-
ate signaling both into and out of cells.
To extend our understanding of the roles of
fibronectins and integrins to intact organisms, we
use genetic analyses in two animal systems. First,
we are analyzing the role of integrins during the
development of Drosophila melanogaster, a
fruit fly that is suitable for genetic analyses. Flies
with mutations in genes encoding integrins have
defects in embryonic development, in muscle
function, and in the development of wings and
eyes. Analyses of these defects provide insights
into the functions of these proteins. We are also
identifying new integrin species in Drosophila.
In the second genetic project, we have gener-
ated strains of mice that are mutant for fibronec-
tins. When both normal copies of the fibronectin
gene are ablated, mouse embryos cannot proceed
normally beyond the early developmental stage
known as gastrulation. We have also made more-
subtle mutations in the fibronectin gene and are
currently analyzing their effects. These mutant
mice, together with transgenic mice expressing
different forms of fibronectin, should allow us to
dissect the functions of the various forms of fibro-
nectin in vivo. Encouraged by the progress on
fibronectin molecular genetics in mice, we have
also begun to generate mutations in integrin re-
ceptor genes and in other cell adhesion mole-
cules, particularly selectins, cell-cell adhesion
molecules involved in the early steps of inflam-
mation. As our analyses proceed, we will be able
to investigate the effects of the various mutations
on hemostasis, thrombosis, wound healing, and
tumor development.
These studies should provide a deeper under-
standing of the molecular basis of cell adhesion
and its involvement in physiological and patho-
logical processes. This understanding, in turn,
should provide opportunities for therapeutic
treatments of diseases such as thrombosis and
cancer.
The work on cultured cells and some of the
work on mice are supported by grants from the
National Institutes of Health.
196
Molecular Genetics of Intracellular
Microorganisms
Ralph R. Isberg, Ph.D. — Assistant Investigator
Dr. Isberg is also Assistant Professor of Molecular Biology and Microbiology at Tufts University School of
Medicine. He received his A.B. degree in chemistry from Oberlin College and his Ph.D. degree in
microbiology and molecular genetics from Harvard Medical School. He conducted postdoctoral work on
bacterial pathogenesis in the laboratory of Stanley Falkow at Stanford University. His honors include a
Searle Scholars Award and a National Science Foundation Presidential Young Investigator Award.
MANY species of bacteria are capable of caus-
ing diseases by colonizing and growing
within human hosts, using tactics that avoid nor-
mal immune responses. As part of a general strat-
egy to establish an infectious niche, a variety of
microorganisms cause diseases by entering and
growing inside human cells soon after encounter.
Bacteria that establish infections in this manner
are called intracellular microorganisms. Among
the diseases they cause are tuberculosis and the
most common types of sexually transmitted and
food-borne diseases found in the industrialized
world. Despite the prevalence of such infections,
there was little information until recent years on
the factors expressed by these microorganisms
that allow them to enter host cells and thrive.
The objectives of our research are to investi-
gate two important aspects of the life-style of in-
tracellular microorganisms. First, we would like
to determine at the molecular level how these
organisms can enter human cells that do not nor-
mally internalize bacteria. Second, we want to
analyze factors they encode that allow them to
survive and grow within the ordinarily hostile en-
vironment of human cells. Our main approach
has been to identify bacterial species that enter or
grow particularly well within host cells and to
develop genetic and biochemical techniques for
analyzing their strategies. The primary rationale
for this approach is that it provides insights into
basic processes that are applicable to numerous
intracellular microorganisms.
To investigate the molecular mechanism of bac-
terial binding and entry into host cells, we have
been analyzing the bacterium Yersinia pseudo-
tuberculosis, an organism that causes an intes-
tinal disease often accompanied by infection of
multiple organ systems.
To investigate intracellular growth, we have
been analyzing Legionella pneumophila, the
causative agent of Legionnaire's disease pneumo-
nia. The intracellular growth process of the bacte-
rium is very similar to that of a wide range of
intracellular microorganisms, and development
of molecular strategies for analyzing it has been
relatively straightforward.
Yersinia pseudotuberculosis Entry Into
Cultured Human Cells
Y. pseudotuberculosis can enter host cells via
three different paths. For each path the microor-
ganism apparently encodes a unique set of pro-
tein factors to be used at different tissue sites dur-
ing the infection process. We have focused on the
path that is promoted by the protein invasin, the
product of the bacterial inv gene. Invasin is a
108-kDa protein on the surface of the bacterium
that allows it to enter human cells by binding
receptor molecules on their surface. We have
shown that a 20-kDa region of invasin binds the
host cells, and this region is sufficient to promote
uptake. Evidence indicates that after the binding
occurs, the host cells do most of the work in in-
ternalizing the bacterium.
Invasin binds at least four different receptors.
Called integrins, these had been previously iden-
tified by investigators interested in a variety of
mammalian cell adhesion processes. The particu-
lar integrin receptors that bind invasin can adhere
to a variety of mammalian proteins, such as fibro-
nectin and molecules that allow adhesion of im-
mune response cells to inflamed tissues.
Although invasin binds these well-character-
ized receptors, there is no obvious sequence simi-
larity between invasin and other proteins that
bind integrins, and mutations that eliminate the
interaction between invasin and its receptors
identify amino acid residues not previously
shown to be involved in integrin binding.
Integrins are clearly not the only host-encoded
factors necessary for internalization. Analysis of
this process has indicated that two additional fac-
tors are required. Mutant studies of one of these
integrin receptors indicate that a cell structure
called a clathrin coat directly interacts with the
integrin receptor during the internalization of
the bacterium. If this interaction is eliminated,
the bacterium cannot be internalized. A second
structure, the host cell cytoskeleton, also is in-
volved in the internalization, but we believe that
this structure performs an indirect role and does
not directly bind the integrin during all stages of
bacterial uptake.
197
Molecular Genetics of Intracellular Microorganisms
Our investigation of invasin/integrin has led to
a model for Yersinia uptake into host cells. Bind-
ing of invasin to its integrin receptor leads to rear-
rangement of the cytoskeleton — rearrangement
requisite to entry. A signal must be sent to cause
the host cell to internalize the microorganism,
and the internalization is facilitated by the ex-
traordinary avidity with which invasin binds its
receptors. Other proteins that bind the identical
integrins cannot produce this signal so effi-
ciently, because they do not bind the receptors
tightly.
Thus invasin appears to promote entry of the
microorganism because it binds an important re-
ceptor that interacts with clathrin coats and com-
municates with the cell cytoskeleton, and be-
cause it binds so tightly to this receptor.
Legionella pneumophila Growth
in Phagocytic Cells
Z. pneumophila causes a variety of diseases in
humans, including Legionnaire's disease pneumo-
nia. The bacterium grows in lung tissues after en-
counter with its human host. Its favorite habitat is
within alveolar macrophages, cells that normally
function to kill invading microorganisms. An im-
portant mechanism for macrophages to kill or
inhibit the growth of a microorganism is to inter-
nalize it and sequester it in a compartment called
a phagosome, which in turn fuses with a lyso-
somal compartment filled with antibacterial fac-
tors. L. pneumophila is able to grow within the
phagosome, convert it into an organelle with a
unique morphology, and prevent the introduc-
tion of the antibacterial lysosomal components
into this site.
We have been interested in determining how L.
pneumophila is able to establish and grow within
this protective niche. We have been taking two
tactics toward analyzing this process. Our first
approach has been to isolate mutations in this bac-
terium that prevent it from growing intracellu-
larly. Our second approach has been to identify
factors that are selectively synthesized by the bac-
terium only during intracellular growth.
Using the first approach, three easily distin-
guishable classes of mutants have been isolated.
The first class causes the bacterium to be internal-
ized by a macrophage via a novel pathway, and
this causes an extreme defect in bacterial growth.
The second class, and most easily isolated, con-
sists of mutants that are no longer able to prevent
the lysosomal contents from being introduced
into the phagosome. The third class appears nor-
mal for uptake as well as for shutting out the lyso-
somal components, but the phagosome contain-
ing the mutant microorganism no longer exhibits
the unique morphology usually found in a Le-
gionella infection.
These classes of mutants indicate that the mi-
croorganism performs a distinct series of steps
within the macrophage, each of which contrib-
utes to the parasite's efficient growth. To investi-
gate the factors missing in these mutants and ana-
lyze the steps in growth performed by this
bacterium, we have identified a small region of
the Legionella chromosome that encodes the fac-
tors missing in the latter two mutant classes. We
are currently analyzing this region of the chromo-
some intensively, with the hope of purifying the
factors encoded by this region in order to de-
scribe their functions in molecular detail.
Our second approach involves using a novel
scheme to identify L. pneumophila genes that are
regulated in a fashion such that they are turned
off when the bacterium is growing outside the
host cell but are rapidly turned on during growth
within macrophages. This tactic involves cloning
fragments of L. pneumophila chromosomal DNA
in front of a reporter gene and introducing these
molecular clones into the bacterium. All molecu-
lar clones constructed in this manner are killed
unless they contain genes that are regulated in the
desired fashion. Using this strategy, we have
cloned at least eight genes that are normally only
expressed by L. pneumophila when the microor-
ganism is within a host cell.
We hope that these two strategies will allow
identification of most of the factors encoded by L.
pneumophila that mediate the striking rearrange-
ment of host cell organelles and facilitate groMT:h
of the microorganism.
198
i
i
1
i
Scanning electron micrograph oflucerifase reporter mycobacteriophages (LRMs ) adsorbed to the
surface o/Mycobacterium bovis — BCG — cells. This is a model for the use of LRMs for rapid detec-
tion o/ Mycobacterium tuberculosis cells in a human clinical sample. The phage particle, once
adsorbed, injects its genome, into which the firefly luciferase gene has been inserted. Expression
of the gene causes the mycobacterial cells to emit light, allowing sensitive and rapid detection of
M. tuberculosis cells and assessment of their drug- susceptibility patterns.
Scanning photograph by Rupa Udani and Frank Macalusa in the laboratory of William Jacobs.
I
r
i
I
!
200
I
Genetic Approaches to the Control
of Mycobacterial Diseases
William R. Jacobs, Jr., Ph.D. — Assistant Investigator
Dr. Jacobs is also Associate Professor of Microbiology and Immunology and of Molecular Genetics at
Albert Einstein College of Medicine. He received a B.A. degree in mathematics at Edinboro University of
Pennsylvania and a Ph.D. degree in molecular cell biology from the University of Alabama at
Birmingham. His doctoral work on Mycobacterium leprae was performed in the laboratory of Josephine
Clark Curtiss and Roy Curtiss III, first at the University of Alabama and then at Washington University,
St. louis. His postdoctoral studies with Barry Bloom focused on developing systems to express foreign
antigens in the tuberculosis vaccine strain BCG (bacille Calmette-Guerin).
TUBERCULOSIS, caused by Mycobacterium
tuberculosis, continues to be a serious
health problem throughout the world. Even in
the developed countries, this plague has been
escalating in recent years. The World Health Orga-
nization (WHO) estimates a worldwide inci-
dence of approximately 8 million new cases and
over 3 million deaths annually. After 32 years of
steady decline in the United States, the number of
new cases has taken a surprising and alarming up-
ward turn for each of the last five years. In 1990
there was a 34 percent increase in New York City
and a 10 percent increase nationally. More re-
cently a strain of multidrug-resistant M. tubercu-
losis has resulted in numerous deaths, demon-
strating the dire reality that tuberculosis caused
by a virtually invincible pathogen is now
spreading.
Other mycobacteria play significant roles in
world health. Mycobacterium leprae is the caus-
ative agent of leprosy, an affliction know to the
ancients that affects over 1 3 million in the world
today. Another agent is Mycobacterium avium, a
major opportunistic pathogen for many individ-
uals with AIDS.
Our ongoing studies are aimed at developing
novel ways of treating mycobacterial disease, par-
ticularly tuberculosis, using molecular genetic
approaches. In addition, we have taken a new
look at the vaccine known as BCG (bacille
Calmette-Guerin), which has been used widely
since 1 924 to prevent tuberculosis. We are genet-
ically engineering BCG into a multivalent vac-
cine vector that can elicit a protective immune
response to a wide variety of bacterial, viral, and
parasitic pathogens. The development of systems
to alter mycobacteria genetically should permit
both goals to be achieved.
Genetic manipulation of these microorganisms
has only been possible in the last few years. M.
tuberculosis, like other mycobacteria, is difficult
to analyze genetically for a variety of reasons.
First, the tubercle bacillus, which multiplies
only once every 24 hours, requires 3 weeks to
form a colony from a single cell. In contrast, Esch
erichia coli yields visible colonies in 8 hours.
The leprosy bacillus has yet to be cultivated in
the laboratory and can only be grown in mouse
footpads or the nine-banded armadillo. Over the
last five years, we have developed a series of
phage- and plasmid-based vectors that have en-
abled the efficient introduction of recombinant
DNA into mycobacteria. Recombinant DNA tech-
nologies have opened exciting new doors to basic
knowledge about these organisms and the ways
they cause infection. In addition, these technolo-
gies offer novel reagents, such as luciferase re-
porter phages, and recombinant vaccines that
could play major roles in combating human
disease.
Epidemiological Analysis
of Tuberculosis Infections
Use of restriction fragment length polymor-
phisms (RFLPs) can play a key role in determin-
ing the mode of transmission of tuberculosis, as
individual isolates can be tracked from one in-
fected person to the next. The recent increases in
the incidence of tuberculosis in the United States
seem to be largely associated with the AIDS epi-
demic. However, it was unclear whether the tu-
berculosis seen in AIDS patients results from
reactivated disease, reflecting pre -AIDS expo-
sure, or from a first-time infection. In collabora-
tions with researchers in San Francisco, including
Gary Schoolnik (HHMI, Stanford University), we
have undertaken analyses to distinguish these two
possibilities.
Different isolates of M. tuberculosis have dif-
ferent RFLP patterns when probed with a particu-
lar DNA element found in M. tuberculosis
strains. In isolates from a recent tuberculosis out-
break among AIDS patients in a group home, RFLP
analysis revealed that 1 1 persons had all been in-
fected with the same strain of M. tuberculosis.
This demonstrates that AIDS patients are highly
susceptible to M. tuberculosis infection and con-
firms that tuberculosis is highly contagious.
These clear results should be translatable into
better public health care policies. Similar analy-
201
Genetic Approaches to the Control of Mycobacterial Diseases
ses are under way to track M. tuberculosis iso-
lates that are multiply drug resistant.
Luciferase Reporter Phages
Accurate diagnosis of M. tuberculosis infec-
tion routinely requires 4-6 weeks to allow culti-
vation of the tubercle bacilli and time to perform
tiie appropriate tests. Assessment of the drug-
susceptibility patterns of clinical isolates can re-
quire another 4-6 weeks of analysis. In light of
the increasing numbers of drug-resistant speci-
mens, we set out to devise a novel diagnostic test
that would shorten the time required for both
procedures. The test uses a mycobacteriophage, a
virus that infects M. tuberculosis, into which we
have cloned a gene that encodes a reporter en-
zyme, the luciferase that makes fireflies glow. Lu-
ciferase reporter mycobacteriophage (LRM) par-
ticles, when mixed with bacterial cells, result in
the production of light.
The LRM test is extraordinarily sensitive, as
photons can be detected at extremely low levels.
In addition, it is exquisitely specific, as the phage
only attaches to mycobacterial cells. Thus the pro-
duction of light reveals the presence of M. tuber-
culosis in the sample. We have undertaken stud-
ies in clinical application. Since the luciferase
reaction requires ATP for photon production, we
also hope to use the test to assess the metabolic
activity of a cell. And since drug treatment of my-
cobacterial cells abrogates this activity, we are
also exploring the use of the test to distinguish
drug-resistant from drug-sensitive cells.
Genetic Analysis of Mycobacterial
Virulence Determinants
We would like to know why pathogens such as
M. tuberculosis or M. leprae cause severe disease
when BCG can elicit an effective immune re-
sponse against these organisms. To approach
these questions, we have focused on developing
systems for defining and characterizing the genes
of mycobacteria.
The first approach is to generate specific inser-
tion mutations within the virulent M. tuberculo-
sis genome and to screen for mutants that are no
longer virulent in our animal models. Toward
this goal, we are doing a number of studies de-
signed to generate random insertions within the
genome by taking advantage of natural mycobac-
terial transposons or recombination systems.
Complementary to this mutant isolation is the de-
velopment of both extrachromosomal and inte-
grating vectors that permit highly efficient in-
troduction of libraries of genes from virulent
mycobacteria into avirulent strains.
The combination of these two strategies has al-
ready allowed us to identify genes necessary for
the biosynthesis of amino acids, purines, and
complex polysaccharides found on the surface of
the pathogenic mycobacteria. We are currently
developing a variety of systems to screen these
libraries for genes that confer virulence charac-
teristics to the avirulent organisms. By identifying
the genes and their products responsible for spe-
cific virulence phenotypes, we hope to make pos-
sible the design of strategies to prevent or control
mycobacterial disease. In addition, the geneti-
cally engineered avirulent mutants should pro-
vide novel vaccine candidates.
Recombinant BCG Vaccines and Novel
Vaccine Strategies
BCG is an attenuated mutant of bovine tuber-
culosis bacillus. It has been used as a vaccine
against tuberculosis in humans for over 50 years.
The bacterium possesses several unique proper-
ties that make it ideal for use as a live vector in
generating multivalent vaccines. BCG is a safe
vaccine, having been used in 2.5 billion individ-
uals with a mortality rate significantly lower than
that of the smallpox vaccine. It is the only live
vaccine other than oral polio that WHO recom-
mends for use in infants. Also, the mycobacterial
cell wall has potent adjuvant properties that can
engender excellent humoral responses; and since
BCG is stored within macrophages, which are key
antigen-presenting cells, it follows that it can also
elicit cellular immune responses.
In the last few years, our laboratory, in collabo-
ration with Barry Bloom (HHMI, Albert Einstein
College of Medicine), Medlmmune, and Graham
Hatfull at the University of Pittsburgh, has devel-
oped a series of expression vectors and transfor-
mation systems whereby foreign genes encoding
antigens from virtually any pathogen can be
cloned and expressed in BCG. Mice immunized
with these recombinant BCG cells have been
shown to elicit both humoral and cellular im-
mune responses to the expressed foreign pro-
teins. We are currently cloning and expressing
genes from pathogens that cause leishmaniasis,
schistosomiasis, and toxoplasmosis in BCG. Im-
munization of mice with recombinant BCG will
provide useful models to test the types of im-
mune responses that can be elicited with these
live bacteria. Ultimately we hope to engineer a
recombinant BCG that could protect humans
from these dread diseases.
202
Mechanisms of Neurotransmitter Storage
and Release
Reinhard John, Ph.D. — Associate Investigator
Dr. Jahn is also Associate Professor of Pharmacology at Yale University School of Medicine. He received his
Ph.D. degree at the University of Gottingen, Germany. His postdoctoral training was with Hans Dieter
Soling at the University of Gottingen and later with Paul Greengard at Yale University and the Rockefeller
University. He was Assistant Professor at the Rockefeller University and subsequently headed a research
group at the Max Planck Institute for Psychiatry, Martinsried.
NERVE cells, or neurons, communicate with
each other and with other cells by means of
small molecules, the neurotransmitters. This
communication occurs at specialized contact
zones, the synapses. Upon arrival of incoming ac-
tion potentials, the sender cell releases its neuro-
transmitters, which cross the synaptic cleft. The
plasma membrane of the receiving cell has spe-
cific receptor molecules that, in turn, translate
the signals into functional changes.
In the resting state, the presynaptic nerve end-
ings of the sender cell store their neurotransmit-
ters in small membrane-enclosed compartments,
the synaptic vesicles. When an action potential
arrives, voltage-gated calcium channels open and
calcium ions (Ca^"^) enter the terminal from the
extracellular space. Within a fraction of a milli-
second, synaptic vesicles fuse with the plasma
membrane, releasing their contents. The vesicle
membrane protein is then retrieved by endocyto-
sis for use in regenerating fusion-competent syn-
aptic vesicles.
The details of this membrane recycling are still
largely unclear, and the enzymes catalyzing the
major steps have not been identified. As a starting
point for detailed functional analysis, we and
others have characterized the major membrane
proteins of synaptic vesicles. This was facilitated
by the fact that synaptic vesicles are abundant and
can easily be purified in large amounts. Due to
the smallness of the vesicles, the number of pro-
tein species per individual vesicle is inherently
limited. Therefore it should be possible to iden-
tify most, if not all, of the major protein constitu-
ents. To date, several families of unique vesicle
proteins have been characterized, and advanced
tools for their study are available. We are
currently involved in a systematic chemical analy-
sis of the vesicle membrane to identify the re-
maining proteins of this organelle.
Parallel to this protein analysis, we have re-
cently begun to establish assay systems for indi-
vidual steps of the vesicle cycle. One of the
model systems accessible for biochemical analy-
sis is the isolated nerve terminal. Upon homogen-
ization of nervous tissue, nerve terminals, though
sheared off their axons, reseal and remain func-
tional for several hours after isolation. Applica-
tion of depolarizing stimuli or Ca^"^ ionophores
causes a massive exocytotic transmitter release
and a parallel increase in membrane turnover.
We found that during exo-endocytosis, one of
the vesicle proteins, the GTP-binding protein
rab3A, dissociates from the vesicle membrane
and reassociates again at a later stage of the mem-
brane cycle. This dissociation-association cycle is
probably necessary for an orderly and sequential
processing of the vesicle membrane. Thus small
GTP-binding proteins may serve in the nerve ter-
minal, as in other organelles, as "status indica-
tors" for the recycling membrane, being asso-
ciated only with a specific step in the life cycle of
the vesicle membrane. The biochemical events
leading to GTPase activation and membrane disso-
ciation and reassociation are presently under
study. In addition, the individual G proteins
should allow an isolation of the compartments
representing separate steps in the membrane
cycle.
Thus we have isolated clathrin-coated vesicles
from nerve terminals by conventional procedures
and have analyzed their membrane and coat com-
position. We have also developed immunoisola-
tion procedures for synaptic vesicle-derived
membrane populations, using monoclonal anti-
bodies directed against individual membrane
components. We hope these studies will aid in
the functional definition of the compartments in-
volved in synaptic vesicle recycling and will form
the basis for the reconstitution of individual steps
of the membrane cycle in cell-free systems.
203
Molecular Studies of Voltage-Sensitive
Potassium Channels
Lily Y.Jan, Ph.D. — Investigator
Dr. Jan is also Professor of Physiology and Biochemistry at the University of California, San Francisco.
During her graduate study at the California Institute of Technology with Jean Paul Revel and Max
Delbriick, Dr. Jan localized the visual pigment rhodopsin at the ultrastructural level. Because she studied
high-energy theoretical physics before becoming a biology student, her Ph.D. degree was in physics and
biophysics. She stayed at CalTech to do postdoctoral research with Seymour Benzer and began to
collaborate with her husband, Yuh Nungjan. Their first collaboration resulted in the identification of the
Shaker locus as a potential structural gene for a potassium channel. Before accepting faculty appointments
at UCSF, the Jans worked in Stephen Kuffler's laboratory at Harvard Medical School.
VOLTAGE-sensitive potassium channels proba-
bly constitute the most diverse and wide-
spread class of ion channels. More than 30 differ-
ent types of potassium channels have been
characterized. They dilFer in their voltage sensi-
tivity, their kinetic properties, and their sensitiv-
ity to second messengers within the cell. Potas-
sium channels have been found in almost every
eukaryotic cell type examined, in both the ani-
mal and the plant kingdoms. They are important
for a wide range of physiological functions, in-
cluding insulin release due to raised glucose lev-
els, proliferation of lymphocytes induced by mi-
togens, and the movements of leaflets in plants or
the opening and closing of leaf stomatal pores. In
the mammalian nervous system, potassium chan-
nels control excitability and the strength of sig-
naling between nerve cells. Indeed, some of the
potassium channels have been implicated as
playing a role in learning and memory.
To study how the diversity of potassium chan-
nels arises and how they serve the wide variety of
cellular functions, one needs to study these chan-
nels biochemically as well as biophysically. How-
ever, they are difficult to purify because they are
rather heterogeneous and inaccessible. For this
reason we have taken advantage of the well-
developed genetic technologies applicable to the
fruit fly Drosophila melanogaster. In this organ-
ism, if a gene (say, coding for a potassium chan-
nel) can be identified by the abnormalities
caused by its mutations, one can clone it for mo-
lecular studies of the gene product.
More than a decade ago, Yuh Nungjan (HHMI,
University of California, San Francisco), Mike
Dennis, and I found that mutations at the Shaker
locus cause prolonged transmitter release from
the motor nerve terminal, probably because of a
defect in potassium channel function. The locus
was subsequently cloned by Diane Papazian, Tom
Schwarz, and Bruce Tempel in our laboratory. It
codes for proteins that contain multiple stretches
of hydrophobic amino acids that can potentially
span the cell membrane; several proteins that
diff^er in sequence flanking these stretches of hy-
drophobic amino acids are generated because of
alternative splicing of the primary transcript.
Leslie Timpe demonstrated that RNA encoding
for these Shaker proteins of known deduced se-
quence, when injected into frog oocytes, causes
the functional expression of potassium channels
of different kinetic properties. Since the different
Shaker proteins are present in different regions
of the fly brain, they are likely to give rise to dif-
ferent subtypes of A channels.
Starting with the Shaker gene in the fruit fly,
Bruce Tempel isolated a gene in the mouse that
codes for a potassium channel in the mouse
brain. This protein is 65 percent identical in its
sequence to the Drosophila potassium channel
protein. In frog oocytes, the gene produces potas-
sium channels that do not inactivate rapidly. Anal-
ysis of the distribution of amino acid residues that
appear to be essential and have been totally con-
served over 600 million years has provided some
clues to the channel's structure. Now over 1 5 dif-
ferent mammalian potassium channel genes have
been characterized by a number of laboratories.
One of these (the rat Shall gene), cloned and
characterized by Tim Baldwin, Meei-Ling Tsaur,
and George Lopez in our laboratory, produces a
rapidly inactivating potassium channel in frog
oocytes and is expressed in the heart as well as
the brain. The high degree of conservation among
these mammalian channels and the fruit fly potas-
sium channels reiterates the point that any experi-
mental organism, as long as it is amenable to the
specific type of experimentation, will reveal in-
formation of medical interest.
Having cloned some of the potassium channel
genes, we can now ask how this channel works.
How does it detect a voltage change across the
cell membrane and, responding, open? How does
it "inactivate" after it opens? How does it discrim-
inate between sodium and potassium ions and
show exquisite selectivity? To probe these ques-
tions, we have altered specific residues of the po-
tassium channel to see how the various functions
are affected. Studies in our laboratory (Ehud
205
Molecular Studies of Voltage-Sensitive Potassium Channels
IsacofF, George Lopez, Diane Papazian, and Leslie
Timpe; this work is supported by a grant from the
National Institutes of Health) and others have
provided evidence for the involvement of spe-
cific structural elements in the detection of volt-
age changes across the cell membrane and in the
subsequent conformational changes that open
the channel and allow the cytoplasmic mouth of
the pore to interact with the inactivation gate,
thereby terminating channel opening.
For studies of the biological functions of potas-
sium channels, we have chosen to concentrate on
the mammalian heart and hippocampus. A variety
of cardiac potassium channels have been charac-
terized biophysically and are important in con-
trolling the rhythmic heartbeat. Molecular stud-
ies of these channels will not only contribute to
our understanding of channel function but will
also be relevant clinically, for example, in the
development of more-specific drugs for cardiac
arrhythmia. The hippocampus is a region of the
mammalian brain that appears to play an impor-
tant role in learning and memory. It has also been
studied extensively in experimental paradigms
that induce epileptic activity. By cloning and ana-
lyzing potassium channel genes that are ex-
pressed in this tissue, we hope to learn about the
involvement of these potassium channels in the
normal function and pathology of the nervous
system.
Using specific probes for individual potassium
channel polypeptides and their messenger RNA,
Morgan Sheng and Meei-ling Tsaur have found
distinct patterns of expression in the mammalian
brain. In addition to spatial regulation at the level
of brain regions as well as specific neuronal
types, dynamic changes in potassium channel
gene expression have also been observed in
the adult brain following pentylenetetrazole-
induced seizure. (This work is supported by a
grant from the National Institute of Mental
Health.) The observed decrease of specific potas-
sium channel transcripts in the excitatory neu-
rons in the dentate gyrus of the hippocampus is
likely to lead to an increase of excitability. These
observations suggest that potassium channel gene
regulation may contribute to long-term neuronal
plasticity.
206
Neural Development in Drosophila
Yuh Nungjan, Ph.D. — Investigator
Dr. Jan is also Professor of Physiology and Biochemistry at the University of California, San Francisco.
Although Dr. Jan went to the California Institute of Technology to study theoretical physics, he instead
became interested in biology and received his Ph.D. degree from Caltech in biophysics and physics. While
there he studied sensory transduction of the fungus Phycomyces with Max Delbriick. Dr. Jan began his
study of the nervous system during postdoctoral research with Seymour Benzer at CalTech and continued
this line of research with Stephen Kuffler at Harvard Medical School. His primary interest remains the
nervous system.
HOW a nervous system is organized during
development is a major unresolved problem
in biology. For several years our laboratory has
been interested in the following questions: How
do neurons arise from undifferentiated ectoder-
mal cells? What gives the neurons their individual
identity in terms of shape and function?
Our long-term goal is to understand these pro-
cesses at the molecular level, and our approach is
a genetic one. We first isolate mutations that af-
fect neurogenesis, neuronal type, or axonal path-
way formation. Identification of these mutations
can lead to the isolation of important genes.
During the last few years, our laboratory has
been engaged in an extensive search and analysis
of mutants affecting neural development in the
fruit fly Drosophila. To identify and analyze such
mutants, we have been using the embryonic sen-
sory nervous system, which has been character-
ized in considerable detail at the single-cell level.
Roughly half of the Drosophila genome has been
screened for mutations that alter the peripheral
nervous system (PNS), resulting in the identifica-
tion of a number of genes that specify cell fate in
the embryo. Analysis of those genes had led us to
propose a "progressive determination of the
PNS" model.
Early during embryogenesis, cells in different
locations within the ectodermal layer acquire un-
equal developmental potential as a result of the
actions of the "prepattern genes." These include
genes that specify dorsoventral and anteroposte-
rior orientation as well as segmentation. Cells
from some of the domains are affected by the ac-
tion of "proneural" genes, which apparently en-
dow cells with the competence to become neuro-
nal precursors. Genes of the achaete scute
complex {AS-C) and daughterless {da) belong
to this group. Both y45-C and da encode proteins
with the helix-loop-helix motif. It is likely that
products of these genes may form homo- or heter-
odimers that bind to DNA and regulate the tran-
scription of target genes in order to initiate neuro-
nal precursor development.
As a neuronal precursor forms, it inhibits neigh-
boring cells from doing so. This "lateral inhibi-
tion" involves the action of six known "neuro-
genic" genes. Removing the function of any of
the six. Notch, Delta (Dl), the Enhancer of split
complex [E(spl)-C\, mastermind (mam), neu-
ralized (neu), and big brain {bib), leads to hy-
pertrophy of both the central nervous system
(CNS) and the PNS, presumably as a result of los-
ing lateral inhibition.
There appear to be at least two independent
cell-cell interaction pathways. One is mediated
by the gene products of Notch and Dl, both cod-
ing for membrane proteins with epidermal
growth factor (EGF)-Iike repeats. E(spl)-C,
mam, and perhaps neu are involved in this path-
way. The second pathway is mediated by bib,
which encodes a membrane protein with signifi-
cant homology to the bovine major intrinsic pro-
tein (MIP), soybean nodulin 26, and Escherichia
coli glycerol facilitator, which allows passive
transport of small molecules such as glycine.
The commitment of neuronal precursors may
involve the actions of a group of "master regula-
tory" genes, which endow the precursor with
certain unique properties of the nervous system.
The identity of a neuronal precursor is further
specified by "neuronal-type selector" genes. For
example, the cut locus is required for external
sensory organs to acquire their correct identity.
In the absence of cut function, these organs are
transformed into chordotonal organs. Normally
cut is expressed in sensory organ precursors but
not in chordotonal organ precursors, and we
think that the cut gene determines which organ
the precursor will develop into. The cm? product
contains a homeodomain and is likely to act as a
transcription factor regulating the expression of
downstream differentiation genes.
Recently we began to work on the problem of
axonal pathfinding and target recognition in Dro-
sophila. Tracing the axonal pathway in the fly
nervous system had been problematic in the past.
Because of the small size of the neurons, it was
difficult to use traditional methods to trace path-
ways, such as filling neurons with fluorescent
dyes and other tracers. However, Ed Giniger, a
207
Neural Development in Drosophila
postdoctoral fellow in our laboratory, found a so-
lution to this difficulty.
He modified the enhancer trap method we
used previously by replacing the /3-galactosidase
(/3-gaI) reporter gene with a kinesin-|8-gal fusion
gene. The idea is that the kinesin part of the fu-
sion protein will drag /3-gaI down the axon to the
terminal. The idea works very nicely. We have
-now generated over a thousand enhancer trap
lines with kinesin-/«cZ as a reporter gene and, in
more than a hundred of those, have labeled sub-
sets of CNS and/or PNS pathways, including sen-
sory and/or motor projections. We would like to
choose a small number of lines that reveal simple
axonal pathways (e.g., sensory neuron or motor
neuron pathways) as a basis for a mutant screen.
Based on sequence information of a number of
genes involved in neural development, it seems
that the majority of these genes contain a previ-
ously identified functional motif — e.g., the EGF
repeat found in Notch, the tyrosine kinase do-
main in sevenless, the homeodomain in cut, and
the helix-loop-helix motif in da and AS-C In
each case the existence of such a motif immedi-
ately provides strong suggestions for the likely
mode of gene action, which may be tested
experimentally.
The existence of readily identifiable motifs in
the majority of genes involved in neural develop-
ment suggests that cells have a limited repertoire
of mechanisms for essential regulatory functions.
These include various signaling systems, such as
ligands, receptors, second messenger systems,
and regulators of gene expression. Perhaps only a
modest number of new tools had to be invented
for the formation of the nervous system. Many of
the cellular mechanisms used in neural develop-
ment may already have been available before the
nervous system evolved. Understanding neural
development may require an understanding of
the usage and manipulations of these basic func-
tional motifs.
It is also apparent that many functional motifs
have been highly conserved during evolution.
Several hundred million years separate the in-
sects and vertebrates, yet the aforementioned mo-
tifs (the EGF repeat, tyrosine kinase domain, ho-
meodomain, helix-loop-helix) are common to
both and remarkably enduring. Thus studying de-
velopment in the simpler organisms with well-
developed genetics, such as the fruit fly and the
nematode, may provide valuable background for
the investigation of corresponding problems in
higher forms.
To follow the development of a nerve
cell, the neurobiologist will typically
inject it with a tracer dye. In view of the
minuteness of the fruit fly embryo, how-
ever, a method was developed to cause
the fly to fill its own neurons with a
marker. A bacterial enzyme was tar-
geted to nerve processes by fusing the
gene for kinesin, a molecular motor, to
the gene encoding fi-galactosidase, an
enzyme for which there are simple col-
orimetric assays, and Drosophila stocks were established that express the fusion protein. The
example shown here displays the pattern seen when the protein is made in all the neurons of the
embryo. The central ladder-like structure comprises the axons of the ventral nerve cord, and the
fine lateral extensions are the peripheral neurons and axons.
Research and photograph by Ed Giniger in the laboratory ofYuh Nungjan.
208
Activation ofCD4 T Cells
Charles A. Janeway, Jr., M.D. — Investigator
Dr. Janeway is also Professor of Immunobiology at Yale University School of Medicine. He is a graduate of
Harvard College (B.A., chemistry) and of Harvard Medical School. He had research training during
medical school at the National Institute for Medical Research in London, England; postdoctoral training
in internal medicine at the Peter Bent Brigham Hospital in Boston; and immunology research experience
at NIH and the Biomedical Center in Uppsala, Sweden. Dr. Janeway was awarded the degree of Doctor
Honoris Causa by the Copernicus Medical School of Jagellonian University in Cracow, Poland.
THE critical event in most immune responses
is the activation of specific CD4 T lympho-
cytes. When these cells are lost, as happens in
AIDS, the immune system can no longer defend
the host against infection. Our laboratory is study-
ing the mechanism by which these cells become
active. This is a complex process involving
changes in both the cell that presents the activat-
ing antigen and the T cell itself.
The process requires two events. First, the
antigen-specific receptor on the T cell must en-
counter its appropriate ligand on the surface of
an antigen-presenting cell. This results in spe-
cific signaling of the T cell that will cause it to
become immunologically inert, or anergic. Aner-
gic T cells cannot mount responses. In order for
antigen recognition to lead to an eff'ective re-
sponse, the CD4 T cell must receive a second sig-
nal or co-stimulator. Studies in our laboratory
have characterized the chemical structure of sev-
eral peptides that are bound by class II molecules
of the major histocompatibility complex (MHC)
and thus recognized by CD4 T cells. Second, we
have identified two molecules expressed on anti-
gen-presenting cells that are capable of co-stimu-
lating the activation of resting, naive CD4 T cells.
Finally, we have shown that a single cell must
present both the peptide-MHC complex and the
co-stimulator in order for normal CD4 T cells to
be activated eff'ectively.
The ligand recognized by the receptor on CD4
T cells consists of a peptide fragment of a foreign
protein antigen bound to a class II MHC-encoded
molecule. Our laboratory has isolated peptides
from such molecules and characterized their
structure. These molecules are approximately 15
amino acids in length and have specific residues
at certain positions for each different MHC class
II molecule. Although the cleavage of these frag-
ments from the native proteins has not been char-
acterized, the results of our analysis suggest that
the MHC molecule protects the peptide fragment
from further proteolytic degradation.
A striking result in these studies was the finding
that 1 2 percent of normal MHC class II molecules
on antigen-presenting cells were in complex
with a particular self peptide. As a number of sim-
ilarly dominant peptides were observed, it seems
possible that the total complexity of self peptides
to which the immune system must be tolerant
may be quite limited. This is important, as it al-
lows the maximum complexity of foreign pep-
tides to be recognized.
Of even greater interest is the finding that this
specific complex is found richly represented on
cells in the thymic medulla, including bone
marrow-derived cells, but is present only at very
low levels on thymic cortical epithelium. These
two tissues are known to carry out distinctive
events involving self MHC-peptide complexes in
the maturation of T cells within the thymus. The
finding that a self peptide is unevenly distributed
between these cell types suggests that T cell de-
velopment occurs under different selecting envi-
ronments for different processes. The antibody
recognizing this processed self peptide bound to
self MHC class II will allow our laboratory to
probe this process in living animals. The same
technology is now being applied to the analysis of
autoantigens and peptides generated by intracel-
lular infectious agents such as Salmonella
typhimurium.
In addition to ligation of the receptor, the same
antigen-presenting cell must also express co-
stimulatory molecules on its surface in order to
activate CD4 T cells optimally. We have con-
firmed the importance of the B7 molecule, a
known co-stimulator of human T cells, in the ac-
tivation of murine CD4 T cells and have discov-
ered that the heat-stable antigen can also serve as
a co-stimulator for murine T cells. Either of these
proteins, transfected into fibroblasts, can confer
their CD4 T cell-activating properties. More-
over, the two proteins act in a strongly synergistic
manner on normal antigen-presenting cells to
promote the proliferation of CD4 T cells.
Finally, we have shown that B7 expression is
regulated in parallel with co-stimulatory activity
when B lymphocytes respond to a variety of mi-
crobial products. This regulated behavior of co-
stimulators allows B cells to present antigen to
CD4 T cells for tolerance in the absence of micro-
209
Activation ofCD4 T Cells
bial infection or for immunity in its presence. We
documented the importance of this by showing
that tolerance to a self protein could be broken by
inducing the expression of co-stimulator mole-
cules on self antigen-reactive B cells. This last
system may help to explain the role of specific
pathogens in the initiation of autoimmunity.
Parts of this work are also supported by grants
from the National Institutes of Health.
The T cell perceives the complex of foreign
peptide and self MHC class II molecule with a
receptor resembling the antigen-binding frag-
ment of an immunoglobulin molecule. Although
the most variable portion of this receptor lies in
its central region, generated by the joining of sev-
eral different gene segments, some variability
is also expressed around the periphery of the
ligand-binding site. This raises the question of
what residues on the peptide-MHC complex are
contacted by these peripheral antigen-binding
loops on the T cell receptor.
We recently showed that at least one of these
loops makes contact with one side of the MHC
molecule binding the foreign peptide. Detailed
mutagenesis studies of both the T cell receptor
and the MHC protein should soon reveal the de-
finitive orientation of the T cell receptor to its
ligand.
The information derived from our studies on T
cell activation is being applied to the analysis
of a model autoimmune disease, the insulin-
dependent diabetes mellitus that occurs sponta-
neously in non-obese diabetic mice. We have iso-
lated cloned T cells capable of invading the islets
of irradiated mice and destroying (8-cells, produc-
ing diabetes. In addition, we have identified
other cells that appear to protect the islet. Analy-
sis of the peptide-MHC ligands recognized by
these various cell types, and of the events in-
volved in their activation, should permit a better
understanding of diabetes mellitus and lead to its
specific immunomodulation. This work is also
supported by a grant from the National Institutes
of Health. Thus the basic science base of the labo-
ratory is being applied to disease models in hopes
of improving therapy in immunological diseases.
210
Control of Cell Pattern in the Developing
Nervous System
Thomas M. Jessell, Ph.D. — Investigator
Dr. Jessell is also Professor of Biochemistry and Molecular Biophysics at Columbia University College of
Physicians and Surgeons and a member of the Center for Neurobiology and Behavior. He received his
Ph.D. degree in neurobiology from Cambridge University, England, and was elected a research fellow of
Cambridge's Trinity College. He was a postdoctoral fellow in Gerald Fischbach's laboratory at Harvard
Medical School. Next he served as Assistant Professor of Neurobiology at Harvard Medical School, before
moving to Columbia University.
OUR research is aimed at understanding the
mechanisms that control cell patterning
within the developing vertebrate nervous system.
The major focus is on deciphering how discrete
neural cell types appear at defined positions in
the embryo. In addition, we are examining the
roles of diffusible and cell surface molecules in
the guidance of developing axons in the spinal
cord. Our studies over the past several years have
provided evidence that the floor plate, a special-
ized group of neuroepithelial cells, has critical
roles in both the control of cell identity and in
axon guidance. Within the past year, further de-
tails of the functions and molecular properties of
the floor plate have become apparent.
Control of Motor Neuron DiflFerentiation
Our previous studies had shown that signals
originating from the floor plate regulate the iden-
tity of specific cell types within the neural tube.
In order to examine in more detail the actions of
the floor plate on neural cell patterning, we have
begun to focus on one well-characterized neuron
class, the spinal motor neuron. There is extensive
information on the mechanisms that control the
pathfinding of motor axons and the formation of
synapses at the neuromuscular junction, but the
events that control the generation of motor neu-
rons remain largely obscure.
Insight into the molecular mechanisms in-
volved in the generation of motor neurons by
floor plate-derived signals requires the identifi-
cation of genes that are expressed at the initial
stages of motor neuron differentiation. In collabo-
ration with Thomas Edlund's laboratory in Umea,
Sweden, we have found that embryonic chick
motor neurons express a homeobox gene called
Islet- 1 (Isl-1). This is a member of the subfamily
of homeobox genes that contain cysteine-rich re-
gions called LIM domains. Other members of the
family include Lin- 11 and Mec-3, which have
been shown to regulate cell fate in Caenorhab-
ditis elegans.
Isl- 1 binds to enhancer elements in the rat in-
sulin gene and is expressed in pancreatic islet
cells and in a subset of neurons, including motor
neurons. In the embryonic chick spinal cord,
Isl- 1 immunoreactivity is first detected in the nu-
clei of cells in the ventral region, lateral to the
floor plate. The number of Isl- 1* cells in the ven-
tral spinal cord increases markedly during early
spinal cord development. We established that the
ventral spinal cord cells that express Isl- 1 are mo-
tor neurons by retrograde injection of horserad-
ish peroxidase (HRP) into motor axons in the
ventral root.
Analysis of the expression of motor neuron
markers by chick spinal cord cells in vivo has
provided evidence that the differentiation of mo-
tor neurons is dependent on inductive signals
from the floor plate. In agreement with this, the
spinal cord of embryos that had received floor
plate grafts contained additional ectopic Isl-1^
cells. The induced Isl-1^ cells also expressed the
SCI glycoprotein and sent axons out of the spinal
cord consistent with their identity as motor
neurons.
These results provide evidence that signals
from the floor plate can induce the expression of
Isl-1 in dorsal neural tube cells. Elimination of
the notochord and floor plate before neural tube
closure results in the development of a spinal
cord devoid of ventral neuronal types, including
motor neurons. Isl- 1 expression in the ventral spi-
nal cord is also dependent on signals from the
notochord and floor plate. These results support
the idea that elimination of the notochord and
floor plate prevents the initial steps in the differ-
entiation of motor neurons.
Homeobox genes are involved in many aspects
of vertebrate development. The expression pat-
tern of Hox genes along the anteroposterior axis
of the neural tube, and of Pax genes along the
dorsoventral axis, together with the phenotypes
that result from inactivation of some of these
genes, suggests that they contribute to the re-
gional patterning of the developing nervous sys-
tem. In contrast, the restricted expression of Isl-1
and the involvement of related LIM-homeodo-
main proteins in the determination of cell fate in
C. elegans suggest that Isl- 1 may be involved in
specifying the fate of specific neuronal subtypes
— in particular, motor neurons.
211
Control of Cell Pattern in the Developing Nervous System
Floor Plate-Specific Genes
in Axon Guidance
After the identity of motor neurons and other
spinal cord neurons has been established, the
floor plate appears to provide both long-range
and local guidance cues that promote the growth
of axons to and across the ventral midline of the
jspinal cord. First, the floor plate secretes a diffu-
sible chemoattractant that can orient the growth
of axons of commissural neurons in vitro and
may account for the homing of these axons to the
floor plate in vivo. Second, the floor plate may
contribute to the change in trajectory of commis-
sural axons from the transverse to the longitu-
dinal plane that occurs immediately after cross-
ing the ventral midline. In support of this
proposal, genetic mutations in mice and zebra
fish that result in the absence of the floor plate
during embryonic development lead to errors in
the pathfinding of commissural axons at the mid-
line of the spinal cord. Third, the floor plate may
promote the fasciculation of commissural axons
that occurs after they cross the midline of the spi-
nal cord by regulating the expression of glyco-
proteins of the immunoglobulin superfamily.
The specialized role of the floor plate in verte-
brate neural development has parallels in inver-
tebrate organisms, in that cells at the midline of
the embryonic Drosophila and C. elegans central
nervous systems have been implicated in neural
patterning and axon guidance.
To identify molecules that may mediate the di-
verse functions of the floor plate during early
neural development, we have used subtractive hy-
bridization techniques to isolate cDNA clones ex-
pressed selectively by the floor plate. One of
these cDNA clones encodes a novel secreted pro-
tein, F-spondin, which is expressed at high levels
by the rat floor plate during embryonic develop-
ment. F-spondin contains domains similar to
those present in thrombospondin and other pro-
teins implicated in cell adhesion and neurite out-
growth. In vitro assays show that F-spondin
promotes neural cell adhesion and neurite
outgrowth, suggesting that its secretion by the
floor plate contributes to the growth and guid-
ance of axons in the developing central nervous
system.
The F-spondin protein may be associated with
the extracellular matrix, since it has several clus-
ters of basic residues that function as glycosami-
noglycan-binding domains in other secreted pro-
teins. The restricted distribution of F-spondin
mRNA in the embryonic nervous system contrasts
with the distribution of other secreted glycopro-
teins that promote neural cell adhesion and neu-
rite outgrov^h. For example, the expression of
F-spondin mRNA is more restricted than that of
thrombospondin and of tenascin, which appear
to be expressed widely in the embryonic central
nervous system.
The prominent expression of F-spondin in the
floor plate suggests that the protein may be in-
volved in the plate's development or functions.
Midline neural plate cells that give rise to the
floor plate undergo marked changes in cell shape
during the closure of the neural tube. Thus one
possible function of F-spondin could be to medi-
ate adhesive plate cell interactions that maintain
the integrity of the floor plate during formation
of the embryonic spinal cord.
The expression of F-spondin mRNA in floor
plate cells is highest at the time of the plate's
suggested role in the chemotropic guidance of
commissural axons. However, recombinant F-
spondin does not mimic the ability of the floor
plate-derived chemoattractant to promote the
outgrowth of commissural axons from dorsal spi-
nal cord explants. This suggests that F-spondin
may not be involved in the long-range guidance
of commissural axons to the floor plate, at least
through chemotropism.
F-spondin is more likely to be involved in the
contact-dependent guidance of commissural
axons once they reach the ventral midline of the
spinal cord under the influence of distinct che-
motropic guidance cues. The growth cones of
commissural neurons cross the midline by grow-
ing between the basal surface of floor plate cells
and the underlying basal lamina. Floor plate-
secreted F-spondin may accumulate at high levels
in association with the basal surface of floor plate
cells or with the underlying basal lamina, thus
generating a difference in adhesive properties of
the floor plate and the lateral neuroepithelium.
The growth cones of commissural neurons may
adhere preferentially to F-spondin, prompting
them to change trajectory at the boundary of the
floor plate and lateral neuroepithelium. It is also
possible that F-spondin has a more active signal-
ing role that induces changes in the properties of
commissural growth cones, permitting them to
respond to other midline guidance cues.
212
Energy-transducing Membrane Proteins
H. Ronald Kaback, M.D. — Investigator
Dr. Kaback is also Professor of Physiology and of Microbiology and Molecular Genetics in the Molecular
Biology Institute of the University of California, Los Angeles. He received his M.D. degree from the Albert
Einstein College of Medicine, interned at Bronx Municipal Hospital Center, and did postdoctoral research
in physiology at Einstein. Subsequently he conducted research in membrane biochemistry at the National
Heart Institute and the Roche Institute of Molecular Biology, chairing at Roche the Department of
Biochemistry. Dr. Kaback is a member of the National Academy of Sciences and the American Academy
of Arts and Sciences. Among his honors is the Distinguished Alumnus Award from the Albert Einstein
College of Medicine.
THE molecular mechanism of energy trans-
duction in the membranes of living cells is an
enigma. Although the immediate driving force
for many seemingly unrelated processes, such as
active transport, oxidative phosphorylation, and
bacterial motility, is a bulk-phase, transmem-
brane electrochemical H"^ or Na^ gradient, the
molecular mechanism (s) by which free energy
stored in these gradients is transduced into work
or into other forms of chemical energy (e.g.,
ATP) remains unknown. In order to gain insight
into this important basic problem, studies in our
laboratory have focused on an enzyme, the lac-
tose (lac) permease of the bacterium Escherichia
coli, as a paradigm.
The ability of E. coli to accumulate the sugar
lactose and other |8-galactosides against a concen-
tration gradient is dependent upon lac permease,
a very hydrophobic plasma membrane protein
that catalyzes the coupled translocation of a sin-
gle sugar molecule with a single H"^ (i.e., symport
or co-transport) . Under physiological conditions,
lac permease utilizes free energy derived from
downhill translocation of to drive accumu-
lated (S-galactosides against a concentration gra-
dient or, conversely, uses free energy released
from downhill translocation of /J-galactosides to
drive uphill translocation of H^. The polarity of
the reaction reflects the direction of the concen-
tration gradient of the substrate. As such, lac per-
mease represents a huge number of machines that
catalyze similar reactions in virtually all biologi-
cal membranes, from archebacteria to the mam-
malian central nervous system.
The permease is encoded by the lacY gene,
which has been cloned into a recombinant
plasmid and sequenced. By combining overex-
pression with the use of a specific photoaffinity-
labeled substrate for the permease and recon-
stitution of transport activity in artificial
phospholipid vesicles (i.e., proteoliposomes),
the permease was solubilized from the mem-
brane, purified to homogeneity, and shown to cat-
alyze all the transport reactions typical of the fi-
galactoside transport system in vivo with similar
turnover numbers. Therefore, a single enzyme —
the product of lacY — is solely responsible for all
of the translocation phenomena catalyzed by the
iS-galactoside transport system. In addition, evi-
dence has been presented that the permease is
functional as a monomer.
Based on circular dichroic measurements indi-
cating that purified permease is about 80 percent
helical and on hydropathy analysis of the de-
duced amino acid sequence, a secondary struc-
ture was proposed. The model predicts that the
protein has a short hydrophilic amino terminus,
12 hydrophobic domains in a-helical conforma-
tion that traverse the membrane in zigzag fashion
connected by hydrophilic loops, and a 17-
residue hydrophilic carboxyl-terminal tail. Spec-
troscopic, biochemical, and immunological data
are consistent with the general features of the
model and indicate that the amino and carboxyl
termini are on the cytoplasmic surface of the
membrane. Studies on an extensive series of lac
permease-alkaline phosphatase chimeras have
provided strong support for the topological pre-
dictions of the 12-helix model.
This report concentrates on current experi-
ments that involve the use of site-directed muta-
genesis to engineer lac permease so as to permit
certain biochemical and biophysical approaches
to structure-function relationships.
Recent studies provide definitive support for
the argument that cysteinyl residues do not play a
direct role in the lac permease mechanism. Thus,
when site-directed mutagenesis is used to replace
each of the eight cysteinyl residues simulta-
neously, the "C-Iess" permease catalyzes active
lactose transport moderately well relative to
wild-type permease (about 35 percent of the
maximum velocity and 55 percent of the steady-
state level of accumulation). Moreover, active
lactose transport in right-side-out vesicles con-
taining C-less permease is not inactivated by
the alkylating agent, 7V-ethylmaleimide, in dra-
matic contrast to vesicles containing wild-type
permease.
213
Energy-transducing Membrane Proteins
Although site-directed mutagenesis is useful
for delineating amino acid residues that are im-
portant for lactose-H"^ symport and/or substrate
binding and recognition, it has become apparent
that high-resolution structure is required to be-
gin to determine the role of these residues in the
mechanism. Moreover, dynamic information at
high resolution will also be required to solve the
mechanism. In this respect, chemical labeling
and spectroscopic approaches in w^hich reactive
cysteinyl residues are tagged v^^ith radioactive
sulfhydryl reagents, electron paramagnetic la-
bels, or fluorescent probes are potentially power-
ful means for examining static and dynamic
aspects of protein structure-function relation-
ships at high resolution. A principle difficulty
with the general approach, however, is the com-
plexity resulting from the presence of multiple
cysteinyl residues in most proteins, eight in the
case of lac permease. Thus, in addition to the im-
portant conclusion that cysteinyl residues do not
play an important role in the mechanism of lac
permease, the construction of a functional per-
mease molecule devoid of cysteinyl residues pro-
vides the basis for an approach to the analysis of
static and dynamic aspects of permease structure-
function relationships.
By using the lacY gtne encoding C-less per-
mease, for instance, it is now possible to design
mutants in which an individual amino acid resi-
due in a putative hydrophilic or hydrophobic
domain is replaced with a cysteinyl residue.
This can then be reacted specifically with ei-
ther permeant or impermeant sulfhydryl re-
agents in right-side-out or inside-out membrane
vesicles, followed by solubilization and im-
munoprecipitation. In addition, single Cys mu-
tants can be solubilized and purified, tagged
with appropriate electron paramagnetic or fluo-
rescent probes, then reconstituted and studied
spectroscopically.
Finally, it should be possible to study proxim-
ity relationships between transmembrane do-
mains by placing single cysteinyl residues in pairs
of helical domains predicted to lie close to each
other within the membrane. In these contexts, it
is encouraging that more than 200 single Cys re-
placements have been constructed in the C-less
permease and that the great majority of the mu-
tants exhibit highly significant transport activity.
Lactose transport in Escherichia coli. A:
Uphill lactose (Lac ) transport in response to
generated either by respiration or A TP hy-
drolysis. B: Uphill transport in response
to an inwardly directed lactose gradient. C:
Uphill transport in response to an out-
wardly directed lactose gradient.
From Kaback, H.R. 1989. Harvey Lect
83:77-105. Copyright© 1989 Alan R. Liss,
Inc. Reprinted by permission ofWiley-Liss, a
division of John Wiley and Sons, Inc.
214
Control of the Immunoglobulin
Heavy-Chain Gene
Thomas R. Kadesch, Ph.D. — Associate Investigator
Dr. Kadesch is also Associate Professor of Human Genetics at the University of Pennsylvania School of
Medicine. He received his Ph.D. degree in biochemistry from the University of California, Berkeley, where
he studied with Michael Chamberlin. His postdoctoral research was done with Paul Berg at the Stanford
University School of Medicine.
EXPRESSION of immunoglobulin genes is lim-
ited to one cell type, namely B lymphocytes.
Only B cells express the functions required for
gene rearrangement, the process whereby mature
immunoglobulin genes are formed from discrete
gene segments; and only B lymphocytes possess
the components necessary for immunoglobulin
gene transcription, the process that creates an
RNA copy of the rearranged genes.
Within these genes, there are at least two major
transcriptional regulatory elements: the pro-
moter, a DNA domain located close to the site
where transcription begins, and the enhancer,
another DNA domain that stimulates initiation
from the promoter. The activities of each of these
elements are restricted to the B cells and are con-
trolled by proteins that bind them. We have fo-
cused our studies on the proteins binding the im-
munoglobulin heavy-chain (IgH) gene enhancer,
in an attempt to understand how they elicit the
enhancer's B cell-specific activity.
Contributions from a number of laboratories,
including our own, led to a detailed map of the
IgH enhancer. The map indicates the protein-
binding locations and gives some information
about the regulatory mechanism. The enhancer,
while relatively small (200 base pairs), proves to
be exceedingly complex. Interestingly, many of
the perhaps nine or more distinct enhancer-
binding proteins are found in multiple cell types,
even those in which the enhancer is normally in-
active. To explain this, it has been argued that
some of the proteins may serve to stimulate the
enhancer's activity, while others may repress it.
During the past few years, our efforts have been
directed toward the isolation of cDNAs that en-
code these enhancer binders. Thus far, we have
identified six. We are presently using segments of
these genes (cDNAs) to manipulate and character-
ize their encoded proteins both structurally and
functionally.
We have shown that two of the IgH enhancer-
binding proteins, E2-5 and TFE3, are involved in
a fascinating transcriptional regulatory scheme.
In B cells the situation is relatively straightfor-
ward, as both proteins bind the enhancer and act
in concert to stimulate its activity. In non-B cells
the situation is more complicated. In vivo exper-
iments suggest the presence of an additional pro-
tein, a repressor, that binds the enhancer and pos-
sibly precludes E2-5 action.
Binding of this putative repressor has two ef-
fects. First, the enhancer is less active as a result
of the absence of bound E2-5 protein. Second, the
repressor can attenuate at a distance the function
of TEE 3. Hence the presence of the repressor in
non-B cells results in the shutdown of both
E2-5- and TFE3-mediated activation. The effects
of the repressor can be overcome in these cells by
artificially overproducing the E2-5 protein. Pre-
sumably such overexpression is sufficient to dis-
place the bound repressor.
A third gene segment isolated in the laboratory
likely encodes the repressor. Although we have
yet to prove this directly, the encoded protein,
denoted Zeb, is structurally distinct from the
aforementioned proteins that activate transcrip-
tion and, moreover, falls into the family of so-
called zinc finger proteins, many of which have
been shown to repress transcription of other
genes. The DNA sites to which Zeb binds in the
test tube overlap, but are distinct from, those that
the activator proteins bind, yet are identical to
those that confer the repressing activity within
cells. Ongoing experiments should further char-
acterize Zeb's mode of action and possible roles
in other gene systems.
E2-5 can be specifically repressed by yet an-
other protein, a structural relative named Id
(identified and initially characterized by Harold
Weintraub [HHMI, Ered Hutchinson Cancer Re-
search Center]). Unlike Zeb, Id binds directly to
E2-5 (and related proteins), prevents its binding
to DNA, and thus keeps it from stimulating tran-
scription. Id expression was observed to fall off
in several cell lines when they were induced to
differentiate, suggesting that it may serve as a gen-
eral antagonist of cellular differentiation by in-
hibiting DNA-binding proteins specifically re-
quired for differentiation.
The importance of E2-5 for IgH enhancer activ-
ity led to our asking whether Id plays a role dur-
215
Control of the Immunoglobulin Heavy-Chain Gene
ing B lymphocyte maturation. In collaboration
with Stephen Desiderio (HHMI, Johns Hopkins
University School of Medicine), we analyzed Id
mRNA expression in a variety of cell lines repre-
senting different stages of B lymphoid-cell dif-
ferentiation. Only two of these lines, represent-
ing the earliest stages of development, were
Jfound to express Id. We confirmed that these
cells were unable to support the activity of the
IgH enhancer, whereas later-stage cells could.
Hence Id is restricted to early B lymphoid-cell
progenitors, and its presence correlates inversely
with IgH enhancer activity. Presumably Id serves
to keep enhancer activity low in the B cell pro-
genitors. Further differentiation of B cells would
require a decrease in Id expression that, in turn,
would allow the IgH enhancer to become active.
We have shown that Id also plays an important
role in the development of another hematopoi-
etic lineage, the myeloid lineage. In collabora-
tion with Giovani Rovera (Wistar Institute), we
demonstrated that Id is expressed in neutrophil
precursors. When induced to differentiate in cul-
ture, the Id levels in these cells were observed to
decrease, but only transiently. Cells bearing an
artificial, constitutively expressed Id gene failed
to differentiate. Hence an appropriate, transi-
ent shutoflf of Id is required for neutrophil
development.
An important corollary to these results is that
myeloid cells must utilize proteins of the E2-5
family to promote their differentiation. We are
presently attempting to identify and characterize
these proteins.
216
Genetic Control of Hemoglobin Synthesis
Yuet Wat Kan, M.D., D.Sc. — Investigator
Dr. Kan is also Louis K. Diamond Professor of Hematology in the Departments of Laboratory Medicine and
Medicine at the University of California, San Francisco. He received his M.D. and D.Sc. degrees from the
University of Hong Kong Medical School. After internship and residency at Queen Mary Hospital, Hong
Kong, he obtained postdoctoral training in hematology at Peter Bent Brigham Hospital, the Massachusetts
Institute of Technology, Royal Victoria Hospital at McGill University, and the Children 's Hospital, Boston.
Dr. Kan has received numerous honors, including the Gairdner Foundation International Award and the
Albert Lasker Clinical Medical Research Award. He is a fellow of the Royal Society (London ) and a member
of the National Academy of Sciences and of the Academia Sinica (Taiwan ).
THE focus of our research is the molecular
basis of genetic diseases affecting the hemato-
poietic or blood-forming cells. The two diseases
we have studied in depth are sickle cell anemia
and thalassemia. Both result from abnormal glo-
bin production and constitute important health
problems in the Mediterranean region, Africa, the
Middle East, and Asia. In the United States these
disorders occur frequently among people of Afri-
can, Italian, Greek, and Asian descent. We have
defined the mutations that give rise to these de-
fects and devised DNA analyses for their detec-
tion. In addition, we are studying the control of
globin gene expression in red cell precursors and
the signals that switch these genes from fetal to
adult globin production.
Previously we demonstrated that the common
genetic defect in a-thalassemia is deletion of the
a-globin structural gene. We also defined some of
the molecular lesions in jS-thalassemia. These
studies led to our ability to detect thalassemia by
analysis of fetal DNA.
We initiated a new method of linkage analysis
using restriction endonucleases to detect poly-
morphism in DNA sequences and applied it to
tracing the evolution of the sickle and thalasse-
mia mutations. Restriction enzyme site polymor-
phism is now an important tool for detecting
many genetic disorders and for mapping the ge-
netic loci of many diseases.
We developed a method for prenatal diagnosis
of sickle cell anemia and thalassemia. Initially,
fetal samples were required. With the advent of
recombinant DNA technology, mutations in the
human genome can be analyzed directly using
DNA obtained by amniocentesis or chorionic vil-
lus biopsy, permitting early in utero diagnosis of
these conditions.
Prenatal Diagnosis of Sickle Cell Anemia
and Thalassemia
The polymerase chain reaction has made it pos-
sible to diagnose many point mutations rap-
idly. For sickle cell anemia, practical nonradio-
active tests are now available. In the case of
(S-thalassemia, however, the need to diagnose
multiple mutations makes these tests somewhat
tedious, especially in the developing countries
where the disease is common. Hence we are now
devising a rapid approach using the reversed dot
blot principle. Oligonucleotides corresponding
to the mutations common in a given area are im-
mobilized on a filter; the test DNA is amplified
and hybridized to the oligonucleotides; and one
of several nonradioactive methods is then used
for detection. We believe this procedure will fa-
cilitate prenatal diagnosis of i8-thalassemia on a
broad scale.
Control of Globin Gene Expression
We are studying the factors that control the tis-
sue- and development-specific expression of the
human globin genes. Synthesis of the globin
chains is precisely coordinated during develop-
ment. The embryonic e- and f-globins, which are
synthesized in the early embryo, are replaced in
the fetus by the a- and 7-globin chains. Prior to
birth, the /3-globin chain is known to become pre-
dominant over the 7-globin chain; but the factors
that control the expression of the /3-globin gene
in the bone marrow cells and coordinate the ex-
pression of the globin genes during development
have not yet been elucidated. We are now study-
ing the DNA sequences and protein factors that
govern the expression of these genes.
Recently DNA elements that are important for
the control of the tissue-specific and develop-
ment-specific expression of the globin genes
have been revealed. This sequence is known as
the locus control region, or LCR. In the jS-globin
gene, the LCR contains four hypersensitive sites,
which contain consensus sequences that bind
trans-acting factors. We have been studying the
cis sequences, which are important for protein-
DNA interactions in two ways.
First, we have determined by mutation analysis
of these sequences, using both DNA-transfection
and transgenic experiments, that important core
217
Genetic Control of Hemoglobin Synthesis
elements called NFE2-AP1 -binding sites are es-
sential for these enhancing activities. In addition,
using in vivo footprint experiments, we have ex-
amined the consensus sequences that are impor-
tant for in vivo interaction. We have studied sev-
eral cell lines that express different amounts of 6-,
7-, and /3-globin genes and have found that the
consensus sequences are bound differently in
these cell lines, depending on the type of fi-
globin gene expressed. We are now characteriz-
ing the factors responsible for this binding in
order to understand the molecular basis of hemo-
globin switching.
Red Cell Membrane Disorders
Many hereditary hemolytic anemias are accom-
panied by spherocytosis or elliptocytosis. The de-
fects in these disorders lie in mutations of various
cytoskeleton proteins, such as spectrin, ankyrin,
and protein 4.1. We have studied families with
hereditary elliptocytosis due to protein 4.1 defi-
ciency and have defined three different defects
due to gene rearrangements.
Protein 4. 1 is found in all cell types and exists
in multiple isoforms generated by alternate splic-
ing in at least five different regions of the gene.
Some isoforms are more abundant in nucleated
cells, others in mature red cells. Although pro-
tein 4.1 maintains the integrity of the cytoskele-
ton in the red cell, its function in nucleated cells
is not yet known. In a patient who has no protein
4.1 in the red cell, severe hemolytic anemia en-
sues, while other cellular functions are normal.
Structural analysis showed that the protein 4.1
gene is rearranged at the region corresponding to
the amino terminal of the erythroid protein 4.1.
Otherwise it continues to produce functional
protein 4.1 by an alternate splicing mechanism,
which skips the mutated region. This example
may point out an advantage of alternate splicing.
218
Cell Biological Studies of Memory
Eric R. Kandel, M.D. — Senior Investigator
Dr. Kandel is also University Professor of Physiology and Psychiatry at the Center for Neurobiology and
Behavior of the Columbia University College of Physicians and Surgeons. He was born in Vienna, Austria.
He graduated from Harvard College, having majored in history and literature, and received his medical
degree from the New York University School of Medicine. He took postdoctoral training with Wade
Marshall in the Laboratory of Neurophysiology at NIH and with Ladislav Tauc at the Institut Morey in
Paris. Dr. Kandel was the founding director of the Center for Neurobiology and Behavior at Columbia.
He is a member of the National Academy of Sciences and counts among his honors the Lasker Award, the
Gairdner Award, and the National Medal of Science.
LEARNING is commonly divided into two ma-
jor types: declarative and reflexive. Declara-
tive learning refers to the acquisition of informa-
tion about people, places, or things. Reflexive
learning refers to the acquisition of motor skills
and strategies. Our laboratory has been studying
elementary forms of reflexive learning in the gill-
withdrawal reflex of the marine snail Aplysia. In
an attempt to compare these mechanisms with
those underlying declarative forms of learning,
we are also conducting a study of long-term po-
tentiation in the mammalian hippocampus. The
present discussion is limited to reflexive studies
in Aplysia.
We have shown that the gill-withdrawal reflex
can be modified by nonassociative and associa-
tive forms of reflexive learning, giving rise to
both short- and long-term memory, whose dura-
tion is a function of the number of training trials.
We have recently focused on sensitization, a non-
associative form of learning in response to a
noxious stimulus. To analyze the relationship be-
tween the memory for short- and long-term sensi-
tization, we have studied in particular detail one
component of the neural circuit of this reflex: the
connections between the siphon sensory neuron
and the gill motor neurons. These connections
can be studied both in the intact animal and in
dissociated cell culture.
With sensitization the connections undergo an
increase in synaptic effectiveness (facilitation),
whose duration is a function of the number of
behavioral reinforcing stimuli to the tail. Simi-
larly, in culture, the duration of the facilitation is
a function of the number of applications of seroto-
nin (5-HT), a modulatory transmitter released by
tail stimuli . A single tail stimulus or a single pulse
of 5-HT produces short-term facilitation lasting
minutes, whereas continuous application of 5-HT
for 1 .5 h or four or five pulses over a 1 .5-h period
elicit long-term facilitation lasting one or more
days.
In both the behavioral reflex and the monosyn-
aptic facilitation, there is a parallel requirement
for protein and mRNA synthesis for long-term but
not for short-term effectiveness. The short-term
process reflects enhanced transmitter release
from preexisting synaptic connections due to co-
valent modification of preexisting proteins. The
long-term process requires new protein synthesis
and leads to the grovvT:h of new synaptic
connections.
Proteins and mRNAs necessary for long-term
memory must either be induced during the brief
(1 .5-h) time window of training or be constituti-
vely expressed and transiently accessible to cova-
lent modification during this period. The experi-
ments described below were designed to
distinguish between these two possibilities.
In earlier experiments, Ari Barzilai, Tim Ken-
nedy, David Sweatt, and I found that repeated
pulses of 5-HT induced changes in the synthesis
of specific proteins in the sensory neurons. Simi-
lar changes in protein synthesis can be produced
by cAMP, a second messenger activated in the sen-
sory neurons by 5-HT. These proteins could there-
fore reflect the transcription of cAMP-inducible
genes. In mammals, genes induced by cAMP share
a control element called the cAMP recognition
element (CRE), which binds transcriptional acti-
vators called the CRE-binding proteins (CREBs) .
Aplysia neurons contain proteins homologous to
mammalian CREBs.
These findings raise three questions that Bong-
Kiun Kaang, Seth Grant, and I attempted to ad-
dress: Can the facilitating transmitter 5-HT in-
duce transcriptional activation of reporter genes
in the sensory neurons that are driven by the CRE?
Does this transcriptional activation correlate
with the graded induction of long-term facilita-
tion? and How does 5-HT activate transcription
by CREB?
Because the CRE will also confer cAMP induc-
ibility when placed upstream of reporter genes
such as (8-galactosidase, these gene constructs can
serve as an assay system for transcriptional induc-
tion by cAMP. Kaang, Grant, and I therefore mi-
croinjected a CKE-lacZ reporter plasmid into
219
Cell Biological Studies of Memory
Aplysia sensory neurons and measured the levels
of expression driven by the CRE in response to
stimulation by 5-HT. We next exposed the neu-
rons to five repeated (5-min) pulses of 5-HT, the
protocol that produces long-term synaptic facili-
tation, and found a fourfold induction of the re-
porter gene. Similarly, cAMP induced expression
by 3-4-fold. Thus repeated pulses or prolonged
exposure to 5-HT stimulates transcription, and
this stimulation can be simulated by elevating lev-
els of cAMP. The increase in lacZ expression is
mediated specifically through the CRE sequence.
A plasmid (ACRE-/«cZ) from which the CRE ele-
ment is deleted shows no expression.
Does the stimulation of transcription mediated
by 5-HT depend on the binding of transcription
factors to the CRE sequence? To address this ques-
tion, we co-injected the CRE reporter construct
with either an oligonucleotide that encodes the
wild-type CRE sequence and binds CREB, or with
a mutant oligonucleotide that does not bind
CREB-like factors. The wild-type oligonucleo-
tide, which binds to and titrates out CREB,
blocked the 5-HT-induced expression of the re-
porter construct, whereas the mutant oligonucle-
otide produced no inhibition.
These experiments suggest that the induction
mediated by 5-HT requires positively acting
CREB-like cellular factors that interact with the
CRE, and provide direct evidence that 5-HT can
stimulate transcription of genes containing the
CRE sequences. This is consistent with results
that Pramod Dash, Binyamin Hochner, and I ob-
tained last year in physiological experiments, in
which we microinjected the CRE into the nucleus
of sensory neurons and selectively blocked the
long-term increase in synaptic effectiveness with-
out affecting short-term facilitation. Both these
sets of results suggest that CREB-like transcrip-
tional activators are required for the induction of
long-term facilitation.
Does the activation of transcription by 5-HT
have a clear threshold or is it graded? To assess
this question, Kaang, Grant, and I were in-
fluenced by our earlier findings discussed above,
that a single (5-min) pulse of 5-HT will only
elicit short-term facilitation. To generate signifi-
cant long-term facilitation, four or five pulses are
required. Using the same pulse protocol, we
found that a single pulse of 5-HT does not, in fact,
stimulate expression of the reporter gene,
whereas four to five pulses of 5-HT/IBMX pro-
duced a sevenfold stimulation. An intermediate
number of pulses (two or three) gave an interme-
diate level (twofold) of stimulation. This correla-
tion, between the number of 5-HT trials and the
level of transcriptional induction and facilita-
tion, suggests that the graded nature of long-term
facilitation may reflect the graded nature of tran-
scriptional induction.
Does activation by 5-HT require that CREB be
phosphorylated? If so, must activation of CREB in
the sensory neuron be mediated by protein kinase
A (PKA)? To address these questions, Kaang,
Grant, and I next microinjected two constructs
into the sensory neurons: a reporter gene as well
as a chimeric transactivation plasmid that ex-
presses CREB. Following co-injection, exposure
to 5-HT produced a 10-fold stimulation of tran-
scription. By contrast, injection of either trans-
activation or reporter plasmid alone showed
no expression. This transcriptional stimulation
depends on phosphorylation of the PKA consen-
sus site at Ser' in CREB. Microinjecting a transac-
tivation plasmid (pSAl 19) containing a mutation
that converts Ser"^ to Ala"^ showed no stimula-
tion with 5-HT.
Thus long-term facilitation induced by 5-HT
leads to activation of CRE-inducible genes. The
induction of these genes by the cAMP cascade is
graded and can be initiated by PKA through phos-
phorylation of CREB. A conventional modulatory
transmitter, 5-HT, can therefore select either a
cytoplasmic or a genomic program of cellular ac-
tion, depending on the number of presentations
of 5-HT. By being able to activate a nuclear signal
(through the phosphorylation of CREBs and its
action on the CREs) , modulatory transmitters can
activate transcription and thereby take on the
properties of growth factors. In the case of sen-
sory neurons, transcriptional activation of CREB-
related proteins seemed to represent one compo-
nent of the switch, which extends the short-term
cytoplasmic process for synaptic facilitation into
the genomic changes characteristic of the long-
term facilitatory process — one that includes the
growth of new synaptic connections.
220
The T Cell Repertoire
John W. Kappler, Ph.D. — Investigator
Dr. Kappler is also a member of the Division of Basic Immunology of the Department of Medicine at the
National Jewish Center for Immunology and Respiratory Medicine, Denver, and Professor of Microbiology
and Immunology and of Medicine at the University of Colorado Health Sciences Center, Denver. He was
educated at Lehigh University and received his Ph.D. degree in biochemistry at Brandeis with Gordon
Sato. He did postdoctoral work at the University of California, San Diego, with Richard Dutton. After
holding faculty positions at the University of Rochester, he moved to his present position at the National
Jewish Center. He was awarded the Wellcome Foundation Prize by the Royal Society and is a member of
the National Academy of Sciences.
AS protection against invasion by foreign or-
ganisms, higher animals have evolved a
complex collection of cells and chemicals
broadly termed the immune system. Components
of this system are able to recognize foreign mate-
rial in the body and give rise to a series of events
that cause the destruction or inactivation of the
invader. Cells arising in the thymus, the T lym-
phocytes, are central to the efficient function of
this system.
T cells bear receptors on their surfaces that are
able to interact specifically with foreign material.
Such interaction stimulates these cells to pro-
duce chemicals that allow other cells, and the T
cells themselves, to respond to the invader. There
are two kinds of T cells, bearing or 76 recep-
tors respectively. The a/? receptors are made up
of several segments — Va, Ja, VjS, D/3, and J(8 —
each of which can differ in structure from one T
cell to another. This is possible because the DNA
of higher mammals contains a number of alter-
nate genes for each of these segments. As each T
cell develops, it selects a different combination
of these genes, and therefore eventually ex-
presses receptors that are not exactly the same
in structure as those of its fellows. It is these varia-
tions in a/3 receptor sequence that enable one T
cell to recognize influenza virus, for example,
and another poliovirus.
In order for T cells to recognize most foreign
materials, they must bear exactly the right combi-
nation of variable segments — Va, Ja, etc. The
proper combinations are usually rare, so when an
animal is confronted with an invading organism,
the cells that can actually recognize the invader
are few, probably about one in 100,000 or one in
a million of all T cells. This fact does not hold
true, however, for special types of foreign mate-
rial called superantigens.
Superantigens bind to special cell-surface mol-
ecules— class II proteins of the major histocom-
patibility complex (MHC). They then interact
with the VjS portion of the T cell receptor, almost
without regard to the composition of its other
variable elements. Since there are only about 75
different sequences for V(8 in humans, any given
superantigen will, theoretically, react with at
least 1 percent of all T cells. In the mouse, with
fewer different V/3 sequences, superantigens
react with at least 5 percent of T cells. In fact, a
particular superantigen can react in some cases
with up to 30 percent of all T cells in either of
these species.
The fact that superantigens can react with so
many T cells causes them to have some important
pathogenic properties. For example, massive
stimulation of T cells by superantigens causes
toxic shock in humans, and there is reason to be-
lieve that these antigens may be involved in cer-
tain autoimmune diseases.
Our laboratory has recently been studying the
interaction among superantigens, the V/3 portions
of T cell receptors, and the class 11 MHC proteins.
A staphylococcal toxin, SEB, has been used as a
model for these experiments. Amino acids have
been identified in SEB that control the binding of
this protein to V/8 or class II.
In many cases proteins that must bind to two
different target molecules express their binding
sites in different domains. That is, the different
functions of the protein are separated spatially.
Surprisingly, this does not seem to be the case for
SEB. The binding sites of this protein for V(8 and
class II appear to be interwoven, as though they
must lie in close proximity.
The structural studies on SEB have led to the
creation of a collection of mutant SEBs, some able
to bind class II MHC but not V/3, some with a more
limited range of V/3 specificities than SEB itself,
and some able to bind neither MHC nor the T cell
receptor. These mutant SEBs are now being
screened as vaccines. Mice preimmunized with
the mutant molecules are no longer sensitive to
the toxic effects of SEB given later.
Our laboratory and others have discovered a
second class of superantigens, encoded by retrovi-
ruses that cause mammary tumors in the mouse.
Although the genes coding for these superanti-
gens have been known for some time, the struc-
221
The T Cell Repertoire
ture and function of their protein products have
only recently been determined. They turn out to
be unusual proteins. Although they are bound to
the surface membranes of cells, they differ from
most surface proteins, in that the amino acids of
their carboxyl terminal rather than amino termi-
nal lie outside the cell. Only a small amino-
terminal stretch of the protein lies inside. Experi-
ments show that this inside portion is not
essential for the superantigenic properties of the
viral proteins, but that the proteins must be mem-
brane bound in order to engage T cell receptors.
The extreme carboxyl-terminal set of amino acids
are among those that bind to T cell receptor V/Js.
In attempts to find out more about how these
viral superantigens operate, monoclonal antibod-
ies have been raised to various parts of the pro-
teins. The antibodies reveal that only a few of
these proteins are expressed on the cell
surface — probably about 1 ,000 per cell. Despite
this small number, the viral superantigens are ex-
tremely effective stimulators of T cell responses.
222
The Genetic Control of Morphogenesis
Thomas C. Kaufman, Ph.D. — Investigator
Dr. Kaufman is also Professor of Genetics in the Department of Biology at Indiana University,
Bloomington, and Adjunct Professor of Medical Genetics in the Department of Medical Genetics at Indiana
University Medical Center. He received his M.A. and Ph.D. degrees from the University of Texas, Austin,
and did his postdoctoral research at the University of British Columbia in Vancouver.
THE long-term goal of our laboratory is to con-
tribute to an understanding of the genetic
basis of the developmental program of higher or-
ganisms. The organism chosen for our studies is
the fruit fly Drosophila melanogaster, and our
principal focus is a set of genes called homeotic,
which play a crucial role in development.
The homeotic genes were first identified by
virtue of the striking phenotypes observed when
flies carried mutations at these loci. Specifically
these homeotic lesions cause one portion of the
animal to be transformed into an identity nor-
mally found in another region. Thus mutations at
the Antennapedia locus cause a transformation
of the antennae of the adult fly into a leg, and
lesions in the proboscipedia gene result in the
development of legs in place of the adult mouth-
parts. Both genes are members of a cluster of five
homeotic genes called the Antennapedia com-
plex (ANT-C), which is found in a restricted
domain at the base of the right arm of chromo-
some 3.
The aggregate results of genetic, developmen-
tal, and molecular analyses of the ANT-C have re-
vealed that the role of the resident loci is best
viewed as a series of developmental switches for
either/or decisions of cellular fate in the embry-
onic and larval stages of the organism. Further-
more, DNA sequence analysis of the homeotic
loci reveals that they encode proteins containing
a motif, dubbed the homeodomain, that endows
the proteins with DNA-binding ability. Indeed,
the homeotic proteins are found complexed with
the nuclear DNA of the cells in which they are
expressed. Thus it appears that the switch activity
of the loci is reflected in their functioning as regu-
lators of specific target genes. Not entirely clear
at this point is how each homeotic locus is re-
stricted to its own unique pattern of expression
and which sets of genes are the targets of the
switches.
In order to investigate these two unknowns, we
have concentrated our efforts on three of the resi-
dent members of the ANT-C: Sex combs reduced
(5cr), proboscipedia (pb), and labial (lab).
Each gene was chosen for the unique properties it
displayed during initial characterization. For ex-
ample, Scr is the only homeotic gene expressed
at the juncture between the head and trunk of the
developing animal, and there were indications
that the regulatory hierarchy of genes expressed
in these two domains is different. Additionally, a
genetic analysis revealed pb and lab to be small
by homeotic standards. This meant that for the
two genes, a complete dissection of the regula-
tory elements of each locus was feasible.
The Sex combs reduced Gene
Our prior genetic and molecular analyses of
the Scr locus had shown that DNA sequences 50
kilobases (kb) distal to the point at which the
RNA product is initiated were necessary for nor-
mal gene expression. Using "enhancer sniffers" —
constructs capable of detecting DNA fragments
that have the ability to regulate gene expression
— we surveyed the entire Scr locus for such ele-
ments. To date we have found at least five frag-
ments that specify the accumulation of gene
product in the posterior head and anterior
thorax. These are scattered over a 30-kb interval;
consistent with our earlier genetic results, the re-
gions of DNA that control head and trunk expres-
sion appear to be physically separate.
Genetic analysis has also shown that Scr is sub-
ject to the regulatory effects of "transvection."
Normally the gene is only expressed in the first,
or most anterior, thoracic segment; however, cer-
tain mutants in the gene allow its product to ac-
cumulate more posteriorly, in the second and
third thoracic segments. We have shown that the
normal restriction of pattern results from nega-
tive regulation and requires that the two copies of
the gene in the cells of the posterior thorax be
paired with each other. If this pairing is
disrupted, the negative effect is removed and ab-
normal ectopic expression occurs.
Using sniffer constructs similar to those above,
we have identified three DNA fragments that ap-
pear to be associated with the transvection effects
at the Scr locus. Two of these three elements are
located approximately 40 and 10 kb upstream of
the gene, while the third is located within an in-
223
The Genetic Control of Morphogenesis
tron about 1 5 kb downstream of the transcription
start site. These analyses of Scr are supported by a
grant from the National Institutes of Health.
The location of these three elements relative to
the identified enhancer elements is intriguing.
Two flank the enhancers, while the third is lo-
cated in the midst of the regulatory elements. Our
current hypothesis is that the transvection ele-
=^-ments serve to define a chromatin domain con-
trolling an on/off state for the locus and that the
resident enhancers can only have their effects in
the on state. We are currently extending our anal-
ysis to define more precisely the boundaries and
sequences associated with the above-identified
fragments. We also hope to identify further the
cellular factors that act through these elements.
The labial Gene
Using enhancer sniffers, we have identified a
majority of the sequences needed for the normal
expression of the lab gene. These are found up-
stream, either within 3 .6 kb of the transcription
start site or within the major 1 4-kb intron. Using a
minigene construct that contains all of the up-
stream sequences but lacks the intronic ele-
ments, we have succeeded in rescuing the lethal-
ity and morphological anomalies associated with
deletion of the lab gene in the embryo. Thus it
would appear that the regulatory elements found
within the intron are not necessary to lab's nor-
mal embryonic functions.
Although the lab minigene rescues the embry-
onic defects associated with /aft-deficient geno-
types, there is no apparent rescue of adults. We
have shown that this failure does not result from
the minigene's inability to be expressed in adult
tissues, but rather because its protein product is
inappropriately and ectopically accumulated.
This abnormal expression pattern only takes
place in lab mutant animals and is not detected in
normal minigene-bearing hosts.
We have found that the normal lab gene actu-
ally encodes two polypeptides, one of which is
six amino acids longer than the other. The mini-
gene is only capable of directing the synthesis of
the shorter protein. It would appear, therefore,
that animals capable of making the long form
show correct adult expression, while short-form
animals cannot, and that normal adult expression
involves autogenous negative regulation.
Our earlier studies revealed that lab expres-
sion in the embryo also involved autogenous regu-
lation. In that case, however, the effect was posi-
tive. That is, lab protein served to keep the lab
gene turned on, not to turn it off. Moreover, the
short-form protein was capable of performing
this function.
We have now demonstrated that the sequences
at the lab locus necessary for the positive and
negative loops are, like the required proteins,
different. The positive sequences are upstream of
the transcription initiation site, while the nega-
tive targets are in the transcribed portion of the
gene in the first exon. We are currently investi-
gating the possibility that the two loops are af-
fected by the direct interaction of the lab protein
isoforms with the two alternate target sites in the
lab gene itself.
The proboscipedia Gene
As in the case of lab, we have used a minigene
derived from the pb locus and have obtained full
rescue of the adult homeotic mutant phenotype.
Moreover, fragments extracted from one of the
introns and placed upstream of the pb promoter
function to direct a normal spatiotemporal pat-
tern of expression. Thus sequences both up- and
downstream of the transcription start site are re-
quired for normal pb expression.
However, tests of the "enhancer" sequences
on a heterologous promoter element in the
sniffer constructs demonstrate no activity. On the
other hand, minigene constructs in which the en-
hancer elements are deleted can be directed by
novel "enhancer" elements derived from other
loci. Therefore it appears that there are specific
promoter/enhancer interactions at this locus and
that the specificity appears to lie primarily with
the "enhancer" elements.
We have also found a second regulatory ele-
ment at the pb locus upstream of the transcrip-
tion start site. However, this element, unlike the
positive enhancer above, appears to be negative
and serves to prevent ectopic expression of the
gene. Moreover, like the Scr locus, it seems to
have a pairing-sensitive component. We are now
beginning a dissection of this system, which
offers an opportunity to make direct comparisons
between two similar regulatory schemes in the
ANT-C. Our work on the proboscipedia gene is
supported by a grant from the National Institutes
of Health.
224
Protein Folding and Macromolecular
Recognition
Peter S. Kim, Ph.D. — Assistant Investigator
Dr. Kim is also Member of the Whitehead Institute for Biomedical Research, Associate Professor of Biology
at the Massachusetts Institute of Technology, and Assistant Molecular Biologist at the Massachusetts
General Hospital, Boston. His undergraduate degree in chemistry was obtained at Cornell University,
where he studied with George Hess. After receiving the Ph.D. degree in biochemistry from Stanford
University, where he studied with Robert Baldwin, Dr. Kim moved to the Whitehead Institute for
Biomedical Research as a Whitehead Fellow.
INFORMATION transfer in biology generally
proceeds from DNA to RNA (transcription) and
then from RNA to protein (translation) . The lin-
ear, unfolded protein chains made during transla-
tion must fold into a three-dimensional shape to
be functional. Although the basic mechanisms of
transcription and translation are understood, at
least in outline, the transfer of information from
one to three dimensions — i.e., protein folding —
remains a major unsolved problem in molecular
biology. To understand protein folding is a prime
objective of this laboratory.
A second effort is aimed at understanding the
principles of macromolecular recognition: spe-
cific protein-protein interactions and interac-
tions between protein molecules and DNA. These
interactions are central to much of molecular
physiology and developmental biology. We have
focused on a structural motif called the leucine
zipper, which occurs in several different DNA-
binding proteins, including the products of some
nuclear oncogenes.
A third and new effort is the de novo design of
peptides and proteins.
Protein Folding
Much of our work in this area is centered on
bovine pancreatic trypsin inhibitor (BPTI) , argu-
ably the protein most thoroughly characterized
in biophysical terms. It is difficult to determine
the structures of protein-folding intermediates,
because protein folding is a cooperative process.
Trapped disulfide-bonded intermediates, such as
those identified in the early folding steps of BPTI,
are often rather insoluble; this hinders detailed
structural characterization by nuclear magnetic
resonance (NMR) . We have developed a peptide
model approach that circumvents the problem of
cooperativity and improves solubility, so that the
structures contained within protein-folding in-
termediates can be characterized in detail.
Peptide models that simulate two crucial early
intermediates in the folding of BPTI have been
designed and synthesized chemically. By using
two-dimensional NMR, we find that the struc-
tures contained within these peptide models are
remarkably native- like, corresponding to subdo-
mains of BPTI. These results suggest that a large
part of the protein-folding problem can be re-
duced to identifying and understanding subdo-
mains of native proteins.
Earlier work by others, however, concluded
that there are well-populated, nonnative states in
the oxidative folding of BPTI. This conclusion
complicates efforts to understand protein fold-
ing. Recently we reexamined the spectrum and
population of intermediates present during the
folding of BPTI, taking advantage of improve-
ments that have been made in separation technol-
ogies in the years since the original BPTI-folding
experiments. In contrast to earlier studies, we
find that all of the well-populated intermediates
in the folding of BPTI contain only native disul-
fide bonds and that the essential features of the
BPTI-folding reaction are determined in large
part by native structure. These results emphasize
the importance of native protein structure for un-
derstanding protein folding.
A recombinant model for an early predominant
intermediate, containing a single disulfide bond
between residues 5 and 55, has been made by
replacing the cysteines not involved in the disul-
fide bond with alanine. Remarkably, this model
folds essentially into the same conformation as
native BPTI, as judged by two-dimensional NMR,
and it inhibits trypsin. These findings provide an
explanation for the properties of this interme-
diate in the folding of BPTI and demonstrate that
the native fold of BPTI can be obtained without
the assistance of nonnative disulfide species. The
recombinant model also provides an attractive
model system for studies of protein folding. This
work is supported by a grant from the National
Institutes of Health.
Other efforts are directed at evaluating electro-
static fields at the ends of a-helices and develop-
ing a model system to evaluate the /3-sheet pro-
pensities of different amino acid residues.
Macromolecular Recognition
In this area, we have focused on the leucine
zipper class of DNA-binding transcriptional acti-
225
Protein Folding and Macromolecular Recognition
vator proteins, originally identified by Steven
McKnight and his co-workers (HHMI, Carnegie
Institution) . The leucine zipper regions of these
proteins are important for homodimer or specific
heterodimer formation.
Our approach in this work is to use "protein
dissection." GCN4, a homodimeric transcription
factor, serves as a prototype protein. A synthetic
peptide corresponding to the 33-residue leucine
zipper region folds as a parallel pair of helices.
This led us to propose that leucine zippers are
actually short coiled coils. X-ray crystal lographic
studies of this peptide (with Tom Alber's group,
University^ of California, Berkeley) confirm that
the leucine zipper of GCN4 is a coiled coil and
provide the first high-resolution structure of a
two-stranded parallel coiled coil. The effects of
amino acid replacements on the stability, struc-
ture, and dynamics of this leucine zipper are be-
ing investigated.
Proper biological function requires that recog-
nition between many different macromolecules
in the cell occurs with exquisite specificity. We
have found that the isolated leucine zipper re-
gions from the nuclear oncogene products Fos
and Jun are sufficient to mediate specific hetero-
dimer formation. This provides a very simple
model system for studying the specificity of
protein-protein interactions: two helices that
prefer to interact with each other rather than with
themselves. By making hybrid leucine zipper
peptides, we found that eight amino acid resi-
dues from each of the leucine zipper sequences
are sufficient to mediate specific heterodimer
formation. The predominant mechanism for spec-
ificity in this system was found to be electrostatic
in nature.
A region of GCN4 rich in basic amino acid resi-
dues, immediately adjacent to the leucine zipper,
is involved in DNA recognition. We find that this
basic region by itself, when dimerized via a flexi-
ble disulfide linker in place of the leucine zip-
per, is also capable of sequence-specific DNA
binding. In addition to simplifying structural
analysis of this new DNA-binding motif, the find-
ing provides a new strategy for the design of DNA-
binding peptides. This work was supported by a
grant from the National Institutes of Health.
Peptide and Protein Design
Knowledge of the rules involved in protein
folding and macromolecular recognition can be
tested by trying to design de novo amino acid
sequences that fold into specific conformations
and/or that interact in a predetermined manner
with other molecules. Toward this end, we have
begun to analyze the structural hierarchies in nat-
ural proteins and to design simple "building
blocks" of peptide structure.
226
RNA Viral Genetics
KarlaA. Kirkegaard, Ph.D. — Assistant Investigator
Dr. Kirkegaard is also Assistant Professor of Molecular, Cellular, and Developmental Biology at the
University of Colorado at Boulder and Adjunct Assistant Professor of Cellular and Structural Biology at the
University of Colorado Health Sciences Center, Denver. After receiving a B.S. degree in genetics from the
University of California, Berkeley, she developed her doctoral thesis in the Department of Biochemistry and
Molecular Biology at Harvard University with James Wang. Her postdoctoral work in virology was in
association with David Baltimore at the Massachusetts Institute of Technology and the Whitehead Institute.
FOR numerous viruses and other subcellular
parasites, RNA rather than DNA is the mole-
cule used for storage and transmission of genetic
information. We are interested in the conse-
quences for a virus of having an RNA genome and
in any mechanistic similarities or differences in
genetic processes between RNA and DNA organ-
isms. In addition, we are exploring the mecha-
nisms of RNA packaging, replication, and recom-
bination in the genome of poliovirus and other
RNA viruses. We are also interested in the interac-
tions between viruses and their host cells, espe-
cially in the realm of RNA-protein biochemistry.
Many of our genetic studies utilize poliovirus,
a small icosahedral virus with an RNA genome of
only 7,500 nucleotides. We have shown, for ex-
ample, that RNA recombination occurs among
poliovirus genomes with sufficient frequency
that 1 out of every 25 is a recombinant. In con-
trast to the breaking and joining of molecules
that leads to DNA recombination, recombination
of RNA occurs during its synthesis. Genetic re-
arrangement results from the switching of paren-
tal templates by the viral RNA polymerase.
Thale Jarvis, a postdoctoral fellow in our labo-
ratory, has developed a sensitive quantitative as-
say for RNA recombination using the polymerase
chain reaction (PGR). Recombinants can be de-
tected at a frequency as low as 1 in 1 0*^ parental
genomes. Using this assay, we have been able to
learn a great deal about the process of RNA recom-
bination in poliovirus-infected cells. For exam-
ple, we discovered that recombination frequency
increases exponentially throughout the course of
a single infectious cycle. It was surprising that
the extensive cytopathic changes during the
course of poliovirus infection do not inhibit the
access of the parental RNA templates to each
other. This work was also supported by a grant
from the National Institutes of Health.
Taken together with other data we have ob-
tained, our results suggest that the process by
which recombinant RNA genomes are generated
may be quite simple, and possibly universal
among RNA viruses. We are using the PGR assay to
test this idea by looking for RNA recombination
among other RNA viruses that are less amenable
to genetic analysis than poliovirus. We hope to
gain a better understanding of the prevalence and
mechanism of genetic recombination among RNA
genomes, a process that is certainly responsible
for much of the variability and rapid evolution of
RNA viruses.
Using x-ray crystallography, James Hogle at
Scripps Glinic has determined the three-dimen-
sional structure of the poliovirion. However, an
appreciation of functional interactions between
the viral RNA and the virion proteins calls for the
application of genetics as well as structural bio-
chemistry. We do not know, for example, exactly
which subviral protein particles package the po-
liovirus RNA into the final virion structure, nor
do we know the structural requirements of the
participants in the packaging reactions. Is the
viral RNA threaded into an intact, preformed ico-
sahedral capsid, or do smaller parts of the capsid
condense around the viral RNA to form the final
icosahedral structure?
We have constructed several mutants in the po-
liovirus RNA genome and have characterized
them in great detail. Two of these mutants have
pointed out a region of the viral capsid, quite
internal to the virion, that is involved both in RNA
packaging and RNA uncoating. We are examining
the RNA-binding properties of subviral particles
from both mutant and wild-type poliovirus-
infected cells in hopes of identifying which sub-
viral particles bind to RNA during intracellular
assembly.
To investigate the RNA uncoating defects dur-
ing cell entry of these mutant viruses, we have
devised an assay to detect interactions between
poliovirions and their cellular receptor directly.
The laboratory of Vincent Racaniello (Golumbia
University) has identified the poliovirus receptor
as a cellular adhesion protein of the immunoglob-
ulin superfamily. Using antibodies prepared
against peptides from the receptor's amino and
carboxyl termini, we can analyze proteolytic di-
gestion patterns of the receptor in the presence
227
RNA Viral Genetics
and absence of bound wild-type and mutant po-
liovirions. In this way we are exploring the spe-
cific protein-protein contacts made betu^een the
viruses and the receptor during binding and the
subsequent conformational changes during entr}'
of the virus into the cell.
We have extended the study of RNA genetics to
she interactions between yeast cells and the small
double-stranded RNA genome of L-A, a cytoplas-
mic virus-like particle. Yeast cells, unlike the pri-
mate cells in which poliovirus and other RNA vi-
ruses of medical interest are propagated, are
amenable to elegant genetic analysis, making it
possible to identif\' quite quickly the cellular
molecules that are involved in any given process.
The virus-like particle L-A has been shown by the
laboratory^ of Reed Wickner (National Institutes
of Health) to replicate in the cytoplasm of yeast
cells in large numbers. Genetic analysis of L-A,
however, will depend on our ability to make de-
fined mutations in its RNA genome.
In the past year, we have developed the techni-
cal capabilities to pursue this goal. First, we had
to develop the technology to introduce RNA di-
rectly into yeast cells, which we accomplished by
using RNA molecules encoding the luciferase
protein of fireflies. Second, we needed an assay
for the successful introduction of the L-A genome
into yeast. Since the particle is so ubiquitous, no
phenotype for its presence or absence in a yeast
cell had been reported. We found that, under
somewhat unusual laboratory conditions, certain
yeast strains require the L-A genome for growth!
This suggests the possibility of mutualistic rela-
tionships between host cells and virus-like RNA
genomes, a possibility we are also pursuing in
other systems.
Most importantly for this work, we now have a
phenotype for the presence of the L-A RNA ge-
nome in yeast cells. We have recently been able
to initiate L-A "infections" in yeast by introduc-
ing RNA made in vitro. We are now in a position
to mutate this RNA genome, to study any defect in
the L-A replication cycle caused by the mutations,
and to observe the effects of yeast mutants on the
replication cycle of mutant and wild-type RNA.
228
Adrenergic Receptor Structure and Function
Brian K. Kobilka, M.D. — Assistant Investigator
Dr. Kobilka is also Assistant Professor of Medicine, Cardiology, and Molecular and Cellular Physiology at
the Stanford University Medical Center. He received his undergraduate degree in biology and chemistry
from the University of Minnesota, Duluth, and his M.D. degree from Yale University. After his residency
in internal medicine at Barnes Hospital, St. Louis, he joined the laboratory of Robert Lefkowitz as a
research fellow in cardiology at Duke University. Four years later he was appointed Assistant Professor in
the Department of Medicine at Duke University, and the following year he assumed his present position
at Stanford.
THE autonomic nervous system serves as the
master control center for the cardiovascular
system. It monitors the effectiveness of the latter
system in providing nutrients and oxygen to the
rest of the body and appropriately adjusts the
heart rate, blood pressure, and blood flow. These
adjustments are made via nerves that serve the
heart, blood vessels, and kidneys.
Adrenergic receptors form the interface be-
tween these nerves (of the sympathetic subsys-
tem) and the organs they innervate. Catechol-
amines released from sympathetic nerve
terminals bind to adrenergic receptors on the sur-
face of target cells, and the activated receptors
modify the function of these cells.
When a catecholamine occupies its binding
site, the receptor activates a GTP-binding protein
(G protein) inside the cell. The activated G pro-
tein may then modulate the activity of a cellular
en2yme or ion channel. The genes (or corre-
sponding cDNAs) for nine types of adrenergic re-
ceptors have been cloned. There are three types
of ttj-adrenergic receptors, three types of
adrenergic receptors, and three types of |8 recep-
tors. All of these receptors are structurally simi-
lar, having seven hydrophobic domains that are
thought to be membrane spanning. These fea-
tures are shared by other receptors that activate G
proteins.
a2-Adrenergic Receptor Subtype Diversity
The role played by the aj-, a2-, and /S-adrener-
gic receptors in the function of the sympathetic
nervous system has been extensively studied.
Adrenergic receptors are involved in blood pres-
sure control and in directing blood flow to spe-
cific tissues. However, the physiological role of
each of the three a2 receptor types is not known.
This is in part due to the lack of highly selective
drugs that can activate or inhibit each type of
adrenergic receptor. We are attempting to iden-
tify distinctive functional and physiological prop-
erties for each of the different a2 receptors. These
studies may provide incentive for the develop-
ment of more highly selective a2'^drenergic re-
ceptor drugs for the treatment of hypertension
and vascular disease.
Adrenergic Receptor Structure
A major focus in my laboratory is to learn more
about the three-dimensional structure of adrener-
gic receptors and to determine how they transmit
signals across the cell membrane's lipid bilayer.
We are taking several approaches to study the re-
ceptor structure. Over the past year our mutagen-
esis studies have identified a specific amino acid
in the /Jj-adrenergic receptor that forms part of
the binding site for a large class of /3 receptor-
blocking drugs. When this amino acid is placed in
an a2 receptor, it confers the ability to bind to
these (8 receptor drugs. We have also used muta-
genesis studies to identify intramolecular interac-
tions within ^2- ^rid a2-adrenergic receptors. This
information is useful in developing models for
the three-dimensional structure of the receptor.
A long-range goal is to study the three-
dimensional structure of the /32-3drenergic re-
ceptor, using techniques that provide high-
resolution pictures of this protein or detect
changes in the structure of the receptor during
signal transduction. These techniques require
large quantities of pure, functional receptor pro-
tein. Using recombinant DNA techniques, we
have made several modifications in the structure
of the /32-adrenergic receptor that lead to in-
creased expression in cultured cells. Our goal
during the next year will be to refine the purifica-
tion procedure to optimize the yield and mini-
mize the cost of large-scale production of recep-
tor protein.
Receptor Biosynthesis
The primary amino acid sequence of a receptor
contains all the essential information needed for
the receptor's proper folding, post-translational
processing, and cellular targeting. Understanding
the process by which receptors are folded and
processed should provide insights into receptor
229
Adrenergic Receptor Structure and Function
structure and may identify factors that will en-
hance the production of functional protein.
Using a cell-free expression system developed
in my laboratory, we have characterized the in-
sertion of specific membrane-spanning domains
into the lipid bilayer. The proper insertion of the
first membrane-spanning domain is relatively in-
efficient. Using recombinant DNA techniques,
we have identified a structural modification that
enhances the integration of this domain and in-
creases the amount of functional ^2 receptor pro-
tein expressed in cultured cells.
We have previously shown that ^2 receptor is
nonfunctional immediately after translation and
translocation into the endoplasmic reticulum.
Our studies indicate that additional cytosolic and
membrane factors are required to process the re-
ceptor to a functional form. We are attempting to
identify structural differences between a func-
tional receptor and newly synthesized receptors
that have not undergone the post-translational
processing necessary to produce functional pro-
tein. We hope to identify the requisite cytosolic
and membrane factors.
Cellular Biology of /82-Adrenergic Receptors
Following prolonged exposure to catechol-
amines, the ;82-adrenergic receptor becomes de-
sensitized and is less efficient in activating adeny-
lyl cyclase. Several mechanisms contribute to the
process of desensitization, including receptor
phosphorylation and the removal of receptors
from the plasma membrane. Over the past year
we have used immunocytochemical techniques
to study agonist-mediated internalization of ^2-
adrenergic receptors. These studies confirm that
these receptors are rapidly internalized into in-
tracellular vesicles called early endosomes. We
plan to continue our efforts to determine the
mechanism by which this occurs and to identify
the cellular proteins that mediate this process.
Furthermore, we hope to gain a better under-
standing of the relative importance of agonist-
activated internalization in the physiology of
182-adrenergic receptors.
230
Molecular Genetics of Lymphocyte Development
and Neoplasia
Stanley J. Korsmeyer, M.D. — Associate Investigator
Dr. Korsmeyer is also Professor of Medicine and Molecular Microbiology at Washington University School
of Medicine, St. Louis. He received his B.S. degree in biology from the University of Illinois, Urbana, and
his M.D. degree from the University of Illinois, Chicago. He did his internship and residency in internal
medicine at the University of California, San Francisco, and his postdoctoral research with Thomas
Waldmann and Philip Leder at NIH, where he became a Senior Investigator at the National Cancer
Institute. His honors include membership in the American Society for Clinical Investigation.
GENES that encode receptors for foreign anti-
gens have provided our most pivotal insights
into early lymphocyte development and lym-
phoid malignancies. The genes for immunoglobu-
lin (Ig), or antibodies, and for the T cell receptor
(TCR) encode the antigen receptors for B cells
and T cells, respectively. During early lympho-
cyte development, recombination at the DNA
level assembles these genes to create a wide re-
pertoire of receptor specificities.
Much of what we know about these genes has
been gleaned from studies of lymphoid tumors.
These malignancies are clonal expansions of a
single cell and provide multiple identical copies
of these genetic events. Provocatively, the char-
acteristic interchromosomal translocations that
typify B cell malignancies break at the Ig genes,
while those of T cell tumors often occur at the
chromosomal home of the TCR genes. We have
exploited this geography to clone the DNA at
these illegitimate interchromosomal junctures.
This serves as a bridge from the antigen receptor
loci to the other chromosomal partner, which has
often introduced a new cancer-promoting gene.
As a prototype, we have cloned the juncture
between chromosomes 1 4 and 1 8 that is present
in the most frequent form of human lymphoma,
follicular-type B cell lymphoma. This transloca-
tion occurs early in the development of a B cell
and introduces a newly discovered gene, Bel- 2,
into the Ig locus. A hybrid Bcl-2-lg fusion gene is
created, resulting in the overproduction of Bcl-2.
Transgenic mice were created that possess a copy
of the abnormal Bcl-2-lg fusion gene in their ge-
netic material. They progressed from an indolent
expansion of resting B cells to high-grade life-
threatening lymphomas, recapitulating the natu-
ral course of the human disease and proving that
this translocation causes malignancy. The Bcl-2
protein is unique among proto-oncogenes by be-
ing located in mitochondria. Moreover, it has a
novel function in that it blocks the programmed
death of cells independent of promoting their
growth. When deregulated, Z?c/- 2 extends the sur-
vival of cells normally destined to die. Bcl-2 con-
stitutes the first member of a new oncogene cate-
gory, regulators of cell death.
In a parallel set of experiments, unanticipated
rearrangements into the 6 TCR locus on chromo-
some 14 have identified the interchromosomal
translocation sites that typify early T cell acute
lymphoblastic leukemias. Two of these new
genes, Ttg-1 from chromosome 11 and Hox-11
from chromosome 10, are not normally ex-
pressed in T cells. Instead their function is di-
verted from their normal sites to T cells. Trans-
genic mice reveal that redirecting these
regulatory genes into T cells causes malignancy.
The majority of chromosomal defects, how-
ever, have no candidate gene at either side of the
juncture. The responsible gene lies a consider-
able distance from known genes. One promising
approach utilizes portions of chromosomes from
yeast organisms to obtain and propagate long
stretches of human DNA. This makes it possible to
clone and map entire segments of human chro-
mosomes, precisely linking known genes, and
generating probes to search for new disease loci.
This approach has pinpointed the chromoso-
mal breakpoint and identified a candidate onco-
gene for an extremely aggressive leukemia of
childhood.
This group of studies aims to improve our un-
derstanding of the genetic pathways of early T
and B cell development, as well as the aberran-
cies that result in malignancy.
231
Bcl-2
,^ Survival Pathway
r
Apoptotic
Death
Centrocytes
r Light Zone
I (). Basal
^J^^^^Light Zone
Bcl-2
Dark^
Zonel
B Blast <][
Plasma
Cell
Memory
Centroblasts
Distribution of the Bcl-2 protein that blocks cell death within the germinal center of the immune
system. Bcl-2 is confined to survival zones. B cells in the follicular mantle are longer lived and
recirculate. The growing centroblasts in the dark zone and the dying centrocytes in the basal light
zone lack the protein. It returns in the apical light zone, where long-lived memory cells emerge to
permit recall responses.
Research of Stanley Korsmeyer.
232
Molecular Genetics of Neuromuscular Disease
Louis M. Kunkel, Ph.D. — Investigator
Dr. Kunkel is also Professor of Pediatrics and of Genetics at Harvard Medical School. He received his B.A.
degree from Gettysburg College and his Ph.D. degree in biology from the Johns Hopkins University. He
took postdoctoral training with Brian McCarthy at the University of California, San Francisco, and with
Samuel Latt at the Children's Hospital, Boston. He held appointments at Children's Hospital/Harvard
Medical School before joining HHMI. His honors include the Gairdner Award and election to the National
Academy of Sciences.
WE continue to study the underlying cause
of Duchenne and Becker muscular dystro-
phies. Our identification of the gene led rapidly
to our description of the encoded protein dystro-
phin, which was found to be an unknown
member of a family of proteins that includes the
spectrins and a-actinins. Located at the inner face
of the plasma membrane of myofibers, dystro-
phin is thought to confer strength to the mem-
brane during muscle contraction and relaxation.
Absence or abnormality of dystrophin at this loca-
tion causes the myofiber degeneration of Du-
chenne and Becker dystrophy. Our work has led
to improved diagnosis of the diseases and to test-
able ideas on therapeutic intervention.
Over the past year the laboratory has concen-
trated on identifying new members of the dystro-
phin family of proteins, in the hope that these can
play a role in mitigating disease caused by dystro-
phin deficiency. These dystrophin-related pro-
teins might themselves be involved in other
neuromuscular diseases. By antigenic cross-
reactivity, we have cloned the human microtu-
bule-associated protein IB (MAP- IB) and have
mapped this locus in close proximity to muta-
tions on chromosome 5 that cause a motor neu-
ron degenerative disease, spinal muscular atro-
phy (SMA).
We have sequenced this human gene and have
begun its direct mutational analysis in SMA pa-
tients. We have also identified at the locus two
separate (CA)n repeat polymorphisms that have
been shown, in collaboration with Conrad
Gilliam, to be linked tightly to SMA mutations. In
this genetic analysis, a few rare recombinations
have been detected between SMA and MAP- IB,
implying that MAP- IB may be very close, but not
involved in the disease. As a result of this observa-
tion, we have cloned large segments of human
DNA from the region of chromosome 5 as yeast
artificial chromosomes (YACs) and have begun a
search for other genes near MAP- IB, any of which
might be the SMA gene.
By reduced stringency hybridization with dys-
trophin cDNA clones, a chromosome 6-encoded
protein was identified by Kay Davies in Oxford.
This dystrophin-related protein (DRP) is highly
similar to dystrophin, and antibodies against it
developed in our laboratory have revealed that
the two proteins are almost identical in size. Im-
munolocalization of DRP has shown that it colo-
calizes with dystrophin in a developing myofiber.
In a mature fiber, however, it is located only at
the neuromuscular and myotendinous junctions
of muscle. We are currently attempting to clone
the entire coding sequence of DRP as cDNA. We
believe that the sequence of this protein might
reveal why it has such a specialized location in
mature muscle and what role, if any, it might play
in mitigating some of the effects of dystrophin
absence in Duchenne dystrophy. We are also
searching for other neuromuscular diseases that
might be caused by DRP abnormalities.
One obvious set of dystrophin relatives for us
to attempt to characterize were the a-actinins. A
smooth muscle form had been cloned and local-
ized to chromosome 14, but there was at least
one other that had been identified from chicken
muscle. Using a conserved motif of the known
a-actinin amino acid sequence, we designed a de-
generate set of PGR (polymerase chain reaction)
primers that amplified a human muscle product
with the sequence of an a-actinin. Using this
small product as a hybridization probe, we
screened a muscle cDNA library and obtained two
classes of hybridizing human cDNA clones. Se-
quence analysis revealed that the clones con-
tained two different a-actinins that were unique
for humans. They are both muscle specific and
are encoded from chromosomes 1 and 1 1 , respec-
tively. We are now developing antisera specific
for each of the muscle a-actinin gene products
and attempting to identify disease phenotypes
that might involve a-actinin.
Abnormalities of dystrophin are easily detected
at the protein level, and nearly 70 percent of the
mutations that cause them have been shown to be
deletions or duplications of some part of this ex-
tremely large locus. We have designed primers
from dystrophin's nucleotide sequence to allow
PGR amplification of specific regions of dystro-
phin's transcript. In analyzing these PGR prod-
233
Molecular Genetics of Neuromuscular Disease
ucts from patients who do not have a detectable
deletion, we have found at least one mutation
that involves splicing of dystrophin's primary
transcript.
PGR amplification yielded a product of abnor-
mal size that lacked exon 5. Sequence analysis of
the patient's DNA from the region of this exon
revealed an identical sequence to that of normals.
We are currently cloning other surrounding DNA
to identify the exact point where a mutation
might effect the splicing of exon 5. We believe
that mutations in splicing signals may lead to a
better understanding of how this huge locus is
processed into a mature mRNA transcript.
During our screening of different dystrophin
antibodies for finding cross-reactive proteins, our
carboxyl domain-directed antibodies detected a
small protein (approximately 70 kDa). At the
same time, David Yafife's group in Israel reported
observing a short dystrophin transcript in non-
muscle tissues. Our laboratory cloned this tran-
script from our human brain cDNA library, and
sequencing of the cDNA clones revealed a diver-
gence from expected dystrophin sequence at ap-
proximately exon 62. We are currently preparing
antibodies against novel peptide sequences to de-
termine if the shorter protein detected in brain
tissue with antidystrophin antibodies is indeed a
product of this novel transcript.
In addition to the published shorter transcript,
we have also detected a slightly larger protein
(approximately 90 kDa) in sciatic nerve. cDNA
clones have been prepared and shown to diverge
from expected dystrophin sequence at exon 54.
Similar studies are under way for this novel tran-
script to determine the role these smaller pro-
teins might play in the phenotype of Duchenne
dystrophy.
Our aim for the future year is to build on work
already in progress. We will continue to identify
dystrophin-related proteins and attempt to deter-
mine their normal function in muscle. We will
determine what role, if any, they play in other
neuromuscular diseases and whether they miti-
gate the effects of abnormal dystrophin. We will
continue our analysis of dystrophin-alternative
transcripts and what role they might play in nor-
mal development and the phenotype of Du-
chenne dystrophy. As more human disease genes
are mapped to specific human chromosomes, we
believe that our candidate gene approach, which
builds upon knowledge of abnormal dystrophin,
should help in the rapid characterization of these
diseases.
234
Structural Studies on DNA-Replication
Enzymes, src- related Oncogene Products,
and Oxidoreductases
John Kuriyan, Ph.D. — Assistant Investigator
Dr. Kuriyan is also Associate Professor of Molecular Biophysics at the Rockefeller University. He received a
B.S. degree in chemistry from Juniata College, Pennsylvania, and a Ph.D. degree from the Massachusetts
Institute of Technology, where he worked with Gregory Petsko and Martin Karplus on the dynamics of
proteins. He continued with Martin Karplus at Harvard University for a year and then moved to the
Rockefeller University as a University Fellow. He is also a Pew Scholar in the Biomedical Sciences.
OUR interests are in understanding the struc-
tural determinants of protein function. To
this end we carry out x-ray diffraction experi-
ments and computer simulations aimed at charac-
terizing the three-dimensional structures of pro-
teins. We apply the knowledge gained from these
studies toward the design of mutations and inhibi-
tors to modify protein activity, and seek, from the
particular cases in hand, to extrapolate to fami-
lies of related proteins. Our current efforts are
focused on three main areas: DNA replication,
5rc-related oncogene products (tyrosine ki-
nases), and redox enzymes and oxidative stress.
Structure of the /S-Subunit
of DNA Polymerase III
DNA polymerases are enzymes that duplicate
the information content of DNA by catalyzing the
template-directed polymerization of nucleic
acids. Polymerases that are involved in chromo-
somal replication, such as DNA polymerase III
(PolIII) of Escherichia colt, are distinguished by
their high processivity — i.e., their rapid replica-
tion (1 ,000 bases/second) of very long stretches
of DNA without dissociation. Processivity is con-
ferred upon the enzyme by one of its subunits, /3
(processivity factor), which acts to clamp the
polymerase onto DNA. The /3-subunit binds very
strongly — indeed, cannot be easily separated
from circular DNA. It has been shown, however,
to slide freely along duplex DNA, consistent with
its role as a clamp that tethers the polymerase
core to the template and advances with the poly-
merase during replication.
We have crystallized the |S-subunit and deter-
mined its structure by x-ray diffraction to 2.5-A
resolution (Xiang- Peng Kong, Rockefeller Univer-
sity; Michael O'Donnell [HHMI] and Rene Onrust,
Cornell University Medical College) . Two mole-
cules of the /3-subunit are tightly associated to
form a donut-shaped structure that forms a closed
ring around duplex DNA. An unexpected feature
of the structure is that it is highly symmetric.
Each monomer consists of three domains of iden-
tical chain topology. Each domain is roughly
twofold symmetric in its architecture, with an
outer layer of two (8-sheets providing a scaffold
that supports two a-helices. Replication of this
motif around a circle results in a rigid molecule
with 1 2 a-helices lining the inner surface of the
ring and with six jS-sheets forming the outer
surface.
The high symmetry of the structure is well
suited to interact with the cylindrically symmet-
ric DNA duplex, and the hole in the middle of the
ring is large enough to accommodate either A or B
forms of DNA with no steric repulsion. Although
the overall charge on the protein is negative, a
number of basic side chains on the inner surface
generate a positive electrostatic field localized in
the hole of the donut, precisely where the nega-
tively charged phosphate backbone of DNA is ex-
pected to be. We are carrying out molecular dy-
namics simulations of a duplex-DNA and
(8-subunit complex, which is expected to provide
information not readily accessible by x-ray crys-
tallography. Other extensions of this project in-
clude crystallization attempts on the eukaryotic
processivity factor PCNA (proliferating cell nu-
clear antigen) , other subunits of PolIII, and other
proteins acting at the replication fork.
Oncogene Products Related to src
Tyrosine kinases such as the product of the src
gene play a central role in signal transduction
pathways in the cell, and abnormalities in their
function can result in dramatic changes in cellu-
lar differentiation and life cycles. Our approach
is to work on these proteins in isolation, and we
are collaborating with Hidesaburo Hanafusa
(Rockefeller University) and Marilyn Resh
(Sloan-Kettering Institute) to crystallize various
functional domains of these oncogene products
and their cellular equivalents. Initial success has
been obtained for the src SH2 domain, which is
responsible for binding to proteins containing
phosphorylated tyrosine residues. Starting from
src-containing plasmids (provided by Resh),
Dorothea Kominos and Gabriel Waksman in our
laboratory have cloned, overexpressed, purified,
and crystallized SH2. The crystals diffract to 2.5-A
resolution, and the x-ray structure determination
235
Structural Studies on DNA-Replication Enzymes, src- related Oncogene
Products, and Oxidoreductases
of a complex of SH2 with a phosphotyrosyl pep-
tide has been completed at 1.5 -A resolution.
Redox Proteins and the Transcriptional
Response to Oxidative Stress*
We have recently solved and refined the three-
dimensional structures of two related redox en-
zymes, thioredoxin reductase and trypanothione
-reductase. Crystallographic investigations of
enzyme-substrate complexes are now in progress
(G. Waksman). Work has also begun on a new
member of the disulfide reductase family. Pro-
teins related to thioredoxin have been implicated
in the process by which disulfide-containing pro-
teins fold up rapidly without scrambling their
disulfide pairings.
James Bardwell, Karen McGovern, and Jon
Beckwith (Harvard Medical School) have re-
cently identified an E. colt protein that is re-
quired for the correct folding of disulfide-
containing proteins in vivo. This 21-kDa protein,
the product of the dsbA gene, has no detectable
sequence similarity to any known protein except
for a short stretch of amino acids with similarity
to the active site of thioredoxin (including the
redox-active disulfide bond). The dsbA protein
has been purified in our laboratory, from cells
provided by Bardwell and Beckwith, and single
crystals that diffract to 2-A resolution have
been obtained. The determination of the three-
dimensional structure is under way, and we are
also pursuing the construction of site-specific
mutants at the active site (Jennifer Martin) .
We are interested in determining the mecha-
nism of the bacterial oxidative stress sensor
OxyR. The OxyR protein responds to oxidizing
agents by elevating the expression of such redox
enzymes as catalase, superoxide dismutase, and
glutathione reductase. The intact OxyR molecule
proved to be unstable in the absence of DNA. In
collaboration with Gisela Storz (National Insti-
tutes of Health), Scott Robertson in our labora-
tory has cloned, overexpressed, and purified the
regulatory domain alone, and its crystallization is
being pursued.
' This work is being supported by a grant from the National
Institutes of Health.
Representations of the structure of
fi-subunit — the sliding clamp — of
DNA polymerase III of Escherichia
coli, determined by x-ray crystallog-
raphy. The atomic structure is
shown in yellow and its electrostatic
potentials in red ( negative ) and
blue (positive). The top two images
are different views of the dimer, and
the bottom is a single monomer. The
electrostatic potential generated by
the protein helps stabilize the dimer
and its interactions with the DNA
molecule ( not shown ), which
threads through the clamp formed
by the protein.
From Kong, X.-P., Onrust, R.,
O'Donnell, M., and Kuriyan, f. 1992.
Cell 69:425-437. Copyright© 1992
by Cell Press.
236
Molecular Analysis of Down Syndrome
David M. Kumit, M.D., Ph.D. — Investigator
Dr. Kumit is also Professor of Pediatrics and Human Genetics at the University of Michigan Medical
School. He received his M.D. and Ph.D. degrees in cell biology from Albert Einstein College of Medicine.
He did his internship and residency in pediatrics at the University of Pittsburgh and his fellowship in
medical genetics at the University of Washington. After seven years as a Harvard Medical School faculty
member at Children's Hospital, he joined the University of Michigan Medical Center.
WE are primarily interested in the etiology
and pathogenesis of Down syndrome.
Why does the risk of having an offspring with
Down syndrome increase dramatically as women
age? There are two competing theories. The
"older egg" model states that the older a woman
becomes, the more abnormal (aneusomic) eggs
she produces. The "relaxed selection" model
states that as women age, they lose the ability to
abort abnormal conceptuses. To determine the
correct hypothesis, or whether a combination of
the two applies, it is necessary to study a large
number of Down syndrome cases.
Given our important (and unexpected) finding
that 94 percent of nondisjunction errors are ma-
ternal in origin, resolution of the above contro-
versy requires an ability to distinguish the stage
of maternal cell division (meiosis I or II) in
which the error occurs. To accomplish this, we
will analyze four distinct probes (H6-5-6, 1 26-4-
1, D21S120, and GT14) that detect polymorphic
oligo d(A,C) sequences in the pericentromeric
region of chromosome 2 1 . Conditions and sizes
have been achieved that allow each of these
probes to be multiplexed in a single lane without
co-interference. Using four different fluorescent
markers, it is feasible to analyze electrophoreti-
cally in a single gel lane a molecular weight stan-
dard and a family consisting of father, mother,
and their child with Down syndrome. Pouring the
gels is the labor-intensive procedure in this pro-
tocol, and the ability to multiplex three family
members and four polymorphisms in a single
lane will yield a 12-fold savings in gel-pouring
effort. This savings is made worthwhile by
the fact that we must study at least 500 fami-
lies to comprehend the biological basis for non-
disjunction. This work was also supported by a
grant from the National Institutes of Health.
To study pathogenesis, we first isolated genes
encoded by chromosome 21, using a novel re-
combination-based methodology. Genomic frag-
ments isolated from yeast artificial chromosomes
(YACs) that map to chromosome 21 are cloned
into a plasmid vector with the genetic marker
sup¥. RNA isolated from a variety of 20-week fetal
human tissues was used to construct bacterio-
phage X cDNA libraries, which are then infected
into cells harboring a supY plasmid carrying a
nonrepeated genomic fragment on chromosome
21 . (Using tissues recently obtained from earlier
abortus specimens, we will now be able to make
complex cDNA libraries from small amounts of
these tissues by polymerase chain reaction tech-
niques.) If any member of the genie cDNA library
in X shares homology with the genomic DNA frag-
ment in the plasmid, then recombination me-
diated by that homology will ensue.
Following recombination between the bacte-
riophage and the plasmid, selection for bacterio-
phages carrying a given plasmid with sup¥ will
result in selection for bacteriophages carrying a
cDNA that is homologous to a genomic DNA
cloned in the plasmid. In other words the system
is designed to select for genomic sequences that
are transcribed. The system is designed to stand
alone or to interdigitate with the genomic initia-
tive as it proceeds. In the latter case, as sequenc-
ing detects open reading frames that represent
candidates for transcription, the recombination-
based assay is designed to delineate analytically
the tissue and timing of transcription and to re-
sult preparatively in isolation of the transcribed
sequence.
The bulk of cDNA (gene) libraries are contami-
nated with small amounts of ubiquitous DNA se-
quences from plasmid pBR322. To circumvent
this problem, we have cloned the sup¥ gene into
an R6K-derived plasmid that lacks homology
with pBR322 and have made other improvements
that should increase the applicability of this pro-
cedure. We should now be able to screen a wide
variety of extant cDNA libraries for transcription
by chromosome-specific elements.
This systematology applied in model experi-
ments has demonstrated that we can perform
both selection and counterselection appropri-
237
Molecular Analysis of Down Syndrome
ately. We are currently analyzing transcribed se-
quences on distal chromosome 21q (where the
Down syndrome phenotype maps) and have iso-
lated multiple candidates. The technique can
also be applied to determine the timing and tis-
sues of transcription of likely-to-be-transcribed
sequences, as determined by other methodolo-
gies. We will collaborate with researchers per-
forming these other techniques to make such
determinations.
238
Replication and Pathogenesis of RNA Viruses
Michael M.-C. Lai, M.D., Ph.D. — Investigator
Dr. Lai is also Professor of Microbiology and Neurology at the University of Southern California School of
Medicine, Los Angeles. He obtained his M.D. degree from National Taiwan University. He studied
retroviruses with Peter Duesberg at the University of California, Berkeley, where he obtained his Ph.D.
degree in molecular biology and continued for postdoctoral work.
MEDICAL history is marked by extraordinary
successes against viral infections, but it is
also punctuated by the continual emergence of
new viruses. Since viruses, in general, contain
very limited genetic information, they must rely
on host cells for their own growth. How they
cause diseases and how they continue to flourish
in nature are not only interesting subjects in
themselves but off^er a lesson in the everyday
workings of normal cells. Our laboratory is inter-
ested in RNA viruses, replicating entities in
which RNA, in contrast to DNA, is the genetic
material. We are exploring how these viruses rep-
licate and cause diseases.
One of those we are studying is the corona-
virus, named for its similar appearance to the cor-
ona of the sun. The virus causes the common cold
in humans and a variety of gastrointestinal and
respiratory problems in animals. It also causes
symptoms very similar to those of multiple sclero-
sis, thus providing a model system for studying
this disease. The virus has an RNA genome of
31,000 nucleotides, which is the longest known
stable RNA. We are interested in learning how
this unusually large RNA expresses its genes and
maintains its genetic stability, despite an over-
whelmingly high frequency of error in RNA syn-
thesis. We have recently determined the com-
plete sequence of the genome, giving us a
glimpse of how the viral genes express
themselves.
The virus utilizes a novel RNA synthesis mecha-
nism, a discontinuous process that fuses a leader
RNA to a gene located some distance from it. This
unusual mechanism allows the leader RNA to
control the expression of viral genes. We have
recently found that the synthesis of one of the
viral surface proteins indeed can be altered by
minor changes in the leader RNA sequences, re-
sulting in the variation of the viral pathogenicity.
Furthermore, this variation can be observed dur-
ing natural viral evolution. These observations
suggest that the enzyme catalyzing coronaviral
RNA synthesis is unusual, which is, indeed, sug-
gested from the sequence of the gene encoding
the enzyme. Our laboratory is investigating this
novel RNA synthesis mechanism. A research grant
from the National Institutes of Health provides
support for this part of the research program.
Another unusual characteristic of coronavirus
RNA has been revealed in our findings: it can un-
dergo genetic exchange (RNA-RNA recombina-
tion) at an extraordinarily high rate. RNA-RNA re-
combination was previously thought to be a rare
event in nature. We demonstrated, however, that
it occurs readily between coronaviruses. This re-
combination can take place almost anywhere in
the RNA genome, both in tissue culture cells and
during animal infections. We recently succeeded
in establishing that with the coronavirus, recom-
bination can occur not only between two replicat-
ing viral RNAs but between viral RNA and a piece
of an RNA fragment existing inside the cell, thus
providing a model system for RNA recombination
between viral and cellular RNAs. This type of re-
combination is one of the mechanisms by which
some animal viruses become pathogenic.
We have demonstrated that a recombinant
virus could become a predominant viral popula-
tion under certain conditions, replacing the pa-
rental virus by a simple natural selection process.
Thus recombination represents a powerful evolu-
tionary tool for RNA viruses. From the standpoint
of viral biology, RNA recombination may be the
genetic mechanism by which coronaviruses weed
out defective RNA sequences generated by errors
in RNA synthesis. Coronavirus is thus able to
maintain an RNA genome larger than was thought
theoretically possible. RNA recombination has
now been demonstrated in many different vi-
ruses, suggesting its important role in virus evo-
lution in nature.
This genetic phenomenon also has an impor-
tant implication in vaccine development for dis-
eases such as AIDS (acquired immune deficiency
syndrome) , since genetic exchanges between vi-
ruses may lead to genetic instability of live, atten-
uated virus vaccines. We are continuing to study
the RNA recombination mechanism and attempt-
ing to use it as a genetic tool in determining how
viruses cause diseases.
Another virus we are studying in our laboratory
239
Replication and Pathogenesis of RNA Viruses
is the hepatitis delta virus (HDV) , a human hepati-
tis virus commonly associated with a severe form
of the disease. HDV, by itself, does not infect hu-
mans, because it is defective and requires another
viral agent, hepatitis B virus (HBV), to supply an
essential envelope protein to infect liver cells.
HDV has been shown to cause epidemics of fulmi-
nant hepatitis in many parts of the world. In the
-United States, it is prevalent among intravenous
drug abusers.
The virus contains a circular, single-stranded
RNA genome of only 1,700 nucleotides. It is the
only animal virus with a circular RNA. This ge-
nome structure is reminiscent of a group of plant
pathogens, viroids or virusoids, that cause a vari-
ety of plant diseases. Indeed, the similarity be-
tween HDV RNA and plant viroid RNAs goes
beyond their circular RNA structure. There are
several structural and biochemical features that
suggest a close evolutionary relationship be-
tween HDV and plant viroid RNAs. Both RNA
groups contain a "ribozyme" activity, in which
the RNA serves as an enzyme that cleaves and li-
gates the RNA itself. Thus HDV RNA stands at a
peculiar place in the evolutionary ladder: it may
have been derived from a plant pathogen by re-
combination with a gene that gave it the ability to
infect human cells and cause diseases. Our labora-
tory is studying the properties of this ribozyme
activity. We have determined its structural and
sequence requirements, which are distinct from
those of other known ribozymes. Furthermore,
we have begun to examine whether ribozyme ac-
tivity is the same in infected cells as in the test
tube. The data obtained thus far suggest the inter-
esting possibility that some cellular factors could
participate in the ribozyme activity. We are pur-
suing these factors.
One important difference between HDV RNA
and plant viroid RNAs is the ability of the former
to synthesize a protein, hepatitis delta antigen
(HDAg), the HDV signature protein. HDAg is re-
quired for RNA synthesis. We have been studying
this protein's properties and functions and have
shown that HDAg is a phosphoprotein that resides
in the nuclei of the cells. It can interact with itself
to form a protein complex, and it can also inter-
act with HDV RNA in a specific way. We have
shown that HDAg utilizes a set of novel sequence
motifs to allow itself to bind to RNA.
What is the role of this protein, HDAg, in HDV
RNA synthesis? — an unusual synthesis because
HDV RNA is so small (1,700 nucleotides) that it
lacks the capacity to provide its own synthesizing
enzymes. Therefore, HDV most likely borrows
cellular enzymes to do the job, which is unchar-
acteristic of RNA viruses. Most RNA viruses have
to make their own enzymes, since normal cells do
not appear to have this type. We are studying how
HDV modifies the cellular enzymes to perform
this rather atypical RNA synthesis and how HDV
RNA synthesis initiates this. It appears that HDAg
participates in these processes.
HDV thus provides a perspective on viral strate-
gies from the small end of the RNA spectrum. Our
laboratory is studying one of the largest RNA vi-
ruses (coronavirus) and the smallest (HDV),
which utilize different principles for viral repli-
cation. Our studies not only offer insights into
how these viruses cause diseases but also into
fundamental mechanisms of RNA synthesis and
RNA evolution.
240
Molecular Biology of Human Papillomaviruses
LaimonisA. Laimins, Ph.D. — Assistant Investigator
Dr. Laimins is also Associate Professor of Molecular Genetics and Cell Biology at the University of Chicago.
He received his Ph.D. degree in biophysics and theoretical biology from the University of Chicago. His
postdoctoral research was done with George Khoury at the National Cancer Institute.
MY laboratory is studying the molecular biol-
ogy of human papillomaviruses types 16
and 18. HPV- 1 6 and - 1 8 are the etiological agents
of the many malignancies of the urogenital re-
gion, and in particular those of the cervix. More
than 70 different types of HPV are known to in-
fect cutaneous and mucosal epithelia in various
body locations. Some types induce warts on the
hands or soles of the feet, but are never found in
genital lesions. Similarly, the viruses that induce
warts in the genital region are never found on the
hands or feet.
The majority of HPVs cause benign warts, and
about one-third of the types are specific for geni-
tal epithelium. A subset of these viruses (HPV- 16,
-18, -31, and -51) infect the cervix and are
strongly associated with the development of
cervical cancer. Although the number of HPV-
positive individuals has increased 10-fold in the
last 10 years, effective monitoring procedures
such as the PAP smear have limited the increase in
cervical cancers in the United States and Europe;
Up to 20 percent of the population in these
countries is infected by some type of genital HPV,
particularly the non-oncogenic types 6 and 1 1 .
However, many cases of infection by HPV- 1 6 and
-18 without visible lesions have been reported. It
therefore appears that although infection by
HPVs is necessary for the development of cervical
cancer, infection alone is not sufficient to induce
malignancy. In low-grade lesions, or condylomas,
viral DNA is present in the cell as episomes, and
progeny viruses are produced. In contrast, in cer-
vical cancers the viral genome is found integrated
into the host chromosome, and no viral particles
are synthesized. One model for progression from
a low-grade lesion to a cancer suggests that inte-
gration is important in the development of the
cancer and that it may act by the removal of a
dominant inhibitor of transformation. Viral inte-
gration into the chromosome may thus be the first
step of a multistep carcinogenic process. My labo-
ratory is currently studying the molecular mecha-
nisms by which HPVs contribute to this process.
A major impediment to the study of HPVs has
been an inability to propagate these viruses in
culture. Papillomaviruses are unusual in that
their life cycle is tightly coupled to the differen-
tiation program of epithelial cells. Although most
viruses infect one cell type and undergo a produc-
tive infection in the same cell, HPVs infect basal
epithelial cells, establish their genomes as epi-
somes, and only replicate upon cellular differen-
tiation, with amplification of viral DNA and ex-
pression of late genes. Inability to duplicate this
differentiation process in vitro long prevented
the successful propagation of virus. We have re-
cently duplicated several features of these pro-
ductive infections of HPV in culture, and current
studies are directed at identifying how these dif-
ferentiation-specific properties are controlled.
This work is supported by a grant from the Ameri-
can Cancer Society.
My laboratory has found that HPV can immorta-
lize epithelial cells derived from either human
foreskin or cervix in tissue culture. Normally pri-
mary cells have only a limited life span in vitro,
but the presence of the HPV E7 protein is suffi-
cient to allow for unlimited grov^h in culture.
The loss of differentiation is usually a charac-
teristic of many epithelial cancers, including
those of the cervix. We have been studying how
HPV viral proteins alter epithelial cell differen-
tiation. In low-grade lesions in which infectious
viruses are produced, these proteins only slightly
alter epithelial differentiation. In cervical can-
cers, on the other hand, infected cells have lost
all ability to differentiate.
We have also used a system that accurately mim-
ics the differentiation properties of epithelial
cells in vitro, to show that HPV sequences alter
the ability of epithelial cells to differentiate. The
morphological changes that are seen in this tissue
culture system are very similar to those seen in
low-grade cervical neoplasias. In tissue culture,
our HPV cell lines quickly lose the ability to dif-
ferentiate and develop the appearance of high-
grade neoplasias or cancers. Two HPV genes, E6
and E7, seem to be required for this transforma-
tion process. We believe this loss of differentia-
tion may be an in vitro model for the develop-
ment of cervical cancer. In the future, we hope to
utilize this model to identify other important fac-
tors involved in controlling the development of
241
Molectilar Biology of Human Papillomaviruses
malignancy. These observations also strongly
support the etiological role of HPV in the devel-
opment of cervical cancer. This work was sup-
ported in part by a grant from the National Insti-
tutes of Health.
My laboratory has been able to identify factors
involved in the cell type-specific regulation of
viral gene expression and replication. We have
-identified viral enhancer sequences that are re-
sponsible for regulating viral expression in ma-
turing keratinocytes. The function of one of these
sequences is dependent on the action of a cellu-
lar protein that is only found in epithelial cells.
We have designated this protein keratinocyte-
stimulating factor, KRF- 1 , and are currently clon-
ing the gene that encodes it.
Eventually we will examine how the action of
cellular transcription factors is altered by the
presence of viral transforming genes. We have
shown that the products of both E6 and E7 can
indirectly regulate the transcription of a variety
of genes. In addition, we believe that a major de-
terminant of the tissue spectrum of papillomavi-
ruses is specified through the regulation of tran-
scription. Understanding the mechanisms of
tissue-specific expression will facilitate our anal-
ysis of why different HPV types only induce le-
sions in particular kinds of epithelial cells. Our
studies on the biology of HPV should provide
information on mechanisms of transformation
and tumor progression, as well as tissue- and
difi'erentiation-specific expression.
242
Genetic Studies in Cardiovascular Disease
Jean-Marc Lalouel, M.D., D.Sc. — Investigator
Dr. Lalouel is also Professor of Human Genetics at the University of Utah School of Medicine. He obtained
a medical doctorate, a master's degree in microbiology and genetics, and a doctorate of sciences in
genetics at the University of Paris, France. He furthered his training as a postdoctoral fellow and a research
associate with Newton Morton at the University of Hawaii and was Professor of Human Biology at the
University of Paris before joining the faculty of the University of Utah.
COMMON cardiovascular disorders such as
coronary artery disease and essential hyper-
tension exhibit a familial tendency. Such broad
clinical classes represent complex etiological en-
tities, where many genes and environmental de-
terminants are likely to be involved. Multiplicity
and heterogeneity in causation stretch the ability
of genetic methods to the limit of their investiga-
tive power. The consideration of biochemical pa-
rameters can reduce this degree of complexity.
Before the global picture can be comprehended,
however, details of the landscape will need to be
scrutinized first. This perspective is highlighted
in two examples from our laboratory. Work on
lipoprotein lipase was led by Mitsuru Emi and
Akira Hata, while work on hypertension was led
by Richard Lifton.
Abnormal lipoprotein concentrations in
plasma are commonly observed in the relatives of
patients with early coronary disease, yielding
various patterns of hypercholesterolemia and/or
hypertriglyceridemia within families. Such com-
plex phenotypes are thought to result either from
the variable expression of a single-gene defect or
from the independent contribution of two or
more genes. The contribution of such genes to
the clinical expression of hyperlipidemia is fur-
ther blurred by the fact that hormonal influences,
diet, and habitus exert major influences on the
regulation of lipid metabolism. A severe form of
familial hypercholesterolemia, which accounts
for about 5 percent of myocardial infarction, has
been linked to molecular defects of a lipoprotein
cellular receptor. More than 90 percent of di-
etary fat is hydrolyzed by lipoprotein lipase (LPL)
in an initial step controlling its delivery to periph-
eral tissues. The enzyme, secreted by adipocytes
and muscle cells, acts at a distance from its site of
synthesis, becoming anchored to the luminal sur-
face of capillaries by an ionic interaction with
heparan sulfate. In the presence of a specific co-
factor, apolipoprotein C-II, the enzyme hydro-
lyzes triglycerides of intestinal or hepatic origin
by binding to the surface of circulating lipopro-
teins, thereby releasing fatty acids for uptake in
the tissues where they can be used as fuel or rees-
terified for storage. Defective functional enzyme
is the diagnostic feature of a rare recessive chylo-
micronemia syndrome, familial LPL deficiency.
This condition is characterized by massive chy-
lomicronemia in the fasting state, episodes of ab-
dominal pain, life-threatening acute pancreatitis,
and eruptive xanthomas. By investigating the rela-
tives of a homozygous subject, we showed pre-
viously that heterozygotes for such molecular
defects tend to express a common form of hyper-
triglyceridemia. However, when we screened
unrelated subjects for molecular defects of the
LPL gene, we found that mutations of this gene
probably account for 3-5 percent of such hyper-
lipidemias. This situation, similar to receptor de-
fects in familial hypercholesterolemia, again
stress that most forms of hyperlipidemia are of
heterogeneous origin.
We and others have now identified a host of
mutations in the LPL gene of subjects with LPL
deficiency, including a large number of simple
atnino acid substitutions. Such mutations may
provide natural, specific probes of functional do-
mains of the enzyme. When they are superim-
posed to the known three-dimensional structure
of the homologous enzyme, pancreatic lipase, it
becomes clear that they are spread over the pro-
tein's first folding domain, which includes the
triad of residues directly involved in catalysis
(see figure) .
We reproduced four such mutations in vitro,
expressed them in cultured cells, and analyzed
the corresponding mutant proteins. A common
feature of these mutations is that they aff^ected the
assembly and stability of the two identical sub-
units characterizing the active enzyme. Hence
the majority of these mutations lead to loss of
catalytic activity through rather nonspecific
mechanisms.
To probe other domains of the enzyme, it be-
came clear that we could not rely solely on natu-
rally occurring mutations. LPL binds to heparan
sulfate, a member of a family of ubiquitous and
abundant complex polysaccharides. Interactions
of this nature play a role in many biological pro-
cesses, such as lipolysis, hemostasis, cell adhe-
sion, cell proliferation, or angiogenesis. The pro-
243
Genetic Studies in Cardiovascular Disease
tein determinants involved in these interactions
are poorly characterized, but the high density of
negative charges on heparan sulfate indicates that
the protein domains should include positively
charged residues.
After generating many artificial mutations at
such positions, we have produced the corre-
sponding proteins by expression in cultured cells
and analyzed their affinity for heparin. We have
identified six basic amino acids that mediate the
high affinity of the active, dimeric enzyme with
heparin. They form a domain of unique structure,
which we are attempting to characterize further.
Essential hypertension presents a yet greater
challenge, for little biochemical insight is
usually available. Glucocorticoid-remediable al-
dosteronism (GRA) is a rare autosomal dominant
disorder marked by severe hypertension and
hyperaldosteronism with high levels of abnor-
mal adrenal steroids. All these manifestations
can be corrected by the administration of
glucocorticoids.
Aldosterone is a steroid involved in the regula-
tion of the balance of sodium and potassium ions
and produced in the zona glomerulosa of the adre-
nal gland under the primary control of the renin-
angiotensin system. By contrast, glucocorticoids
exert their effects on carbohydrate metabolism,
are produced by the zona fasciculata of the adre-
nals, and are regulated by the adenohypophyseal
hormone adrenocorticotropin (ACTH). Their syn-
thetic pathways share several enzymes, including
lljS-hydroxylase. Aldosterone synthesis, how-
ever, requires a unique enzymatic step catalyzed
by aldosterone synthase, which is normally ex-
pressed only in the zona glomerulosa.
In a subject with GRA, we found that unequal
crossing over between 1 1 (8-hydroxylase and aldo-
sterone synthase, in close proximity on chromo-
some 8, had created a new chimeric gene com-
prising regulatory sequences of 1 lj8-hydroxylase
and sequences responsible for the catalytic speci-
ficity of aldosterone synthase. This observation
explains the ectopic production of aldosterone in
the adrenal tissue responsible for synthesis of
glucocorticoids, and the corresponding hor-
monal control observed in GRA. Although this
finding provides a clear interpretation of the
complex physiology of a rare form of hyperten-
sion, its significance for our understanding of the
more common forms of essential hypertension
remains to be determined.
Mutations of human lipoprotein lipase that
impair the enzyme's ability to regulate lipid
metabolism are shown superimposed upon a
schematic rendering of the backbone of a ho-
mologous enzyme, human pancreatic li-
pase. The three circled residues define the
triad directly involved in catalysis.
From Lalouel, f.-M., Wilson, D.E., and Iver-
ius, P.-H. 1992. Curr Opin Upidol 3:86-95.
244
Structure and Replication of Influenza Virus
and Paramyxoviruses
Robert A. Lamb, Ph.D., Sc.D. — Investigator
Dr. Lamb is also John Evans Professor of Molecular and Cellular Biology and Professor of Microbiology-
Immunology at Northwestern University. He received his undergraduate degree reading biochemistry at
the University of Birmingham, England, and his Ph.D. and Sc.D. degrees from the University of Cambridge.
He came to the United States to do postdoctoral work with Purnell Choppin at the Rockefeller University,
where he later became a faculty member. Ten years ago he joined the faculty of Northwestern University.
ANIMAL viruses provide a unique tool with
which to study the complex biochemical
processes involved in the biosynthesis and main-
tenance of eukaryotic cells. Our laboratory is in-
vestigating the molecular structure and the mech-
anism of replication of two enveloped RNA
viruses, influenza virus and the paramyxovi-
rus SV5.
Influenza virus causes important diseases in
humans and animals. It has tremendous socioeco-
nomic consequences, for influenza continues to
occur in regular epidemics and occasional pan-
demics and is a leading cause of morbidity and
mortality. Paramyxoviruses cause many biologi-
cally and economically important diseases of hu-
mans and lower animals. Besides SV5, these vi-
ruses include parainfluenza virus types 1-5,
mumps virus, measles virus, canine distemper
virus, Newcastle disease virus of chickens, and
rinderpest of cattle.
We have been elucidating the wide range of
mechanisms used by these RNA viruses to maxi-
mize the amount of encoded protein in their
compact genomes. We have identified overlap-
ping reading frames, splicing of mRNAs, the use
of bicistronic mRNAs, transcriptional stuttering
to add nontemplated nucleotides to an RNA tran-
script (and hence yield a separate mRNA) , and a
coupled stop-start translation of tandem cistrons.
Influenza virus and SV5 were selected for study
not only because of their importance as the caus-
ative agents of major diseases but also because
they provide excellent models for examining a
variety of properties of integral membrane pro-
teins. Since these proteins are the major antigenic
determinants of the viruses, knowledge about
their structure should enhance our understand-
ing of how they act as immunological targets,
thus aiding in developing new vaccines. In addi-
tion, some of the biochemical activities of the
influenza virus are specialized to the virus, mak-
ing them attractive as points of intervention in
the virus life cycle to which rationally designed
therapeutic agents can be developed. We are ana-
lyzing biochemical properties of the viral
glycoproteins.
We are also investigating the mechanism by
which integral membrane proteins are trans-
ported to the cell surface in the exocytotic path-
ways and are internalized from the surface by the
endocytotic pathways. We are studying the seven
integral membrane proteins encoded by influ-
enza virus and SV5 — three of which were discov-
ered in our laboratory — because they provide a
diverse group of membrane proteins that span the
cell membrane once.
Virus Cation Channels
Influenza virus protein M2 is a small (97-resi-
due) type-Ill integral membrane protein that
forms a disulfide-linked tetramer. The sensitivity
of influenza virus to the drug amantadine hydro-
chloride, the coupling of antiviral action to the
M2 transmembrane domain, and the premature
acid-induced conformational change in the viral
hemagglutinin in the presence of the drug sug-
gest that M2 is an ion channel, that it is essential
for virus uncoating in secondary endosomes, and
that it can alter the intracellular pH of the trans
Golgi network. In collaboration with Lawrence
Pinto, Northwestern University, we have tested
the M2 protein for ion channel activity by inject-
ing M2 mRNA into Xenopus oocytes and measur-
ing surface currents with a two-electrode patch-
clamp apparatus. We have shown that expression
of the protein is associated with an ion channel
activity selective for monovalent ions.
Amantadine hydrochloride significantly atten-
uated the inward current induced by hyperpolar-
ization of oocyte membranes, and mutations in
the M2 membrane-spanning domain that confer
viral resistance to amantadine produced currents
that were resistant to the drug. Thus we have pro-
vided direct data on the antiviral drug's mecha-
nism of action.
The M2 protein does not have the molecular
structure of most ion channels cloned to date. We
had to perform a large number of experiments to
eliminate the possibility that the M2 protein was
not a regulator that activates a normally silent
channel endogenous to oocytes. Our analysis of
distinguishing characteristics of the currents pro-
245
Structure and Replication of Influenza Virus and Paramyxoviruses
duced by various M2 proteins with changes in the
transmembrane domain suggests that the domain
forms the channel pore, and that the M2 protein is
therefore a channel per se.
Our experiments have also indicated that the
M2 protein channel activity is activated by lov^
pH, suggesting that the channel is only switched
on in endosomes and the trans Golgi network —
-intracellular compartments with lowered pH.
We are currently beginning a detailed structure-
function analysis of this channel to characterize
further the residues involved in drug sensitivity
and ion specificity.
We have also shown that the influenza B virus
NB glycoprotein, which has a similar overall
structure but no obvious amino acid homology to
the influenza A virus M2 protein, has ion channel
activity when expressed in oocytes. However, the
conductance of the NB ion channel activity is spe-
cific for chloride ions and is regulated by calcium
ion concentration.
The proposed pivotal role of ion channels in
the replicative cycle of the influenza viruses sug-
gests that the proteins present an important target
for a point of intervention by drugs (in addition
to amantadine) in the prophylaxis and therapy of
virus infections. We are currently testing whether
other viral proteins with similar structures to M2
and NB have ion channel activities — e.g., the
paramyxovirus SH protein, the vpu integral mem-
brane protein of the human immunodeficiency
virus I, and the C3 protein of the coronavirus
avian infectious bronchitis.
Intracellular Transport of Glycoproteins
To elucidate the rules that govern protein ori-
entation in the lipid bilayer, we are examining
how polypeptides are initially inserted into the
endoplasmic reticulum (ER) and are determining
the signals necessary for the protein-bilayer inter-
action. One of the major factors is the presence of
positively charged residues flanking the hydro-
phobic membrane-spanning domain to retain a
region of the protein in the cytoplasm. We have
also been focusing on the factors and signals
needed to fold the primary polypeptide chain
once it has been translocated across the mem-
brane of the ER.
The cellular glucose-regulated protein GRP78-
BiP is a member of the HSP70 stress family of
gene products and is a resident component of the
ER, where it is thought to play a role in the fold-
ing and oligomerization of secretory and mem-
brane-bound proteins. GRP78-BiP also binds to
malfolded proteins and this may be one mecha-
nism for preventing their intracellular transport.
During folding, the SV5 hemagglutinin-neuramin-
idase (HN) glycoprotein specifically and tran-
siently associates with GRP78-BiP. This complex
formation can only be detected prior to oligo-
merization of the immature HN molecules to
form the native tetramer, suggesting that GRP78-
BiP acts as a chaperone to promote correct fold-
ing of the molecule.
Internalization and Degradation
of Glycoproteins
The SV5 HN glycoprotein is extensively inter-
nalized from the virus-infected cell surface and
degraded in lysosomes. We are making an exten-
sive study of the mechanism of HN internaliza-
tion. This is of considerable interest because HN
lacks an aromatic amino acid in its cytoplasmic
tail that has been found necessary for the internal-
ization of several well-characterized receptor
molecules internalized by the clathrin-coated ves-
icle pathway. Examination of chimeric mole-
cules constructed between HN and another type-
II integral membrane protein that is not
internalized, influenza virus neuraminidase, sug-
gests that the HN transmembrane domain signals
internalization from the cell surface and specifies
targeting to lysosomes.
Some aspects of research in our laboratory on
influenza virus and paramyxoviruses are sup-
ported by grants from the National Institutes of
Health.
246
Cancer and Genetic Modification
Philip Leder, M.D. — Senior Investigator
Dr. Leder is also John Emory Andrus Professor in the Department of Genetics at Harvard Medical School.
He received his M.D. degree from Harvard Medical School. He has also received three honorary D.Sc.
degrees. Dr. Leder held several positions at NIH before returning to Harvard. His many honors include the
Albert Lasker Medical Research Award, the National Medal of Science, and the Heinekin Prize awarded by
the Royal Netherlands Academy of Arts and Sciences. He is a member of the National Academy of Sciences.
THE growth of cells in an organism is far too
delicate a process to be left to chance.
Rather, as with all biologic processes, it is subject
to a very stringent set of rules that are pro-
grammed into the makeup of the organism. The
basis for the control of growth is genetic. The
genes set the parameters that allow, say, the liver
to take the shape it does or the kidney to assume
its particular size and function. Genes establish
the rules whereby an organ grows in an orderly
fashion and reaches a prescribed and limited size.
Thus growth can proceed so far but no farther,
attaining a programmed equilibrium compatible
with life.
Cancer as a Disease of Genes
Cancer is a profound disorder of cell growth.
The delicate balance established by a genetically
encoded program is overturned. Instead of reach-
ing an equilibrium, the cancer cell no longer re-
sponds to signals that would limit its ability to
divide. It is out of control, and its unlimited
growth has profoundly dangerous consequences
for the organism.
Over the past decade or so, it has become in-
creasingly clear that many cancers can be ac-
counted for, at least in part, by damage occurring
to genes that encode the rules for control of cell
growth. Genetic damage, or mutation, can inacti-
vate a gene or cause it to function at the wrong
time or in the wrong place or, indeed, even cause
it to make the wrong product. The set of genes
whose mutation can give rise to cancer is often
just those that normally regulate cell growth. Ge-
neticists refer to the damaged genes that contrib-
ute to the development of malignancy as onco-
genes (from the Greek ovkoc, or tumor).
Transgenic Mice and the Genetic Basis
of Cancer
For some time my colleagues and I have been
interested in genes that control cell growth. Our
work has been considerably advanced by the tech-
nique of introducing active oncogenes into the
hereditary makeup of special strains of laboratory
mice. Called "transgenic," such mice carry on-
cogenes created in the laboratory, pass on these
cancer-causing genes to offspring, and therewith
transmit a strong tendency to develop cancer.
Thus, in many ways, transgenic mice become use-
ful models of human malignancy.
For example, we have designed specific mice
that develop cancer of the breast and others that
develop cancer of the blood cells — specific leu-
kemias and lymphomas. Some of these mice even
develop benign prostatic hypertrophy, a condi-
tion that aff^ects up to 85 percent of men by the
eighth decade of life. These experiments have
taught us that certain cancers can be caused by
specific oncogenes and that many, but not neces-
sarily all, cancers are the result of a collaboration
between two or more oncogenes. This suggests
that cancer is often a "multihit" process, one that
requires several activating events.
A Binary System for Activating
and Silencing Transgenes
During the past year we have extended the
power of transgenic technology by creating a sys-
tem that gives us much better control over the
transgene we have introduced. For example, we
often introduce genes that dramatically increase
the incidence of certain cancers in our mice.
Cancers obviously influence the ability of our an-
imals to pass their genes on to succeeding genera-
tions, as such genes often preclude survival. To
overcome this problem and to assure that no
more cancer-prone mice are produced than we
need for our experiments, we have designed a
binary system in which "target" genes can be
held in an inactive state in one line of mice and
become active only when the mice are bred to a
second line that carries an "activator" gene. The
system is suggestive of epoxy cement that is held
in two tubes, the contents of which must be
mixed to become functional.
The binary system has the further advantage
that it can be "multiplexed," or used in a variety
of combinations, such that a target can be one of
many different transgenes that is in turn com-
bined with one or more activator genes. For exam-
ple, the activators could specify expression of a
247
Cancer and Genetic Modification
target in many different organs at different times.
A further interesting aspect of this system is that
the unity of biology permits elements of the sys-
tem that have been derived from the regulatory
machinery of a simple unicellular organism,
yeast, to be used quite readily in mammals.
Host Defenses Against Cancer
While transgenic mice are very useful in analyz-
ing the action of oncogenes, they are also useful
in exploring the host defense mechanism mobi-
lized to prevent the development and spread of
cancer. The immune system is one of the organ-
ism's chief instruments against the spread of in-
fectious disease and for the rejection of foreign
tissues. For instance, the body's immune system
must be neutralized to accomplish effective heart
or kidney transplants. The role that the immune
system plays in tumor rejection is poorly under-
stood, but important discoveries in immunology
indicate that the immune response is regulated
by an array of hormone-like agents called lym-
phokines. These are released by cells of the im-
mune system to influence the growth and devel-
opment of other cells.
We are particularly interested in how lympho-
kines influence the host's response to cancer. In
the course of this work, we have focused on the
action of two particularly interesting lympho-
kines, IL-4 and IL-7. (These biologic response-
modifying agents are frequently called IL, de-
rived from "interleukin," an agent that mediates
signals between white blood cells, or leuko-
cytes.) Several cell-signaling functions have been
recognized in IL-4. Although its precise role in
the body has not been proved, IL-4 is suspected of
playing a role in modulating the immune re-
sponse. IL-7 is thought to be important for the
orderly development of antibody-producing
cells, the so-called B cells.
We have been able to show that both IL-4 and
IL-7 are potent antitumor agents, acting to induce
host defense mechanisms. Further studies of the
action of IL-4 have allowed us to identify two par-
ticular cell types that may mediate this antitumor
effect. One of these is the eosinophil, a cell in-
volved in many allergic responses. Another is the
macrophage, a scavenger cell concerned with
cell-killing functions. Our most recent work has
focused on identifying the active regions of the
IL-4 molecule, with a view to distinguishing
those structures required for antitumor activity
from those that give rise to unwanted side effects.
New Directions: Distinct Genetic
Contributions from Mothers and Fathers
Mendel's vision of genetics held that a particu-
lar gene carried by an organism behaves in ex-
actly the same way whether it is inherited from
the mother or the father. For the most part, this is
true. Nevertheless, in mammals, the highest or-
ganisms, it has proved impossible to induce de-
velopment artificially from the egg alone (a pro-
cess known as parthenogenesis) , although this is
possible in many lower forms. Some experiments
using transgenic mice have helped us to under-
stand this phenomenon and why it is that, at least
for mammals, mother and father (that is, both egg
and sperm) are necessary for development of the
offspring.
It turns out that a small number of genes are
expressed differently if they are inherited from
one or the other parent. In a particular example
that we created, a transgene is only expressed if it
is inherited from the father. The very same gene
inherited from the mother is silent. We have now
correlated this so-called "parentally imprinted"
expression with a chemical modification of the
gene whereby it is heavily altered (but not ex-
pressed) if inherited from the mother or not al-
tered (but expressed) if inherited from the fa-
ther. During the past year we have worked out the
rules that govern the modification of this gene
during embryogenesis. Our most recent work is
directed toward identifying the encoded signal
that evokes this parental effect.
248
From Molecular Biology to Therapy
of Human Disease
Fred D. Ledley, M.D. — Assistant Investigator
Dr. Ledley is also Associate Professor of Cell Biology and Pediatrics at Baylor College of Medicine. He
received his B.S. degree in physical sciences from the University of Maryland, College Park, and his M.D.
degree from Georgetown University. He trained in pediatrics and medical genetics at the Children 's
Hospital, Boston, and Harvard Medical School. His postdoctoral research was conducted with David
Baltimore at the Massachusetts Institute of Technology and with Savio Woo at Baylor College of Medicine.
SINCE its inception this laboratory has focused
on genetic deficiency of the enzyme methyl-
malonyl CoA mutase (MCM) as a model for molec-
ular genetic investigations of human disease. We
have cloned and sequenced human and murine
cDNAs for MCM, have mapped and cloned the
genomic locus in mice and humans, and have de-
scribed a series of mutations causing interesting
and informative phenotypes of this disorder.
While continuing these studies, we have begun
to address the technologies that will be essential
in moving toward somatic gene therapy for MCM
deficiency. We have focused our efforts on un-
derstanding the consequences of metabolic engi-
neering of MCM activity in human cells, on devel-
oping methods for efficient manipulation and
transduction of human hepatocytes, and on as-
sessing methods for hepatocellular transplanta-
tion in large animals and humans.
Retroviral-mediated Correction
of MCM in Vitro
We constructed a high-titer, amphotropic re-
troviral vector containing the full-length normal
human MCM gene. MCM-deficient fibroblasts
were transduced with this vector, and the effi-
ciency of transduction was assessed by semiquan-
titative identification of the recombinant pro-
virus. The restoration of MCM activity was
measured by the ability of cells to metabolize
['^C]-propionate. The metabolic capacity of fibro-
blasts was restored to normal levels by transduc-
tion, even though only 1 0-30 percent of the cells
were effectively transduced. No further increase
in metabolic capacity was evident when MCM
apoenzyme activity was increased above normal
levels by varying transduction or transfection
conditions. Transduction of normal fibroblasts or
hepatoma cells increased MCM apoenzyme activ-
ity, but not the capacity for propionate metabo-
lism, suggesting that other steps on these path-
ways are normally limiting.
Metabolic cooperation between cells was
shown to increase the flux of propionate through
subpopulations of metabolically competent
cells. The capacity for propionate metabolism of
hepatic cells was also found to be more than 10-
fold greater than the maximal capacity of fibro-
blasts, suggesting that the phenotypic impact of
metabolic engineering in hepatic cells would be
greater than in other potential targets for gene
therapy.
Transduction of Primary Human
Hepatocytes
One strategy for targeting gene therapy to the
liver involves harvesting and culturing hepato-
cytes from patients, transducing these cells ex
vivo with retroviral vectors, and returning them
to patients by hepatocellular transplantation. We
demonstrated the feasibility of hepatocellular
harvest from human liver segments preserved in
UW (University of Wisconsin) solution, demon-
strated proliferation of hepatocytes and preserva-
tion of differentiated hepatocellular functions in
cells cultured in hormonally defined media, and
established methods for transducing human cells
with recombinant retroviral vectors.
The efficiency of transduction of human cells
with amphotropic vectors (1-10 percent) was
significantly less than that observed in various an-
imal models. Vectors containing MCM were
shown to constitute expression of recombinant
mRNA from the 5' long terminal repeat (5'LTR)
promoter. Higher-efficiency transduction was
obtained using retroviral vectors with xenotropic
or gibbon ape enf determinants. Ongoing studies
are aimed at optimizing conditions for hepato-
cyte cultivation and transduction.
Hepatocellular Transplantation
One of the major factors limiting the applica-
tion of ex vivo strategies for hepatic gene therapy
is that hepatocellular transplantation has never
been attempted in clinical practice. We have as-
sessed the feasibility in large animal models, us-
ing a novel method for tracking transplanted
cells. The cells are stained with a fluorescent dye,
Dil. This dye is not metabolized or exchanged
between cells, and engraftment can be identified
in recipient tissues by fluorescent microscopy or
flow cytometry.
249
From Molecular Biology to Therapy of Human Disease
Studies in mice and rats confirm previous re-
ports that the efficiency of engraftment is rela-
tively poor (less than 1 percent) . We have stud-
ied autologous hepatocellular transplantation in
baboons as a model for human anatomy and surgi-
cal methods. Transplantation resulted in consti-
tution of approximately 5 percent of the host
liver from the hepatocellular graft without
, complications.
We have explored the possibility of hepatocel-
lular transplantation and gene therapy in utero.
Heterologous hepatocellular transplantation was
performed in fetal lambs at 80-85 percent gesta-
tion by infusion of cells into the umbilical vein.
Flow cytometry of hepatocytes recovered from
transplanted animals demonstrated that 1.5-4.5
percent of hepatocytes originated from the graft.
Transient Gene Therapy by in Vivo
Gene Delivery
Other laboratories have reported that DNA
coupled to asialoglycoproteins can be targeted to
the liver in vivo, but that expression of these
genes is transient (days) . We are interested in the
possibility that transient expression might be the
most efficacious approach to treating MCM defi-
ciency, which is characterized by periods of rela-
tive stability punctuated by life-threatening epi-
sodes of acidosis. We have demonstrated in vivo
delivery of MCM to the livers of experimental
mice using asialoorosomucoid-polylysine-DNA
conjugates and are currently studying the conse-
quences of this gene delivery. In particular, we
are concerned with demonstrating that DNA is
completely eliminated after the period of tran-
sient expression, since damage to chromoso-
mal DNA from inadvertent integrations is one
of the major potential risks. Vectors will be con-
structed with suicide sequences to eliminate in-
tegrated DNA.
Future Directions
Successful gene therapy requires attention not
only to methods for gene delivery and gene ex-
pression, but consideration of the metabolic, cel-
lular, clinical, and social consequences of ge-
netic manipulation. We are attempting to
establish a broad base of expertise and experi-
ence, using MCM deficiency as a model. This
should enable rational development of clinical
trials involving somatic gene therapy in the
future.
250
Molecular Biology of Hormone and Drug Receptors
in Health and Disease
Robert J. Lefkowitz, M.D. — Investigator
Dr. Lefltowitz is also James B. Duke Professor of Medicine and of Biochemistry at the Duke University
Medical Center. He received his B.A. ( chemistry) and M.D. degrees from Columbia University and clinical
and research training at Columbia- Presbyterian Medical Center in New York, NIH, and Massachusetts
General Hospital. As a molecular pharmacologist he has focused on the molecular structure and
regulatory mechanisms controlling the function of the adrenergic receptors that mediate the actions of
catecholamines. Dr. Lefkowitz has received numerous awards, including the Gairdner Foundation
International Award. He is a member of the National Academy of Sciences.
OUR research program is concerned with the
molecular properties and regulatory mecha-
nisms that control the function of plasma mem-
brane receptors for hormones and drugs under
normal and pathological circumstances. Recep-
tors are the cellular macromolecules with which
biologically active substances (e.g., hormones,
drugs, neurotransmitters, growth factors, viruses,
lipoproteins) initially interact. Such receptors
perform two essential functions: 1) They receive
or bind these biologically active substances at the
surface of the cell. 2) They transmit the sub-
stance's message into the cell, thus influencing
its metabolic activity and function.
We have utilized the receptors for epinephrine
(adrenaline) and related compounds, which are
generally termed adrenergic receptors, as models
for the study of receptors. Such receptors are
found throughout the brain, heart, smooth mus-
cle cells, and most other cells of the human organ-
ism. There are at least nine distinct subtypes of
adrenergic receptors (aj, a-^, 182, etc.). These
receptors interact not only with endogenous epi-
nephrine and norepinephrine but with a variety
of clinically important drugs.
We have studied these particular receptors for
several reasons: they are more widely distributed
than perhaps any other type of receptor, they are
exemplary of each of the major biochemical
pathways by which receptors are known to signal
to the interior of the cell, and they are clinically
and therapeutically significant. These receptors
mediate physiological responses as diverse as
changes in blood pressure, changes in heart rate
and contractility, and alterations in the metabo-
lism of glucose. Drugs that interact with these
receptors are among the most effective agents
used to treat various forms of heart disease, hy-
pertension, asthma, shock, and depression. Re-
search on these exemplary receptors has impor-
tant implications for understanding hormone and
drug receptor interactions generally and the
mechanisms by which they are regulated.
Our research is focused on several intimately
related goals. First, we wish to understand, in de-
tailed molecular terms, the biochemical nature
of the receptors. This is being accomplished by
the application of recombinant DNA or gene-
cloning techniques. These permit isolation of the
genes for the receptors, which in turn permits
their complete primary amino acid sequences to
be deduced.
We have isolated the genes for all of the known
adrenergic receptors, as well as a number of
closely related receptors, and determined their
complete sequences. Remarkably, the structures
of these receptors are similar to each other and to
that of the visual light receptor rhodopsin. This
insight is helping to clarify the general mecha-
nisms by which signals as divergent as a photon of
light and a drug molecule like epinephrine elicit
their characteristic biochemical and physiologi-
cal responses. By varying the structure of the gene
and hence the receptor protein that it produces,
we can define which structural features of the
receptor molecule determine its characteristic
functions.
An unexpected result of the cloning of the
known types of adrenergic receptor genes was
the discovery of the genes for several novel sub-
types of adrenergic receptors not previously
known to exist. Knowledge of these new recep-
tors opens the way for the development of new
drugs with greater selectivity and fewer side ef-
fects. Such drugs might conceivably have appli-
cations in the treatment of such disparate diseases
as hypertension, asthma, diabetes, and prostat-
ism, or even as novel anesthetic agents.
A second current research goal is to elucidate
the patterns of receptor regulation. One of the
most important insights to come from our studies
of receptors is that their properties are not fixed.
Rather, the properties of the receptors are in-
fluenced by the hormones and drugs with which
they interact, as well as by a variety of disease
states.
There are important clinical implications of
the ever-changing nature of the receptors. For ex-
ample, this provides a basis for beginning to un-
derstand the phenomenon of drug tolerance or
desensitization, the diminishing eff^ect of drugs
251
Molecular Biology of Hormone and Drug Receptors in Health and Disease
over time. This phenomenon markedly compro-
mises the therapeutic efficacy of epinephrine and
many other drugs. When drugs like epinephrine
combine with their receptors, they not only stim-
ulate them but also produce changes that impair
their function, thus leading to desensitization.
These changes involve an actual loss of receptors
from the cell surface (they move inside the cell
where they cannot function) and a chemical
change of those receptors remaining at the cell
surface so that they function less effectively. With
fewer functioning receptors present at their sur-
face, cells are less able to respond to drugs or
hormones.
Our recent research is increasing our under-
standing, in molecular terms, of how the recep-
tors become functionally desensitized. We have
discovered two new proteins that function to de-
sensitize the receptors. The first is an enzyme, the
jS-adrenergic receptor kinase (/3ARK) , which mod-
ifies the structure of the receptors by introducing
a phosphate group when the receptors are stimu-
lated. The second is the protein /3-arrestin, which
binds to the phosphorylated receptors and pre-
vents them from acting. Both proteins are widely
distributed at synapses throughout the central
nervous system, suggesting their actions are not
limited to the j8-adrenergic receptors.
The implications of such fundamental research
on receptors for clinical medicine are profound.
Elucidation of the detailed structure of the recep-
tors will allow the precise design of drugs that are
more potent and specific. Unraveling of the mo-
lecular basis of desensitization will allow the de-
velopment of strategies for interdicting the basic
reactions that lead to loss of hormone and drug
effect. An example is the design of specific en-
zyme inhibitors for ^ARK that could block the
reactions leading to desensitization. Successful
conclusion of such research may lead to methods
for greatly prolonging and augmenting the thera-
peutic actions of diverse types of drugs.
252
Axis Formation and Germline Determination
in Drosophila
Ruth Lehntann, Ph.D. — Assistant Investigator
Dr. Lehmann is also Associate Member of the Whitehead Institute of Biomedical Research, Associate
Professor of Biology at the Massachusetts Institute of Technology, and Assistant Molecular Biologist at
Massachusetts General Hospital, Boston. She received her M. Sc. degree from the University of Freiburg,
where she worked with Jose Campos-Ortega on early neurogenesis in Drosophila. She received her Ph.D.
degree from the University of Tubingen, where she worked with Christiane Ntisslein-Volbard at the Max
Planck Institute for Developmental Biology on the genetics of pattern formation in the Drosophila embryo.
After postdoctoral training in Tiibtngen and at the MRC in Cambridge with Michael Wilcox, she joined
the Whitehead Institute.
HOW does a developing embryo know where
to form a head and where to put a tail? In
Drosophila, basic information about the "coordi-
nates" of the embryo is supplied to the egg cell
by the mother. Mutations in maternal genes have
led to the identification of a small number of
genes that are required for the establishment of
anterior-posterior (head-to-tail) polarity. In these
mutants the lack of a particular gene product in
the mother is lethal to its progeny.
Three signals are required for the establish-
ment of pattern along the anterior-posterior axis.
The anterior signal controls the development of
head and thorax, the posterior signal determines
the abdominal region, and the terminal signal is
required at both ends of the embryo. The anterior
and posterior signals are localized to the respec-
tive poles (see figure) .
We have concentrated on the dissection of the
pathway leading to normal development of the
posterior region. Nine maternal genes, called the
posterior group, set the basic posterior pattern.
These genes share the abdominal phenotype: ho-
mozygous mutant females produce offspring that
lack abdominal segmentation. One gene, nanos,
encodes the localized signal required for the de-
velopment of the abdomen. Another gene, pumi-
lio, regulates the activity of NANOS protein,
while the remaining seven genes — oskar, vasa,
tudor, valois, staufen, cappuccino, and spire —
are required for the localization of nanos RNA to
the posterior pole.
These seven genes are also required for germ
cell formation. Embryos from mutant females
lack the specialized posterior pole plasm that
normally contains the polar granules. These em-
bryos therefore fail to produce pole cells, the
Drosophila germ cell precursors. Our molecular
analysis of the posterior group is aimed at under-
standing how nanos and other RNA species be-
come localized to the posterior pole and how the
pole plasm, composed of RNA and protein, is
assembled.
Synthesis and Function
of the Abdominal Signal
Through genetic experiments as well as cyto-
plasmic transfers between wild-type and mutant
embryos, we concluded that the specialized cyto-
plasm at the posterior pole is the source of an
abdominal signal, and that embryos derived from
females with mutant posterior group genes lack
this signal in the abdominal region. The signal is
encoded by nanos. We have isolated and cloned
this gene and have shown that nanos mRNA is
transcribed during oogenesis and becomes local-
ized to the posterior pole plasm of the mature
oocyte. The role of nanos as a signal for posterior
pattern formation is demonstrated by the finding
that injection of synthetic nanos mRNA into the
anterior pole of early embryos leads to suppres-
sion of head formation and induction of a second
abdomen in mirror image to the normal "poste-
rior" abdomen.
Seven of the posterior group genes affect the
localization of nanos mRNA to the posterior
pole. Since these seven mutants also lack pole
plasm, we can conclude that their abdominal seg-
mentation defect is a consequence of a lack of
localized nanos mRNA. We have identified se-
quences within the nanos mRNA required for its
localization. In future experiments we will deter-
mine which of the posterior group genes are di-
rectly involved in tethering nanos mRNA to the
posterior pole plasm.
Regulation of Posterior Activity
Several lines of evidence indicate that the pu-
milio gene is required to regulate the activity of
NANOS protein. After fertilization of the egg,
NANOS is translated from the localized mRNA and
emanates anteriorly toward the abdominal region
to form a gradient of protein concentration. Cyto-
plasmic transplantation experiments and analysis
of the distribution of nanos RNA and NANOS pro-
tein in pumilio mutant embryos show that pumi-
253
Axis Formation and Germline Determination in Drosophila
Ho does not interfere with nanos RNA localiza-
tion or NANOS protein distribution, synthesis,
and stability. We conclude that pumilio is re-
quired for optimal NANOS activity in regions of
the embryo where NANOS protein concentrations
are low.
We have cloned the pumilio gene and have
shown that pumilio mRNA is localized to the pos-
terior pole in early embryos. This observation is
consistent with a possible interaction between
the pumilio and nanos gene products. Localiza-
tion of pumilio mRNA, like that of nanos mRNA,
is disrupted in mutants that lack the specialized
posterior pole plasm (see below) . It is thus possi-
ble that the same mechanism of RNA localization
acts on both mRNAs.
Assembly of the Pole Plasm
The remaining seven genes in the posterior
group are associated with pole plasm defects. We
are studying these genes from two perspectives.
One is their role in pattern formation and body
segmentation (see above) ; the other, their role in
pole plasm assembly and germ cell formation
(see below) .
During the first hour of embryogenesis, the
pole plasm appears as a distinct clear zone at the
posterior end of the newly fertilized egg. Closer
examination reveals that this zone contains spe-
cialized cytoplasm packed with mitochondria
and numerous donut-shaped organelles, called
polar granules, that do not occur anywhere else
in the embryo. The zone of specialized cytoplasm
coincides with the site of germ cell formation,
and polar granules may therefore contain factors
that control germ cell fate. Polar granules were
first recognized in Drosophila embryos 30 years
ago, but until recently little was known about
their biochemical structure or function. Now the
link between defects in abdomen formation and
the absence of germ cells has provided a new ave-
nue for research.
We have begun extensive studies of the pole
plasm in mutant and normal embryos, using
probes designed to recognize mRNA and proteins
derived from the posterior group genes. We sus-
pect that oskar, one of the genes required for
abdominal and germline development in the Dro-
sophila embryo, is an essential component of the
pole plasm. We have isolated this gene. In situ
hybridization to whole-mount embryos reveals
that oskar mRNA distribution is restricted to the
posterior pole of early embryos. Genetic studies
suggest that pole plasm formation occurs in a se-
ries of steps, with each step dependent on the
previous one, and that oskar is required early in
the assembly pathway. Future experiments will
address questions about the role of oskar mRNA
and protein for pole plasm formation and for lo-
calization of the abdominal signal.
We are optimistic that new information about
the assembly and composition of the pole plasm
and polar granules will lead to better understand-
ing of their functions in the early embryo. Their
primary role may be to protect or sequester infor-
mation required by the future germ cells. In addi-
tion, however, they could provide a convenient
anchor for substances like the nanos mRNA that
must be retained in the posterior region. Struc-
tures similar to the polar granules in Drosophila
have been observed in association with germ
plasm in many different invertebrate and verte-
brate animals. Thus the Drosophila model could
reveal basic concepts underlying the establish-
ment of germline tissues in all species.
The National Institutes of Health provided fi-
nancial support for two projects that are summa-
rized here. These projects focus on the molecular
and functional characterization of the nanos and
oskar genes.
Localization of positional signals in the Dro-
sophila embryo. In situ hybridization experi-
ment, using biotinylated cDNA probes, local-
izes RNA of the anterior signal bicoid and of the
posterior signal nanos respectively to the ante-
rior and posterior poles of an early ( stage 2 )
embryo. Hybridization to nanos at the posterior
also demarcates the position of the germ plasm.
Research by Laura Dickinson. Photograph by
Ruth Lehmann.
254
Regulation of Gene Expression During Cellular
Differentiation and Activation
Jeffrey M. Leiden, M.D., Ph.D. — Associate Investigator
Dr. Leiden is also Associate Professor of Internal Medicine at the University of Michigan Medical School. He
received his M.D. and Ph.D. degrees from the University of Chicago. His residency training was at Brigham
and Women's Hospital, Boston, and his postdoctoral fellowship was in the laboratory of Jack Strominger
at Harvard University.
THE processes of cellular differentiation and
activation are accompanied by complex and
precisely orchestrated changes in gene expres-
sion. Abnormalities in the patterns of expression
of these specific genes may be involved in the
etiology of a number of pathologic states, includ-
ing autoimmune disease and malignancy. My labo-
ratory is studying gene regulation during T lym-
phocyte and muscle cell differentiation in order
to increase understanding of the molecular mech-
anisms that regulate gene expression during both
normal and pathologic development.
Regulation of the T Cell Receptor Gene
During T Cell Development
Human T lymphocytes recognize foreign anti-
gens, such as virus- infected cells and tumor cells,
via a specific cell-surface molecule, the T cell
receptor (TCR) . T cells can be divided into two
subsets based on their expression of two distinct
types of antigen receptor. The majority of T cells
that circulate in peripheral blood, including all
helper and cytotoxic T cells, express the a//3
heterodimer receptor, while a small but distinct
T cell subset of unknown function expresses the
7/6 receptor. The a//3 and 7/6 T cells appear to
develop as separate lineages during thymic
ontogeny.
During the last several years, my laboratory has
been interested in identifying and characterizing
the molecular mechanisms that regulate the ex-
pression of these different TCR genes during T
cell development. In an initial set of studies, we
identified the transcriptional enhancer elements
that control the expression of the TCR a and /5
genes. These two enhancers were shown to be
required for the expression of the a and genes
and to function equally well in both mouse and
human cells. The identification and localization
of the human TCR enhancers led us to propose
that certain T cell tumors that had previously
been shown to contain chromosomal trans-
locations into the human TCR a and loci
might be caused by the apposition of the
TCR gene enhancers with translocated human
proto-oncogenes.
Our more recent studies have focused on pre-
cisely identifying and characterizing the en-
hancer DNA sequences and the nuclear proteins
they bind that are responsible for controlling
TCR a and /3 gene expression. These experiments
have demonstrated that both enhancers contain
4-5 different binding sites for nuclear proteins.
At least two of these sites in each enhancer are
absolutely required for enhancer activity. Several
of the nuclear protein-binding sites in each en-
hancer correspond to previously defined en-
hancer motifs, while others represent novel se-
quence elements. At least one site in each
enhancer was shown to bind T cell-specific nu-
clear proteins.
Both enhancers were shown to contain a bind-
ing site that was identical to the previously de-
scribed cAMP response element (CRE). To learn
more about the function of this element in regu-
lating T cell gene expression, we have cloned
two novel CRE-binding proteins, CREB-2 and
CREB-3, that specifically bind to the TCR a CRE
and to CREs from several other eukaryotic pro-
moters. Both of these new CREB proteins contain
similar basic DNA-binding domains and a leucine
zipper region that allows them to form dimeric
complexes both with themselves and, poten-
tially, with other members of the CREB protein
family.
During the past year, we have shown that ets-\,
a. previously described human proto-oncogene, is
actually a DNA-binding protein that specifically
recognizes one of the nuclear protein-binding
sites in the TCR a enhancer. These studies have
helped to define the Ets proteins as transcrip-
tional regulators that may play an important role
in controlling T cell gene expression. More re-
cently, we have cloned several new members of
the Ets family of transcription factors. One of
these, which we call Elf- 1 , regulates a set of genes
during the process of T cell activation. In addi-
tion, it appears to play an important role in regu-
lating the expression of HIV-2, one of the AIDS
viruses, in T cells. Finally, in collaboration
with Stuart Orkin (HHMI, Children's Hospital,
Harvard Medical School), we have identified a
255
Regulation of Gene Expression During Cellular Differentiation and Activation
novel T cell-specific transcription factor called
hGATA-3 that binds to a third nuclear protein-
binding site within the TCR a enhancer. Ongoing
studies are designed to determine the role of each
of these transcription factors in T cell tumors, as
well as in normal T cell development and
activation.
Genetically Engineered Myoblasts
as a Recombinant Protein Delivery System
A variety of acquired and inherited diseases are
currently treated by repeated intravenous or sub-
cutaneous infusions of recombinant or purified
proteins. In addition to hemophilia A, which is
treated with intravenous infusions of human fac-
tor VIII, these include diabetes mellitus, treated
with subcutaneous or intravenous injections of
insulin, and pituitary dwarfism, treated with sub-
cutaneous injections of growth hormone. The de-
velopment of cellular transplantation systems
that could stably produce and deliver such recom-
binant proteins into the systemic circulation
would represent an important advance in our abil-
ity to treat such diseases.
The ideal recombinant protein delivery system
would utilize a cell that could be easily isolated
from the recipient, grown and transduced with
recombinant genes in vitro, and conveniently
reimplanted into the host organism. Such a cell
should produce large amounts of secreted recom-
binant protein, and following secretion, this pro-
tein should gain access to the circulation. Finally,
these implanted, genetically engineered cells
should survive for long periods and continue to
secrete the transduced protein product without
themselves interfering with the function of the
tissue into which they were implanted. Several
different cellular systems have been used to pro-
duce recombinant proteins in vivo. These in-
clude keratinocytes, skin fibroblasts, hepato-
cytes, lymphocytes, and bone marrow. Although
some of these systems have yielded detectable lev-
els of circulating proteins briefly, stable physio-
logical levels of circulating proteins have proved
difficult to produce in normal animals.
Genetically engineered myoblasts represent a
potentially useful system for the in vivo delivery
of recombinant proteins into the circulation.
Myoblasts can be readily isolated from a muscle
biopsy and expanded in vitro to very large cell
numbers. Cultured myoblasts can be transfected
in vitro and will synthesize large amounts of re-
combinant proteins. Most importantly, previous
studies have demonstrated that cultured myo-
blasts can be injected intramuscularly and will
survive and fuse into adjacent normal muscle
fibers at the site of injection. Finally, skeletal
muscle is a highly vascular tissue. Thus proteins
secreted from myoblasts should readily enter the
circulation.
We have recently explored the feasibility of us-
ing genetically engineered myoblasts as a recom-
binant protein delivery system. To this end,
stable transfectants of the murine C2C12 myo-
blast cell line were produced that synthesize and
secrete high levels of human growth hormone
(hGH) in vitro. These stably transfected myo-
blasts were injected intramuscularly into normal
syngeneic C3H mice, and serum and muscle lev-
els of hGH were measured 5 days to 3 weeks after
injection.
Mice injected with the growth hormone-
transfected myoblasts produced significant and
physiological levels of hGH both locally in mus-
cle and in serum, as compared with control mice
injected with nontransfected myoblasts. Human
growth hormone levels in both muscle and serum
were stable for at least three months following
injection and exceeded those measured in
serum from normal human volunteers. Histologi-
cal examination of muscles injected with /3-
galactosidase-expressing C2C12 myoblasts dem-
onstrated that many of the injected cells had
fused to form multinucleated myotubes. Thus
these studies demonstrated that genetically engi-
neered myoblasts represent a novel and powerful
system for the stable delivery of recombinant
proteins into the circulation.
Dr. Leiden is now Professor of Medicine and
Pathology at the University of Chicago.
256
Chemical Communication
Michael R. Lemer, M.D., Ph.D. — Associate Investigator
Dr. Lemer is also Associate Professor in the Departments of Internal Medicine and of Pharmacology and
the Child Study Center at Yale University School of Medicine. He obtained his B.A. degree in chemistry
from the University of Pennsylvania and his M.D. and Ph.D. degrees from Yale. His doctoral research, with
Joan Steitz, was on small nuclear ribonucleoproteins (snRNPs). He interned in internal medicine at
Barnes Hospital, St. Louis, and did postdoctoral research in neurobiology at Washington University with
Gerald Fischbach before returning to Yale. His honors include the George Herbert Hitchings Award for
innovative methods in drug design.
INTRASPECIES communication via specific
chemical messengers is widely employed
throughout the animal kingdom. Among the com-
mon uses of chemical communication are mark-
ing of territory, signaling danger, and indicating
sources of food. A particularly striking example is
the use of sex pheromones. Here, animals release
a defined blend of related molecules that trigger
distinct mating behaviors in members of the op-
posite sex. For detection to occur, three criteria
must be satisfied. Molecules of the pheromone
must reach olfactory receptors, must interact
with them, and must be inactivated so that subse-
quent molecules can be detected.
Past research has focused on the problems of
chemical transport to olfactory receptors and in-
activation of these molecules. Moths — particu-
larly Manduca sexta and Antheraea polyphe-
mus — have provided excellent models. Many of
the olfactory receptor cells of the male M. sexta
and most of those from the male A. polyphemus
are specialized for detecting sex pheromone. For
both animals, the pheromone-binding proteins,
which solubilize molecules of the pheromone
blend and carry them to receptors, have been
characterized. In addition, a family of general
odorant-binding proteins, which are related to
the ones employed by the moths to carry phero-
mone, has been discovered and characterized.
Likewise, the enzymes that rapidly and specifi-
cally inactivate pheromone molecules and are ap-
parently requisite to the sensory apparatus em-
ployed by males to locate females, have been
investigated at the biochemical level. For M.
sexta, in which both major components of the
pheromone blend are aldehydes, a single en-
zyme, an aldehyde oxidase, suffices, while A.
polyphemus employs both an aldehyde oxidase
and an esterase because its pheromone compo-
nents include both an aldehyde and an ester. Our
research has now turned to developing methods
for investigating olfactory receptors themselves.
Over the past few years many laboratories have
conducted biochemical, electrophysiological,
and molecular cloning experiments concerned
with the nature of signal transduction by olfac-
tory receptors. Depending on the animal whose
sense of smell is being investigated, the results
indicate that either cAMP, IP3/DAG (IP3, inositol
1,4,5-trisphosphate; DAG, diacylglycerol) , or
both second messenger systems are involved. It
now appears that the study of olfaction, and
hence a major aspect of chemical communication
between animals, is part of the general problem
of how G protein-coupled receptors work.
At the present time, several methods are em-
ployed to study how receptors that regulate intra-
cellular concentrations of cAMP or IP3/DAG
work, such as adenylate cyclase assays, radioim-
munoassays, measurements of IP3 or DAG, moni-
toring the flow of ions through channels in frog
oocyte membranes, or looking at changes in in-
tracellular calcium concentrations via appro-
priate fluorescent indicators. To study G pro-
tein-coupled receptors, and ones relevant to
olfaction in particular, we are developing a new
method for monitoring receptor stimulation that
can track changes in intracellular concentrations
of cAMP or IP3/DAG in over 10,000 individual
cells simultaneously.
To follow changes in cAMP or IP3/DAG levels
in many individual cells at the same time, we
have turned to that ability of some animals to
change colors rapidly. In nature, color changes
are used for such purposes as camouflage and the
communication of states of emotion. Among ver-
tebrates, quick color alterations are brought
about by the controlled movement of pigment
granules within chromatophores. When pigment
granules in melanophores (a particular type of
chromatophore) are aggregated, the animal ap-
pears light, and when pigment is dispersed, the
animal appears dark. It turns out that the pigment
translocation apparatus is controlled via second
messenger systems that are themselves regulated
by G proteins. As a result, the state of pigment
disposition within melanophores reflects the
state of activity of G protein-coupled receptors.
Recently the laboratory has successfully har-
nessed frog melanophores as the center of an
257
Chemical Communication
assay for monitoring the activity of G protein-
coupled receptors that act to raise or lower intra-
cellular cAMP or that raise intracellular IP3/DAG.
The ability of exogenous receptors to couple to
and control the pigment translocation appa-
ratus within melanophores should lead to im-
proved methods for studying G protein-coupled
receptors.
258
Structural Determinants of Human a-Glohin
Gene Expression
Stephen A. Liebhaber, M.D. — Investigator
Dr. Liebhaber is also Professor of Genetics and Medicine (Hematology) at the University of Pennsylvania
School of Medicine. He received his B.A. degree in chemistry from Brandeis University and his M.D. degree
from Yale University. He took clinical training in internal medicine and hematology at Case Western
Reserve, the University of Colorado, Washington University, and the University of California, San
Francisco. As a postdoctoral fellow with David Schlessinger at Washington University, Dr. Liebhaber
examined ribosomal RNA processing, and with Yuet Wai Kan at the University of California, San
Francisco, he studied human globin gene expression and genetic defects in a-thalassemia. Before moving
to Philadelphia, he was a faculty member of the Department of Medicine at UCSF.
OUR laboratory has largely concentrated on
studying the expression of the human glo-
bin genes. These genes encode hemoglobin, the
major red cell protein responsible for transport of
oxygen from the lungs to peripheral tissues.
Since the hemoglobin molecule, a2i^2> is com-
posed of an equal number of a- and ;8-globin
chains, normal synthesis demands balanced ex-
pression of both sets of genes, which are located
on different chromosomes. Defects in either set
result in an imbalance of expression and conse-
quent hereditary anemia: a- or (S-thalassemia.
Thalassemias result from more than 150 different
mutations in the globin genes, affecting the
health of millions worldwide.
Certain characteristics of globin gene expres-
sion make it particularly interesting for study.
The extremely high level of globin mRNA in the
differentiating red cell (over 95 percent of total
cellular mRNA) has no equal in any other cell
type. This abundance reflects both high levels of
globin gene transcription and an unusual stability
of the mature globin mRNA.
An additional interesting aspect of globin ex-
pression is that the genes in the a- and |S-globin
gene clusters follow an orderly sequence of ex-
pression during embryologic development. This
results in a well-defined switch from embryonic
to adult globin gene expression during develop-
ment of the fetus. The switching results in the
synthesis of successive hemoglobins with oxygen
affinities that match changes in the uterine envi-
ronment. The active transcription, the clearly de-
fined pattern of developmental switching, and
the unusual mRNA stability are areas of special
focus in our laboratory.
The loss of a-globin expression observed in
a-thalassemia usually results from deletion or ab-
normal structure of one or more of the a-globin
genes. We have recently studied three unrelated
patients from Germany, Portugal, and Italy who
have an unusual form of a-thalassemia. In each
case the loss of a-globin synthesis reflects loss of
expression of a structurally normal a-globin clus-
ter. In other words, one of the a-globin gene clus-
ters in each of these patients is not functioning,
even though the genes in these clusters are struc-
turally normal and synthesize normal levels of
a-globin when isolated and expressed in tissue
culture cells. Although initially puzzling, the
mechanism of this remarkable expression defect
was eventually defined by extensive DNA map-
ping. In each of the three independent cases we
demonstrated a large deletion 5' to the inacti-
vated cluster itself. In one case this deletion be-
gan as much as 50,000 bases 5' to the silenced
a-globin genes. By comparing the maps of each of
these deletions, a region of common overlap was
noted. These studies define a new category of
a-thalassemia and demonstrate a critical determi-
nant of a-globin gene expression located entirely
external to the a-globin gene cluster. A similar set
of transcriptional signals has been localized adja-
cent to the j8-globin gene cluster. One can there-
fore speculate that such signals serve coordi-
nately to activate and balance the expression of
the a- and ;8-globin clusters in the red cell.
The human a-globin gene cluster contains a
globin gene expressed specifically in the embryo
and two a-globin genes, al and a2, expressed in
the fetus and the adult. The switch from embry-
onic f-globin to adult a-globin expression occurs
at the end of embryonic development (7-8
weeks of gestation) . This critical developmental
switch, which occurs widely in mammals, pres-
ents a well-defined model system for studying de-
velopmental control of gene expression.
To establish a system in which to study switch-
ing within the human a-globin gene cluster, we
have injected the human embryonic and adult
a-globin genes into fertilized mouse eggs to gen-
erate transgenic mice. The red cells of these
transgenic mice appropriately express the human
transgenes during development. In the embry-
onic period, there is a parallel expression of the
human and mouse f-globin genes and, by day 1 2
of development, parallel expression of the
a-globin genes. These data suggest that 1 ) the hu-
259
Structural Determinants of Human a-Globin Gene Expression
man transgenes contain the necessary informa-
tion for appropriate developmental control and
2) the factors responsible for developmental
switching in mouse red cells have been suffi-
ciently conserved during evolution to substitute
in the control over the human transgenes. By gen-
erating transgenic mice that carry only the human
a-globin gene in the absence of the f-globin gene,
we have further demonstrated that the appro-
priate expression of this gene is not dependent on
competition between the two developmentally
specific genes. By introducing into the mouse ge-
nome more limited gene fragments, as well as
genes with specific alterations, and by studying
their pattern of developmental expression in the
transgenic model system, we should be able to
define the signal (s) critical to globin gene
switching.
Selective stabilization of globin mRNA is essen-
tial to its accumulation in terminally differentiat-
ing red cells to 95 percent of total cellular mRNA.
We are attempting to define the structural basis
for the stability of a-globin mRNA through inves-
tigation of an a-thalassemia mutation in which
this stability is lost: a Constant Spring (aCS) . This
mutation, the most common cause of nondele-
tional a-thalassemia in Southeast Asia, is a CAA for
UAA substitution at the normal termination codon
of the a2-globin gene. As a result of this single-
base substitution, the ribosome translates into the
normally untranslated 3' region and destabilizes
the mRNA.
To study the basis for this destabilization in de-
tail, we have established an experimental system
that reproduces the selective instability of aCS
mRNA in tissue culture cells. Remarkably, we
find that the instability of the aCS mRNA relative
to normal a-globin mRNA observed in the patient
is recapitulated faithfully when the a- and aCS-
globin genes are expressed in an erythroid tissue
culture cell line. In contrast, we find that the
mRNAs from these two genes are expressed at
equivalent levels when expressed in nonery-
throid cell lines. These data suggest that the sta-
bility of globin mRNA may depend on interaction
with one or more erythroid-specific factors.
By specifically altering the structure of the a-
globin genes prior to expression, we have demon-
strated that the region critical to the stability of
the a-globin mRNA is located in a segment of the
3'-nontranslated region just past the translation
termination codon. This determinant can be de-
stabilized by the translating ribosome if it is al-
lowed to enter this region, as is the case in the
aCS mutation, or by site-specific alterations in
the primary sequence of this segment. These data
suggest that it should now be possible to map the
limits of this determinant and subsequently iden-
tify critical cellular factors mediating this
response.
260
The Heat-Shock Response
Susan L. Lindquist, Ph.D. — Investigator
Dr. Lindquist is also Professor in the Department of Molecular Genetics and Cell Biology and in the
Committees on Developmental Biology and Genetics at the University of Chicago. She received her B.S.
degree in microbiology from the University of Illinois, where she worked with John Drake on
bacteriophage T4. Her graduate research was done with Matthew Meselson at Harvard University, where
she began her studies on the heat-shock response. She continued this work during her postdoctoral
research with Hewson Swift at the University of Chicago.
THE causes of heat-induced lethality and the
mechanisms that cells employ to protect
themselves from heat damage are poorly under-
stood. Over the past decade, a great deal of re-
search has focused on a small group of mole-
cules, the heat-shock proteins (HSPs). These are
induced in response to temperature elevation
and a wide variety of other stresses. No known
genetic induction is more highly conserved in
evolution, which underscores its fundamental
importance in biology. Archaeobacteria, eubac-
teria, plants, and animals all produce similar pro-
teins. Several of these proteins show very high
levels of conservation, commonly with 40-50
percent amino acid identity between HSPs of hu-
man and bacterial cells.
The induction of HSPs 1) allows cells to grow
at the upper end of their normal growth range, 2)
potentiates survival during long exposures to tem-
peratures just beyond the normal growth range,
and 3) protects cells from lethality at tempera-
ture extremes. Interestingly, different proteins
are required for each of these functions. HSPs
also protect cells from heavy metal ions, ethanol,
and many other sources of stress. The importance
of different proteins in protecting against differ-
ent forms of stress also varies.
The stress inductions of HSPs are of interest to
human biology and medicine for several reasons.
First, studies of cultured cells in vitro and of tu-
mors in vivo demonstrate that many cancer cells
are more readily killed by heat than are untrans-
formed cells. For this reason, hyperthermia, in
conjunction with radiation and chemotherapy, is
emerging as an important new tool in cancer ther-
apy. Second, high temperatures are associated
with a number of developmental anomalies in a
wide variety of plants and animals, including
spina bifida in humans. In those organisms that
have been subjected to experimental manipula-
tion, mild preheat treatments, which induce the
HSPs, provide protection. Third, the induction of
HSPs is associated with a variety of human patho-
logical states, including strokes, heart attacks,
and kidney disease. Interest in the proteins in-
cludes both their putative protective functions in
affected tissues and the possibility of quantifying
them as disease markers. Fourth, the proteins in-
teract with and potentiate the function of many
other vital proteins in the cell.
The heat-shock response also provides a superb
model system in which to study the cellular
mechanisms involved in regulating protein syn-
thesis. Because HSPs are required for survival, a
number of regulatory mechanisms are employed
to ensure that the proteins will be produced as
rapidly as possible after exposure to stress. Thus
studies of the response have provided funda-
mental insights on the nature of nuclear and cy-
toplasmic regulation in both eukaryotes and
prokaryotes.
The recent discovery that the HSPs themselves,
or close relatives produced at normal tempera-
tures, play vital roles during normal growth and
development has opened up a whole new field of
investigation. The specific molecular functions
of the HSPs are only beginning to be elucidated,
but they play a role in a remarkable number of
basic cellular processes, including secretion, sig-
nal transduction, and ribosome assembly. Deter-
mining the roles that HSPs play in these processes
will provide fundamental insights in cell biology.
We are investigating the regulation and func-
tion of these proteins. Our research focuses on
the yeast Saccharomyces cerevisiae and the fruit
fly Drosophila melanogaster, because tech-
niques of genetic manipulation and molecular
analysis are so advanced in these organisms. For
the past few years our investigations of the regula-
tion of the response have concentrated on post-
transcriptional mechanisms that are employed to
maximize the synthesis of the HSPs during heat
shock or to shut off synthesis after heat shock.
Tom McGarry and Bob Petersen found that
heat-shock mRNAs in Drosophila cells are prefer-
entially translated during heat shock by virtue of
sequences in their 5'-untranslated leaders and are
preferentially repressed during recovery through
sequences in their 3'-untranslated tails. The latter
sequences are shared by certain normal cellular
messages, which have the common property of
being rapidly degraded at normal temperatures.
261
The Heat-Shock Response
Heat-shock regulation takes advantage of this
common pathway to control HSP expression. The
mechanism is inactivated during heat shock and
restored during recovery.
Joseph Yost demonstrated that heat shock
blocks the processing of mRNA precursors,
which explains why heat-shock genes generally
do not have intervening sequences. (If they had,
the block in splicing would prevent expression.)
Sudden high-temperature heat shocks also inhibit
transcription termination (discovered by Robert
Dellavalle). By some mechanism we do not yet
understand, heat-shock genes are more likely to
be terminated correctly than normal cellular
genes.
Although our studies of HSP function are in
many ways independent of our studies on regula-
tion, in one important respect they overlap. HSPs,
and hsp70 in particular, play an important role in
restoring normal gene expression patterns after
heat shock. They are required at the level of
translation, RNA turnover, RNA processing, and
transcription.
To investigate the function of the HSPs, we cre-
ated a series of mutations in the genes of both
yeast and Drosophila. Kathy Borkovich found
that hsp82 is essential for growth at all tempera-
tures in yeast cells, but is required at higher con-
centrations for growth at high temperatures. Thus
induction is required for cells to grow at the up-
per end of their normal temperature range. We
believe the protein is needed to regulate the activ-
ity of a wide variety of other cellular proteins and
that it is needed at higher concentrations at high
temperatures in order to drive the equilibrium of
these interactions toward complex formation.
In collaboration with Keith Yamamoto's labora-
tory, Bushra Khursheed and Marc Fortin demon-
strated that hsp82 interacts with the steroid hor-
mone family of receptors and helps these
proteins fold into an active conformation. Most
recently, Yang Xu found that hsp82 is also re-
quired for the maturation of oncogenic proteins
in the src family.
Yolanda Sanchez created mutations in the
HSP 104 gene of yeast. The mutations have no ef-
fect on growth at high or low temperatures. How-
ever, the cells are unable to acquire tolerance to
extreme temperatures when given a mild preheat
treatment. Thus this mutation confirms the long-
standing assumption that HSPs play a vital role in
establishing thermotolerance. Moreover, it pro-
vides protection from many other forms of stress,
such as exposure to ethanol and sodium arsen-
ite. Dawn Parsell found that hspl04 is highly
conserved in mammals and in prokaryotic cells
and contains two essential nucleotide-binding
domains.
In Drosophila our mutational analysis has con-
centrated on hsp70. Janice Rossi found that vary-
ing the level of hsp70 expression in Drosophila
cells varies the rate at which the cells recover
from heat shock. Jonathan Solomon found that
expressing hsp70 from independently regulated
promoters, in the absence of heat shock, helps
cells to survive extreme temperatures but in-
hibits their growth. Thus hsp70 helps to protect
cells from the ravages of extreme temperatures
but is actually disadvantageous at normal
temperatures.
To study the role of hsp70 in whole flies, Kent
Golic developed a new system for manipulating
the Drosophila genome. He created flies that ex-
press the FLP recombinase gene of yeast under
the control of heat-shock regulatory sequences.
When flies that also carry a recombinase target
sequence are given a very mild heat shock, the
recombinase is induced and catalyzes rearrange-
ments of the target sequence.
Michael Welte and Joan Tetrault have em-
ployed this site-specific recombination system to
study the role of hsp70 in induced thermotoler-
ance. Strains that carry several extra copies of the
hsp70 gene were constructed. Embryos from
these lines survive heat treatments much better
than wild-type embryos. This suggests that it will
be possible to alter the stress tolerance of even
developmentally complex species.
262
T Cell Surface Glycoproteins in Development
and Viral Infections
Dan R. Littman, M.D., Ph.D. — Associate Investigator
Dr. Littman is also Associate Professor of Microbiology and Immunology and of Biochemistry and
Biophysics at the University of California, San Francisco. Following undergraduate studies on the
structure of microtubules in Marc Kirschner's laboratory at Princeton University, he completed his
M.D./Ph.D. program at Washington University in St. Louis, working with Benjamin Schwartz and Susan
Cullen on the function of histocompatibility molecules in antigen presentation. His postdoctoral research
was done in Richard Axel's laboratory at Columbia University, where he isolated the genes for CD4
and CDS.
THE shaping of a mature repertoire of T lym-
phocytes capable of responding to patho-
genic microorganisms involves a complex pro-
cess of differentiation within the thymus. In this
process, self-reactive cells are eliminated and
cells that can react to foreign antigen complexed
to molecules of the major histocompatibility
complex (MHC) are selected to survive and mi-
grate to peripheral lymphoid organs.
The signals involved in the different phases of
thymic selection require the interaction of sev-
eral thymocyte surface proteins with components
of the thymic microenvironment. These cell sur-
face molecules include the clonally restricted T
cell receptors and the CD4 and CDS glycopro-
teins. Both CD A and CDS are expressed on imma-
ture thymocytes, but the gene for one or the other
is shut off upon maturation. Cells that have re-
ceptors for self-MHC class I molecules continue
to express CDS, but shut off CD4; cells v^^ith re-
ceptors for MHC class II molecules express CD4
and shut off CDS. Our laboratory is studying the
mechanism of regulation of the CD4 and CDS
genes as well as the molecular basis of the cell-
cell interaction resulting in the selection of cells
that have appropriate specificity.
The CD4 and CDS molecules interact directly
with MHC class II and class I molecules, respec-
tively. Effective T cell activation requires co-
recognition of MHC by both the T cell receptor
and either CD4 or CDS. To test whether corecog-
nition is also required in transmembrane signal-
ing during T cell development, we have analyzed
transgenic mice that express a mutant form of
MHC class I incapable of binding to CDS but in-
teractive with T cell receptors. These mice were
shown to be defective in developing a mature T
cell repertoire specific for the mutant MHC class
I molecule.
This study indicates that CDS:MHC binding is
required for both intrathymic deletion of self-
reactive T cells and positive selection of useful T
cells. Moreover, since endogenous MHC class I
molecules were competent to bind CDS but un-
able to rescue the defect, coordinate recognition
of antigen/MHC by a complex of the T cell recep-
tor and CDS is required for both positive and neg-
ative selection.
To study the roles of CD4 and CDS in greater
detail, we are preparing animals defective in the
expression of these molecules. We have utilized
gene-targeting technology to mutate the CD4
gene in embryonal stem cells. These have been
injected into mouse blastocysts, resulting in the
birth of chimeric animals. The mutant CD4 gene
has been propagated in the mouse germline, and
we are currently studying the immune system of
mice lacking CD4 expression. These mice lack
helper T cells and have essentially no antibody
response to T-dependent antigens and no T cell
response to antigen presented by MHC class II
molecules.
Transgenic mice expressing mutant forms of
CD4, predicted to be defective in signal-
transducing functions (as a result of defective in-
teraction with cytoplasmic tyrosine kinases or
with the T cell receptor complex), have been
prepared and are being used to analyze the role of
CD4 signaling during development in the ab-
sence of endogenous CD4 expression. We have
also prepared transgenic mice that express the
human CD4 gene and will examine the ability of
the human molecule to function during T cell
development in mice that lack murine CD4. Such
animals may be especially useful for studies of
pathogenesis and therapy in autoimmune dis-
eases and AIDS.
The CD4 and CDS glycoproteins have been
shown to be involved in the activation of periph-
eral T lymphocytes. For example;, artificial cross-
linking of CD4 or CDS to the T cell receptor com-
plex results in T cell activation. Moreover, T cells
that have lost expression of either CD4 or CDS
but continue to express the T cell receptor are
usually no longer responsive to antigen. We are
performing a variety of structure/function stud-
ies to determine the mechanism through which
CD4 and CDS facilitate signal transduction in T
cells.
It is known that CD4 and CDS are associated
263
T Cell Surface Glycoproteins in Development and Viral Infections
with a cytoplasmic tyrosine kinase, p56'''*, a
member of the src kinase family. The sites of
interaction between these molecules map to
cysteine-containing regions in the cytoplasmic
domains of CD4 and CDS and in the unique
amino-terminal domain of p56'''*. We have dem-
onstrated that only CD4 molecules that can asso-
ciate with the kinase are functional in antigen-
specific T cell hybridomas, which normally
produce interleukin-2 upon stimulation with the
appropriate antigen.
Since interaction of CD4 with the Ick kinase is
essential for T cell activation, we are now begin-
ning to characterize the proteins that serve as
substrates for tyrosine phosphorylation. One of
the substrates appears to be a tyrosine kinase as-
sociated with the T cell receptor complex. Cross-
linking of p56'''* to the receptor-associated kinase
results in tyrosine phosphorylation and activation
of phospholipase-C7 1 , an enzyme that is also as-
sociated with the T cell receptor complex. Cleav-
age of phosphoinositol-containing phospholipid
by this activated enzyme generates second mes-
sengers that activate the transcriptional machin-
ery in the T cell.
The laboratory is also studying the mechanism
involved in the developmental switch from
double-positive (CD4"^CD8'^) to single-positive
thymocytes. The switch may be due to differen-
tial signaling through CD4 or CDS (hence due to
instruction) or to random (stochastic) shutoff of
either CD4 or CDS, followed by selection of cells
that have the appropriate T cell receptor and co-
receptor. Several experiments are being per-
formed to discriminate between these two basic
mechanisms of differentiation.
An understanding of the mechanism involved
should also shed light on the transcriptional regu-
lation that thymocytes employ, first to turn on
both the CD4 and CDS genes and subsequently to
shut off only one of the genes. We have character-
ized a sequence upstream from the CD4 pro-
moter with T cell-specific transcriptional en-
hancing activity. The ability of this and other
sequences to provide subset-specific regulatory
signals in thymocyte differentiation is being stud-
ied in transgenic mice.
The CD4 glycoprotein is doubly important be-
cause it is the receptor for the human immunode-
ficiency virus (HIV) . We have found that several
cell types that express CD4 can bind virus but
cannot be infected, indicating that host factors
other than CD4 are involved in viral entry. Resis-
tance of these cells to infection is due to the in-
ability of the viral envelope to fuse to the cellular
plasma membrane. It is therefore likely that
plasma membrane molecules other than CD4 are
required for fusion of virus to target cells. We are
using genetic and biochemical approaches to
identify such molecules.
Identification of additional molecules in-
volved in HIV entry may permit design of novel
agents to interfere with the spread of HIV. In ad-
dition, we hope that this information will facili-
tate the design of a mouse model system for HIV
disease. The currently available mice that express
the human CD4 transgene are resistant to infec-
tion, but expression of additional genes involved
in HIV entry may permit infection of these ani-
mals with HIV.
Using a genetic system for studying HIV entry,
we have shown that the envelope glycoprotein of
HIV can be replaced by that of another patho-
genic human retrovirus, human T cell leukemia
virus (HTLV), forming HIV(HTLV) pseudotypes.
HTLV causes T cell leukemias and lymphomas
and myelopathies. Individuals infected with both
HIV and HTLV have more rapid progression of
HIV disease than those infected with HIV alone.
Since both viruses infect T lymphocytes, it is
likely that mixed viral particles can form in vivo.
We have demonstrated that mixed particles
(pseudotypes) that form in vitro have an ex-
panded host range — i.e., HIV particles, endowed
with the HTLV-I envelope glycoprotein, can
readily infect CD4-deficient cells. Such mixed
particles may have an important role in HIV
pathogenesis, particularly in infection of cells
that do not have HIV receptors, such as cells of
the central nervous system. In addition to investi-
gating HIV infection, we are using the hybrid par-
ticles to study the yet uncharacterized HTLV
receptor.
264
The Biology of T Lymphocyte Development
Dennis Y.-D. Loh, M.D. — Associate Investigator
Dr. Loh is also Professor of Medicine, Genetics, and Molecular Microbiology at Washington University
School of Medicine and Chief of the Division of Allergy and Immunology and Associate Physician at
Barnes Hospital, St. Louis. He received his undergraduate degree in biology and chemical engineering from
the California Institute of Technology and his medical degree from Harvard Medical School. After finishing
his clinical residency in internal medicine at the Peter Bent Brigham Hospital, Boston, he studied as a
postdoctoral fellow with David Baltimore at the Massachusetts Institute of Technology.
THE immune system is involved intimately in
our body's defense against invading microor-
ganisms and tumors. In addition, it plays a central
role in organ graft rejection and autoimmune dis-
eases such as systemic lupus erythematosus,
rheumatoid arthritis, and diabetes. Its critical
role in maintaining health is best manifested in
the acquired immune deficiency syndrome
(AIDS) , in which destruction of a specific portion
of the immune system results in a potentially fatal
disease.
Our investigation is focused on the molecular
mechanisms that allow normal and abnormal de-
velopment of the antigen-specific T cells. T cells
are those lymphocytes (a type of white cell) that
depend on the presence of the thymus gland for
maturation. An antigen may be viewed as any
marker that these cells recognize. T cells are
thought to play a central role in the regulation of
the immune response. T cells recognize antigens
by means of a cell surface structure called the T
cell receptor (TCR) . The genes that are responsi-
ble for the expression of the TCR undergo DNA
gene rearrangement and gene activation specifi-
cally in the thymus during the individual's early
development. Once the TCR is expressed as pro-
tein, it is the interaction of the TCR with its anti-
gen that triggers the activation of T cells, result-
ing in an immune response. The ultimate result of
such a response may be either defense against
invading organisms or tissue destruction, as
seen in transplantation rejection and autoim-
mune phenomena.
My laboratory initially concentrated on identi-
fying the genetic elements that encode the TCR
genes. We then shifted our efforts to study the
function of T cells in the intact animal, especially
during its development. Two important ques-
tions were addressed. 1) Why are we tolerant of
our own tissues and organs? 2) Why are trans-
planted organs rejected readily (unless they are
carefully cross-matched)? We have used both re-
combinant DNA technology and our ability to
create transgenic mice (mice with cloned genes
incorporated in their own chromosomes) to
study these questions. Two kinds of transgenic
mice have been created. One kind bears trans-
genic TCR genes; the other has transgenic major
histocompatibility complex (MHC) genes (a
marker that distinguishes us individually during
transplantation) . By introducing these genes back
into the mouse itself and into the mouse germ-
line, we can determine how normal T cells de-
velop by studying how the TCR and MHC interact
during development.
This strategy has been very successful. By creat-
ing mice of appropriate genetic background, we
discovered that T cells that are self-recognizing
and hence self-reactive are deleted in the thymus
during development. This implies that part of
self-tolerance is accomplished by physical elimi-
nation of self-reactive T cells. Using mice con-
taining transgenic MHC, we showed a second
mechanism of self-tolerance that does not involve
physical deletion. In this case, self-reactive cells
are not physically eliminated but are functionally
paralyzed. These studies allow us to lay the
foundation to study how T cells acquire self-
tolerance. Since distinguishing what is self and
nonself is a central problem in immunology, we
hope that these studies will lead to a better un-
derstanding of transplantation rejection and au-
toimmune phenomena.
The detailed study of T cell development also
allows us to investigate the cellular and molecu-
lar requirements of normal cellular developmen-
tal processes. For example, we still do not under-
stand how certain T cells are eliminated while
others are selected to survive. To elucidate the
exact molecular mechanisms that underlie this
highly regulated process, we have most recently
focused on the structural and signaling basis that
controls differential cell fate of the developing
thymocytes. To do this we have combined the
two powerful technologies of transgenic mice
and gene knock-out mice. In the latter mice, se-
lected genes can be targeted to be destroyed in
the germline so that mice bearing selected muta-
tions can be created. Once these mice have been
prepared, mutant molecules can be introduced to
replace the "knocked-out" genes. Current efforts
are concentrated on signaling molecules such as
CD4 and CDS that are important in T cell devel-
265
The Biology of T Lymphocyte Development
opment. By such an approach, we hope to under- mechanisms by which T cells undergo the pro-
stand the signaling requirements that determine grammed cell death that leads eventually to self-
cell fate during development as well as the exact tolerance.
266
Molecular Genetics of Mammalian
Glycosyltransferases
John B. Lowe, M.D. — Assistant Investigator
Dr. Lowe is also Associate Professor of Pathology at the University of Michigan Medical School. He received
his bachelor's degree in mathematics from the University of Wyoming and his M.D. degree from the
University of Utah College of Medicine, Salt Lake City. He was trained in clinical pathology and molecular
genetics at Washington University School of Medicine, St. Louis. He was later Assistant Professor in the
Departments of Pathology and Medicine at Washington University and also served as Assistant Medical
Director of the Barnes Hospital blood bank, St. Louis, before moving to Michigan.
THE primary long-range goal of our research is
to understand the functions of oligosaccha-
rides that are found on the surface of mammalian
cells and to explain how the cells regulate their
expression. Oligosaccharide molecules consist
of many different single-sugar structures linked
together in complex linear and branching arrays.
Quantitative and structural changes in such mole-
cules have been shown to correlate with morpho-
logic changes that occur during the embryonic
development of animals and in association with
neoplastic transformation. These and other ob-
servations suggest that cell surface oligosaccha-
rides may function as information bearers in me-
diating interactions between cells during the
developmental process.
Mammalian cells, in constructing these mole-
cules, use special proteins called glycosyltrans-
ferase enzymes. With few exceptions, a unique
glycosyltransferase is responsible for the synthe-
sis of each linkage between the sugar molecules
in an oligosaccharide. The enormous number of
different oligosaccharides dictates that many dif-
ferent glycosyltransferases will enter the con-
struction of the complex cell surface carbohy-
drates on any particular cell or tissue.
In many instances, changes in cell surface car-
bohydrate structure observed during differentia-
tion or in association with malignant transforma-
tion have been shown to correlate with changes
in the glycosyltransferase repertoire. The mecha-
nisms by which cells coordinate and regulate the
expression of these enzymes, and thus the ex-
pression of oligosaccharide structures at the cell
surface, are largely unknown. During the past few
years, the main focus of our work has been in
establishing systems that will allow molecular
analysis of the mammalian genes responsible for
glycosyltransferase synthesis.
The human ABO, H, and Lewis blood group an-
tigens are actually cell surface oligosaccharides.
The determinant genes encode particular glyco-
syltransferases that are able to construct the
"blood group" molecules. These glycosyltrans-
ferases provide convenient genetic and biochemi-
cal models for studying how the processes regu-
late cell surface oligosaccharide expression. The
blood group antigens are not restricted in their
expression to blood cells. They are found on a
number of other tissues in the body, suggesting
that tissue-specific mechanisms regulate their ex-
pression. Moreover, their expression changes
during human embryonic development and is of-
ten altered in malignancy.
Our initial efforts focused on developing sys-
tems to isolate glycosyltransferase genes without
the benefit of purified enzyme protein. Using
gene transfer approaches, we have been able to
isolate several of these genes. They include hu-
man genes encoding the H blood group a(l ,2)fu-
cosyltransferase and the Lewis blood group
q;(1,3/1,4) fucosyltransferase .
The cloned gene segments in each case repre-
sent tools for investigating the genetics of these
enzymes and for studying the function and regula-
tion of their corresponding cell surface oligosac-
charides. For example, we have recently used the
H blood group gene to investigate the molecular
basis for the Bombay blood group phenotype. In-
dividuals with this blood group are extraordi-
narily rare and are cross-match incompatible with
virtually all other humans, excepting other Bom-
bay individuals. This incompatibility is due to the
fact that these persons apparently lack a func-
tional H blood group locus. As a consequence,
they are unable to construct A, B, or H blood
group determinants, and thus maintain high titers
of antibodies directed against the ABH blood
group structures found on red cells from virtually
all other humans.
The molecular basis for the defect in Bombay
individuals had not been defined. By analyzing
the structure of the H gene in Bombay pedigrees,
we identified point mutations in both alleles of
the gene in affected individuals. We subse-
quently demonstrated that these mutations inac-
tivate the enzyme encoded by the gene and are
thus responsible for the Bombay phenotype. We
have also analyzed this gene in para-Bombay indi-
viduals, whose red cells are deficient in ABH
structures but whose secretory tissues express es-
sentially normal levels of these molecules (under
267
Molecular Genetics of Mammalian Glycosyltransferases
the control of the Secretor blood group locus).
We also found inactivating point mutations in the
H gene in these persons, thus providing strong
evidence that the human Secretor blood group
locus corresponds to a distinct a(l ,2)fucosyl-
transferase gene.
We have also used glycosyltransferase gene
segments to identify specific cell surface oligosac-
charide molecules that play pivotal functional
roles in the inflammatory process. One of the pri-
mary events in inflammatory conditions involves
a process whereby circulating white cells leave
the interior of blood vessels and become local-
ized in inflammatory foci outside the vascular
system. This process begins when the endothelial
cells lining the blood vessels become "activated"
by substances that accompany an incipient in-
flammatory condition. Circulating white cells
known as neutrophils adhere tightly to activated
endothelium, insinuate themselves into the endo-
thelial cell pavement lining the blood vessel, and
ultimately come to occupy areas outside the vas-
cular tree.
Neutrophil adhesion to activated endothelium
is mediated in part by a protein known as E-
selectin, or endothelial leukocyte adhesion mole-
cule 1 (ELAM- 1 ) , which is found on the surface of
activated endothelium. Structural features exhib-
ited by this protein suggested that it might inter-
act with an oligosaccharide molecule specific to
the surface of neutrophils. By transfecting diff'er-
ent glycosyltransferase gene segments into mam-
malian host cells, we were able to recapitulate
the biosynthesis of several distinct sets of cell sur-
face oligosaccharide molecules and to demon-
strate that one set allowed transfected cells to ad-
here to E-selectin. We further demonstrated that
the oligosaccharide molecules were one or
more members of a family of oligosaccharides
containing sialic acid and fucose and were rep-
resented by a molecule known as the sialyl Lewis
X determinant. More recently, we have shown
that E-selectin maintains a relatively high specific-
ity for oligosaccharide ligands with a(l,3)-
linked fucose residues in terminal positions, ver-
sus subterminal locations. We have also shown
that terminal «(! ,3)fucosylation may be cata-
lyzed by some, but not all, human a(l,3)-
fucosyltransferases .
Aberrant or overexuberant recruitment of neu-
trophils to sites of inflammation can contribute,
after tissue hypoxia and in other pathological cir-
cumstances, to undesirable tissue damage in au-
toimmune disease. Initial events in this process
require adhesive interactions between E-selectin
and its oligosaccharide counter-receptors on the
surfaces of neutrophils. It thus seemed possible
that purified carbohydrate counter-receptors
might function as anti-inflammatory molecules
by preventing pathologic neutrophil recruit-
ment. We have tested this notion in collaboration
with Peter Ward here at the University of Michi-
gan. We have prepared therapeutic quantities
of sialyl Lewis X-containing oligosaccharides,
using recombinant «(! ,3)fucosyltransferases.
When tested in an animal model of E-selectin-
dependent lung inflammation, these carbohy-
drates exhibit potent anti-inflammatory activity.
This work suggests that a new class of antiin-
flammatory pharmaceuticals may be developed
from such molecules or from their chemical
analogues.
These studies have been aided by our recent
identification of several novel human a(l,3)fu-
cosyltransferase genes. We have shown that these
encode enzymes with shared and unique primary
sequence domains, as well as distinctive and use-
ful catalytic properties. For example, one of
these works extremely well for in vitro synthesis
of the sialyl Lewis X tetrasaccharide, and others
can efficiently construct analogues of this mole-
cule. Work is in progress to identify sequence
domains within the enzymes that dictate their
distinctive substrate specificities.
Circumstantial evidence gathered by other in-
vestigators suggests that oligosaccharides are im-
portantly involved in cell adhesion during mam-
malian embryogenesis. We are now directing our
efi'orts toward exploring this hypothesis and
characterizing the genes that determine these in-
teractions, through genetic manipulation of the
murine genome.
268
Mechanisms of Embryonic Induction
in Vertebrates
Richard L. Matts, M.D., Ph.D. — Assistant Investigator
Dr. Maas is also Assistant Professor of Medicine at Harvard Medical School and Associate Physician at
Brigham and Women's Hospital, Boston. He received his A.B. degree in chemistry from Dartmouth College
and an M.D. -Ph.D. degree from Vanderbilt University School of Medicine. Following his thesis work with
John Oates, he trained as a medical house officer at Brigham and Women's Hospital and completed a
postdoctoral fellowship in Philip Leder's laboratory in the Department of Genetics at Harvard Medical School.
THE goal of our research is to understand the
role that the homeobox genes play in control-
ling vertebrate organogenesis. These genes are
defined by their expression of a 60-amino acid,
helix-turn-helix, DNA-binding domain. Highly
conserved in evolution, they are present in spe-
cies as divergent as Drosophila, yeast, and hu-
mans. Mutations in homeobox genes of fruit fly
and mouse result in specific developmental de-
fects. Our work thus far has focused on the char-
acterization of two such genes that appear to play
important roles in the formation of the mamma-
lian kidney and eye, respectively. A long-term
goal is to understand the target genes that these
homeobox genes interact with, using a combina-
tion of biochemical, embryologic, and genetic
techniques.
Murine Homeobox Genes Expressed in
Mouse Embryonic Kidney
To determine which homeobox genes are ex-
pressed in the developing mouse kidney, we un-
dertook a polymerase chain reaction (PGR)
screen of reverse-transcribed, microdissected
kidney RNA from mouse embryos at day 1 5 . This
experiment yielded 27 different homeobox-
containing genes, some 11 of which correspond
to new genes. Among these novel sequences, we
identified several with 85-98 percent sequence
similarity to known murine Hox genes at the nu-
cleotide level. In addition, we identified two
other genes, closely related to each other, that
appear to be new members of the Hox-1 and
Hox- 3 clusters in the mouse. Mapping experi-
ments indicate that the Hox-1 member is
Hox- 1.8.
The structures of several cDNA clones of Hox-
1.8 are being determined. Thus far, five different
alternatively processed forms have been identi-
fied. Surprisingly, all these forms share a com-
mon feature: due to the presence of upstream ter-
mination codons, none would actually encode a
translatable homeodomain. We suspect that a
homeodomain-encoding form exists, because the
homeodomain is preserved intact at the sequence
level. Current efforts are aimed at securing the 5'
end of the Hox- 1.8 gene in order to determine
whether splice forms exist that would encode a
functional homeodomain.
A current working hypothesis of our laboratory
is that many murine Hox genes may, as a general
rule, encode both homeobox-containing and
homeobox-less forms. Such forms may interact in
heterodimeric combinations with one another, or
with other Hox genes, to alfect the capacity of the
homeobox-containing form to bind to DNA.
The expression of the //ox- 7. S transcripts has
been analyzed as a function of mouse embryogen-
esis. As determined by Northern blot analysis, sig-
nificant expression appears at day 1 0 of embryo-
genesis, peaks at day 13, and subsides by day 15.
Expression in adults is confined to the kidney and
to skeletal muscle. Interestingly, the adult kidney
appears to express a smaller transcript form of
approximately 1.5 kb, in addition to the larger
class of 2.8-3.9 kb observed in both embryos and
kidney. This size range corresponds to extensive
alternate processing, as noted above.
We have further analyzed the expression of
Hox- 1 .8 duTing embryogenesis by in situ hybrid-
ization. Regionally restricted expression is ob-
served in the submucosa of the foregut and mid-
gut and also in somites. Of particular interest to
the potential role of this gene in nephrogenesis is
its expression in the condensing collecting duct
system, in the region that comprises the develop-
ing calyces. Three-dimensional reconstruction of
the Hox- 1.8 expression pattern shows that ex-
pression is localized to this part of the developing
kidney at embryonic day 1 3 •
Identification of a Pax Gene Involved
in Formation of the Vertebrate Eye
The formation of the vertebrate eye has long
served as an attractive model system for studying
basic features of embryonic induction. The eye
forms as a consequence of outgrowth of the dien-
cephalon and a subsequent interaction of this
neuroectoderm-derived structure with the sur-
face ectoderm, resulting in an invagination of the
latter to form the lens vesicle. Additional neuroec-
toderm and mesodermal ingrowth anterior to the
269
Mechanisms of Embryonic Induction in Vertebrates
lens results in formation of the anterior structures
of the eye, specifically the iris, cornea, and ciliary
body. An additional interaction, between the de-
veloping lens and retina, is suggested by experi-
ments involving transgenic mice: ablation of the
developing lens results in an abnormal prolifera-
tion of retina, perhaps suggesting the presence of
an inhibitory factor from lens.
In both mouse and human, there are naturally
occurring mutations that affect eye development.
In the mouse, a mutation on chromosome 2,
called Small eye {Sey), results in deficient eye
formation. The portion of mouse chromosome 2
to which Sey maps is homologous to a portion of
human chromosome 11, llpl3. Interestingly, a
semidominant human disorder called aniridia,
which also affects basic eye development, is lo-
cated in this region of 1 1 . The observation that a
mouse paired box and homeobox gene. Pax- 6,
maps to the Sey locus prompted us to examine
the role of this gene and its human counterpart in
basic ocular development. Recently it has been
shown that point mutations in the mouse Pax- 6
gene are in fact responsible for the Sey mutation.
We have cloned the human PAX6 gene, which
maps to within 200 kb or less of the human ani-
ridia locus, as defined by various translocation
breakpoints. This gene is more than 95 percent
conserved over 422 amino acids with a related
gene cloned from zebra fish, which is expressed
in both the diencephalon and lens vesicle during
ocular development. Thus both the chromosomal
location and the expression pattern of this gene
are consistent with its involvement in aniridia.
We have determined the genomic structure of
the PAX6 gene, which consists of at least 13
exons spread over some 30 kb of genomic DNA,
and have sequenced the intron-exon boundaries.
This has permitted the design of PGR primers to
each exon, which are conveniently sized for anal-
ysis by PGR. Thus far we have identified a non-
sense mutation located in helix 2 of the homeo-
box of the human PAX6 gene in one patient and,
preliminarily, a small insertion in the PAX6 gene
of another patient. Although analysis of more mu-
tations will be required to establish that muta-
tions in PAX6 indeed account for aniridia, these
findings argue persuasively that they do.
270
Cell Cycle Control
James L. Mailer, Ph.D. — Investigator
Dr. Mailer is also Professor of Pharmacology at the University of Colorado School of Medicine. He received
his B.S. degree in biochemistry from Cornell University and his Ph.D. degree in molecular biology from
the University of California, Berkeley, where he worked with John Gerhart. He then carried out
postdoctoral studies with Edwin Krebs at both the University of California, Davis, and the University
of Washington before moving to Colorado.
TWO events mark the reproductive life of a
cell: replication of the DNA, and its distribu-
tion to daughter cells at mitosis. Because of the
central importance of cell reproduction to or-
dered cell growth and to birth of the next genera-
tion, cells have evolved rigorous controls to en-
sure that both events are carried out with high
fidelity and at the appropriate time. My labora-
tory is interested in understanding the nature and
regulation of these controls with respect to how a
cell commits itself to replicate its DNA and how it
knows when to divide.
The cell cycle has four main phases: Gi, S (syn-
thesis), G2, and mitosis. The decision to synthe-
size DNA (to enter the S phase) is made in Gj , and
the decision to begin cell division (to enter the M
phase of mitosis) is made in G2. There is abun-
dant evidence that these decisions are made at
checkpoints, or restriction points, in the cycle.
The nature of these Gj and G2 decision-making
periods in the cell cycle underlies fundamental
processes operative in early embryonic develop-
ment and in malignant cells.
G2 ^ M Regulation
Our laboratory developed a G2-phase extract
from frog eggs in which synthetic nuclei entered
mitosis at the addition of mitotic signals. We then
purified the mitosis-signaling enzyme (called
maturation-promoting factor, or MPF) and found
that it was composed of a protein kinase com-
plexed to a G2 cyclin. Kinases have the ability to
attach a phosphate group to many different cellu-
lar proteins, modifying their function and caus-
ing profound changes in cellular biochemistry.
The protein kinase was identified as a vertebrate
homologue of the cdc2 gene, which had been
genetically implicated in the control of mitosis
by the study of certain mutants in yeast.
G2 cyclins are proteins that accumulate during
interphase, reach high levels in late G2, and are
then degraded near the metaphase anaphase
transition in mitosis. This degradation is required
in order for cells to complete mitosis successfully
and enter Gj . In most cells there are two classes of
G2 cyclins, termed A and B cyclins, that differ in
sequence similarity and have different kinetics of
accumulation and degradation. Both bind cdc2
kinase, but A- type complexes are activated much
earlier in the cell cycle than B-type complexes,
and only B-rype cyclins are found in puri-
fied MPF.
To investigate the role of cyclin A, we utilized
extracts from metaphase-arrested eggs that are
able to exit mitosis in vitro, to undergo DNA syn-
thesis, and then to reenter mitosis. These extracts
retain the characteristic dependence of mitosis
upon completion of DNA synthesis — that is, will
not enter mitosis if DNA synthesis has not been
completed, which can result from an excess of
DNA in the system or the presence of aphidicolin,
an inhibitor of DNA polymerase. By using anti-
sense oligodeoxynucleotides to ablate cyclin A
mRNA from the system, we were able to show that
activation of MPF (cyclin B/cdc2) occurred even
when DNA synthesis had not been completed.
Readdition of recombinant cyclin A protein to the
antisense-ablated extracts restored the depen-
dence of mitosis on DNA synthesis by causing a
lengthening of S phase until DNA synthesis was
complete.
This provides evidence that one of the func-
tions of cyclin A is in the crucial checkpoint that
prevents the activation of cyclin B/cdc2 (MPF)
until DNA synthesis has been completed. It also
explains why cyclin A/cdc2 complexes are acti-
vated earlier in the cell cycle than cyclin B/cdc2
complexes and why cyclin A is degraded before
cyclin B. The latter phenomenon, long known, is
clearly appropriate, since cyclin A exerts an inhib-
itory function on cyclin B. A question that now
merits attention concerns what substrates exist
for phosphorylation by cyclin A/cdc2 that are in-
volved in this feedback control mechanism.
We are interested in the mechanism of activa-
tion of MPF in oocytes during the cell cycles of
meiosis I and II. In these cycles the synthesis of
proteins other than cyclin are required for MPF
activation. One protein required for meiosis I and
II is the product of the mos proto-oncogene.
Proto-oncogenes are the normal cellular counter-
part of mutated oncogenes found in cancer cells,
271
Cell Cycle Control
suggesting that oncogenes act by perturbing nor-
mal cellular pathways. In general, very little is
known about how proto-oncogenes work, but the
specific involvement of mos in cell cycle control
is the clearest example of a specific function for
any proto-oncogene in a defined cellular process.
The mos gene encodes a serine/threonine pro-
tein kinase, indicating the existence of a substrate
for phosphorylation by mos that can lead to acti-
vation of MPF as well as stabilization of cyclin in
the metaphase arrest of meiosis II. The transition
between meiosis I and II is also perhaps the only
well-documented case in which cdc2 kinase ac-
tivity declines without accompanying cyclin B
degradation.
This year Linda Roy observed that introduction
of protein synthesis inhibitors into the system at
the transition between meiosis I and II leads to
the immediate destruction of cyclin B, which
would otherwise be stable, thereby establishing
that the stability of cyclins between meiosis I and
II requires continuous synthesis of protein. Ini-
tially we expected that the protein synthesis re-
quired for cyclin stability would be exemplified
by the synthesis of the mos proto-oncogene ki-
nase itself, since that kinase is known to comprise
one of the major newly synthesized proteins dur-
ing maturation. However, Dr. Roy found that, in
fact, the mos kinase remains fully active even
when B-type cyclins are degraded. This implies
that there is another protein whose synthesis is
required for cyclin stability besides the mos
proto-oncogene .
Dr. Roy also did the converse experiment by
microinjecting antisense oligodeoxynucleotides
against the mos kinase mRNA into cells just prior
to entry into meiosis I. By this procedure she was
able to ablate greatly the level of mos proto-
oncogene expression. However, under these con-
ditions, cyclin B does not undergo degradation
between meiosis I and II, thereby establishing
that mos is not necessary for the stability of cyclin
B between meiosis I and II. These results suggest
that other components besides the mos kinase are
required for the unusual cyclin stability between
meiosis I and II, and clearly the identification of
these other components merits attention.
Gj -*■ S Regulation
The cell cycle restriction point in Gj governing
the Gi ^ S transition that involves cdc2-like ki-
nases has been termed START in yeast and the R
point in mammalian cells. In both budding yeast
and fission yeast, it is quite clear that the genuine
cdc2 gene product mediates both the Gj and G2
control points. Recently it has become evident
that the regulation of the Gj ^ S transition in
vertebrate cells is considerably more compli-
cated than in yeast. One aspect of this complexity
involves the presence of another cdc2-like gene
in the Gj phase that may mediate events at the
Gj -* S transition. This gene was originally discov-
ered in Xenopus and given the name Egl but has
since been renamed cyclin-dependent kinase 2,
or cdk2. The idea that cdk2 might be a form of
cdc2 specialized for Gj control has come from
the finding by others that cdk2 itself will not
complement mutations in the Saccharomyces cere-
visiae cdc2 cognate gene CDC28 that affect the
G2 ^ M restriction point, but will substitute at least
partially for the Gj function of when coex-
pressed with a Gj cyclin from human cells.
This year we have made a major effort to study
the biochemistry and regulation of cdk2 in the
Xenopus embryonic cell system. Toward this end
we have developed an antibody against the pro-
tein and used it to show that the protein kinase
activity of cdk2 oscillates in the cell cycle with a
periodicity similar to cdc2. In the Xenopus em-
bryonic cell cycle, cdk2 is not associated with
either cyclin A or cyclin B, but instead with two
proteins of 36 and 48 kDa. This suggested the
possibility that cdk2 would be regulated by phos-
phorylation, perhaps in ways similar to the
known regulation by phosphorylation of cdc2 ki-
nase itself.
Recently we showed that in fact cdk2 is a phos-
phoprotein, that it is phosphorylated on tyrosine
and serine residues, and that phosphorylation of
these residues changes during the cell cycle. In
particular, the presence of phosphotyrosine cor-
relates with less-active forms of the protein, but
the major site of phosphorylation was determined
by tryptic phosphopeptide mapping to be in a
phosphopeptide distinct from the one containing
tyrosine 15, so characteristic of cdc2 kinase. De-
spite this diff^erence in the sites of phosphoryla-
tion, however, it would appear that the same en-
zyme responsible for the dephosphorylation of
cdc2 is also active on cdk2.
By continuing to study both cdc2 kinase in the
G2 ^ M transition and cdk2 kinase in the Gj ^ S
transition, it should be possible to achieve a com-
prehensive understanding of cell cycle control at
the two main restriction points present in eukar-
yotic cells.
272
The Role of T Cells in Health and Sickness
Philippa Marrack, Ph.D. — Investigator
Dr. Marrack is also a member of the Division of Basic Immunology of the Department of Medicine at the
National Jewish Center for Immunology and Respiratory Medicine, Denver, and Professor of Biochemistry,
Biophysics and Genetics, of Microbiology and Immunology, and of Medicine at the University of Colorado
Health Sciences Center, Denver. She took her Ph.D. in biological sciences at the MRC Laboratory for
Molecular Biology in Cambridge, England, and then did postdoctoral work with Richard Dutton at the
University of California, San Diego. From there she moved to the University of Rochester, and, after seven
years, to her present position. Dr. Marrack is a member of the National Academy of Sciences and was
recently awarded the Christopher Columbus Discovery Award for Biomedical Science.
T cells are essential to the ability of higher ver-
tebrates to resist disease. These cells are not
only able to destroy invading organisms by killing
cells in which such organisms live, but they also
produce hormone-like substances that contribute
to protection against invasion, by stimulating pro-
duction of protective antibodies, for example.
T cells bear receptors for antigen on their sur-
face. There are about 20,000 receptors on each
cell. These are composed of tw^o polypeptide
chains, ol and (8, each made up of a number of
variable elements: Va, Ja, V/3, D/?, andJjS. In each
T cell the receptors are composed of different
combinations of these elements, so that each T
cell has receptors with a somewhat different
structure.
As far as we know, receptors are assembled ran-
domly from the available Vas, Jas, and so on,
while T cells are developing. Consequently there
is a distinct possibility that some T cells will bear
receptors able to interact with self antigens, i.e.,
components of the individual containing the T
cells. These cells are a potential threat, since they
could attack and destroy their own host. Usually,
however, they themselves are destroyed or inacti-
vated before they become mature enough to
cause damage.
There is reason to believe that some potentially
self-reactive T cells avoid the processes of toler-
ance and are allowed to mature. Even though they
could attack tissues of their host, they seem not
to, perhaps because the self antigens with which
they could interact are sequestered in a tissue
that they cannot reach. Occasionally, however,
these self-reactive cells are activated by en-
counter with an environmental antigen. After
stimulation the cells become more motile and ac-
tive and are then able to interact with and destroy
the tissues of their host, causing a so-called au-
toimmune disease.
A number of human diseases — juvenile dia-
betes, lupus erythematosus, and multiple sclero-
sis among others — may be due to T cell malfunc-
tion similar to that described above. Our own
work, in collaboration with Brian Kotzin (Depart-
ment of Pediatrics, National Jewish Center), has
concentrated on rheumatoid arthritis. About 2
million people in the United States suffer from
this disease, which seems to be due to an immune
attack on material in joints. We have shown that
many of those afflicted have unexpectedly low
levels of T cells bearing V|8 1 4 in their blood, and
some of the missing cells are present in the fluid
bathing the rheumatic joints.
We have suggested that rheumatoid arthritis in-
volves chronic invasion of the host by a foreign
antigen able to interact with T cells bearing
V/314. This antigen would first stimulate target T
cells, and then cause their disappearance, proba-
bly by the processes of tolerance described
above. Perhaps the subset of V(8 14 -bearing cells
that can recognize self antigens is waylaid and
rescued from death by sequestration in the joints.
Here they cause inflammation and the symptoms
of arthritis.
At the moment all these are simply ideas upon
which to base future experiments on the cause,
treatment, and prevention of rheumatoid arthri-
tis. Meanwhile, we have begun a search for the
foreign antigen that may interact with V|8l4-
bearing T cells and thus start the disease. There is
every indication that this is a superantigen, a spe-
cial kind of antigen that interacts with T cells,
primarily via the VjS portion of their receptors. It
is also likely, as mentioned above, that the anti-
gen is produced by a chronic infectious agent,
such as a chronic virus.
Until recently the only infectious agents
known to produce superantigens were myco-
plasma and bacteria, such as staphylococci and
streptococci. These agents do not usually infect
their hosts in a chronic fashion. Early in 1991,
however, we and several other groups showed
that certain viruses, a collection of retroviruses
present in mice, can also express superantigens.
Encouraged by this result, we decided to screen
human viruses for such expression. Our prelimi-
nary results suggest that Epstein-Barr virus, the
273
The Role of T Cells in Health and Sickness
cause of a common chronic infection and of in- that Epstein-Barr virus is the causative agent of
fectious mononucleosis in humans, may indeed rheumatoid arthritis, it is possible that other
encode a superantigen. While we do not think members of the Herpesviridae family may be.
274
Cell Regulation by Transforming
Growth Factors
Joan Massague, Ph.D. — Investigator
Dr. Massague is also a member of the Cell Biology and Genetics Program at Memorial Sloan- Kettering
Cancer Center and Professor of Cell Biology at Cornell University Graduate School of Medical Sciences,
New York. He received his Ph.D. degree in biochemistry from the University of Barcelona, Spain, and was
a postdoctoral fellow with Michael Czech at Brown University. He was Assistant and Associate Professor
of Biochemistry at the University of Massachusetts Medical School before assuming his present position.
THE proliferation of cells is controlled by a
balance of positive and negative signals. The
machinery that conveys growth inhibitory signals
is similar in design to that which signals cell
growth. Both involve 1) factors that circulate be-
tween cells and 2) membrane receptors that are
coupled to signal transduction circuitry inside
the cell. The signals carried by growth-promoting
factors have been extensively studied for the past
two decades. The growth inhibitors, however,
have come to the attention of biologists only re-
cently. Yet they include some of the most wide-
spread and versatile regulators of cell growth and
phenotype. Some of them are implicated in pro-
cesses of development, tissue repair, and recy-
cling, and their study may show us ways to con-
strain the unrestricted growth of cancer cells.
Multifunctional Growth Inhibitors
and Their Receptors
Ranking high in the list of growth inhibitors,
and in our research interests, is the polypeptide
TGF-(S (transforming growth factor-/?) . In reality,
TGF-(S represents a large family of growth and dif-
ferentiation factors that also includes the acti-
vins, the bone morphogenetic proteins, the
Miillerian inhibiting substance, and others. The
evolutionary conservation of these factors is un-
usually strict, and they are broadly multifunc-
tional. For example, TGF-)8 can inhibit cell prolif-
eration, regulate cell differentiation, affect how
cells organize tissue structures, and perform
various other functions in cells from virtually
every lineage. Likewise, other members of the
family, such as activin and decapentaplegic, are
involved in body axis formation during embryo
development.
Over the past year our research program has
centered on identifying and isolating genes that
encode receptors for TGF-(8 and related factors.
We had previously determined that TGF-|8 binds
to various types of receptor proteins on the cell
membrane, and we have recently cloned genes
for two classes of such receptors. One class com-
prises receptor membrane proteins that can trig-
ger a biochemical reaction inside the cell upon
binding a factor on the outside. The receptors in
this class probably mediate most cellular re-
sponses to these factors. Like the factors they
bind, these receptors exist in many variants, each
probably representing a discrete adaptation to
achieve optimal control of cell functions.
One member of this receptor class binds the
TGF-/3-related factor activin. The biochemical re-
action carried out by this receptor is the transfer
of phosphate from ATP to serine and threonine
groups present in certain intracellular proteins.
Phosphorylation of serines and threonines repre-
sents a significant departure from previously
known receptor signals. In contrast to the recep-
tors for activin and TGF-|8, those for growth-
promoting factors typically trigger phosphoryla-
tion of tyrosine groups in proteins. Thus the tyro-
sine phosphorylation signals of mitogens are
challenged by the serine/threonine phosphoryla-
tion signals of antimitogens. These findings offer
new insight into the counteracting signals that
preserve the normal balance of cell growth. This
work has been supported by a grant from the Na-
tional Cancer Institute.
Receptor Accessory Molecules
The other TGF-/? receptor class recently cloned
by our group is interesting for other reasons. This
protein, called betaglycan, is thought to act as a
helper of the signaling receptors. Rather than me-
diate cell responses directly, betaglycan seems to
regulate the access of cells to TGF-|8 by either
helping present this factor to the signaling recep-
tors or storing it for later use by the cell.
The structure of betaglycan is unusual for a
growth factor-binding protein. It consists of a
core protein that carries a large mass of negatively
charged carbohydrate. TGF-/3 binds the core pro-
tein, whereas the carbohydrate can bind the so-
called heparin-binding growth factors. Work is
under way to map the portion of this molecule
that binds TGF-jS and to test its ability as a modula-
tor of TGF-|8 activity.
With the cloning of these genes, it is now possi-
275
Cell Regulation by Transforming Growth Factors
ble to explore their properties in detail and to
identify key pieces of the machinery that transfers
growth inhibitory signals from the membrane to
the nucleus. Furthermore, it should be possible
to determine to what extent the genetic loss of
TGF-jS receptors might cause loss of constraint in
cell proliferation and thus incite tumor cell
outgrowth.
Cell-Cell Stimulation by Membrane-bound
Growth Factors
Paracrine growth factors and polypeptide hor-
mones are generally synthesized as larger soluble
precursors that are later fragmented to yield the
bioactive forms. In a recently found variation of
this theme, factors such as TGF-a (no structural
relationship to TGF-jS) are generated from mem-
brane-anchored proteins rather than from soluble
precursors. The TGF-a precursor can accumulate
on the cell membrane and bind to receptors lo-
cated on the surface of adjacent cells. This inter-
action can sustain cell-cell adhesion and stimu-
late DNA replication by cell-cell contact.
In work supported by a grant from the National
Cancer Institute, we have shown that generation
of TGF-a by cleavage of its precursor occurs at
the cell surface by a highly regulated enzymatic
system. This system is strongly activated by
tumor-promoting phorbol esters and growth fac-
tors via mechanisms involving protein kinase C
and calcium influx into the cell. Thus the precur-
sor cleavage process functions as a regulated
switch between two active forms of the grouT:h
factor, one membrane-bound and the other diffus-
ible. The membrane-bound forms could be im-
portant in tissue development processes whose
guidance depends on discrete cell-cell interac-
tions incompatible with the diffusible nature of
soluble factors. The regulated nature of the pro-
TGF-a cleavage process renders it susceptible to
exogenous control with pharmacologic agents.
Furthermore, it provides a way to identify the
pieces of the general machinery controlling the
release of this and other membrane proteins that
mediate cell-to-cell interaction.
In this molecular model, a benzene molecule
(in yellow) is shown buried within the core of
T4 lysozyme in a cavity created by replacing
leucine 99 with alanine. The inner dotted form
represents the van der Waals surface of the ben-
zene, and the outer envelope, the van der Waals
surface of the cavity.
Research and photograph by A. Elisabeth
Eriksson and Xue-jun Zhang in the laboratory
of Brian Matthews.
276
Structural Basis of Interactions Within
and Between Macromolecules
f
Brian W. Matthews, Ph.D., D.Sc. — Investigator
Dr. Matthews is also Professor of Physics and Director of the Institute of Molecular Biology at the
University of Oregon and Adjunct Professor of Biochemistry at the Oregon Health Sciences University,
Portland. He received his undergraduate and graduate training at the University of Adelaide, Australia.
He did postdoctoral research at the MRC Laboratory of Molecular Biology, Cambridge, England (with
David Blow) and at the National Institutes of Health ( with David Davies ). Dr. Matthews is a member
of the National Academy of Sciences.
OUR laboratory uses x-ray crystallography, in
concert with other techniques, to try to ad-
dress some of the fundamental problems in biol-
ogy: How do proteins spontaneously fold into
their biologically active three-dimensional con-
figurations? What determines the stability of
these folded proteins, and can stability be im-
proved? How do proteins interact with each
other? How do they interact with DNA? How do
enzymes act as catalysts?
The Protein-folding Problem
An area of long-standing interest is the so-called
protein-folding problem. How does a newly syn-
thesized, extended peptide chain "know" how
to fold spontaneously into its active three-
dimensional shape?
Although it has long been recognized that the
amino acid sequence of a protein determines its
three-dimensional structure, recent work from
several laboratories has made it clear that certain
amino acids are more important than others in the
folding process. At some positions, typically the
solvent-exposed mobile sites in the folded pro-
tein, amino acids can be interchanged almost at
random with little apparent effect on folding or
stability. On the other hand, interchange of
amino acids in buried or rigid parts of a folded
protein can destabilize it, suggesting that the
amino acids at these positions are important in
determining the folded conformation.
One of the encouraging developments has
been the relative freedom with which amino acid
replacements can be introduced in a protein of
interest. To try to simplify the complexity of the
protein-folding problem, we are attempting to
replace some of the "nonessential" amino acids
in phage T4 lysozyme with alanine. Such a
"polyalanine protein" would, in principle, trun-
cate all nonessential side chains and allow one to
focus on those parts of the amino acid sequence
that are critical for the folding process.
In experiments to date, a series of alanines has
been introduced within two different of-helices of
T4 lysozyme. The somewhat surprising result is
that alanines are not only tolerated at most posi-
tions in the a-helix; they can sometimes increase
the protein's stability. In an extreme case it has
been found that 1 0 alanines can be introduced in
sequence, yet the protein still folds normally and
has full activity. This illustrates that the informa-
tion in the amino acid sequence of a protein is
highly redundant.
Understanding the Interactions That
Stabilize Protein Structures
It is generally agreed that the major factor in
stabilizing the folded structures of globular pro-
teins is the hydrophobic effect. Until recently it
has also been generally agreed that the strength of
the hydrophobic effect — i.e., the energy of stabi-
lization provided by the transfer of hydrocarbon
surfaces from solvent to the interior of a protein
— is about 25-30 cal mol~' for each square ang-
strom of surface area buried within the protein.
However, some recent studies using site-directed
mutagenesis and protein denaturation have sug-
gested that the strength of the hydrophobic effect
might be much higher.
A principal difficulty in addressing this prob-
lem has been the lack of relevant structural data.
How does a protein structure respond when a
bulky hydrophobic residue such as leucine is re-
placed by a smaller residue such as alanine? Does
the protein structure remain essentially un-
changed or is there structural rearrangement to
avoid the creation of a cavity? If cavities are cre-
ated, do they contain solvent?
To address these questions, six "cavity-creat-
ing" mutants in which a large hydrophobic
amino acid was replaced by a smaller one were
constructed within the hydrophobic core of
phage T4 lysozyme. All variants were crystallized
and the structures determined at high resolution.
The structural consequences of the mutations
differ from site to site. In some cases the protein
structure hardly changes at all. In other cases,
however, both side-chain and backbone shifts up
to 0.8-1.0 A were observed. In every case re-
moval of the wild-type side chain allowed some
of the surrounding atoms to move toward the va-
cated space, but a cavity always remained.
277
Structural Basis of Interactions Within and Between Macromolecules
This suggests a way to reconcile the different
values for the apparent strength of the hydropho-
bic effect. One can imagine two extreme situa-
tions. In one case a leucine alanine replace-
ment is constructed, and the protein structure
remains completely unchanged. In this situation
the size of the created cavity is large, and the mu-
tant protein is maximally destabilized. In the
other extreme, the protein structure relaxes in
response to the leucine -•- alanine substitution,
fills the space occupied by the leucine side
chain, and so avoids the formation of any cavity
whatsoever. In this case the decrease in energy of
the mutant protein relative to wild type drops to a
constant energy term that is characteristic for a
leucine alanine replacement.
Ligand Binding Within Cavities
We have shown by crystallographic and thermo-
dynamic analysis that the cavity created by the
replacement leucine 99 alanine in T4 lyso-
zyme is large enough to bind benzene and that
ligand binding increases the melting temperature
of the protein by 5.7°C. This shows that cavities
can be engineered within proteins and suggests
that such cavities might be tailored to bind spe-
cific ligands. The binding of benzene at an inter-
nal site 7 A from the molecular surface also illus-
trates the dynamic nature of proteins, even in
crystals.
Receptor-Ligand Interaction
To develop an understanding of the mode of
action of growth factors and their interactions
with their receptors, we have crystallized and de-
termined the high-resolution structure of human
fibroblast growth factor. The structure is very sim-
ilar to that of interleukin-I/3. Clearly, many
growth factors have similar overall structures.
but the exact relationship of these factors in the
vicinity of their receptor-binding regions remains
to be clarified. Structural studies of human nerve
growth factor are in progress.
Protein-DNA Interaction
We have been interested for some time in the
interaction between proteins and nucleic acids.
In 1981 we determined the structure of the Cro
repressor protein of X bacteriophage (bacteria-
infecting virus), one of the prototypical exam-
ples of a DNA-interacting protein. The structure
of Cro, as determined crystallographically, sug-
gested that a characteristic part of the protein,
now known as the helix-tum-helix motif, is espe-
cially important in DNA binding. The helix-turn-
helix unit can be considered as a "reading head"
that fits into the grooves of the DNA and matches
the DNA structure at the specific recognition site.
The helix-turn-helix motif is now known to occur
in a large number of DNA-binding proteins, and
its functional role has been confirmed by struc-
tures of several DNA-protein complexes.
We have subsequently determined the crystal
structure of Cro protein in complex with a tight-
binding 1 7-base pair DNA operator. In general
terms the structure of the complex supports the
model for Cro-DNA interaction that was proposed
on the basis of the uncomplexed protein, al-
though the Cro dimer undergoes a substantial
conformational change relative to the uncom-
plexed crystal structure.
Recently we have determined the structure of
the biotin repressor from Escherichia coli. This is
a more complicated protein that requires the
presence of an effector molecule to bind DNA. It
also acts as an enzyme.
Studies of protein stability and protein-DNA in-
teraction were supported in part by grants from
the National Institutes of Health.
278
What Viruses Are Telling Us About Gene Regulation
in Mammalian Cells
Steven Lanier McKnight, Ph.D. — Investigator
Dr. McKnight is also a staff member in the Department of Embryology at the Carnegie Institution of
Washington, Baltimore, and Adjunct Professor in the Departments of Biology and of Molecular Biology
and Genetics at the Johns Hopkins University School of Medicine. He earned his Ph.D. degree in biology
from the University of Virginia and, except for four years with the Fred Hutchinson Cancer Research
Center in Seattle, has been with the Carnegie Institution ever since. Dr. McKnight was recently elected to
the National Academy of Sciences.
VIRUSES that attack mammalian cells rely on
preexisting enzymes, factors, and cellular
functions to negotiate their infectious cycle. Mo-
lecular studies of virus infection have thereby
provided key insights into normal cellular pro-
cesses. For example, studies of the processing
and transport of membrane glycoproteins that
form the exterior coats of influenza virus and ve-
sicular stomatitis virus have helped to explain
\\oy<i proteins are selectively transported to the
appropriate cellular compartment.
Studies of viruses have also illuminated com-
plex phenomena regarding selective gene ex-
pression. RNA splicing was first discovered in
studies of human cells infected by adenovirus.
Likevi'ise, the capacity of DNA segments known as
enhancers to regulate gene expression from re-
mote locations was discovered in studies of sim-
ian virus 40.
Work from our laboratory has focused on the
mechanisms of herpesvirus gene regulation. The
herpesvirus chromosome contains roughly 50-
100 genes that are expressed in a tightly con-
trolled temporal cascade. Early during the in-
fectious cycle, five immediate-early (IE) genes
are activated. The IE genes encode protein prod-
ucts, termed transcription factors, that act to regu-
late subsequent viral gene expression. Several
hours later about 25 delayed-early (DE) genes are
activated. Transcription of DE genes is strictly de-
pendent on the prior production of IE proteins.
DE genes encode proteins required for replica-
tion of viral DNA. Following viral DNA replica-
tion, about 25 late (L) genes are activated. Her-
pesvirus L genes encode structural proteins that
form the intact virus particle, including a com-
plex set of membrane glycoproteins and struc-
tural proteins of which the viral capsid is
composed.
Interestingly, one of the L gene products en-
capsidated in the mature virus is a potent and
specific transcription factor dedicated to the acti-
vation of IE genes in the subsequent infectious
cycle. Roughly 1,000 molecules of this protein,
termed viral protein 16 (VP 16), are packaged
into the mature virus particle. Upon infection of
an otherwise healthy cell, VP 16 is released from
the infecting virus and comes to be associated
with enhancer elements located upstream from
each IE gene. Thus emplaced, VP 16 acts as a po-
tent "trigger" for the rapid and prolific expres-
sion of IE genes.
Early studies in our laboratory were focused on
the herpesvirus DE gene encoding thymidine ki-
nase (TK) . In order to probe the mechanisms of
activation by IE proteins, the TK gene was system-
atically mutated with the aim of defining specific
regulatory switch points within or around the
gene. A region of 35 base pairs encompassing the
site of transcription initiation was identified as
being responsive to activation by the IE transcrip-
tion factors. Surprisingly, however, such studies
identified three additional regulatory DNA se-
quences located upstream of the TK gene. These
supplementary regulatory elements were identi-
fied as binding sites for host cell proteins, includ-
ing the Spl (selectivity protein 1) transcription
factor discovered by Robert Tjian (HHMI, Univer-
sity of California, Berkeley) and his colleagues.
The involvement of host cell transcription fac-
tors in viral gene expression has since been ob-
served in numerous cases. An exciting recent dis-
covery along such lines has come from Joseph
Nevins (HHMI, Duke University Medical Center).
Dr. Nevins and his colleagues have identified a
human transcription factor, termed E2F (early re-
gion 2 factor) , that plays a pivotal role in the tran-
scriptional induction of certain adenovirus
genes. Remarkably, the E2F factor has been
shown to form a specific complex with the prod-
uct of a recessive oncogene encoded by the reti-
noblastoma locus. Such observations are begin-
ning to provide mechanistic insight into the role
of the retinoblastoma protein in growth control
and cancer.
More recent studies in our laboratory have fo-
cused on the activation of herpesvirus IE genes
during the early stage of infection of cultured
mammalian cells. As mentioned previously, IE
gene activation is stimulated by VP 16, a virally
encoded L protein that is encapsidated in mature
virus particles. Like the soldiers sequestered in
279
What Viruses Are Telling Us About Gene Regulation in Mammalian Cells
the proverbial Trojan horse, VPl6 molecules are
deposited in the newly infected cell as "ready-
made" transcription factors. In an elegant series
of experiments, Bernard Roizman and his col-
leagues at the University of Chicago showed that
the activating specificity of VP 16 for herpesvirus
IE genes is dictated by regulatory DNA sequences
located upstream of each IE gene. Such IE "en-
hancers," when linked onto another gene other-
wise unresponsive to VP 16, conferred a direct
and specific response.
On its own, VP 1 6 is incapable of direct associa-
tion with the IE enhancers. In order to develop an
understanding of the molecular mechanisms con-
trolling transcriptional activation by VP 16, ef-
forts have been undertaken to identify and purify
cellular proteins that bind to IE enhancers. At
least four different host cell proteins are required
for maximal activation of IE gene expression by
VPl 6. The most recent work in our laboratory has
entailed the identification and purification of a
DNA-binding complex that associates avidly and
specifically with a critical switch point within IE
enhancers. This particular switch point, termed
in the jargon of the transcription field a "cis-
regulatory element," is largely composed of gua-
nine (G) and adenine (A) residues. The cellular
DNA-binding protein that interacts specifically
with the GA-rich cis-regulatory element has been
termed GA-binding protein (GABP) .
Kelly LaMarco, a postdoctoral fellow in our lab-
oratory, purified GABP and found it to be com-
posed of two polypeptide subunits. After deriv-
ing the partial amino acid sequence from
proteolyzed fragments of each subunit. Dr. La-
Marco, along with postdoctoral associates Cath-
erine Thompson and Tom Brown, succeeded in
cloning the genes encoding each GAPB subunit.
The amino acid sequence of one subunit, termed
GABPa, exhibits significant similarity to the prod-
uct of the ETS proto-oncogene. Indeed, the ETS-
related region of GABPa represents the part of the
protein that mediates direct contact with DNA.
This same region of the protein is also neces-
sary for protein-protein contact with the other
subunit of the complex, termed GABP/?. Fortu-
nately, the sequence of GABP^ also exhibits a re-
gion of amino acid sequence similarity with previ-
ously studied proteins. GABP/5 contains four
imperfect repeats, 33 amino acids in length, that
are related to similarly sized repeats present in
the products of a number of interesting proteins.
One such relative is the product of the Notch
gene of fruit flies studied by Spyridon Artavanis-
Tsakonas (HHMI, Yale University). The Notch
gene product is a membrane protein that plays an
important role in cell-cell communication dur-
ing fruit fly development. Another protein that
contains the 33 amino acid repeats is ankyrin, a
cytoskeletal protein in red blood cells discovered
by G. Vann Bennett (HHMI, Duke University Med-
ical Center). Drs. Thompson and Brown found
that the 33 amino acid repeats of GABP/? consti-
tute the part of the protein required for direct
interaction with its matching subunit (GABPa),
thus providing the first conclusive evidence for
the mechanist role of this protein structural
motif.
Many questions regarding the properties and
function of this protein complex remain unre-
solved. For example, how does binding of the
complex facilitate activation of herpesvirus IE
genes? What role does GABP play in the control of
cellular gene expression? Might GABP in some
way influence the decision of herpesvirus to exe-
cute its lytic cycle (as it does in epithelial cells)
or the latent state it enters when the virus infects
neuronal cells? Given a critical set of molecular
reagents, including recombinant DNA clones and
specific antibodies, it should now be possible to
address these questions directly.
280
Fundamental Mechanisms of Ion Channel Proteins
Christopher Miller, Ph.D. — Investigator
Dr. Miller is also Professor of Biochemistry at Brandeis University and Adjunct Professor of Molecular
Biology at Massachusetts General Hospital, Boston. He received his B.A. degree in physics from
Swarthmore College and his Ph.D. degree in molecular biology from the University of Pennsylvania. He
carried out postdoctoral work in membrane biochemistry with Efraim Racker at Cornell University for
two years and then joined the Graduate Department of Biochemistry at Brandeis.
ION channels are the most fundamental ele-
ments of molecular hardware in the nervous
system. They are the membrane-spanning pro-
teins that directly mediate the transmembrane
ionic fluxes by which electrical signals are gener-
ated, propagated, and integrated in neurons,
muscle, and other electrically active cells. By
forming aqueous pores through the heart of the
channel protein (and hence through the mem-
brane that the protein spans), channels act as
"leakage" pathways for ions down their preestab-
lished thermodynamic gradients. These proteins
make intelligent leaks. Channels can discrimi-
nate fiercely among the different species of inor-
ganic ions present in the aqueous solutions bath-
ing the cell membrane. They can also rapidly
open and close their conduction pores in re-
sponse to external signals, such as binding of neu-
rotransmitters or changing of the transmembrane
electric field.
Work here is directed toward questions of basic
molecular mechanisms of ion channel operation
and of the underlying protein structures. Since no
high-resolution structures have been obtained
(or are soon likely to be) for this class of proteins,
it is necessary to draw structural inferences from
close examination of ion channel function. This
can be done because ion channels, unique among
proteins, can be studied quantitatively at the
single-molecule level. In this laboratory, heavy
use is made of the technique of "single-channel
reconstitution," in which individual ion channel
molecules are inserted into an artificial mem-
brane under simple, chemically controllable
conditions.
This approach has allowed us to develop crude
physical pictures of several ion channels in
which crucial dimensions have been deduced:
the conduction pore's width and length, the dis-
tance of its entryway from the lipid bilayer sur-
face, and the number of ions inside the channel
during the conduction process. We are currently
complementing these purely functional and
mechanistic studies with recently developed
methods of membrane protein biochemistry and
manipulation of ion channels at the genetic level.
Use of Peptide Neurotoxins as Probes
of Channel Structure
Charybdotoxin (CTX) is a scorpion venom-
derived peptide that blocks a small family of
K^-specific channels. Having shown that it acts by
physically plugging the channel's externally fac-
ing mouth, we are now utilizing CTX as a probe
of this important region. We are using two differ-
ent channels for these efforts: the high-
conductance Ca^^-activated channel from skel-
etal muscle reconstituted into planar lipid
bilayers, and the Shaker channel expressed in
Xenopus oocytes. Each of these channels has a
particular advantage, the former lending itself to
close mechanistic study and the latter to molecu-
lar genetic manipulation.
Using a synthetic gene for CTX expressed in
Escherichia colt, in combination with the known
structure of the peptide, we have mapped the in-
teraction surface of the CTX, using both types of
channels as CTX receptors. As a result, we
know that the peptide makes intimate contact
with the channel on a well-defined surface built
from 7-8 residues. One of these residues inter-
acts electrostatically with a ion residing in the
channel's conduction pore.
In parallel with mapping the toxin, we have
used site-directed mutagenesis with the Shaker
channel to identify channel residues making
up the CTX receptor located in the channel's ex-
ternal mouth. With this information in hand, we
are currently trying to identify channel-toxin in-
teraction partners by constructing complemen-
tary mutants of peptide and receptor. This will
allow us to deduce physical distances between
residues lining the mouth of the K"^ channel.
Purification and Reconstitution
of CI" Channels
The electric ray Torpedo californica carries in
its electric organ a CP-specific channel with an
unusual structural characteristic. The channel is
built as a dimeric, or "double-barreled" com-
plex, with two identical CI" diffusion pathways
in a single molecular unit. We have developed a
functional assay for this channel protein in a solu-
281
Fundamental Mechanisms of Ion Channel Proteins
bilized state and are performing conventional
membrane purification studies. The channel is
expressed in its natural tissue at high density, so
milligram quantities should be easily obtained.
The isolation of this channel will enable us to
study the molecular family of voltage-dependent
anion channels at the protein-biochemical level.
-Structure-Function Relations in a Minimal
K+ Channel
We are beginning a structure-function analysis
on a K^-specific channel, first cloned from a kid-
ney cDNA library, that shows a remarkable molec-
ular property: a very small polypeptide size of
only 130 amino acids, some 1 0-fold smaller than,
for example, the voltage-dependent Na"^ channel.
We have constructed and expressed a fully syn-
thetic gene for this channel, using the degeneracy
of the genetic code to build a large number of
unique restriction sites throughout the coding se-
quence. Thus prepared to perform routine cas-
sette mutagenesis, we are initiating a search for
functional domains of the channel and develop-
ing direct tests to determine whether this gene
does in fact code for an ion channel at all, a basic
question that has yet to be answered rigorously.
We have found point mutations in the single
hydrophobic domain that specifically alter ion se-
lectivity among close analogues as well as the
affinities of several channel blockers. The results
not only demonstrate that this cDNA does code
for a structural gene for a channel but also
provide initial intimations about the way the Re-
conducting pore is formed.
282
Neural Foundations of Vision
J. Anthony Movshon, Ph.D. — Investigator
Dr. Movshon is also Professor of Neural Science and Psychology at New York University and Adjunct
Professor of Physiology and Biophysics at New York University Medical Center. He received his B.A. degree
and his Ph.D. degree in experimental psychology from Cambridge University, where he worked with Colin
Blakemore. After joining the faculty of NYU, Dr. Movshon has remained there except for a sabbatical year
at Oxford University. He was founding director of the NYU Center for Neural Science. Among his honors
are the Young Investigator Award from the Society for Neuroscience and the Rank Prize in Optoelectronics.
OUR research concerns the function and de-
velopment of the visual system, especially
the visual areas of the primate cerebral cortex.
Our main experimental tools are electrophysio-
logical recording and quantitative analysis of the
visually evoked activity of single neurons. We
also draw importantly from related work in visual
psychophysics, computational modeling, and
complementary neuroanatomy.
Presently we are involved in two broad groups
of studies. The first concerns the functional prop-
erties of single neurons in the extrastriate visual
areas of the macaque monkey's cerebral cortex,
with special emphasis on the processing of infor-
mation about visual motion, space, form, and
color. The second group of studies concerns the
development of cortical visual function in mon-
keys and the way that development is affected by
abnormal early visual experience.
An important organizing theme derives from
the discovery of two functional streams in the
monkey's geniculo-cortical visual pathway. One
stream, the P system, originates in the dense and
numerous P|8 ganglion cells of the retina, contin-
ues through the parvocellular layers of the lateral
geniculate nucleus (LGN), and extends to layer
4Ci8 of the striate (or primary visual) cortex, VI .
A second stream, the M system, originates in the
large, fast-conducting but relatively sparse Pa ret-
inal ganglion cells, continues through the mag-
nocellular layers of the LGN, and enters the
striate cortex through layer 4Ca.
Signals from the P system are passed preferen-
tially into a set of cortical areas that seem to be of
special importance for the processing of form and
color, especially visual areas 2 and 4, and into the
inferior temporal cortex. Signals from the M sys-
tem pass rather selectively into another set of cor-
tical areas that seem to be essential for the analy-
sis of visual motion and visual space, especially
the middle temporal area (MT or V5), and into
the posterior parietal cortex. Our working hy-
pothesis is that these streams subserve different,
albeit overlapping visual functions, and also that
different forms of developmental visual disorder
may reflect abnormalities primarily affecting one
stream or the other.
To study the functions of cortical visual areas,
we analyze the responses evoked in single neu-
rons by visual stimuli carefully selected to permit
formal characterization of underlying neuronal
mechanisms. The class of properties in which we
are generally interested concerns the selectivity
with which neurons respond to variations along
one or another visual dimension. We also try to
examine the neuroanatomical distribution and
functional properties of neurons providing affer-
ent signals to a particular area, so that we can
attempt to understand the computational trans-
formations of the visual signal executed by the
circuits in each area.
An important concern is to establish the partic-
ular dimensions of the visual stimulus for which
neurons in that area show an invariant selectivity
— that is, for which their selectivity is unaffected
by parametric variation in other, unrelated di-
mensions. For example, neurons in VI have in-
variant selectivity for the spatial, temporal, and
chromatic structure of visual stimuli. Neurons in
MT transform afferent spatiotemporal signals into
invariant representations of an object's speed and
direction. Neurons in V4, on the other hand, may
transform simple afferent chromatic signals into
invariant representations of the object's surface
properties.
A critical issue in cortical sensory physiology is
to relate perceptual experience and judgment to
the activity of neurons and neuron assemblies. In
collaboration with William Newsome at Stanford
University, we have used statistical methods
based on the theory of signal detection to com-
pare the performance of single neurons with psy-
chophysical measures of performance obtained
concurrently from an awake, behaving monkey.
The goal is to deduce the associations between
the computation of perceptual features and the
activity of particular groups of neurons. The re-
sults suggest that small groups of neurons in area
MT may carry the signals upon which behaving
monkeys make judgments of the motion content
of visual targets. This approach allows us to form
a common language in which to consider psycho-
physical, computational, and neurobiological
283
Neural Foundations of Vision
analyses of visual cortical function. We are also
applying methods derived from statistical deci-
sion theory to the question of the absolute effi-
ciency of visual representations in the brain.
In addition to its purely visual functions, the M
stream provides signals that drive eye movements
of pursuit, the slow, smooth eye movements with
which primates stabilize on the retina the image
of a moving visual target. In collaboration with
Stephen Lisberger (University of California, San
Francisco), we have studied several aspects of the
relationship between visual and visuomotor pro-
cesses. In a series of neurophysiological studies,
we have explored the responses of MT neurons to
the dynamic motion profiles used to characterize
pursuit and have documented the suitability of
the motion-related signals in these neurons for
the task of initiating pursuit. In psychophysical
work, we have begun to explore the kinds of vi-
sual signals that pass into the oculomotor system
by examining the relationship between the de-
tectability of particular visual patterns and the
pursuit eye movements they elicit.
Our overall ambition for these studies is to
"turn the sensory-motor corner" and relate the
particulars of visual processing to the higher
mechanisms that produce voluntary motor com-
mands. To this end, we are developing computa-
tional models designed to explain the signal
transformations that take place at a series of stages
between the initial registration of the visual
image and the formulation of the final oculomo-
tor command.
To analyze development, in a project sup-
ported by the National Eye Institute, we study the
vision of monkeys reared either with an artificial
strabismus (deviation of one eye) or anisometro-
pia (difference in the refractive state of the two
eyes) . Both of these manipulations lead to condi-
tions resembling human amblyopia, a common
visual deficit of central nervous system origin. In
behavioral experiments, we learn how experi-
mental amblyopia affects perceptually defined
mechanisms that support visual sensitivity to
form, contrast, and position. Neurophysiological
studies in the same animals then reveal alter-
ations in cortical neuron properties that seem to
be related to the psychophysical ly measured vi-
sual deficits.
Using this strategy, we seek to uncover the rela-
tionship between the neural changes that under-
lie amblyopia and the perceptual consequences
of the disorder. We are currently pursuing the
idea that the relatively mild type of amblyopia
typically produced by anisometropia (having un-
symmetric parts) involves a deficit in the P sys-
tem, while the more complex syndrome that of-
ten follows strabismus also involves important
deficits in the M system.
284
Human Retroviral Gene Expression
and Cellular Transcription
Gary J. Nabel, M.D., Ph.D. — Associate Investigator
Dr. Nabel is also Associate Professor of Internal Medicine and Biological Chemistry at the University of
Michigan Medical School. He received his bachelor 's degree from Harvard College, his M.D. degree from
Harvard Medical School, and his Ph.D. degree in cell and developmental biology from Harvard University.
He was a research fellow at the Whitehead Institute, Massachusetts Institute of Technology, in the
laboratory of David Baltimore, before moving to the University of Michigan.
T lymphocytes protect the body from invasion
by foreign organisms but can also serve as tar-
gets of infection by viruses. One such agent is the
human immunodeficiency virus (HIV), which
causes the acquired immune deficiency syn-
drome (AIDS). Under normal circumstances, T
cells become activated in response to infection
and begin to synthesize a set of proteins that acti-
vate the immunologic defense system. In T cells
that contain the AIDS virus, cellular activation
provides a signal to HIV to stimulate viral replica-
tion. Our laboratory has characterized regulatory
proteins that stimulate T cell and retroviral gene
expression. These cells provide a model to study
coordinate gene expression during development
and following viral infection.
We have identified proteins that, in binding to
control regions, regulate the expression of HIV
and other immune system proteins. At the same
time, we have begun to use our knowledge of
cellular and viral transcription to deliver recom-
binant genes in vivo. This system has allowed us
to learn more about the biology of these genes,
particularly in endothelial and vascular smooth
muscle cells of the vessel wall, and has provided
new opportunities for studies on gene transfer.
Regulation of HIV Gene Expression
in T Cells and Monocytes
The expression of HIV can be activated in T
cells treated with phorbol esters or other immune
system activators. We have previously shown that
stimulation of these cells increases the binding
activity of a protein that binds to a DNA control
region, called NF-kB (nuclear factor that recog-
nizes a sequence in the k immunoglobulin light
chain of B cells) . NF-kB is responsible for stimula-
tion of HIV transcription in activated T cells. The
DNA sequence recognized by this transcription
factor is twice repeated in the HIV control region,
and mutation of these sites abolishes inducibility
of HIV. NF-kB acts in synergy with HIV products,
such as the tat-\ gene, to enhance HIV gene ex-
pression in an infected cell.
The transcription factor NF-kB is a protein com-
plex composed of a DNA-binding subunit and an
associated transactivation protein (of relative mo-
lecular masses 50 and 65 kDa, respectively).
Both subunits have similarity with the rel onco-
gene and the Drosophila maternal-effect gene
dorsal. The 5 0-kDa DNA-binding subunit was pre-
viously thought to be a unique protein, derived
from the 105-kDa gene product (pi 05). We have
recently reported the isolation of a cDNA that en-
codes an alternative DNA-binding subunit of NF-
kB. It is more similar to pi 05 NF-/cB than other
family members and defines a new subset of rel-
related genes. It is synthesized as a protein of
about 100 kDa (pi 00) that is expressed in differ-
ent cell types, contains cell cycle motifs, and like
pi 05, must be processed to generate a 50-kDa
form.
A 49-kDa product (p49) can be generated inde-
pendently from an alternatively spliced tran-
script. It has specific /cB DNA-binding activity and
can form heterodimers with other rel proteins. In
contrast to the 50-kDa protein derived from
pl05, p49 acts in synergy with p65 to stimulate
the HIV enhancer in transiently transfected Jurkat
cells. Thus p49/plOO NF-kB could be important
in the regulation of HIV and other KB-containing
genes. The above studies are supported in part by
grants from the National Institutes of Health.
In addition to HIV type 1 (HIV-1), a second
related virus, HIV type 2 (HIV-2), can induce
AIDS. HIV-2, a distinct retrovirus, shares nucleic
acid and protein similarity with HIV-1. First de-
scribed in West Africa, HIV-2 has begun to appear
throughout the world. Although HIV-1 and -2
both cause AIDS, the length of the asymptomatic
period following infection differs for the two
viruses.
Because increased viral replication is asso-
ciated with progression of HIV-related disease,
the rate of disease progression may be influenced
by host cell regulatory proteins that activate virus
replication. Such proteins could be regulated by
distinct cofactors that selectively stimulate cellu-
lar activation pathways. These T cell activation
pathways regulate specific transcription factors
that may contribute to the regulation of the latent
phase of HIV infection.
285
Human Retroviral Gene Expression and Cellular Transcription
We recently defined the transcriptional regula-
tion and induction of these retroviruses and
found that their regulation differs. A distinct T
cell activation pathway — triggering of the CD3
component of the T cell antigen receptor com-
plex— stimulates HIV- 2 gene expression but
does not affect HIV- 1 . The response to T cell re-
-<:eptor stimulation in HIV- 2 is mediated by an up-
stream regulatory element, CD3R, which a se-
quence-specific DNA-binding protein of the ets
family recognizes.
In addition, at least three other cis-acting regu-
latory sequences contribute to HIV- 2 gene ex-
pression, including /cB, another ets binding site,
and an associated element. Jurkat T leukemia cell
lines containing HIV-2 provirus also show in-
creased viral replication following stimulation of
the T cell receptor complex, in contrast to HIV-1 .
These findings suggest that HIV-2 and HIV-1
differ in their transcriptional regulation and in-
duction. These studies also raise the possibility
that different cofactors contribute to the activa-
tion of AIDS associated with HIV-1 and HIV-2.
Expression of Cellular and Retroviral
Vector Genes in Vivo
Despite recent advances in the understanding
of eukaryotic gene regulation, a major obstacle to
the therapeutic management of human disease
remains the site-specific expression of genes in
vivo. Using our knowledge of retroviral gene ex-
pression, we have developed systems that utilize
viral vectors to express biologically active pro-
teins in cells and tissues in vivo. We have devised
methods that allow a recombinant gene to be ex-
pressed efficiently at a specific site in vivo by
direct introduction of genetic material at the time
of catheterization. A recombinant jS-galactosidase
gene was expressed in a specific arterial segment
in vivo by direct infection with a retroviral vector
or by DNA transfection using liposomes. Several
cell types in the vessel wall have now been trans-
duced with recombinant genes, including endo-
thelial and vascular smooth muscle cells. Site-
specific gene expression can therefore be
achieved by direct gene transfer in vivo and
could be applied to the treatment of such human
diseases as atherosclerosis, cancer, or AIDS.
These studies have been supported by grants from
the National Institutes of Health and the Ameri-
can Heart Association.
Biologically active proteins are now being in-
troduced into cells, including disparate histocom-
patibility antigens, growth factors, growth inhibi-
tors, or immune system proteins. The goal of this
research is not only to understand basic mecha-
nisms of gene regulation, transcriptional activa-
tion, and viral gene expression, but also to define
the biological significance of factors that regulate
gene expression in complex organisms and to de-
velop novel molecular interventions for human
disease.
286
Molecular Analysis of Mtiscle Development
and Function
Bernardo Nadal-Ginard, M.D., Ph.D. — Investigator
Dr. Nadal-Ginard is also the Alexander S. Nadas Professor of Pediatrics and Professor of Cellular and
Molecular Physiology at Harvard Medical School and Cardiologist- in- Chief at the Children's Hospital,
Boston. He received his M.D. degree from the University of Barcelona, Spain, and his Ph.D. degree in
biology from Yale University. After training in internal medicine and cardiology, he was a student and
postdoctoral fellow with Clement Markert at Yale. He was a professor of cell biology at Albert Einstein
College of Medicine before assuming his present position.
OUR laboratory continues to be interested in
elucidating the molecular mechanisms that
regulate the production and function of the con-
tractile system in muscle cells. This apparatus
converts the chemical energy generated by food-
stuffs and stored in the form of ATP, and is thus
the molecular motor for locomotion and for the
heartbeat. One of its fundamental roles is in
maintaining cell shape. There are variations of
the contractile system in every cell of the
organism.
The functional unit of the contractile system is
the sarcomere, which is composed of a precisely
arranged set of proteins. These are produced by a
small family of genes, each making a different
isoform. In addition, a single gene can in many
cases produce several kinds of proteins by a pro-
cess called alternative splicing. Combinations of
the different isoforms arising from these two
mechanisms can lead to millions of different
types of sarcomeres. Thus significant functional
differences among sarcomeres are produced by
changing their components, either through gene
switching at the transcriptional level or alterna-
tive splicing from the same gene. In addition, the
performance of a given sarcomere can be affected
by changing the availability of ions in the muscle
cells. These three aspects of muscle cell biology
continue to be the focus of our research.
Transcriptional Regulation of Contractile
Protein Genes
Which sarcomeres assemble in a cell depends
on which member of the multigene family of
contractile protein genes is expressed at that par-
ticular time. To analyze the mechanisms involved
in switching from one gene to another in the same
gene family, we have concentrated on those cod-
ing for the myosin heavy chain (MHC) . This is the
most important protein of the sarcomere, since it
contains the enzymatic activity responsible for
generating force. We are currently exploring two
main questions in the regulation of the MHC
genes: What determines which one is expressed
in a given cell type at a particular developmental
stage? What determines the level of expression?
To date, the best-characterized muscle-specific
regulatory factors are the myogenic basic helix-
loop-helix (bHLH) proteins of the MyoD family.
Muscle-specific induction by these proteins de-
pends on specific DNA binding to a particular se-
quence, called an E box, present in many muscle
enhancers and promoters. The MyoDs interact
with the DNA in conjunction with other HLH pro-
teins that are distributed more widely and are
present in limiting amounts.
During the past year we have cloned and char-
acterized two new proteins that function as he-
terodimeric partners of MyoD. These are alterna-
tively spliced products of gene £2.2 and are
particularly abundant in skeletal muscle, heart,
and brain. Several lines of evidence indicate that
in vivo these proteins are the partners of MyoD in
muscle and that they also play an important role
in other tissues. The fact that these proteins inter-
act with the product of the retinoblastoma gene
suggests that they are involved in the regulation
of cell proliferation.
One of the paradoxes of the MyoD paradigm for
muscle-specific gene regulation is that not all
muscle genes contain E boxes, although they are
uniformly required for efficient muscle-specific
expression. In addition, many of the genes in-
duced by MyoD in skeletal muscle are also ex-
pressed in cardiac and, in some cases, smooth
muscle, where myogenic bHLH proteins have not
been found. It was expected, therefore, that other
transcription factors, which in skeletal muscle
might be regulated directly or indirectly by the
MyoD family, mediate activation and high-level
expression of these genes. A candidate for such a
factor is the myocyte-specific enhancer-binding
factor 2 (MEF2), which binds to a specific DNA
sequence known to be important for high-level
expression in skeletal and cardiac muscle.
During the past year we have cloned and char-
acterized a family of MEF2 factors. In humans
these factors are encoded by a family of at least
four genes, which through alternative splicing
generate a larger number of proteins. All these
proteins share domains that are strikingly similar
to the DNA-binding and dimerization domains of
287
Molecular Analysis of Muscle Development and Function
2l recently identified MADS gene family, which
includes yeast transcription factors, plant homeo-
tic genes involved in floM^er morphogenesis, and
the serum response factor.
MEF2 mRNAs accumulate preferentially in skel-
etal muscle, heart, and brain, but post-transcrip-
tional mechanisms must also contribute to the
, MEF2 tissue specificity. Remarkably, cardiac and
smooth as well as skeletal muscle contain saturat-
ing levels of MEF2 trans-activating factors that are
absent in nonmuscle cells. We have shown that
MEF2 is induced in skeletal muscle cells by the
MyoD gene family, but that this factor, by itself, is
insufficient to produce the whole muscle pheno-
type. Therefore, although MEF2 is induced in
skeletal muscle by bHLH proteins, other lineage-
determining pathways must lead to MEF2 expres-
sion in nonskeletal muscle tissues.
In this context, the finding that both muscle
and brain have high-level expression of the same
isoforms strongly suggests that MEF2 also has an
important role in neuronal gene regulation.
Whether bHLH or other factors induce MEF2
genes in cardiac and smooth muscle, as well as
nerve, the regulatory sequences of these genes
will serve as powerful tools for the identification
of the lineage-determining pathways in these cell
types.
Regulation of Protein Diversity
by Alternative Splicing
One of the fundamental aspects of pre-mRNA
splicing that is particularly important for the gen-
eration of different proteins from a single gene is
the mechanism by which splice sites are identi-
fied. We have previously suggested a scanning
model for the location of the 3' splice site of
mammalian introns. We proposed that this site is
located by a scanning mechanism searching for
the first AG downstream of the branch point/
polypyrimidine tract. During the past year we
have further confirmed and extended this model.
Recent experiments show that scanning for the
3' splice site starts at the branch point, not the
pyrimidine tract, and proceeds until an AG is rec-
ognized. Failure to recognize the most proximal
AG can arise from extreme proximity to the
branch point, or the AG can be sequestered
within a hairpin loop. Once the AG has been en-
countered, scanning stops, but the spliceosome
can still see a stretch of about 1 2-35 nucleotides
downstream. The most competitive AG within
this scanning window is then selected as the 3'
splice site. The strongest determinant of compe-
titiveness remains the proximity to the branch
point. Thus, in many respects, the scanning
model for the 3' splice site closely resembles the
model for translation initiation, both in its sim-
plest formulation and in the predictable excep-
tions to the general rule.
To analyze the mechanisms involved in alterna-
tive splicing, we have continued to focus on the
a-tropomyosin (TM) gene. This gene generates a
minimum of 1 0 different isoforms that are tissue
specific and developmentally regulated. We have
concentrated on the mutually exclusive exons 2
and 3 to elucidate the elements involved in this
type of regulation.
In the past we identified the mechanisms in-
volved in the production of the default pattern,
which results in the exclusion of exon 2 and in-
clusion of exon 3 in all cell types but smooth
muscle cells. During the past year, we have con-
centrated on the production of the regulated pat-
tern in smooth muscle cells that results in the
inclusion of exon 2 and exclusion of exon 3- This
pattern is the result of negative regulation by inhi-
bition of the default pattern and is attributable to
two well-defined sequences located 5' and 3'
from exon 3- These sequences are the binding
site for protein factors present in smooth muscle
but not in HeLa cells. Experiments are in progress
to clone this factor.
The search for splicing factors that interact
with the polypyrimidine tract has resulted in the
cloning and characterization of a novel essential
splicing factor. This is a protein of about 1 00 kDa
that is closely associated with the polypyrimidine
tract-binding protein previously described by
our laboratory and others. Experiments are in pro-
gress to determine the particular stage of spliceo-
some assembly in which this factor is involved.
288
The Genomic Response to Growth Factors
Daniel Nathans, M.D. — Senior Investigator
Dr. Nathans is also University Professor of Molecular Biology and Genetics at the Johns Hopkins University
School of Medicine. He received his B.S. degree in chemistry from the University of Delaware and his M.D.
degree from Washington University. His postdoctoral research was done at the National Cancer Institute
and the Rockefeller University. Dr. Nathans is a member of the National Academy of Sciences and serves
on the President's Council of Advisers on Science and Technology. He received the Nobel Prize in medicine
or physiology in 1978 for the application of restriction enzymes to problems in molecular genetics.
THE proliferation of mammalian cells is regu-
lated by extracellular proteins called growth
factors. When a growth factor interacts with its
specific cell surface receptor, a cascade of bio-
chemical reactions leads to the sequential activa-
tion of specific genes. Research in my laboratory
concerns the analysis of this induced genetic pro-
gram in a mouse fibroblast cell line.
Among the genes expressed in the first wave of
gene activation induced by growth factors are
many that encode transcriptional regulatory pro-
teins. We have previously described several of
these "immediate-early" transcription factors,
including members of the Jun and Fos families,
zinc finger proteins, and a helix-loop-helix pro-
tein. During the past year we have continued our
characterization of these proteins and their
genes.
Several of the immediate-early transcription
factor genes are activated by platelet-derived
growth factor (PDGF) and other ligands through
one or more serum response elements (SREs) up-
stream of each gene. In the case of jun-B, the
upstream sequence near the start of the gene has
no functional SRE, nor other signals essential for
activation of the gene. By analyzing a series of
mutants, a graduate student in the laboratory,
Evelio Perez-Albuerne, has found that jun-B has
regulatory elements downstream of the gene, in-
cluding a functional SRE and a cyclic AMP re-
sponse element that mediates the activation of
jun-B by agents that elevate intracellular cAMP.
Thus the mechanism of activation of jun-B by
serum or PDGF appears to be similar to that previ-
ously found for a number of other immediate-
early genes, except that the response elements
are farther away from the start of transcription
and downstream of the gene.
The Jun and Fos family of proteins are DNA-
binding transcription factors that form dimers
through interacting domains called leucine zip-
pers. Pierre Chevray, another graduate student,
has initiated a search for other proteins that inter-
act with Jun and Fos and regulate their activities.
For this purpose he used a previously described
yeast genetic system that allows one to detect
protein-protein interactions and to clone the
gene for an interacting protein. By this means he
has identified several proteins that interact with
the Jun segment that forms a leucine zipper.
Among these proteins are another leucine zipper
transcription factor, two previously unidentified
proteins, and the cytoskeletal proteins a- and
/3-tropomyosin, which are known to form leucine
zippers. We are now exploring the physiological
significance of these interactions.
After the appearance of immediate-early tran-
scription factors in growth factor-treated cells,
another set of genes comes into play. Activation
of these "delayed-early" genes is thought to be
mediated by immediate-early transcription fac-
tors. Associates Anthony Lanahan and John
Williams have cloned and characterized a num-
ber of cDNAs corresponding to delayed-early
genes. Some of the genes are induced by mitogens
in nonfibroblastic cell lines also. Among the pro-
teins they encode are a chromosomal protein, a
transmembrane channel protein, an enzyme in-
volved in adenine nucleotide biosynthesis, a pro-
tein related to a known cytokine, and several pre-
viously unidentified proteins. We are further
characterizing some of these. In addition, we are
studying the role of immediate-early transcrip-
tion factors in activating their genes.
Research on the Jun proteins is supported by a
grant from the National Institutes of Health.
289
I
The commonest rhodopsin mutation among patients with autosomal dominant retinitis pigmen- |
tosa has a substitution of the amino acid histidine for proline 23- Antibody-stained human opsin |
(the apoprotein of rhodopsin), expressed by transient transfection in tissue culture cells, is i
shown here. A: Wild-type opsin accumulates in the plasma membrane. B: The mutant opsin is \
predominantly in the endoplasmic reticulum. |
From Sung, C. -H., Schneider, B.G., Agarwal, N., Papermaster, D.S., and Nathans, f. 1991. Proc I
Natl Acad Sci USA 88:8840-8844. j
i
I
i
I
I
I
I
I
I
290
Molecular Biology of Visual Pigments
Jeremy Nathans, M.D., Ph.D. — Associate Investigator
Dr. Nathans is also Associate Professor of Molecular Biology and Genetics and of Neuroscience at the Johns
Hopkins University School of Medicine. His undergraduate work was in biology and chemistry at the
Massachusetts Institute of Technology. He received a Ph.D. degree in biochemistry and later his M.D.
degree at Stanford University. Before joining the staff at Johns Hopkins, Dr. Nathans spent a year as a
postdoctoral fellow at Genentech.
VISUAL pigments are the light-absorbing pro-
teins that initiate phototransduction. Each
consists of a chromophore, 1 l -cis retinal, joined
to an integral membrane protein, opsin. The vi-
sual pigments constitute one branch of a large
family of cell surface receptors that transduce ex-
ternal stimuli by activating G proteins. In the vi-
sual system, the activated G protein stimulates a
cGMP phosphodiesterase, and the resulting tran-
sient decline in cGMP closes plasma membrane
cation channels.
Photon absorption by 1 \ -cis retinal causes it to
isomerize from W-cis to all-fraws. The attached
protein then undergoes a series of conforma-
tional changes, leading ultimately to a form that
interacts with the G protein. The changes under-
lying visual pigment activation are likely to re-
semble those that accompany hormone-receptor
binding among the other members of this recep-
tor family.
Our laboratory is interested in three general
areas related to the visual pigments: their struc-
ture and function, the control of their expression,
and their variation within the human population.
Structure/Function Studies
Several years ago we developed a system for
producing large quantities of bovine rhodopsin
by expression of cloned cDNA in tissue culture
cells. As described below, we are using this sys-
tem in conjunction with site-directed mutagene-
sis to define those residues involved in protein
conformational changes, in protein-chromophore
interactions, and in protein stability.
Because the retinal chromophore is sensitive to
changes in its immediate environment, the
various conformations of light-activated rhodop-
sin each possess distinctive absorption spectra.
By following the changes in spectral absorbance
following photoactivation, one can determine
the quantity and rates of formation and decay of
each conformational intermediate. Charles Weitz,
a postdoctoral fellow, has used this assay to map
those amino acids that play an important role in
rhodopsin's transition to the active conforma-
tion, i.e., the conformation that interacts with the
G protein.
In one set of experiments, we sought to exam-
ine the mechanism responsible for the pH depen-
dence of this transition. Thirty years ago George
Wald and his colleagues observed that low pH
favors this transition and that the pH dependence
was consistent with a mechanism in which pro-
tonation of one or more histidines was tightly
coupled to the transition. We therefore mutated
one-by-one each of the six histidine residues to
phenylalanine and monitored the ability of the
mutant rhodopsin to form the active conforma-
tion. All of the mutant proteins could bind to 11-
cis retinal and form a light-sensitive pigment, but
mutants in which histidine^'' was replaced
either by phenylalanine or cysteine were unable
to assume the active conformation.
The simplest interpretation of this experiment
is that the histidine'^" is the site at which proton-
ation drives rhodopsin into its active conforma-
tion. Dr. Weitz has pursued this observation by
identifying mutants at other sites that have the
converse property: they form the active confor-
mation more efficiently than the normal protein.
This second type of mutant therefore identifies
amino acids that normally act to keep the protein
in the inactive conformation. As most amino acids
have little or no effect on this conformational
transition, it should be possible to identify the
handful of critical residues that control it.
In the human retina, rhodopsin mediates vision
in dim light, whereas a related set of visual pig-
ments, the cone pigments, mediate bright light
vision as well as color vision. The spectral proper-
ties of the cone pigments have been of great inter-
est to physiologists and psychologists and have
been the object of investigation for over a cen-
tury. Unfortunately, the instability and scarcity of
these proteins make them difficult to study. Sev-
eral years ago we isolated the genes that encode
the human cone pigments. Using these reagents,
Shannath Merbs, a graduate student, has recently
succeeded in producing large quantities of the
pigments and determining their precise absorp-
tion spectra.
291
Molecular Biology of Visual Pigments
Interestingly, the human gene pool contains
two versions of the red pigment. One carries an
alanine at position 180 and absorbs maximally at
552 nm, whereas the second version carries a ser-
ine at position 180 and absorbs maximally at 557
nm. This spectral difference explains a number of
long-standing observations regarding differences
in color vision among otherwise color-normal in-
dividuals. We are now using this system to study
hybrid pigments encoded by hybrid genes that
were generated by recombination between the
red and green pigment genes. Hybrid genes of
this type are carried by 7 percent of human X
chromosomes and account for most of the com-
mon forms of red-green color blindness.
Control of Visual Pigment Gene Expression
We have been interested for some time in the
general question of how different cells in the ret-
ina assume their correct identities. As an initial
approach, we have examined the control of vi-
sual pigment gene expression. Each photorecep-
tor cell appears to produce only a single type of
visual pigment: rhodopsin in the rods, and the
red, green, and blue pigments in their respective
cone types. As a working model, we assume that
this specificity in protein production reflects a
corresponding specificity at the level of gene
transcription.
One region of DNA that is important for activa-
tion of the red and green pigment genes has re-
cently been identified. Over the past several
years, we have studied families with a rare X-
linked disorder called blue cone monochromacy,
in which both red and green cone sensitivities are
absent. In many families a DNA deletion is ob-
served in which sequences are removed adjacent
to the cluster of red and green pigment genes on
the X chromosome. This set of deletions defines a
region of 0.6 kilobases that appears to be re-
quired for normal red and green visual pigment
gene function, even though the start sites of tran-
scription of these genes are, respectively, 3 kilo-
bases and 42 kilobases away.
Yanshu Wang, a graduate student, has recently
shown that a segment of human DNA encompass-
ing this essential region and including the red
pigment gene promoter directs expression of a
linked reporter gene to cone cells in the mouse
retina. An otherwise identical construct from
which the 0.6-kilobase segment has been re-
moved is inactive. We have observed that within
this essential segment, there is a smaller region
that has a high degree of DNA sequence homol-
ogy across species. Most likely, this region repre-
sents a binding site for one or more transcription
factors present in the red and green cones.
Retinitis Pigmentosa
During the past several years, we have begun to
work on a group of inherited retinal diseases
called retinitis pigmentosa (RP). The hallmarks
of RP are night blindness and a slow progressive
loss of peripheral vision, leading in most cases to
complete blindness by the fifth or sixth decades
of life. RP affects one person in 4,000 in all popu-
lations examined. Last year Ching-Hwa Sung, a
postdoctoral fellow, reported finding rhodopsin
mutations in one-quarter of patients with autoso-
mal dominant RP, an inheritance pattern that is
found in approximately one-fifth of the patient
population. In a group of l6l unrelated patients,
13 different mutations were identified, and in all
cases the mutations co-inherited with the disease
in affected families.
Dr. Sung has produced each of the mutant op-
sins, as well as normal human opsin, in tissue cul-
ture cells and has analyzed their biochemical
properties. The mutant proteins fall into two
classes. Members of one class resemble wild-type
opsin in yield, ability to bind 1 l-cis retinal, and
efficient transport to the cell surface. By contrast,
members of the second class are produced in low
yield, bind 1 1 -cis retinal variably or not at all, and
are transported inefficiently to the cell surface.
These characteristics suggest that the second
class of mutant proteins are either incorrectly
folded or unstable. It seems reasonable to sup-
pose that production of large quantities of an un-
stable opsin may be deleterious to the photore-
ceptor. The biochemical defect in the first class
of mutant proteins is not apparent from the ex-
periments performed thus far and is currently
under investigation.
292
Gene Regulation in Animal Cells
Joseph R. Nevins, Ph.D. — Investigator
Dr. Nevins is also Professor of Genetics and of Microbiology and Immunology at Duke University Medical
Center. He received his Ph.D. degree in virology at Duke University, where he studied with Wolfgang Joklik.
His postdoctoral studies as a Jane Cofftn Childs fellow focused on the mechanisms of mRNA biogenesis
and were conducted with James Darnell at the Rockefeller University, where he later became a faculty
member.
REGULATION of gene expression is central in
the determination of cellular phenotype and
in the complex transformations that take place
during such events as oncogenesis. The goal of
our laboratory is to elucidate the molecular mech-
anisms of gene control pathways.
Adenovirus Transcriptional Regulation —
A Role in Oncogenesis
Complex cellular events are often best studied
through the use of simple model systems. The
control of transcription (transfer of DNA-coded
information to RNA to guide protein synthesis)
mediated by viral regulatory proteins is instruc-
tive as regards the v^^orkings and control of tran-
scription factor activity in eukaryotic cells.
Work in our laboratory has focused on the elu-
cidation of transcription control by viral pro-
teins, using in vitro assays. We have found that a
cellular transcription factor termed E2F is regu-
lated by other cellular proteins and that these in-
teractions prevent E2F from being co-opted by
the viral genome. The adenovirus regulatory pro-
tein El A, however, can dissociate these com-
plexes, releasing E2F, which can then be utilized
by another viral product, the E4 protein. This re-
directs the cellular transcription factor for viral
purposes.
These findings are significant for several rea-
sons. First, they provide insight into the evolu-
tion of viral regulatory events that manipulate the
host cell for the virus's benefit. In particular, they
highlight the interplay between two viral pro-
teins (El A and E4) that together redirect the E2F
factor to a viral-specific role.
Second, this activity of ElA correlates with the
oncogenic activity of the protein. The sequences
in ElA that are responsible for this activity are
shared with other viral oncogene products, such
as the SV40 T antigen and the human papilloma-
virus E7 protein. Each of these viral proteins has
been shown to activate transcription, dependent
on E2F.
Third, our recent experiments have shown that
the viral T antigen and E7 protein can also dissoci-
ate complexes containing the E2F transcription
factor. In short, this latest discovery has defined a
common biochemical activity of these viral regu-
latory proteins that is likely part of their onco-
genic activity.
The identification of the cellular proteins that
are complexed to the E2F factor has provided im-
portant insight into the role of ElA as an onco-
gene. We have found that there are multiple such
E2F complexes and that their formation is regu-
lated during the cell cycle. One complex con-
tains the cell cycle-regulated protein cyclin A,
and another the retinoblastoma gene product.
The product of the retinoblastoma susceptibil-
ity gene {RBI) is a 1 10-kDa nuclear phosphopro-
tein that is expressed at equivalent levels in all
cell types examined except certain tumor cells in
which the RBI gene has been inactivated by mu-
tation or deletion. Given that it is the loss of RBI
function that is correlated with the development
of certain human tumors, it is widely believed
that the Rb protein functions to limit or constrain
cell proliferation.
It now appears that the ability of E 1 A, as well as
T antigen and E7, to dissociate the E2F-Rb com-
plex, releasing free E2F, is at least one of the
events of Rb inactivation. The E2F-Rb complex,
like the E2F-cycIin complex, is also regulated by
cell proliferation. Moreover, the E2F-Rb complex
is absent in cells that express a nonfunctional
form of the Rb protein.
This latter result strongly argues that E2F is a
functional target for Rb action. Thus the conse-
quence of either ElA action, the deletion or mu-
tation of Rb, or the phosphorylation of Rb is the
loss of the E2F-Rb interaction and the generation
of free E2F molecules. In short, "inactivation" of
Rb can be viewed as the loss of the E2F-Rb inter-
action. Moreover, it would appear that the conse-
quence of disruption of either the E2F-cyclin A
complex or the E2F-Rb complex is the release of
transcriptionally active E2F and that this event
may be important for stimulating the transcrip-
tion of genes whose products are critical for en-
hancing cell proliferation.
Molecular Mechanisms for Polyadenylation
The generation of the 3' terminus of the mature
mRNA, commonly termed poly(A) site formation.
293
Gene Regulation in Animal Cells
is a critical event in the biogenesis of most mRNA
molecules. Moreover, since many transcription
units encode multiple mRNAs that utilize alterna-
tive poly(A) sites, there is a potential regulatory
role for polyadenylation. Indeed, analysis of the
formation of the immunoglobulin heavy-chain
transcripts during B cell differentiation suggests
-that alternative poly(A) site choice is a contribut-
ing factor in the switch from production of the Ig
mRNA that encodes the membrane-bound form of
the protein to the mRNA encoding the secreted
form of the protein. Thus an understanding of the
mechanism of poly(A) site utilization, including
the sequences directing the processing event and
the factors involved in processing, is an impor-
tant focus of our work.
We have recently used a biochemical approach
to investigate the molecular mechanism of poly-
adenylation. Factors have been purified from
HeLa cell nuclear extracts that can reconstitute
accurate and sequence-specific poly(A) site pro-
cessing in vitro. Two of these factors interact spe-
cifically with the pre-mRNA, and an analysis of
this interaction suggests a pathway of assembly of
a functional poly (A) site-processing complex.
The interaction of one factor with the con-
served AAUAAA element is specific, but the RNA-
protein complex that forms is very unstable and
rapidly dissociates. The interaction of a second
factor with the RNA requires a distinct sequence
element but also requires the prior interaction of
the first factor with the AAUAAA element. Most
importantly, the resulting ternary complex is
stable and does not readily dissociate.
These observations suggest a pathway of assem-
bly of poly(A) site factors that involves an initial
recognition of the AAUAAA element followed by a
commitment step in which the second factor sta-
bilizes the overall complex. Since the relative sta-
bility of this complex reflects the relative effi-
ciency of processing, it appears that these
interactions may be important regulatory targets
for the control of polyadenylation.
This work on the mechanisms of polyadenyla-
tion is supported by a grant from the National
Institutes of Health.
294
Molecular Genetics ofX-linked Disease
Robert L. Nussbaum, M.D. — Associate Investigator
Dr. Nussbaum is also Associate Professor of Human Genetics, Pediatrics, and Medicine at the University of
Pennsylvania School of Medicine and Consultant in Clinical Genetics at the Children's Hospital of
Philadelphia. He received his undergraduate training in applied mathematics at Harvard College and his
M.D. degree at Harvard Medical School in the Harvard-MIT Joint Program in Health Sciences and
Technology. After his residency in internal medicine at Barnes Hospital, Washington University School of
Medicine, he moved to Baylor College of Medicine, first for a genetics fellowship with Thomas Caskey and
Arthur Beaudet and later as a faculty member. He then moved to the University of Pennsylvania, where
he developed his research program in molecular genetics and its application to the diagnosis and
elucidation of human genetic disease.
THE research in my laboratory is directed to-
ward elucidating the molecular bases for a
number of human genetic diseases. Each disease
under investigation is known to be caused by a
gene on the X chromosome, but the molecular
mechanism, the gene involved, and the nature of
the underlying mutations have been hitherto un-
known. Recombinant DNA techniques have been
employed to isolate the responsible genes, with
the aim of furthering our understanding of the
normal processes that when disrupted result in
each of these diseases.
Choroideremia
Choroideremia is a rare X-linked disease of the
retina that produces blindness in affected males.
The gene responsible and the mechanism of reti-
nal damage have until recently been unknown.
Our laboratory is using information about where
the choroideremia gene is to identify it and ex-
plain why mutations in this gene cause the
disease.
We have isolated a gene from the region around
a chromosomal translocation in a female patient
with choroideremia. Her disease resulted from
disruption of the choroideremia gene by the chro-
mosomal break in the X chromosome in this re-
gion. A transcribed gene that is disrupted by this
chromosomal translocation has been identified
and found to be very similar, although not identi-
cal, to one isolated in the laboratory of Frans
Cremers by his study of males with choroi-
deremia and submicroscopic deletions. The ex-
pression of this gene, at the level of mRNA,
is abnormal in 75 percent of patients with
choroideremia.
The predicted protein sequence of the gene
identified in both laboratories has a subtle similar-
ity with a protein isolated from platelets that may
be involved in regulation of the G proteins, a very
large and heterogeneous group involved in carry-
ing certain intracellular signals. The cellular
mechanism by which vision occurs, known as vi-
sual transduction, is the entire pathway by which
light, the initial signal, is translated first into bio-
chemical reactions and then nervous impulses in
the retina. A growing body of evidence suggests
that abnormalities in the proteins that carry out
visual transduction contribute to a variety of de-
generative retinal diseases that lead to blindness.
The plan is to investigate the exact role of the
choroideremia gene product in visual transduc-
tion, in an attempt to understand how abnormali-
ties in that gene lead to retinal degeneration. This
work is supported by a grant from the National
Institutes of Health.
Lowe's Syndrome
Lowe's syndrome is an uncommon X-linked
disease that causes mental retardation, cataracts,
and kidney dysfunction. The cause is unknown.
As with choroideremia, our strategy is to identify
the Lowe's syndrome gene through information
about its location.
We have isolated two cDNA sequences from the
region around a translocation breakpoint in a fe-
male with Lowe's syndrome. The transcript de-
tected by one of these cDNAs is disrupted by the
translocation and is absent or abnormal in nine
unrelated males with the disease. This gene en-
codes a protein involved in the metabolism of
inositol phosphates, a very heterogeneous but
important class of compounds with a complex
biochemistry. These structural components of
normal cell membranes also play an important
role in anchoring proteins to cell membranes and
in intracellular signaling pathways.
Our goal is to understand the biochemical pro-
cesses that, when defective, lead to brain, lens,
and kidney dysfunction and damage. Insights into
normal lens formation and normal brain and kid-
ney function could result, and methods for im-
proved diagnosis and therapy for the disease may
be found. This work is supported by a grant from
the National Institutes of Health.
295
Expression pattern of theWnt-5 gene in the developing mouse brain. In
situ hybridization at 10 days of gestation shows '^m-i expression in
one of the compartments of the diencephalon, D2, and in the develop-
ing midbrain-hindbrain junction that give rise respectively to the dor-
sal thalamus and the cerebellum.
Research and photograph by Patricia Salinas in the laboratory of
Roel Nusse.
Function of Oncogenes in Early Emhryogenesis
Roel Nusse, Ph.D. — Associate Investigator
Dr. Nusse is also Associate Professor of Developmental Biology at the Stanford University School of
Medicine. He obtained his Ph.D. degree from the University of Amsterdam and was a postdoctoral fellow
with Harold Varmus at the University of California, San Francisco, before returning to Amsterdam, where
he became head of the Department of Molecular Biology at the Netherlands Cancer Institute. Several years
ago he moved to the Beckrnan Center of Stanford University and the Howard Hughes Medical Institute.
Dr. Nusse is a member of the European Molecular Biology Organization (EMBO ).
IT is now well established that genes whose al-
tered expression can lead to cancer (generally
called oncogenes or proto-oncogenes) are indis-
pensable regulators of normal cell proliferations.
Many oncogenes participate in the cells' deci-
sions whether to divide or remain quiescent.
Apart from the regulation of cell proliferation,
growth in a normal organism obviously needs
control at another level: during the formation of
patterns in which cells become properly
arranged. Over the past few years, it has become
increasingly clear that oncogenes are among the
key regulators of this aspect of growth control
also.
The evidence for such controlling functions
has primarily come from the genetic dissection of
morphogenesis in organisms such as Drosophila
and Caenorhabditis elegans. Many of the devel-
opmental genes isolated from those animals are
highly homologous to mammalian oncogenes, in
particular those that encode signaling molecules
involved in cell-to-cell communications during
embryogenesis. Now, with the increasing possi-
bilities to identify developmental genes in the
mouse as well, additional examples have
emerged for a link between cancerous growth
and the control of normal development.
The Wnt/wingless gene family is a paradigm
for the connections between tumorigenesis and
embryogenesis. The prototypic member of this
group, Wnt-l, is an oncogene frequently acti-
vated in mouse mammary cancer. The Wnt-l
gene is normally not expressed in the mammary
gland or in most other adult tissues. In tumors,
however, its transcription is induced by nearby
insertion of proviral DNA of a retrovirus, the
mouse mammary tumor virus. Proof that Wnt-l is
an oncogene came from transfection experiments
and from the finding that introduction of Wnt-l
as a transgene into the germline of mice can lead
to tumor induction. Wnt-l is part of a family of
genes that in the mouse consists of at least 10
members. All of these genes encode secreted pro-
teins rich in cysteine residues.
Over the past few years, evidence that the Wnt
genes are prime determinants for early develop-
ment has accumulated from many different
corners. For example, most of these Wnt genes
have a very restricted pattern of expression dur-
ing early developmental stages, in organisms
ranging from mice to fruit flies. More telling are
the findings that Wnt gene mutations prevent
normal development of the mouse brain and nor-
mal segmentation of Drosophila embryos. More-
over, it has been shown that ectopic expression
of Wnt genes induces axis duplication in frog
embryos.
The work in our group is focused on the role of
several members of the Wnt gene family in the
development of the mouse, along with investiga-
tions of Wnt genes in the fruit fly. In the mouse
we perform detailed in situ RNA hybridization
analysis of the expression of Wnt-5 and a highly
related gene, Wnt-5A. Both genes are expressed
in the developing neural tube.
In particular, the anterior boundary of expres-
sion of Wnt-5 and Wnt-5A is interesting. The
genes are expressed in the diencephalon and in
the cerebral hemispheres, suggesting that they
play important roles in establishing these com-
partments in the developing brain. For example,
Wnt-5 is expressed in the D2 neuromere of the
developing diencephalon during day 9 and 1 2 of
embryogenesis. The expression precedes the sub-
division of the diencephalon into the ventral and
dorsal thalamus (see figure), suggesting a func-
tional role for Wnt-5 in this process. Another re-
markably restricted expression pattern of Wnt-5
is seen in the cerebellum, where the gene is ex-
clusively expressed in the Purkinje cells.
Understanding the mechanism of action of the
Wnt-l gene family during embryogenesis has
been complicated by difficulties in characteriz-
ing the Wnt proteins. We wish, for example, to
identify the receptors for these secreted mole-
cules, but as no biologically active Wnt gene
product is available, the nature of the receptors
remains elusive. In parallel to these biochemical
experiments, we study the interactions between
Wnt and other genes by taking advantage of the
extensive genetic analysis of Drosophila embryo-
genesis. Some years ago we made the surprising
finding that the Drosophila Wnt-l homologue is
297
Function of Oncogenes in Early Embryogenesis
identical to the segment polarity gene wingless.
This observation has allowed us to study the in-
teractions of Wnt-\/ wingless with other segmen-
tation genes.
The basic body plan of the fruit fly is estab-
lished by several classes of genes that progres-
sively divide the embryo into smaller compart-
^-ments: the gap genes, the pair-rule genes, and the
segment polarity genes. Whereas the gap and
pair-rule genes encode nuclear proteins and are
active before the Drosophila embryo becomes
cellularized, the segment polarity genes are the
first that control cell-cell interactions. The wing-
less gene is a good example, encoding a secreted
factor, but other segment polarity genes are
thought to interact with wingless. To study the
properties of the wingless protein, we have made
antibodies that recognize it in whole-mount em-
bryos and in individual cells. The protein is seen
on the surface of cells and in intracellular vesicu-
lar structures. We have characterized the distri-
bution of the wingless protein in embryos that
are mutant for wingless itself and have obtained
evidence that lack of secretion of the protein is
the primary defect in these mutants.
In an additional approach, we have overex-
pressed the wingless gene in embryos, from a
heat-shock promoter. The phenotype of the heat-
shock wingless embryos is the opposite of the
phenotype of embryos lacking wingless. Surpris-
ingly, heat-shock wingless embryos are very simi-
lar to embryos that are mutant for two other
segment polarity genes, called naked and zeste-
white-3, suggesting common biochemical path-
ways of wingless and those genes.
We have also found that wingless in Drosoph-
ila is part of a gene family, with at least two addi-
tional members. These genes, called DWnt-2 and
T>Wnt-5 for the time being, are also expressed
during early embryogenesis, in characteristic pat-
terns that differ from wingless. Mutants at these
genes have not yet been obtained, but the un-
usual character of the DWnt-5 protein, which is
three times as long as other Wnt proteins, may
give us new clues as to the biochemical mecha-
nism of action of the genes.
In the future we hope to extrapolate our in-
creasing understanding of the function of the
Wnt genes in embryogenesis to the mechanism of
action of the genes in cancer.
298
Molecular Mechanisms That Regulate
B Cell Development
Michel C. Nussenzweig, M.D., Ph.D. — Assistant Investigator
Dr. Nussenzweig is also Assistant Professor and Head of Laboratory at the Rockefeller University. He
received his undergraduate and M.D. degrees from New York University and his Ph.D. degree from
Rockefeller. He completed his residency and clinical fellowship at Massachusetts General Hospital
and conducted postdoctoral research at Harvard Medical School with Philip Leder.
WE are interested in understanding the mo-
lecular mechanisms that regulate B lym-
phocyte activation and differentiation. Our ap-
proach has been to focus on one important
transition in B lymphocytes, allelic exclusion.
The immune system is responsible for protect-
ing vertebrates from both invasion by infectious
organisms and deregulated growth of endoge-
nous malignancies. To accomplish this task, the
immune system must be able to distinguish self
from non-self. Evolution has solved this problem
in higher vertebrates by providing a network of
cell types and humoral agents. It is the lympho-
cytes— T cells and B cells — that direct the speci-
ficity of immune responses.
Although the mechanism of antigenic recogni-
tion differs for these two cell types, the genera-
tion of receptor diversity is achieved in a similar
fashion. In both cases the business end of the re-
ceptor is created in individual somatic cells by a
series of genetic recombinations at a minimum of
two loci. For example, immunoglobulin mole-
cules that serve as the B cell receptor are com-
posed of two sets of rearranging genes that en-
code the heavy- and light-chain immunoglobulin
proteins. Furthermore, the same genes that en-
code the B lymphocyte receptor also direct the
production of secreted antibodies that are an
important component of humoral immune re-
sponses. Thus the regulation of T cell and B cell
receptor rearrangements is a central feature of
the generation of immune responses.
The joining events that bring together the im-
munoglobulin segments occur in an ordered and
regulated fashion. In B lymphocytes, rearrange-
ments begin at the heavy-chain locus with the re-
combination of D and J segments. This is followed
by joining of DJ with one of 100-1 ,000 variable-
region segments. After a functional immunoglob-
ulin heavy-chain transcription unit is created, the
light-chain genes undergo a similar set of
rearrangements .
One poorly understood aspect of these events
is the ability of lymphocytes to limit themselves
to the production of a single receptor. Since pro-
ductive rearrangements could occur in two
heavy-chain and four light-chain alleles, a single
B lymphocyte could potentially make several
types of receptors, including hybrid molecules.
The mechanism that ensures that only one recep-
tor is produced is referred to as allelic exclusion.
It is an important safeguard for the immune sys-
tem, since production of multiple receptors by a
lymphocyte would dilute the specificity of any
given immune response.
Much of the early work in the area of allelic
exclusion was based on examining the status of
immunoglobulin genes in transformed B cells.
The transformed cells are frozen in one stage of
lymphocyte development and for this reason
offer only a static picture of important regulatory
events. Unfortunately there is no in vitro system
that faithfully reproduces regulated immuno-
globulin gene rearrangements. To study how im-
munoglobulin genes can regulate allelic exclu-
sion, we turned to transgenic mice.
Our approach has been to introduce into the
germline of mice human immunoglobulin genes
that have been modified to direct the synthesis of
either membrane-associated or secreted immuno-
globulin heavy chains. We found that the expres-
sion of human membrane-bound immunoglobu-
lin M (IgM) results in the exclusion of most
endogenous mouse immunoglobulins. The se-
creted version of the same transgene has little
effect.
This initial observation raises two important
questions. First, how does membrane immuno-
globulin signal? Second, how is exclusion regu-
lated at the genetic level? During the past year we
have made significant progress toward resolving
both questions. To understand signaling by the
membrane-associated immunoglobulin mole-
cule, we have developed a system for complete
functional reconstitution of the immunoglobulin
receptor from cloned components in heterolo-
gous cells. Transport of IgM to the surface of T
cells requires coexpression of the immunoglobu-
lin heavy and light chains with B29, an immuno-
globulin-associated polypeptide. In addition, the
transfected receptor is fully active in the pres-
ence of B29. MBI , a second IgM-associated poly-
299
Molecular Mechanisms That Regulate B Cell Development
peptide, is not required for transport or signal
transduction. Progress in understanding the func-
tional role of the immunoglobulin receptor sub-
units in signaling has been hindered by the multi-
subunit nature of the receptor. In addition,
transfected immunoglobulin expressed on the
surface of fibroblasts does not appear to be func-
tional for signal transduction, even in the pres-
ence of B29 and MBl. Our ability to produce a
functional receptor by transfection in T cells es-
tablishes that there is enough structural similarity
between the T and B cell signal transduction
pathways to allow recognition of the IgM-B29
complex in T cells. This observation should sim-
plify the structural and functional analysis of
the IgM antigen receptor and ultimately allow
us to understand how membrane immunoglobu-
lin signals for antigen recognition and allelic
exclusion.
In addition to our ability to study the structural
requirements for immunoglobulin function, our
experiments also have potential clinical implica-
tions. One major difference between the T cell
receptor and immunoglobulin antigen receptors
is the nature of the antigen recognized. Immuno-
globulins recognize antigens directly; the major
histocompatibility complex (MHC) restricts rec-
ognition of antigen by the T cell receptor. The
requirement for MHC recognition severely limits
the targets recognized by T cells and restricts
transfer of cellular immunity. Our reconstitution
experiments offer one potential solution to this
difficult clinical problem — the production of T
cells that utilize immunoglobulins as antigen re-
ceptors. In T cells that recognize antigen with
immunoglobulin receptors there is no MHC re-
striction, and thus this barrier to transfer of cellu-
lar immunity is potentially abrogated.
Our second goal has been to understand allelic
exclusion at the genetic level. We have docu-
mented that chromosomal position plays an im-
portant role in the regulation of gene rearrange-
ments in the immunoglobulin locus. In addition
we have started to investigate the regulatory
function of the recombinase genes RAG-1 and
RAG-2 in allelic exclusion. Deregulating the ex-
pression of the RAG genes in the lymphocytes of
transgenic mice has profound effects on lympho-
cyte differentiation and function. Animals that
carry /MG- 1 and /2/4G'-2transgenes develop hepa-
tosplenomegaly and a profound lymphoprolif-
erative disorder that is rapidly lethal. We are
pursuing a molecular understanding of the patho-
genesis of this disorder.
300
Mechanism of DNA Replication
Michael E. O'Donnell, Ph.D. — Assistant Investigator
Dr. O'Donnell is also Associate Professor of Microbiology at Cornell University Medical College, New York
City. He received his Ph.D. degree from the University of Michigan, Ann Arbor, where his research on
electron transfer in the flavoprotein thioredoxin reductase was conducted with Charles H. Williams, Jr.
He performed postdoctoral work on Escherichia coli replication with Arthur Kornberg and then on herpes
simplex virus replication with Bob Lehman, both in the Biochemistry Department at Stanford University.
MY laboratory is studying the duplication of
genetic information. By understanding the
fundamental mechanisms of cell growth, or the
replication of DNA, we may obtain insights into
the development of abnormal cells, including tu-
mor cells.
The genetic material of our cells, the chromo-
somes, is a library with all the information
needed for the multitude of duties required to
maintain the cell's life. Included in these duties
is the buildup of complete, new cellular machin-
ery for the synthesis of another cell (reproduc-
tion). The chromosome library is made of two
long interwound helical fibers of DNA (deoxyri-
bonucleic acid polymers). Before a cell can di-
vide to form two new cells, it must duplicate the
genetic library so that each cell has a complete
copy of instructions on how to live.
The process of duplicating DNA is intricate,
and the cell has evolved a precision machine to
carry out this important task. Its several protein
parts are like gears of a machine, which coordi-
nate their actions to unzip and unwind the dou-
ble-helical strands of DNA. The machinery then
uses the separated single strands as templates to
synthesize two double-helical daughter chromo-
somes. Subsequently these will segregate in two
newly formed cells.
Our goal is to understand, at a molecular level,
the workings of proteins in the mechanics of DNA
duplication. The system we are studying is the
bacterium Escherichia coli, a relatively simple
organism. The protein machine that duplicates
the E. coli chromosome is called DNA polymer-
ase III. The DNA polymerase III of E. co/i has nine
accessory proteins plus the polymerase. Like the
E. CO/? polymerase III, the DNA polymerases that
replicate the chromosomes of higher organisms
such as yeast, Drosophila, and humans are also
composed of a DNA polymerase protein and sev-
eral other "accessory" proteins.
The function of the DNA polymerase protein
(the a-subunit) is to synthesize the DNA. One of
the accessory proteins, the e-subunit, is an exonu-
clease that "proofreads" the product of the poly-
merase protein. Very little is known about the
functions of the other eight accessory proteins.
However, since the several accessory proteins to
the DNA polymerase are conserved in evolution
from bacteria to humans, it seems reasonable to
expect their individual functions to serve very
important roles in the process of chromosome
duplication. Analysis of the E. coli DNA polymer-
ase III system will likely extend and generalize
the understanding of the replication process in
all organisms.
We have recently developed methods to obtain
pure preparations of each protein, or subunit, of
the E. coli DNA polymerase III, and from these
the whole complex can be reassembled. We have
studied the individual subunits for biochemical
activities and for their physical interactions. Two
subunits, 7 and b, bind to each other to form a
complex that, upon binding to primed DNA, hy-
drolyzes ATP. In the presence of the /3-subunit,
the 75 heterodimer couples the hydrolysis of ATP
to clamp a dimer of the |8-subunit onto primed
DNA. One molecule of the 76 heterodimer can
clamp many (8-dimers onto primed DNA. Our bio-
chemical studies indicated that the |8-dimer is
clamped to DNA by encircling it like a doughnut.
The x-ray structure of this /3-clamp has recently
been solved in a collaboration with John Kuriyan
(HHMI, Rockefeller University). It appears as a
thin disk with a hole through the middle to ac-
commodate the DNA. These studies on the /?-
clamp and the 76 heterodimer were funded by
the National Institutes of Health.
The |8-clamp on DNA binds the polymerase sub-
unit, tethering it to the DNA template. Whereas
the polymerase alone is slow (20 nucleotides/
second), it is greatly accelerated upon binding
the jS-clamp (700 nucleotides/second) and repli-
cates an entire 8-kb single-strand circular DNA
without coming off (processive). This fits nicely
with the fact that the E. coli cell duplicates its 4
million-base chromosome within 30 minutes.
Another function of the 7-, 6-, and j8-subunits is
to rapidly deliver the polymerase subunit from a
completely replicated DNA molecule to a new
primed DNA template. This rapid delivery of poly-
merase is important because one strand of the
301
Mechanism of DNA Replication
DNA duplex (lagging strand), as a result of the
geometry of the DNA helix, must be replicated in
fragments. Synthesis of these fragments requires
that the polymerase subunit be used over and
over every 1-2 seconds by repeatedly delivering
it from a finished fragment to a new one. Studies
on the mechanism of rapid polymerase delivery
-j;o new primed templates were supported by a
grant from the National Institutes of Health.
Two T-subunits bind tightly to each other (di-
meric), and each binds a polymerase molecule.
Hence the r-subunit dimer serves as a scaffold to
form a twin polymerase. Since the chromosome
has two strands of DNA, both of which must be
replicated, the twin polymerase likely serves the
function of coordinately replicating both DNA
strands at the same time. These studies on the
dimeric polymerase were also supported by a
grant from the National Institutes of Health.
These studies on four (7, 5, /?, t) of the eight
accessory proteins of E. coli DNA polymerase III
have been greatly aided by having the genes (the
informational area in the DNA) for three of them
(7, fi, r). This has provided large quantities of
proteins for studies via molecular cloning and
overproduction techniques. We have recently dis-
covered the genes for each of the remaining five
proteins (5, 5', x, 0) of DNA polymerase III and
have used these genes to produce and purify large
quantities of these proteins. Identification of the
genes encoding 6', x, 4^, and 6 was supported by a
grant from the National Institutes of Health.
Toward a goal of understanding the overall
structure of the holoenzyme, we are using the
pure subunits to define the various subunit con-
tacts within the holoenzyme and to assemble the
entire replicating machine from its separate
parts. Our plans to identify the individual func-
tion of each subunit include biochemical analysis
and also use of the newly discovered genes for the
holoenzyme subunits to construct genetic mu-
tants of E. coli for further clues to the function of
each subunit. In addition to the polymerase holo-
enzyme, the helicase and priming proteins are
also central to the process of chromosome repli-
cation, and we have initiated studies to examine
how these proteins coordinate their actions with
the holoenzyme.
Model of the 13-clamp of the en-
zyme DNA polymerase III. The
function of the clamp is to
tether the enzyme to the DNA,
greatly accelerating DNA syn-
thesis. The two parts of the ^-
subunit dimer, represented in
red and yellow, encircle the
DNA ( in green and gray ),
which has been modeled in the
central cavity.
From Kong, X.-P., Onrust, R.,
O'Donnell, M., and Kuriyan, J.
1992. Cell 69:425-437. Copy-
right © 1992 by Cell Press.
302
Large-Scale Analysis of Yeast and Human DNA
Maynard V. Olson, Ph.D. — Investigator
Dr. Olson is also Professor of Genetics at the Washington University School of Medicine. He was trained as
a chemist, receiving his B.S. degree from the California Institute of Technology and his Ph.D. degree from
Stanford University. After five years on the faculty at Dartmouth College, he moved to the University of
Washington and changed fields from chemistry to genetics. He has served on the National Research
Council Committee on the Mapping and Sequencing of the Human Genome and presently serves on the
NIH Program Advisory Committee of the National Center for Human Genome Research.
MOST human cells contain 6 billion base
pairs of DNA. Embedded therein are an un-
known number of genes, perhaps 100,000, that
direct the biochemical events in the cells. At pres-
ent, well over 95 percent of this DNA is
unexplored.
Geneticists have developed powerful methods
with which to study DNA in small packets. Gene-
splicing techniques, DNA sequencing, and meth-
ods of reintroducing altered DNA molecules into
cells allow the detailed structural and functional
analysis of DNA molecules containing up to tens
of thousands of base pairs. Our laboratory seeks
to extend these approaches to encompass mole-
cules ranging up to millions of base pairs in size.
In the short run, this research should allow the
analysis of larger functional units of DNA — for
example, large human genes, clusters of coregu-
lated genes, and such structures as centromeres
and telomeres, which govern the behavior of hu-
man chromosomes during the cell division cycle.
In the long run these methods should allow the
systematic analysis of the whole human genome
— the entire complement of DNA sequences —
thereby creating tools, such as detailed maps, that
would be of permanent value in biology.
Our approach has been to build on progress in
the genetic analysis of microorganisms, particu-
larly yeast. The common laboratory yeast Sac-
charomyces cerevisiae, familiar from its use in
baking and wine making, is an ideal model for
studies of human cells, because its genetic organi-
zation and biochemical pathways are similar to
those in higher organisms. It is a powerful tool as
well, because human DNA can be altered by
gene-splicing techniques into a form that is stably
propagated in yeast.
We are just completing a long-term project to
map the 1 4 million base pairs of DNA present on
the 16 natural yeast chromosomes. The yeast
chromosomes have been mapped at a resolution
(i.e., the average spacing between landmarks) of
3,000 base pairs, and nearly 200 genes have been
localized on the map. New genes can now be
placed on this map in only a few hours. This pro-
cess, which replaces mapping techniques that re-
quired weeks of effort, is now in use in more than
100 laboratories.
Despite this success, it has long been apparent
that no straightforward extension of these meth-
ods would be successful on the greatly expanded
scale of the human genome. For this reason, we
developed a method to import manageable seg-
ments of the human chromosomes into yeast,
where they can be propagated as yeast artificial
chromosome (YAC) clones. One advantage of
YACs over previous cloning systems is that there
is no absolute upper size limit to the clones. We
can now prepare large collections of YAC clones,
each containing a different segment of human
DNA, averaging hundreds of thousands of base
pairs in size, a 10-fold improvement over the ca-
pacities of previous systems. Another advantage is
that the methods of packaging, maintaining, and
replicating DNA are more similar in yeast than in
bacteria to the analogous processes in human
cells. Consequently, a higher fraction of the hu-
man genome can be successfully propagated in
yeast than in the more-conventional bacterial
hosts.
Methods of recovering any desired segment of
human DNA as a YAC have become standard dur-
ing the past three years. YAC clones played a cen-
tral role in two recent major successes in human
genetics: discovery of the molecular basis of the
fragile X syndrome, carried out in part in the labo-
ratories of Thomas Caskey (HHMl, Baylor College
of Medicine) and Stephen Warren (HHMI, Emory
University) ; and the discovery of the gene that is
mutated in familial adenomatous polyposis,
carried out in part in the laboratory of Raymond
White (HHMI, University of Utah). The fragile X
syndrome is a common heritable cause of mental
retardation; familial adenomatous polyposis is a
genetic predisposition to colon cancer.
Now that YAC cloning is a proven method for
recovering large blocks of human DNA, our atten-
tion has turned to the analysis of YACs. A need for
efficiency arises from the sheer scale of human
chromosomes: approximately 10,000 YACs
would be required to recover the DNA present in
all the human chromosomes. To meet this chal-
lenge, it will be necessary to develop a new area
503
Large-Scale Analysis of Yeast and Human DNA
of analytical biochemistry. Although the ultimate
goal is fully automated methodology, the princi-
ples that underlie DNA analysis are too poorly
understood to support immediate instrumenta-
tion development. One key will be to obtain a
better grasp of the information flow that accom-
panies the mapping and sequencing of DNA. In
collaboration with Will Gillett (Washington Uni-
versity), we have analyzed the computational
problems posed by DNA-mapping methods in
which large blocks of DNA are broken down into
smaller segments by subcloning. Powerful, flexi-
ble software devoted to this problem is under de-
velopment. Parallel efforts to achieve a better un-
derstanding of the experimental steps are also
under way.
It is difficult to overestimate the extent to
which the analysis of DNA — cloning, mapping,
and sequencing — has come to underlie progress
in biology and biomedical research. Biological
research has been driven by advances in method-
ology, and the systematic analysis of DNA is one
of the field's critical technologies. However, in
proportion to the challenges posed by the genetic
complexity and diversity of organisms, the devel-
opment of this technology is still in its infancy.
Dr. Olson is now Professor of Molecular Bio-
technology at the University of Washington,
Seattle.
304
Molecular Genetic Studies of Hematopoietic Cells
Stuart H. Orkin, M.D. — Investigator
Dr. Orkin is also Leland Pikes Professor of Pediatric Medicine at Harvard Medical School. He received his
B.S. degree in biology from the Massachusetts Institute of Technology and his M.D. degree from Harvard
Medical School. His postdoctoral research was in the Laboratory of Molecular Genetics at NIH under the
supervision of Philip Leder. Upon returning to Harvard, Dr. Orkin received specialty training in pediatric
hematology at the Children 's Hospital, where he later joined the faculty. His many honors include the
Clinical Investigator Award from the American Federation for Clinical Research and the Dameshek prize
of the American Society of Hematology. He is a member of the National Academy of Sciences.
ALL mature blood cells are derived from plur-
ipotem hematopoietic stem cells, which
constitute a rare population in the bone marrow.
The decision of stem cells to differentiate leads to
the production of a heterogeneous array of cells
with varying developmental potentials and with
commitment to expression of lineage-specific
protein products. A major goal of this laboratory
is an improved understanding of hematopoietic
cell development and the expression and func-
tion of specific genes that relate to the normal
biology of hematopoietic cells.
Efforts are directed to analyses of both red and
white blood cells. These cell types are important
in severe, clinically significant genetic disorders
in which the capacity to produce specific pro-
teins is impaired by mutation. In these studies we
seek to describe the molecular basis of inherited
disorders, understand the normal regulation of
the affected genes, and utilize the findings from
this basic work to formulate novel treatments
based on molecular biologic considerations.
One of the major, classical disorders of red
cells is ;8-thalassemia (also known as Cooley's ane-
mia) , in which the synthesis of hemoglobin is de-
fective. Through molecular cloning and gene ex-
pression, the molecular basis of the disease was
determined in this laboratory several years ago.
Now the unsolved problems are related to how
globin genes are normally regulated in develop-
ing erythroid precursor cells. Specifically, how
are the globin genes activated only in red cells?
How are different globin genes regulated in devel-
opment? How do erythroid precursor cells arise
during development from progenitor cells that
have the potential to yield either red or white
cells?
To approach these general problems, we have
concentrated on identifying and characterizing
unique DNA-binding proteins that appear to be
major transcriptional regulators in erythroid
cells. A prominent, apparently erythroid-specific
DNA-binding protein was first discovered that rec-
ognizes a small DNA motif (GATA) found in
the promoters or enhancers of all erythroid-
expressed genes. Through molecular cloning,
mammalian, avian, and amphibian homologues
were characterized.
The protein is modular, consisting of a novel
two-finger structure required for DNA-binding
and other domains that serve as potent positive
activators of gene transcription. Expression of the
protein in two other hematopoietic cell types,
megakaryocytes and mast cells, suggests that it is
first expressed in a multipotential progenitor cell
and may regulate genes in those cell types as
well. Recent data have supported these conclu-
sions. Attention has been directed to how this
transcription factor is itself regulated in hemato-
poietic cells. An improved understanding may
provide important insights into the initial events
involved in erythroid decision-making and
maturation.
Studies of the gene revealed an element in the
promoter region that serves as a site for positive
autoregulation. In this manner, expression of the
factor tends to maintain its own expression and
stabilize the differentiated state. Furthermore,
the promoter for the receptor for the erythroid-
specific growth factor, erythropoietin, is under
control by this transcription factor. By such cir-
cuitry, expression of the factor tends to guarantee
subsequent erythroid development and viability.
Site-specific gene disruption in mouse embryo-
derived stem cells and generation of chimeric an-
imals has also revealed that the protein is essen-
tial for normal erythroid development and that
related proteins that bind the GATA motif cannot
compensate for its absence. Using in vitro differ-
entiation of embryo-derived stem cells into hema-
topoietic cell types, we have developed an exper-
imental system that permits assessment of the
role of GATA-transcription factor in erythroid de-
velopment and systematic testing of various
aspects of the function and/or regulation of the
protein.
In separate but conceptually related studies, a
gene that encodes an essential component of the
white blood cell (phagocytic) system responsi-
ble for killing ingested microorganisms is being
examined in an effort to understand how this
505
Molecular Genetic Studies of Hematopoietic Cells
clinically important host defense system is regu-
lated and, more generally, how cell-specific gene
expression is achieved in this lineage, also de-
scendent from the pluripotent stem cell. The
gene under study encodes a subunit of a unique
cytochrome that is defective in an X-linked con-
dition, chronic granulomatous disease. By posi-
tional cloning, we previously isolated the gene,
determined its structure, and demonstrated the
presence of the protein product in the cy-
tochrome complex of phagocytic cells.
In addition, since interferon-7 stimulates
phagocytic cell function generally as well as ex-
pression of the cytochrome, it was also possible
to show that this lymphokine is clinically effec-
tive in chronic granulomatous disease. Studies
have identified several point mutations in the cy-
tochrome that interfere with protein function in
vivo and mutations in an associated cytochrome
subunit encoded by an autosome in rarer cases of
chronic granulomatous disease.
More current efforts are directed toward defin-
ing the elements of the gene responsible for ap-
proaching regulation in phagocytic cells.
Through the use of transgenic mice, we identi-
fied a DNA fragment sufficient for targeting re-
porters and oncogenes to a subset of phagocytes
in vivo. Coupled to an oncogene, this fragment
leads to the consistent development of an in-
herited malignancy of phagocytes in mice.
In the promoter we identified a negative regula-
tory site that binds a protein whose concentration
falls dramatically during phagocyte differentia-
tion. This factor corresponds to a putative repres-
sor termed CCAAT displacement protein (CDP) .
Through molecular cloning, we have character-
ized human CDP and shown that it is highly re-
lated to a novel Drosophila homeobox protein, a
product of the cut gene, which controls cell fate
decisions in several tissues during fly develop-
ment. Although this repressor does not explain
the restriction of cytochrome expression to
phagocytic cells, its loss combined with the pres-
ence of unknown positive, white cell-specific
regulators very likely accounts for the temporal
pattern of cytochrome expression during cellular
maturation.
More generally, the remarkable similarity of
human CDP and the fly cut protein suggests func-
tional correlates and predicts that CDP will partic-
ipate in cell fate decisions during mammalian
embryogenesis. Current efforts are directed to
analysis of the role and function of CDP in mam-
malian cells and to identification of the critical
cis- and trans-components responsible for white
cell-specific gene expression.
306
Albinism and Tyrosinase
Paul A. Overbeek, Ph.D. — Assistant Investigator
Dr. Overbeek is also Associate Professor of Cell Biology at Baylor College of Medicine and has adjunct
appointments in the Institute for Molecular Genetics, the Division of Neuroscience, and the Department
of Ophthalmology at Baylor College of Medicine. He received his B.A. degree in chemistry from Kalamazoo
College, his Ph.D. degree in cellular and molecular biology from the University of Michigan, and an M.B.A.
degree from the University of Chicago. His postdoctoral research was done in the laboratory of Heiner
Westphal at NIH.
THE genetic disorder of albinism, or loss of
pigmentation, has been identified in many
species. Albinism typically is caused by the loss
of production of the black pigment termed mela-
nin. Specialized cells known as melanocytes are
responsible for melanin synthesis in the skin,
hair, and iris. In addition to melanocytes, there
are cells at the posterior of the retina, referred to
as retinal pigment epithelial cells, that normally
synthesize melanin. When these two types of
cells lose their melanin-synthesizing ability, al-
binism results. Since melanin helps protect the
skin from ultraviolet radiation and the visual sys-
tem from bright light, albinism is often associated
with secondary health problems, including in-
creased risk of skin cancer and visual system
deterioration.
Melanin is derived from the amino acid tyro-
sine. An enzyme known as tyrosinase converts ty-
rosine to dopaquinone, which is then polymer-
ized to produce melanin. Studies of albino
individuals, including humans and laboratory
mice, have revealed that albinism in often asso-
ciated with a decrease or loss of tyrosinase enzy-
matic activity. This observation has led to the hy-
pothesis that albinism may be caused by a
mutation in the gene that encodes tyrosinase. A
mutation could result in synthesis of an abnormal
tyrosinase protein that no longer has enzymatic
activity. We have recently undertaken a series of
experiments to test this hypothesis in albino labo-
ratory mice.
One important prediction of this hypothesis is
that the nucleic acid sequence of the tyrosinase
gene in pigmented mice will be different from
the sequence in albino mice. A procedure known
as the polymerase chain reaction was used to am-
plify specific regions of the tyrosinase gene from
more than 20 different pigmented and albino
mouse strains. The nucleic acid sequences were
then determined, and a computer program was
used to compare them. Interestingly, every pig-
mented strain was found to encode a cysteine at
amino acid 103 of the tyrosinase gene, while the
albino mice all had a single base pair change that
would cause amino acid 103 to become a serine.
These results suggested that conversion of cys-
teine to serine at amino acid 103 was sufficient to
inactivate the tyrosinase enzyme and produce
albinism.
In order to confirm this prediction, we used
recombinant DNA techniques to construct two
miniature versions of the tyrosinase gene: one
with a cysteine, the other with a serine at amino
acid 103. The two constructs were tested in an
albino strain of mice. Embryos of the albino strain
were harvested at the one-cell stage and injected
under the microscope with purified DNA. After
injection the embryos were reimplanted into the
reproductive tracts of pregnant females and al-
lowed to develop to birth.
The newborn mice were then screened to iden-
tify those mice (termed transgenic) in which the
injected DNA had become stably incorporated
into the genome. Mice that had incorporated the
cysteine 103 version of tyrosinase became pig-
mented, while all of the mice that had integrated
the serine 103 version of tyrosinase were still al-
bino. These experiments have confirmed that al-
binism can be caused by a mutation in the tyro-
sinase gene and that a single base pair
substitution is sufficient to inactivate the gene.
These studies of albinism have provided an im-
portant new strategy for the identification of
transgenic mice. Prior to the design of the minia-
ture tyrosinase gene, transgenic mice were iden-
tified by techniques that required genomic DNA
isolation. The miniature tyrosinase gene allows
transgenic mice to be recognized by simple vi-
sual inspection for pigmentation.
To test the general usefulness of this system,
experiments were done in which two different
recombinant DNAs were co-injected into em-
bryos. One was the miniature tyrosinase gene; the
other was designed to generate a mouse model
for predisposition to kidney cancer. The newborn
mice were first checked for pigmentation, then
genomic DNA was isolated and tested for integra-
tion of the co-injected DNA. Altogether five pig-
mented mice were obtained, and all five had inte-
grated both DNAs. Breeding studies showed that
the two different DNAs were always transmitted
together.
When the kidneys were checked for cancer in-
307
Albinism and Tyrosinase
duction, only the pigmented mice were found to
have kidney lesions. Neoplastic changes were lim-
ited to the kidney and did not occur in the skin.
These co-injection experiments demonstrate that
two unrelated genes can become located in close
proximity to each other in the genome and still
function in an independent fashion. Moreover,
this strategy simplifies long-term maintenance
-of experimental transgenic mice, since the de-
sired mice can be readily identified by their
pigmentation.
An additional question was whether the synthe-
sis of tyrosinase in some cell type other than mela-
nocytes or retinal pigment epithelial cells would
lead to melanin synthesis and pigment formation.
To answer this question, regulatory sequences
that were known to be active specifically in lens
cells were linked to the tyrosinase coding se-
quences using recombinant DNA techniques, and
transgenic mice were generated in an albino
strain of mice. The transgenic mice all devel-
oped black eyes. Synthesis of tyrosinase in the
lens led to pigment production in the lens, con-
firming that the reason lens cells do not nor-
mally make melanin is because they do not
normally make the enzyme tyrosinase. Previous
studies had suggested that pigment synthesis in
inappropriate cell types might be harmful for
those cells. In the transgenic mice, the pig-
mented lens cells show histological evidence of
cellular injury and inhibition of normal growth.
When transgenic DNA integrates into the ge-
nome, it can insert into the middle of an endoge-
nous gene, causing insertional inactivation of the
gene. Since our transgenic tyrosinase mice were
easy to identify and breed, 80 families were
tested for the presence of insertional mutations.
Eight families were found that have defects, rang-
ing from embryonic lethality to male sterility to
anemia and premature kidney failure.
One of the mutations is particularly fascinating
because the homozygous transgenic mice all have'
an inversion of the left-to-right organization of
their internal organs. This condition in humans is
known as situs inversus. Since the factors that
control polarity in the developing mammalian
embryo have not yet been identified, we have be-
gun experiments that will make use of the trans-
genic insert to try to isolate the regulatory factor
for situs inversus. The identification and charac-
terization of such a factor could greatly enhance
our understanding of embryonic development
and guide future efforts to help prevent analo-
gous birth defects in humans.
308
Structural Studies of DNA-binding Proteins
Carl O. Pabo, Ph.D. — Investigator
Dr. Pabo is also Professor of Biophysics and Structural Biology at the Massachusetts Institute of
Technology. He received his undergraduate degree from Yale University, where he majored in molecular
biophysics and biochemistry. He did his graduate work in Mark Ptashne's laboratory at Harvard
University, where he continued his research as a Jane Coffin Childs fellow in the laboratories of Stephen
Harrison and Don Wiley. Prior to accepting a position at MIT, Dr. Pabo was Professor of Molecular
Biology and Genetics and of Biophysics at the Johns Hopkins University School of Medicine.
WE are interested in understanding how pro-
teins recognize specific sites on double-
stranded DNA and how the bound proteins regu-
late gene expression. We would like to know
what structural motifs are used by DNA-binding
proteins, what side chains make sequence-spe-
cific contacts, and whether there are any recur-
ring patterns or rules for recognition of sites on
double-stranded DNA. Much of our current re-
search has focused on characterizing the major
structural motifs found in DNA-binding proteins.
We hope to use this information to design novel
DNA-binding proteins for research, diagnosis,
and therapy.
Prokaryotic repressors provide useful model
systems for the study of protein-DNA interactions,
and we are continuing to study several bacterial
repressors. The repressor from the bacteriophage
X uses a helix-turn-helix motif and an extended
amino-terminal arm to contact sites in the major
groove. The «rc repressor from 5a/mo«e//a bacte-
riophage P22 uses a |8-sheet for site-specific rec-
ognition. The major developments in our labora-
tory during the past two years, however, have
involved studies of two of the key motifs — the
homeodomain and the zinc finger — that are used
by eukaryotic regulatory proteins.
Crystal Structures of Homeodomain-DNA
Complexes
The homeodomain is a conserved structural
motif found in many eukaryotic proteins that reg-
ulate development and cell fate. To understand
how this motif recognizes DNA and how this is
related to the helix-turn-helix motif seen in pro-
karyotic repressors, we have determined the
crystal structures of two homeodomain-DNA
complexes.
We began by studying the homeodomain from
the engrailed protein, which plays a key role in
Drosophila development. (This project is a col-
laboration with Thomas Kornberg at the Univer-
sity of California, San Francisco.) We were able to
grow good cocrystals of the homeodomain-DNA
complex, and Chuck Kissinger solved the struc-
ture of this complex. The homeodomain makes
contacts in both the major and minor grooves.
The helix-turn-helix unit makes critical contacts
in the major groove, but the orientation of this
helix-turn-helix unit with respect to the DNA is
different than the arrangements observed with
the prokaryotic repressors. Residues near the
amino-terminal end of the homeodomain form an
extended "arm" that fits into the minor groove
and makes additional site-specific contacts.
We also have been studying a complex contain-
ing the homeodomain from the a2 protein,
which helps to regulate mating type in yeast.
(This project is a collaboration with Alexander
Johnson at the University of California, San Fran-
cisco.) Cynthia Wolberger recently solved this
structure, and comparison with engrailed re-
vealed that 1 ) the structures of these two homeo-
domains are very similar (despite a 3 -residue in-
sertion in a2 and despite significant amino acid
sequence differences) and 2) the orientation of
the helix-turn-helix unit with respect to the DNA
also is conserved. This conserved docking ar-
rangement is maintained by side chains that are
identical in a2 and engrailed. Because these resi-
dues tend to be conserved among all homeodo-
mains, these structures may provide a general
model for homeodomain-DNA interactions.
Our studies of the homeodomain are supported
by a grant from the National Institutes of Health.
Crystal Structures of Zinc Finger-DNA
Complexes
The zinc finger domain, which contains about
30 amino acids, is another key DNA-binding motif
that is found in a large family of eukaryotic regula-
tory proteins. Studies from other groups have
shown that each finger contains an antiparallel
jS-sheet and an a-helix, but little has been known
about how these fingers recognize DNA. Nikola
Pavletich recently solved the structure of a com-
plex containing three zinc fingers from a murine
transcription factor. Starting with cDNA for the
zif 268 gene (provided by Daniel Nathans, HHMI,
Johns Hopkins University), he cloned and ex-
pressed a three-finger peptide and crystallized
the peptide-DNA complex. The zinc fingers rec-
309
Structural Studies of DNA- binding Proteins
ognize B-DNA and fit into the major groove. Each
finger makes its primary contacts with a 3 -base
pair "subsite," and side chains near the amino-
terminal end of the a-helix make the critical con-
tacts with the bases.
Since the zinc fingers are used in a modular
fashion, they may be the ideal motif to use as we
try designing novel DNA-binding proteins. How-
ever, we need to determine the structure of addi-
tional zinc finger-DNA complexes so that we can
see how this motif is used in other proteins. (Do
all zinc fingers dock against the DNA in the same
way?) Nikola Pavletich has cloned the zinc finger
regions from a number of other proteins and has
recently solved the structure of a complex that
contains the five zinc fingers from GLI, a protein
that is amplified in a subset of human tumors.
Finishing this structure and comparing it with the
zif complex should provide a firmer basis for the
design projects.
Structural Studies of Other Protein-DNA
Complexes
It is important to obtain structural information
about the other major DNA-binding motifs that
occur in eukaryotic regulatory proteins. We are
focusing on the helix-loop-helix proteins and the
POU domain, which have very important roles in
development. We also are working with the
TFIID protein (because of its central role in tran-
scription) and with the p53 protein, which is the
most common site of mutations in human tumors.
Because structural analysis often is limited by
the ability to obtain suitable crystals, we are try-
ing to improve methods for the cocrystallization
of protein-DNA complexes. Our initial approach
involved systematic changes in the length of the
DNA site and required that the entire site be
resynthesized for each experiment. We now have
encouraging preliminary results with a linker co-
crystallization scheme that combines the protein,
the binding site, and a library of DNA linkers that
can be used with any complex. This strategy may
allow a dramatic increase in the number of co-
crystallization conditions that can be tested.
Design of Novel DNA-binding Proteins
We are attempting to use the zinc finger struc-
tures as a basis for designing novel DNA-binding
proteins. Two major approaches are being tested:
1 ) genetic strategies for selecting zinc finger pro-
teins that recognize desired target sequences and
2) strategies for computer-aided protein design.
These programs, which can systematically con-
sider a large number of sequences and conforma-
tions, are being used in our attempts to design
zinc finger proteins that will recognize novel
binding sites. This combination of structural anal-
ysis, computer-aided protein design, and genetic
selection should provide a better understanding
of protein-DNA recognition and allow rapid de-
sign of zinc finger proteins that recognize novel
target sites.
310
Image of an x-ray crystal structure showing how regulatory proteins of a key class known as zinc
finger bind to DNA. The protein contains three "fingers" ( shown in orange, yellow, and purple)
that bind to the double-helical DNA (blue). Each finger has an a-helix that fits into the major
groove of the DNA and makes critical contacts that help the protein recognize its proper bind-
ing site.
From Pavletich, N.P., and Pabo, CO. 1991. Science 252:809-817. Copyright© 1991 by the
AAAS.
311
Hypothesis: XX Males and XY Females Result
from X-Y Interchange at Paternal Meiosis
' ^ TDF
X
1/
I I
g +TDF -TDF
I
XX male XY female
The X-Y interchange hypothesis: XX males and XY females receive reciprocal
products of similar aberrant XY exchanges occurring in the fathers during
the production of sperm ( meiosis). TDF is testis-determining factor, the sex-
determining gene normally located on the Y chromosome.
Research of David Page.
312
The X and Y Chromosomes in Mammalian
Development
David C. Page, M.D. — Assistant Investigator
Dr. Page is also Associate Member of the Whitehead Institute for Biomedical Research, Associate Professor
of Biology at the Massachusetts Institute of Technology, and Assistant Professor at the Harvard
University-MIT Division of Health Sciences and Technology. He received his undergraduate degree in
chemistry from Sivarthmore College and a medical degree from Harvard Medical School and the
Harvard-MIT Health Sciences and Technology Program. After training with Raymond White, at the
University of Massachusetts, and David Botstein, at MIT, Dr. Page became one of the first Fellows
of the Whitehead Institute. He subsequently joined the faculties of Whitehead and MIT.
TO a large degree, human individuals differ in
physical characteristics because of the im-
pact of genetic variation on the course of embry-
onic development. The human genome is orga-
nized into 23 pairs of chromosomes, each
believed to carry, on average, about 5,000 genes.
When considering genetic differences among in-
dividuals, it is important to distinguish between
variation in a single gene, called Mendelian, and
massive multigene variation, as found in chromo-
somal disorders.
As discussed elsewhere in this volume, single-
gene defects are responsible for certain condi-
tions such as color blindness, cystic fibrosis, and
muscular dystrophy. Other conditions, such as
Down syndrome, appear to be the result of
"wholesale" abnormalities affecting an entire
chromosome. As reductionists and molecular bi-
ologists, we proceed on the assumption that the
developmental consequences of chromosomal ab-
normalities will ultimately be understood in
terms of individual genes and their particular
functions.
My colleagues and I are seeking to understand
how massive variability in one chromosome pair
— the sex chromosomes — dramatically affects
the course of development. Embryos normally in-
herit one sex chromosome from each parent. The
mother contributes an X chromosome, and the
father contributes either an X or a Y. Thus normal
embryos have one of two sex chromosome consti-
tutions, XX or XY.
Sex Determination
In both humans and mice the presence or ab-
sence of the Y chromosome determines whether
an embryo develops as a male or as a female. XX
embryos become females, XY embryos males. For
years scientists wondered whether the Y chromo-
some carried few or many sex-determining genes
and how those were distributed along the
chromosome.
We found that the entire sex-determining func-
tion can be traced to one tiny portion of the hu-
man Y chromosome. This sex-determining region
was identified by studying DNA from "XX males"
and "XY females." XX males have small testes
and are sterile. XY females are also sterile and do
not develop secondary sexual characteristics. We
found that almost all XX males had inherited a
small bit of the Y chromosome attached to one of
their X chromosomes. Conversely, some XY fe-
males lacked the same segment of the Y that was
present in XX males. On the basis of the chromo-
somal deletions found in such patients, we con-
structed a map of the Y chromosome. It was then
we came to recognize that the presence or ab-
sence of one small region, about 0.4 percent of
the chromosome, correlated well with gender.
Detailed analysis of XX males suggests that one
or more genes within this relatively small seg-
ment of the Y chromosome determine the out-
come of sexual development. Laboratories
around the world have scoured this relatively
small segment of the Y chromosome searching for
such sex-determining genes. We are now con-
ducting a variety of experiments to characterize
the functions of two genes in the region: ZFY,
which we identified in 1987, and SRY, a. gene
described by British scientists in 1990.
Both ZFY and SRY appear to encode DNA-
binding proteins that are likely to regulate the
transcriptional activity of particular but un-
known target genes. Experiments clearly demon-
strate that SRY is a sex-determining gene. Much
less clear is the role, if any, that ZFY plays in the
process. We hope to understand better the func-
tion of the ZFF gene by simultaneously analyzing
a closely related gene, ZFX, that we identified on
the X chromosome.
In some human XY females, we have identified
mutations near or within the SRY gene. By con-
trast, other XY females, some human and some
mouse, appear to have intact Y chromosomes but
may have mutations elsewhere, perhaps in auto-
somal genes that play important roles in sex deter-
mination. Identification of such autosomal sex-
determining genes is a future goal. Our studies of
sex determination arc supported by a grant from
the National Institutes of Health.
313
The X and Y Chromosomes in Mammalian Development
Turner Syndrome
As mentioned earlier, embryos normally have
two sex chromosomes. However, about 1-2 per-
cent of all human embryos have only one. The
vast majority of such XO embryos are lost to spon-
taneous miscarriage, but a few survive. The sur-
viving XO embryos develop as females with a par-
ticular set of physical features known as Turner
syndrome, which includes short stature, webbing
of the neck, puffiness of the hands and feet, and
failure of secondary sexual development. It had
been postulated that Turner syndrome might be
the result of having a single copy of one or more
genes common to the X and Y chromosomes.
Nothing was known as to the number or nature of
these hypothetical Turner genes.
We began to focus our attention on this dis-
order when it was noticed that certain XY females
exhibit the same anatomic abnormalities as XO
females. A pivotal finding was that all such XY
Turner females lacked a portion of the Y chromo-
some. We postulated that the Y chromosomal de-
letions in these individuals might encompass not
only a sex-determining gene or genes but also a
nearby Turner gene or genes.
Pursuing this hunch, we discovered two candi-
date Turner genes, one on the Y chromosome and
one on the X chromosome. These genes, named
RPS4Y2in<S. RPS4X, appear to encode slightly dif-
ferent forms of a protein constituent of the ribo-
some, a structure required for protein synthesis
and vital to all cells. In embryos lacking a second
RPS4 gene (i.e., having a single RPS4), the rate at
which ribosomes are constructed maybe slowed,
in turn reducing the embryo's capacity to synthe-
size other proteins. We are currently testing the
highly speculative hypothesis that such a reduc-
tion in protein synthetic capacity is the cause
of some of the physical features of Turner
syndrome.
An interesting analogy can be found in the fruit
fly Drosophila melanogaster. There, deficien-
cies in ribosomal protein genes are associated
with a particular "syndrome" called the "Mi-
nute" (pronounced mi-NUTE) phenotype, which
includes reduced body size, diminished viability
and fertility, and specific anatomic abnormali-
ties. The Drosophila Minutes may serve as a use-
ful model system in which to study effects of
RPS4 gene dosage. We are also exploring poten-
tial mouse models of Turner syndrome.
The very existence of related but nonidentical
ribosomal protein genes on the X and Y chromo-
somes raises the possibility that the ribosomes of
human males may differ slightly from those of
females. Experiments still in their early stages
suggest that this is the case. Thus the differences
between the sexes may extend all the way down
to the most fundamental and vital of intracellular
machines!
The Human Y Chromosome
As mentioned above, we constructed a map of
the human Y chromosome by characterizing natu-
rally occurring deletions, such as those found in
XY females and XX males. We are continuing to
refine this map, which is useful not only in study-
ing sex determination and Turner syndrome, but
also in examining the role of Y chromosomal
genes in other processes, including the develop-
ment of certain cancers and the making of sperm.
We recently set out to "clone" the human Y
chromosome as a series of overlapping segments,
each segment constituting about 1 percent of the
chromosome. Such an ordered array of cloned
segments should facilitate the process of locat-
ing and characterizing all genes on the Y chromo-
some. Our efforts to map and clone the Y chro-
mosome are supported by a grant from the Na-
tional Institutes of Health.
314
Mammalian Development and Disease
Richard D. Palmiter, Ph.D. — Investigator
Dr. Palmiter is also Professor of Biochemistry at the University of Washington. He received his Ph.D. degree
from Stanford University and did postdoctoral work at Stanford, Searle Research Laboratories in England,
and Harvard University. Prior to his current work with transgenic animals. Dr. Palmiter studied the
mechanism of steroid hormone action in the chick oviduct and the regulation of metallothionein gene
expression in mice. He is a member of the National Academy of Sciences and the American Academy
of Arts and Sciences.
ABOUT 1 0 years ago we began a fruitful col-
laboration with Ralph Brinster's laboratory
at the University of Pennsylvania. Together we
helped develop methods for introducing func-
tional genes into all cells of the mouse. The genes
under study are manipulated and grown in bacte-
rial plasmids, using standard recombinant DNA
techniques. Then the regions of interest are ex-
cised from the plasmid, and a few hundred copies
are injected into the pronucleus of a fertilized
mouse egg (or that of any other mammal) .
Remarkably, the DNA integrates about 30 per-
cent of the time into one of the chromosomes
prior to replication, and the genes are inherited
by all daughter cells, as any other gene would be.
Furthermore, many of the genes are functional,
imparting new genetic characteristics to the ani-
mal. Mice and other animals carrying foreign DNA
are referred to as transgenic. Because the new
genes are also in the germ cells, they are usually
transmitted to subsequent generations.
One of our goals has been to discover what
parts of a gene determine when, where, and how
efficiently it will be utilized. We often start by
testing a large piece of DNA that includes the
gene of interest. In transgenic animals, the gene
will usually be expressed in the appropriate time
and place, even though it has integrated at an ab-
normal chromosomal location and may be de-
rived from a different mammalian species. Then
we delete various regions of the genes, and with
each variant we make transgenic mice to deter-
mine what regions are essential for appropriate
expression.
For example, we have delineated a small re-
gion (125 base pairs) of the rat elastase I gene
that is essential for the expression of the gene in
acinar cells of the pancreas. Furthermore, this se-
quence (often called an enhancer) can be used to
direct the expression of another gene (e.g., the
growth hormone gene) to the acinar cells, and
the sequence will function when positioned al-
most anywhere in the vicinity of the growth hor-
mone gene. In similar experiments we have been
identifying sequences responsible for directing
appropriate expression of globin genes in red
blood cells, albumin in hepatocytes, and prota-
mine I in male germ cells.
Because regulatory elements from one gene
can often be used to control another; the expres-
sion of many interesting genes can be directed to
a specific cell type and the consequences on the
development and function of those cells can be
assessed. For example, using the elastase en-
hancer element, we have been able to make
strains of mice that reproducibly develop pancre-
atic cancer as a consequence of expressing the
transforming gene from simian virus 40, the
mouse myc gene, or the human Yi-ras oncogene.
Similarly, we have developed models of liver
cancer by directing the expression of these genes
to hepatocytes with the albumin enhancer.
Significantly, each of these genes results in a
characteristic morphological transformation of
the organ, which probably reflects the particular
cellular events that these genes mediate. By
means of simple genetic crosses, mice carrying
any pair of these transforming genes can be cre-
ated. They develop tumors that appear more rap-
idly and grow more aggressively than those in
mice carrying a single gene, suggesting that these
genes act cooperatively.
Some genes that are not generally considered
oncogenes may also predispose cells to malignant
transformation and cancerous growth. In one ex-
ample, we expressed the surface antigens of hepa-
titis B virus (HBV) in the liver, using the albumin
enhancer. In transgenic mice, expression of this
gene resulted in the synthesis of the viral surface
antigen and envelope protein, which aggregated
within the secretory apparatus of the liver cells,
causing cellular injury and death. This is accom-
panied by liver cell regeneration. However,
when the mice were more than a year old, they
developed liver cancer. Because HBV infects mil-
lions of people worldwide, and the incidence of
liver cancer among them is very high, this result
may indicate that chronic expression of HBV sur-
face antigens may be a contributing factor. Simi-
larly, expression of plasminogen activator in the
liver using the albumin enhancer results in defec-
tive blood clotting and liver injury. In this case,
315
Mammalian Development and Disease
liver cells that have deleted the transgene repopu-
late the liver, and when the mice are a few
months old they seem normal. However, they
also succumb to liver cancer when they are one to
two years old. In both cases we postulate that
liver regeneration in the toxic environment of
liver injury results in genetic damage that predis-
poses the liver cells to malignant transformation.
It is also possible to develop transgenic mice
that mimic some human genetic diseases. For ex-
ample, we made a model of human sickle cell
disease. By introducing into mice both human a-
and /8-globin genes under control of the locus
control region (a newly discovered genetic ele-
ment essential for high-level expression of globin
genes) , we have made mice that produce as much
human hemoglobin as mouse hemoglobin. When
the mutant /3-globin gene from people with sickle
disease is substituted for the normal gene in these
experiments, the red blood cells of the mice
sickle under appropriate conditions. These mice
may be a valuable resource for testing experimen-
tal therapies.
A long-range goal is to use transgenic mice to
study aspects of neural development. We have
started by cloning the genes involved in the syn-
thesis of the catecholamine neurotransmitters:
dopamine, norepinephrine, and epinephrine.
The control elements from these genes are being
tested in conjunction with genes whose products
can be easily visualized, to assess when and
where they are expressed during development.
For example, we have shown that the regulatory
elements from the gene responsible for mak-
ing norepinephrine direct the expression of ^-
galactosidase to certain neurons of the central
nervous system, the peripheral nervous system,
the enteric nervous system, and the adrenal me-
dulla (see figure). Because this marker gene is
expressed very early during neural development,
it allows us to visualize the cells while they are
still migrating to their final destinations and be-
fore they acquire properties of mature neurons.
By mating these mice to mice carrying a genetic
defect that affects the innervation of the bowel
and results in a condition similar to Hirsch-
sprung's disease in humans, we have shown that
neuronal precursors fail to migrate into the distal
portion of the gut. The lack of innervation of the
colon results in chronic impaction of fecal mate-
rial and ultimately death.
We are also using the control elements from
these genes to direct the expression of other
genes (e.g., encoding neurotransmitters, growth
factors, hormones, proteases, or oncogenes) to
these neurons with the aim of affecting the deci-
sions they make during the process of forming
functional connections with target cells.
Blue staining reveals the location of cells that ex-
press a ^ galactosidase transgene in neural precur-
sors in a 10. 5- day mouse embryo.
Research and photograph by Raj Kapur in the lab-
oratory of Richard Palmiter.
316
Regulation of Gene Expression
in Steroid Hormone Biosynthesis
Keith L. Parker, M.D., Ph.D. — Assistant Investigator
Dr. Parker is also Associate Professor of Medicine and Biochemistry at Duke University Medical Center.
After attending Williams College, he earned his M.D. and Ph.D. degrees in genetics at Washington
University, studying with Donald Shreffler. He served as intern and resident in internal medicine at
Parkland Memorial Hospital, Dallas. He then moved to the Department of Genetics at Harvard Medical
School, where he was a postdoctoral fellow with Jonathan Seidman. Dr. Parker 's next move was to the
faculty of Duke University Medical Center.
THE adrenal gland plays an essential role in the
body's ability to respond to stress. Two dif-
ferent parts of the adrenal gland, an outer cortex
and an inner medulla, produce distinct compo-
nents of this response. The medulla produces epi-
nephrine and norepinephrine, which are re-
leased very rapidly, preparing the organism for
immediate physical activity. In contrast, the cor-
tex produces steroid hormones, which are re-
leased more slowly and exert prolonged effects.
These adrenal steroids constitute two major
classes: glucocorticoids, which are made in the
inner zone of the cortex and control carbohydrate
metabolism, and mineralocorticoids, which are
made in the outer zone and regulate salt and
water balance. Both classes of steroid hormones
are formed from cholesterol by the sequential ac-
tion of a related group of steroidogenic enzymes.
One of these, the cholesterol side-chain cleavage
enzyme, is expressed in all steroidogenic tissues.
A second, 2 1 -hydroxylase, is expressed through-
out the adrenal cortex. Finally, there are distinct
forms of 1 1/3-hydroxylase: one form produces
mineralocorticoids in the outer zone, and the
other produces glucocorticoids in the inner
zone. The physiological regulators of these two
classes of adrenal steroids differ markedly, de-
spite the shared role of the same enzymes in their
biosynthesis.
We are interested in defining the events that
control the expression of the adrenal steroido-
genic enzymes. These studies have addressed two
major questions. First, what mechanisms direct
the expression of these related genes within adre-
nocortical cells? Second, what determines the
functional differentiation of the adrenal cortex
into mineralocorticoid- and glucocorticoid-
producing zones?
Our studies of gene regulation have focused on
the 5'-flanking regions of these genes. This part,
the promoter region, contains most sequences
important in transcriptional regulation of other
genes. These promoter analyses identified a pro-
tein, steroidogenic factor 1, that plays a major
role in regulating the expression of all three ste-
roidogenic enzymes. We only found this protein
in cells that made steroid hormones, suggesting
that it contributes to the cell-selective expression
of these genes. The global role of this protein in
the expression of three distinct genes suggests
that it coordinates the adrenocortical expression
of a network of enzymes.
We next used cow adrenal glands, providing
much greater amounts of protein, and purified
the bovine form of steroidogenic factor 1. The
presently available amounts of purified protein
should be sufficient to determine its amino acid
sequence and to raise specific antibodies. The
combination of specific antibodies and amino
acid sequence data should allow us to clone the
gene encoding this key regulatory protein. By
comparing the primary structures of steroido-
genic regulatory protein and the previously de-
scribed transcriptional regulatory proteins (such
as the steroid hormone receptor proteins), we
may gain new insights into the mechanisms that
regulate the adrenal steroidogenic enzymes. The
specific probes and antibodies to steroidogenic
factor 1 will further permit us to study the mecha-
nisms that regulate its expression. These studies
will provide new insights into the basis for tissue-
specific differences in the production of steroid
hormones and may identify important avenues for
therapeutic intervention in clinical settings of
abnormal adrenal steroidogenesis.
In a related effort, we are trying to define the
potential role of the steroidogenic enzymes in hy-
pertension. Extremely prevalent, hypertension
affects approximately 20 percent of the adult
population and is a major risk factor for heart at-
tacks and strokes, the leading causes of death in
developed nations. Although the underlying de-
fect is unknown in most cases, family studies indi-
cate a significant genetic component. Certain an-
imal models of hypertension have directly
implicated disordered adrenal steroidogenesis as
an important contributor to hypertension. We are
therefore investigating in more detail the role of
adrenal steroids in hypertension.
Initial studies implicated a single 1 1(8-
hydroxylase protein in the biosynthesis of both
mineralocorticoids and glucocorticoids, but we
317
Regulation of Gene Expression in Steroid Hormone Biosynthesis
subsequently defined two mouse 1 1 jS-hydroxy-
lase genes located on chromosome 1 5 . Although
the overall structures of the two genes are quite
similar, selected regions diverge significantly,
suggesting that the proteins encoded by these
two genes might have different activities. More-
over, the 5'-flanking regions of the two genes di-
verge significantly, suggesting that their regula-
tion may differ.
To assess the enzymatic activities of the pro-
teins produced by these two genes, we analyzed
their activities following gene transfer into COS-
7 cells, which normally do not make any steroid-
metabolizing enzymes. One of the two proteins,
designated 1 l/S-hydroxylase, was able to produce
glucocorticoids but could not make mineralocor-
ticoids. In contrast, the other protein, designated
aldosterone synthase, performed all of the reac-
tions needed to make aldosterone. Thus there are
significant differences in the enzymatic activities
of the two mouse 1 1 |S-hydroxylase homologues.
To see where these two proteins are made in
the adrenal cortex, we used the technique of in
situ hybridization to examine sites of expression
in the sections of the mouse adrenal gland. The
aldosterone synthase protein is only present in
the outer zone of the cortex, where mineralo-
corticoids are made. In contrast, the 11(8-
hydroxylase protein is only present in the sites of
glucocorticoid production. These results docu-
ment an exquisite coupling between enzymatic
activities and sites of expression. The ability to
maintain this segregation of the two proteins is
undoubtedly an important part of the adrenal
gland's ability to regulate separately the produc-
tion of mineralocorticoids and glucocorticoids.
Based on these results, we plan to use these
promoter regions in transgenic mice to target
gene expression to specific cortical zones. Ini-
tially we are using the two promoters to direct the
expression of renin, a gene previously shown to
cause genetic hypertension when expressed in
multiple tissues of transgenic animals. If success-
ful, these experiments will validate the zone-spe-
cific expression of the two promoter regions and
will establish that adrenal expression of renin is
relevant to this hypertensive state.
We will also express the aldosterone synthase
protein in the inner, glucocorticoid-producing
zone. To this end, we have prepared a hybrid
gene with the inner-zone-specific promoter
driving expression of the aldosterone synthase
gene. Based on the gene transfer experiments de-
scribed above, we anticipate that this hybrid gene
will synthesize large amounts of mineralocorti-
coids, thereby creating a genetic form of hyper-
tension. Moreover, treatment with glucocorti-
coids should alleviate the hypertension by
suppressing the expression of the hybrid gene in
the inner zone. Recent studies strongly suggest
that just such a mechanism is responsible for a
subset of human patients with a glucocorticoid-
remediable hypertension.
318
Molecular Neuroimmunology
Donald G. Payan, M.D. — Assistant Investigator
Dr. Payan is also Associate Professor of Medicine and of Microbiology and Immunology at the University
of California, San Francisco. He received his B.S. degree in physics and mathematics from Stanford
University. He went on to do graduate work in physics at the Massachusetts Institute of Technology and
then returned to Stanford Medical School, where he received his M.D. degree. His medical residency at
Massachusetts General Hospital, Boston, was followed by fellowships in infectious diseases at MGH and
in allergy-immunology at Brigham and Women's Hospital.
MY laboratory continues to study the interac-
tions between the nervous and immune sys-
tems at the molecular level. Ongoing work is fo-
cusing on the signal transduction pathways that
are activated in cells into which we have trans-
fected a number of neuropeptide receptors, in
particular the tachykinin receptor for substance
P. An additional effort is directed at understand-
ing the biochemical properties of agrin, the nico-
tinic acetylcholine receptor-clustering mole-
cule, which was cloned and sequenced during
sabbatical studies with Richard Scheller (HHMI,
Stanford University) .
The two main projects currently under way are
the analysis of the signal transduction capabilities
of the substance P receptor (SPR) and accessory
molecules that mediate that signaling, and an
analysis of agrin's protease inhibitor domains and
their potential function in modulating neuronal
plasticity in the developing brain. With Julie
Sudduth-Klinger, Christine Christian, and Mark
Gilbert, we have transfected SPR cDNA into a
number of different immune cells, in particular
Jurkat and Reh, in order to study the effects of this
receptor's expression on the cells' immunologic
properties. We can get approximately 100,000
functional receptors per cell and can demon-
strate a very brisk mobilization of intracellular
calcium and inositol phosphate (IP3) metabo-
lites following stimulation with substance P.
Of great interest are many differences we have
observed in comparing the Jurkat SPR with the
Jurkat muscarinic receptor. Activation of the mus-
carinic receptor results in desensitization of sub-
sequent T cell receptor activation with an anti-T
cell receptor antibody; conversely, activation of
the T cell receptor does not result in desensitiza-
tion of the transfected muscarinic receptor. In
contrast, we find that activation of Jurkat SPRs by
substance P does not desensitize subsequent T
cell activation, but activation of the T cell recep-
tor does result in desensitization of SPR. We are
currently investigating whether some heterolo-
gous desensitization mechanism is taking place
and whether a particular kinase is involved.
In collaboration with Phyllis Gardner at Stan-
ford University, we have also been able to demon-
strate, using patch-clamp techniques, that activa-
tion of the SPR in these transfected lymphocytes
results in the opening of a chloride channel. In
addition, these changes can be nullified by in-
jecting the cells with a calcium/calmodulin ki-
nase inhibitor peptide.
Our early observations that stimulation of these
cell lines by substance P resulted in both IP3 and
cyclic AMP activation have now been extended.
Looking at the cells' respective nuclear regula-
tory elements, we have demonstrated that sub-
stance P stimulation results in increased expres-
sion of the AP-1- and the cAMP-responsive
elements in these cells. Furthermore, using West-
ern and Northern blotting techniques, we have
also been able to demonstrate that the proto-
oncogenes /os and^wn are also up-regulated fol-
lowing substance P stimulation. Consequently,
signal transduction analysis suggests that a dual
activation signal pathway may be activated when
the SPR is expressed in lymphoid cells.
We have studied the functional consequences
of stimulating these cells with the peptide sub-
stance P by examining the expression of a num-
ber of cell surface antigens and their modulation
on Jurkat SPRs. We find that in Jurkat SPR-positive
cells, substance P stimulation results in a down-
regulation of the LFA- 1 (integrin) surface antigen
and an up-regulation of the CD2 surface antigen.
In addition, when combined with the mitogen
phytohemagglutinin, substance P stimulation re-
sults in an increased expression of the interleu-
kin-2 receptor. We continue to examine these re-
sults in order to delineate further how tachykinin
peptides may modulate immune responses.
Joseph Fisher and Sandra Biroc have now begun
an extensive in situ hybridization study of the
expression of agrin and alternate agrin transcripts
in the developing rat. Preliminary results suggest
that agrin is extensively expressed in unique sites
within the developing brain. Moreover, Dr.
Fisher has now expressed full-length recombi-
nant agrin in a number of cell types and is examin-
ing the molecule's biochemical properties. In
particular, the amino terminus of the molecule
319
Molecular Neuroimmunology
has some unique sequences characterized by pro-
tease inhibitor domains. He has begun a study of
the molecule's protease inhibitor activity to see if
it plays a role during the remodeling process in
neuronal development. We expect the studies re-
garding receptor signal transduction mechanisms
and early developmental remodeling processes in
the nervous system to help us further understand
how the immune and nervous systems develop
and interact.
Cell surface receptors on lymphocytes some-
times share intracellular signaling mecha-
nisms. In this case, the receptor for a T cell
growth factor (interleukin-2 or IL-2) and the
protein molecule CD4, which assists these cells
in the recognition of foreign antigens, both in-
teract with a lymphocyte- specific protein-
tyrosine kinase, p56^'^^ — a potent signaling
molecule that induces proliferation. A highly
acidic region of the IL-2 receptor's fS-chain me-
diates this interaction, while a very different
domain in the CD4 protein, containing two
critical cysteine residues (c ), performs a simi-
lar function for this receptor.
Diagram from Roger Perlmutter, based on
work in his laboratory.
IaCIDIC
:- - =7/-- ' proline-rich
\\ tv Signal
V:~y Transduction
320
I
Molecular Basis of Lymphocyte Signaling
Roger M. Perlmutter, M.D., Ph.D. — Investigator
Dr. Perlmutter is also Professor of Immunology, Medicine, and Biochemistry at the University of
Washington School of Medicine. He received his B.A. degree from Reed College and his M.D. and Ph.D.
degrees from Washington University, St. louis, where he studied with Joseph Davie. After clinical training
in internal medicine at Massachusetts General Hospital, Boston, and the University of California, San
Francisco, he became Senior Research Fellow and later Instructor in Biology at the California Institute
of Technology, where he worked with lee Hood.
IMMUNE recognition of potentially injurious
foreign macromolecules requires the elabora-
tion of an enormous repertoire of clonally re-
stricted receptors (antigen receptors) on the sur-
faces of lymphoid cells. These receptors are
sufficiently heterogeneous to permit recognition
of virtually the entire universe of infectious or-
ganisms. Interaction of these receptors with cog-
nate antigen provokes a stereotyped response
leading to cell proliferation and the production
of soluble mediators of inflammation. During the
past several years, analysis of the mechanisms re-
sponsible for antigen receptor diversification has
stimulated interest in a related question: How is
the signal from a lymphocyte antigen receptor
transmitted to the cell interior? Our laboratory
has adopted a molecular genetic approach to the
dissection of signaling pathways in immune cells.
Initially we identified a lymphocyte-specific
enzyme, similar in structure to proteins known to
transmit growth-promoting signals in nonlym-
phoid cells, that modifies the behavior of target
proteins by catalyzing the addition of phosphate
groups onto certain tyrosine amino acids in sub-
strate proteins. The gene that encodes this lym-
phocyte-specific kinase was identified by virtue
of its overexpression in a murine lymphoid malig-
nancy. Moreover, we were able to demonstrate
that a single point mutation in this Ick gene en-
abled it to confer malignant properties on cells
maintained in culture. Thus the Ick gene encodes
a protein that is capable of altering the growth
properties of at least some cell types. Since Ick is
normally expressed only in lymphocytes, there is
reason to believe that its product assists in regu-
lating lymphocyte proliferation.
Biochemical studies support this view. In par-
ticular, we and others have recently demon-
strated that the Ick-cncoded kinase is physically
associated with proteins that form part of the an-
tigen receptor on T lymphocytes. Our studies
also enabled us to identify two additional pro-
tein-tyrosine kinases that are specifically ex-
pressed in immune cells. In each case there is
reason to believe that the kinase is physically
coupled to a cell surface receptor involved in im-
mune recognition.
To investigate the functional importance of
these protein-tyrosine kinase signaling elements,
we have developed methods for manipulating the
expression of each gene in its appropriate cellu-
lar context. Using this approach, we have begun
to dissect the hierarchy of signal transduction
events precipitated by normal immune recogni-
tion. For example, we found that overexpression
of an activated Ick gene leads to extraordinarily
rapid development of thymic tumors in mice.
Hence altered expression of the Ick gene can di-
rectly affect lymphocyte proliferation. In a re-
lated series of experiments we learned that inhibi-
tion of the function of the Ick gene completely
disrupts normal mechanisms that permit develop-
ment of T lymphocytes. In fact the level of Ick-
encoded protein must be maintained within a
very narrow range. Even very modest (twofold)
changes in the abundance of this kinase are in-
compatible with normal T lymphocyte develop-
ment. These observations probably reflect the
fact that the /cfe-encoded kinase participates in a
large number of receptor signaling pathways,
including some pathways that are activated
by lymphocyte-specific growth factors such as
interleukin-2.
Disturbances in lymphocyte signaling almost
certainly contribute to the pathogenesis of lym-
phoproliferative and immunodeficiency diseases
in humans. A detailed understanding of immune
cell signaling mechanisms should permit the de-
sign of more-effective therapeutic strategies for
the treatment of immune system dysfunction.
321
wildtype
"'-iVilv.
corkscrew
Darkfield preparation of a wild-type and a corkscrew embryo. The twist in
the mutant embryo is due to a perturbed establishment of cell fates at the
termini of the blastoderm.
From Perkins, L.A., Larsen, I., and Perrimon, N. 1992. Cell 70:225-
236. Copyright © 1992 by Cell Press.
Genetic Dissection of a Signal Transduction
Pathway in Drosophila melanogaster
Norbert Perrimon, Ph.D. — Assistant Investigator
Dr. Perrimon is also Assistant Professor of Genetics at Harvard Medical School. Of French nationality, he
tvas educated at the University of Paris VI, where he majored in biochemistry. His thesis, with Madeleine
Gans as advisor, was on Drosophila genetics. He moved to Case Western Reserve University as a
postdoctoral research fellow with Anthony Mahowald. He became a Lucille P. Markey Scholar in
Biomedical Sciences while in Cleveland. He then assumed his present position at Harvard Medical School.
INTERCELLULAR communication is a major
player in the establishment of developmental
patterns. For example, in the early Drosophila
embryo, determination of cell fates at the termini
requires the normal activities of genes from two
different cell types, the maternal follicle cells
and the oocyte. The current model is that the
transmembrane tyrosine kinase receptor encoded
by the gene torso is locally activated at the egg
termini by cues emanating from the follicle cells.
This localized activation of torso is believed to
trigger a phosphorylation cascade in the egg
which ultimately controls the expression of the
transcription factors tailless and huckebein.
Knocking out this signaling pathway has detri-
mental effects on embryonic development. Since
cell fates at both termini are perturbed, the re-
sulting embryos lack most head and all tail
structures.
Our laboratory has focused on identifying the
genetic components involved in transduction of
the signal from torso, the membrane-bound tyro-
sine kinase, to the nucleus. Thus far we have
characterized two genes, 1(1 )pole hole and
l( 1 )corkscrew, that are involved in this process.
Genetic epistasis experiments have demonstrated
that both these genes act downstream of the torso
protein activity. Furthermore, l( l)corkscrew
acts by up-regulating the activity of 1(1 )pole
hole.
We previously showed that the l( 1 Jpole hole
gene product is the homologue of the mamma-
lian Raf- 1 proto-oncogene and encodes a serine/
threonine kinase. Recently we discovered that
l( 1 ) corkscrew encodes a protein-tyrosine phos-
phatase similar to the mammalian PTPIC protein.
Identification of the l( 1 )pole hole and 1(1 Jcork-
screw gene products has strengthened the
current model that torso signaling involves a
phosphorylation cascade, since both genes en-
code proteins that have the ability to affect
the level of phosphorylation of intracellular
components.
To identify additional molecules involved in
torso signaling, we have taken a genetic ap-
proach. Screens for second-site suppressors and
enhancers of l( 1 Jpole hole and l( 1 Jcorkscrew
mutations have successfully identified a number
of loci involved in the torso signaling pathway.
Future work will involve a detailed characteriza-
tion of these suppressors.
In addition to the genetic approach described
above, we are utilizing a biochemical screen to
identify and characterize other components of
this signal transduction pathway. Involved is the
cloning of genes encoding proteins that bind di-
rectly to activated receptor tyrosine kinase. In
this way we are isolating proteins that respond
directly to the ^or50-encoded receptor.
Establishment of cell fate at the termini of the
embryo provides a unique genetic system to dis-
sect the cascade by which activation of a receptor
tyrosine kinase controls the expression of tran-
scription factors. A combination of classical ge-
netics and molecular and biochemical tech-
niques will allow characterization of the
components involved in the various steps of re-
ceptor tyrosine kinase signaling.
In addition, the homology between torso and
the mammalian gene for platelet-derived growth
factor (PDGF), between l( 1 Jpole hole and the
mammalian Raf l proto-oncogene, and between
l( 1 Jcorkscrew and the human PTPIC enzyme
suggests that biochemically this signal transduc-
tion pathway may have been conserved in evolu-
tion between organisms as diverse as Drosophila
and humans. Thus characterization of this path-
way in Drosophila may help elucidate the func-
tions of the homologous mammalian proteins.
525
Gene Regulation and Immunodeficiency
B. Matija Peterlin, M.D. — Associate Investigator
Dr. Peterlin is also Associate Professor of Medicine and of Microbiology and Immunology at the University
of California, San Francisco. He obtained his undergraduate degree in chemistry and physics at Duke
University and his M.D. degree from Harvard Medical School. His postdoctoral work was performed with
facob Maizel and Philip leder at NIH and with Hugh McDevitt at Stanford University. As a rheumatology
fellow at Stanford, he chanced upon a family with the bare lymphocyte syndrome, which stimulated his
interest in this genetic disorder. He is a member of the American Society for Clinical Investigation.
SOME years ago we described a variant of the
genetic disorder called the bare lymphocyte
syndrome (BLS), in which the patient's lympho-
cytes fail to express either class I, class II, or both
major histocompatibility determinants on their
cell surfaces. These transplantation antigens are
essential for the development of the immune sys-
tem, for tumor surveillance, for eradication of
viral infections, and for normal immune re-
sponses. Thus it is not surprising that BLS patients
are severely immunocompromised, fail to make
antibodies, or have autoimmune diseases. In ad-
dition, this autosomal recessive syndrome is one
of only two known inherited deficiencies of a reg-
ulatory gene in humans.
By fusing defective cells from different patients
and those obtained by mutagenesis in tissue cul-
ture, four genetic complementation groups of
BLS were found. The isolation of their defective
genes should make possible prenatal diagnoses
through use of specific genetic probes and possi-
bly lead to the cure of BLS by the targeting of
normal genes into the bone marrow of alfected
patients.
To study the defective gene in BLS, we first ex-
amined regions that regulate B cell-specific and
interferon-7 (IFN-7) -inducible expression of
class II genes. Next, we looked at proteins that
bind to these DNA sequences and compared class
Il-specific factors in various cell types. Distinct
patterns of DNA-binding proteins were found in B
cells, IFN-7-inducible cells, and T cells.
We cloned several cDNAs that code for proteins
that bind to B cell-specific and IFN-7-inducible
sequences in class II promoters. One cDNA codes
for Jun, which forms active Jun/Fos heterodimers
in cells that do not express class II determinants.
Of the remaining two cDNAs, one codes for a B
cell-specific helix-loop-helix protein and the
other for an ETS-like protein. We are currently
studying their genetic organization, expression,
structure, and function. By expressing one of
these full-length cDNAs in human cells, we hope
to rescue class II gene expression in one type of
BLS. In parallel with direct biochemical studies,
we are also using genetic approaches to rescue
regulatory defects in BLS.
In setting up these genetic approaches, we first
tested a well-known viral trans-regulatory system
— namely trans-activation of the human immuno-
deficiency virus (HFV) by the virally encoded Tat
protein. The precise mechanism of Tat action had
not been defined. We discovered that Tat acts
slightly downstream from the promoter to modify
HIV transcription so that efficient copying of the
viral genome ensues. Factors assembled near the
site of initiation of HIV transcription tether the
transcription complex to the promoter. The addi-
tion of Tat, which binds to an RNA stem-loop in
the process of nascent transcription, releases this
transcription complex. Efficient elongation of
transcription and clearance of the promoter fol-
low. New transcription complexes can then as-
semble, interact with Tat, and move quickly
through the viral genome. Interactions between
Tat, the RNA stem-loop, and cellular proteins
have been defined. For example, using a heterolo-
gous RNA-tethering mechanism (that of the coat
protein of bacteriophage Rl 7 that binds to its op-
erator), we mapped activation and RNA-binding
domains of Tat. By studying rodent cells and so-
matic cell hybrids between rodent and human
cells, we defined a cellular RNA-binding com-
plex that facilitates interactions between Tat and
TAR and is encoded on human chromosome 12.
We hope that interfering with trans-activation by
Tat will lead to new therapies for AIDS (acquired
immune deficiency syndrome) and AIDS-related
disorders.
Since upstream promoter sequences are also
essential for HIV replication, we clarified inter-
actions between host cell factors and viral se-
quences (long terminal repeat, LTR). Increased
rates of initiation of HIV transcription were ob-
served in activated T cells and macrophages.
These result from actions of nuclear proteins that
are also required for T cell and macrophage ef-
fector functions and for T cell proliferation.
Differences between LTRs of HIV types 1 and 2
were observed that might explain the longer la-
tency and attenuated clinical course of HIV-2 in-
fection. Furthermore, effects of trans-activators
encoded by several DNA viruses on HIV transcrip-
325
Gene Regulation and Immunodeficiency
tion were examined. Since effects of these ago-
nists and Tat were synergistic, infection by DNA
viruses might be an important cofactor in progres-
sion from latent disease to clinical AIDS.
Thus the quest to rescue the expression of class
II major histocompatibility genes in a rare human
congenital disease has led to genetic approaches
to the study of transcription by RNA polymerase
II and to the elucidation of a potent new tran-
scriptional regulatory mechanism.
326
Mechanism of Action of Polypeptide
Growth Factors
Linda J. Pike, Ph.D. — Associate Investigator
Dr. Pike is also Associate Professor in the Department of Biochemistry and Molecular Biophysics at
Washington University School of Medicine, St. Louis. She received her B.S. degree in chemistry from the
University of Delaware and her Ph.D. degree in biochemistry from Duke University, where she studied with
Robert Lefkowitz. Her postdoctoral training was done in the laboratory of Edwin Krebs at the University
of Washington, Seattle.
A number of low-molecular-weight polypep-
tides have been shown to regulate cell
growth. These growth factors bind to specific re-
ceptors on the surface of cells. Through a com-
plex series of reactions, the binding of the growth
factor to its receptor stimulates the cell to grow
and divide. Until recently, little was known of the
mechanism by which growth factors induce cell
division. It is now recognized, however, that the
receptors not only bind the appropriate growth
factor but also have an enzymatic activity.
The receptors catalyze the transfer of a phos-
phate group from adenosine triphosphate to tyro-
sine residues in selected protein substrates. The
enzyme possessing this activity is called a tyro-
sine protein kinase. Typically the phosphoryla-
tion of a protein by a kinase leads to changes in
the activity of the protein. Although much is
known about the growth factor receptor kinases,
the substrates for these enzymes have not been
identified.
My laboratory is involved in studies of the
mechanism by which the binding of epidermal
growth factor (EGF) to the outside of the cell
elicits a biological response inside. This is re-
ferred to as signal transduction. Because they are
extremely responsive to EGF, the A431 line of
human epidermal carcinoma cells is used as the
model system in most of our studies.
Phosphatidylinositol Metabolism
One of the earliest responses of A431 cells to
EGF is an increase in the metabolism of a particu-
lar phospholipid, phosphatidylinositol. This im-
portant lipid serves as a precursor for the genera-
tion of two intracellular compounds that activate
various enzymes and thereby mediate the effects
of EGF within the cell. One of the enzymes in-
volved in phosphatidylinositol metabolism is
phosphatidylinositol kinase. Since the activity of
this enzyme is stimulated by EGF, it represents a
potential substrate for phosphorylation by the
EGF receptor tyrosine protein kinase.
We have purified the phosphatidylinositol ki-
nase from both A431 cells and human placenta.
This 55-kDa enzyme is active as a monomer —
that is, as a single polypeptide chain. It phos-
phorylates phosphatidylinositol on the 4 position
of the inositol ring and hence is distinct from an-
other phosphatidylinositol kinase that phosphor-
ylates the ring on the 3 position. Information re-
garding the sequence of the amino acids that
make up the phosphatidylinositol 4-kinase (PI
4-kinase) was obtained. Molecular biology tech-
niques were used to isolate and characterize the
cDNA encoding the PI 4-kinase. From this cDNA
the amino acid sequence of the entire PI 4-kinase
was deduced and was found to be unique. The
sequence showed limited similarity to other pro-
teins that bind inositol phosphate or carry out the
phosphorylation of other sugars, suggesting that
the PI 4-kinase may be derived from enzymes that
metabolize sugars. Analysis of the types of mRNAs
present in a variety of tissues that encode the PI
4-kinase indicated that at least two types of mRNA
are present and that the amount of message is reg-
ulated by physiological stimuli. Efforts are now
directed toward producing a cell line in which
this PI 4-kinase cDNA is overexpressed, leading
to elevated levels of PI 4-kinase activity within
the cell. The effects of this overexpression on
the growth properties of the cells will be investi-
gated. This work is supported by a grant from the
National Institutes of Health.
Another enzyme involved in phosphatidylino-
sitol metabolism is a phosphatidylinositol mono-
phosphate phosphatase. This enzyme catalyzes
the reverse of the reaction catalyzed by the PI
4-kinase — that is, it removes the phosphate from
the 4 position of the inositol ring of phosphatidyl-
inositol monophosphate. Although this enzyme
has not been studied previously, its position in
the metabolic pathway of phosphatidylinositol
suggests that it may be important in the overall
regulation of the pathway. We have characterized
it with respect to its kinetic properties, substrate
specificity, and response to various inhibitors.
We have purified the enzyme to a high degree and
have shown that it is a 95-kDa glycoprotein. Un-
like other enzymes in this pathway, it appears to
have an extracellular domain. The possibility that
its activity is regulated through the binding of an
extracellular mediator is being investigated.
327
Mechanism of Action of Polypeptide Growth Factors
It has recently been postulated that the inositol
phospholipids may function in the regulation of
cell shape via interactions with proteins that con-
trol the polymerization of actin. Using EGF and
A43 1 cells, we have shown that there is no corre-
lation between EGF-induced changes in cell
shape and changes in the levels of the classical
phosphoinositides, phosphatidylinositol 4-phos-
phate and phosphatidylinositol 4,5-bisphos-
phate. Further experiments have also failed to
document the involvement of novel inositol phos-
pholipids, phosphorylated on the 3 position of
the inositol ring, in the regulation of cell shape
by EGF. These data suggest that EGF may control
the actin cytoskeleton simply by a mechanism
that involves phosphorylation of cytoskeletal
proteins rather than indirectly by causing alter-
ations in inositol phospholipid levels.
Desensitization of the EGF Receptor
When A43 1 cells are treated with large doses of
EGF, washed, and subsequently rechallenged
with EGF, they fail to respond to the grovvT:h fac-
tor. This phenomenon is known as desensitiza-
tion. Our studies have shown that when the EGF
receptor becomes desensitized, it is no longer in-
ternalized into the cells, and EGF no longer stimu-
lates phosphatidylinositol metabolism. This EGF-
induced desensitization is specific for the EGF
receptor, because the responsiveness of other re-
ceptors is not decreased after EGF treatment.
The EGF receptor itself is a monomeric pro-
tein, a single chain. Upon binding of EGF to its
receptor, two of the receptor monomers come
together to form an EGF receptor dimer. This
dimer is the form that is active in signal transduc-
tion. We have shown that desensitized EGF re-
ceptors do not transduce a signal because they
cannot undergo this EGF-induced dimerization.
Our data suggest that the desensitization of the
EGF receptor results from its phosphorylation by
a protein kinase. We have identified a protein ki-
nase in A43 1 cell cytosol that is activated by EGF
and appears to be involved in receptor desensiti-
zation. The kinase catalyzes the phosphorylation
of the EGF receptor in vitro. Consistent with
what has been observed in whole cells, phos-
phorylation of EGF receptor monomers by this
kinase leads in vitro to an inhibition of the ability
of the phosphorylated monomers to dimerize.
The kinase phosphorylates the EGF receptor on
a serine residue in the second half of the receptor
molecule. Using techniques of molecular biol-
ogy, we have altered this site and are in the pro-
cess of characterizing the properties of cells ex-
pressing this mutated form of the EGF receptor.
328
Protein Structures, Molecular Recognitions,
and Functions
Florante A. Quiocho, Ph.D. — Investigator
Dr. Quiocho is also Professor of Biochemistry and Structural Biology and of Molecular Physiology and
Biophysics at Baylor College of Medicine. He obtained his Ph.D. degree in biochemistry at Yale University
and then did postdoctoral research in chemistry at Harvard University. He was a member of the Rice
University faculty before joining the Baylor faculty. Dr. Quiocho has been a visiting research scientist at
Oxford University, a research fellow of the European Molecular Biology Organization (EMBO), and a
Guggenheim fellow.
LIGAND specificity and the activity of pro-
teins are derived from their precise three-
dimensional structures. Using mainly x-ray
crystallographic techniques, our laboratory is en-
gaged in atomic-level elucidation of the struc-
tures and functions of several proteins (including
enzymes) involved in biologically important pro-
cesses. To complement our work, w^e also employ
biochemical and recombinant DNA techniques.
Adenosine Deaminase
Adenosine deaminase (ADA) is one of the major
enzymes in purine metabolism, catalyzing the ir-
reversible hydrolysis of adenosine or deoxyaden-
osine to the respective inosine product and am-
monia. The enzyme is found in nearly all
mammalian cells and plays a central role in main-
taining competency of the immune system,
among several other functions. Lack or deficiency
of ADA is associated with severe combined immu-
nodeficiency disease (SCID), a genetically in-
herited disorder usually fatal within two years of
birth if left untreated.
Last year we determined the crystal structure of
ADA with bound 6-i?-hydroxyl-l,6-dihydropu-
rine ribonucleoside (HDPR) , a nearly ideal tran-
sition-state analogue inhibitor. We have since
elucidated the structures of complexes of the en-
zyme with the following ligands: 1 -deazaadeno-
sine, a substrate analogue; 2'-i?-deoxycoformycin,
a potent transition-state analogue and a chemo-
therapeutic agent for the treatment of hairy cell
leukemia; and inosine, the product of the deami-
nation of adenosine. All these structures have
provided us with a molecular anatomy of the
various steps associated with ADA's catalytic
activity.
All of the four crystal structures of the deami-
nase indicated above have been determined and
refined at pH 4.2, where the enzyme is only 20
percent active. We have also carried out the re-
finement of the structure of ADA complexed with
HDPR at pH 6, where the enzyme is fully active.
The structures at either pH are essentially the
same.
Antibody- Antigen Interactions
Because monoclonal antibodies against extra-
cellular polysaccharide antigens exhibit very
stringent specificity, they have been used in
blood-grouping and in differentiating bacterial
serogroups and serotypes. We previously ob-
tained crystals of the Fab fragment of the antibody
raised against the surface polysaccharide O-anti-
gen of Shigella flexneri. In the past year we de-
termined and refined the structures of the Fab
and its complexes with a trisaccharide, a-Rha(l-
3)Q!-Rha(l-3);8-GlcNAc, and a pentasaccharide,
a-Rha(l-2)a-Rha(l-3)a-Rha(l-3)i3-GlcNAc-(l-
2)a;-Rha. Both oligosaccharides contain determi-
nants of the O-antigen serotype of the bacteria.
This structural work is in line with our interest
in protein-carbohydrate interactions. Moreover it
is relevant to clinical problems, as oligosaccha-
ride epitopes of bacterial and tumor cell surfaces
are considered to be disease markers and targets
for therapeutic antibodies.
Aldose Reductase
Aldose reductase catalyzes the NADPH-depen-
dent reduction of a wide variety of carbonyl-
containing compounds to their corresponding al-
cohols, with a broad range of catalytic efficien-
cies. Steroids are the best substrates and sugars
the least favorable. Although the enzyme is found
in a variety of cells, its physiological function has
not been firmly established. A role in reducing
the hyperglycemia of diabetes mellitus has been
linked to diabetic complications aff"ecting the
lens, retina, peripheral nerves, and kidney. Drugs
designed to control these complications have not
been clinically successful to date because of lack
of specificity or inefficacy.
In collaboration with Kurt Bohren and Kenneth
Gabbay of the Baylor College of Medicine, we
have obtained excellent diffracting crystals of re-
combinant aldose reductase (from human pla-
centa) with bound NAD PH. We have determined
and refined the enzyme's three-dimensional,
1.65-A resolution structure. The enzyme has a
parallel jS/a-barrel motif, with eight central /?-
strands connected by eight peripheral a-helices.
329
Protein Structures, Molecular Recognitions, and Functions
This establishes a new motif for NADP/NAD-bind-
ing oxidoreductases and the first structure of the
superfamily of aldo-keto reductases.
The substrate-binding site is located in an ex-
tremely hydrophobic elliptical pocket at the car-
boxyl-terminal end of the (8-barrel. The nicotin-
amide group of the NADPH is at the bottom of the
,deep pocket. Although the hydrophobic nature of
the active site greatly favors aromatic (e.g., ste-
roids) and apolar substrates, it is not wcW suited
for binding of highly polar monosaccharides,
which are believed to figure in the pathogenesis
of diabetic complications. The determination of
the structure of aldose reductase paves the way
for rational design of specific inhibitors that
might provide molecular understanding of the
catalytic mechanism, as well as possible thera-
peutic agents for the prevention of diabetic
complications.
Periplasmic Receptors for Active Transport
and Chemotaxis
The family of binding proteins that serve as ini-
tial periplasmic receptors for bacterial active
transport and chemotaxis continues to be a gold
mine for detailed study of protein structure and
molecular recognition of a variety of ligands. We
continue to push the structure refinements of the
seven different periplasmic receptors to much
higher resolutions — the sulfate-binding protein
(at 1 .7-A resolution), phosphate-binding protein
(1.17 A), L-arabinose-binding protein (1.7 A),
D-galactose/D-glucose-binding protein (1.5 A),
maltodextrin-binding protein (1.7 A), leucine/
isoleucine/valine-binding protein (1.7 A), and
leucine-specific-binding protein (1.7 A).
Electrostatic Interactions in Molecular
Recognition of Ligands
Electrostatic interactions are among the key
factors determining the structure and function of
proteins. The refined structure of the liganded
form of sulfate-binding protein shows that the
bound sulfate dianion is completely buried and
bound by hydrogen bonds and van der Waals con-
tacts. The bound sulfate is adjacent to the amino
termini of three helices and is coupled via a pep-
tide unit to a positively charged His residue. Nev-
ertheless, using site-directed mutagenesis and
theoretical analysis, we have shown that helix
macrodipoles and the His residue contribute al-
most nothing to ligand fixation and charge com-
pensation. It is the collection of local dipoles im-
mediately surrounding the sulfate that is
responsible for charge compensation.
330
Molecular Approaches to Olfaction
Randall R. Reed, Ph.D. — Associate Investigator
Dr. Reed is also Associate Professor in the Departments of Molecular Biology and Genetics and of
Neuroscience at the Johns Hopkins University School of Medicine. He received his bachelor's degree in
biophysics from Johns Hopkins and his Ph.D. degree from Yale University. His postdoctoral research was
done with Philip Leder at Harvard Medical School.
OLFACTION is among the oldest of the sen-
sory systems. All multicellular and many un-
icellular organisms have evolved sensitive che-
mosensory systems able to detect and identify
natural chemical substances. The olfactory sys-
tem of vertebrates and analogous systems for the
other senses — vision, hearing, taste, and touch
— allow the conversion of external stimuli into
nerve impulses. In mammals, the olfactory sys-
tem is exquisitely sensitive, capable of detecting
some odorants present at a concentration of only
a few parts per trillion.
The ability of the olfactory system to discrimi-
nate thousands of different odorants suggests a
complex coding mechanism. However, the bio-
logical basis for this coding remains a mystery.
Unlike the visual and auditory systems, which
need only encode information on frequency (or
wavelength) and intensity, the olfactory sense re-
quires multidimensional information. These con-
siderations suggest a complex signal transductory
process.
The signal transductory pathway for olfaction
can be divided temporally and spatially into sev-
eral distinct steps. The first consists of the solubi-
lization and concentration of airborne odorants.
Considerable experimental data suggest that com-
ponents of the mucus are able to concentrate
odorants several thousandfold. We previously
identified cDNA clones encoding proteins that
are present at high concentration in the mucus
and appear to bind odorants. These proteins,
from rat and frog, are members of the retinol-
binding protein family, many of which have been
shown to solubilize hydrophobic ligands in
serum. They are likely to play a similar role in the
olfactory system.
The recognition of the chemical structure of an
odorant and transduction of that information
across the plasma membrane is a poorly under-
stood process. Some investigators hypothesize
that odorants interact directly with the lipid
membrane, but it is difficult to see how the abil-
ity to discriminate stereoisomeric compounds
could be accommodated by such a system. Sev-
eral years ago we began to test an alternative hy-
pothesis: that binding of odorants to specific
membrane-associated proteins leads to intracel-
lular changes in the primary sensory neuron.
The detection of odorant-stimulated activation
of second messengers in olfactory neurons sug-
gested an analogy to sensory systems. In visual
transduction, which is the best characterized of
these systems, sensitivity is achieved through a
second messenger cascade consisting of the
membrane-bound receptor rhodopsin, a rod
photoreceptor-specific GTP-binding protein,
transducin, and a cyclic GMP phosphodiesterase.
Light-stimulated decrease in the concentration of
intracellular second messenger leads to modula-
tion of the cellular membrane. Activation of
adenylyl cyclase in preparations of rat olfactory
cilia depends on the presence of guanine nucleo-
tides. The observation of odorant-stimulated
GTP-dependent adenylyl cyclase activity argues
strongly for a similar GTP-binding protein-
coupled signal transductory pathway in olfaction.
We have identified a GTP-binding protein as
well as an adenylyl cyclase expressed exclusively
in olfactory sensory neurons. Moreover, these
components are localized to the olfactory cilia,
where the initial events in olfactory signal trans-
duction are thought to occur. The Go,f protein,
which is highly homologous to a GTP-binding
protein that stimulates adenylyl cyclase in other
systems, is expected to interact directly with the
olfactory receptors. Additionally, we have demon-
strated that Golf can couple receptor activation to
increases in intracellular cAMP in cell lines defi-
cient for the stimulatory G protein, G^.
Identification of genes encoding olfactory re-
ceptors may reveal how these structures are able
to detect thousands of different odors. Recently,
Linda Buck and Richard Axel (HHMI, Columbia
University College of Physicians and Surgeons)
have described a gene family that encodes pro-
teins expressed in olfactory tissue. We have iden-
tified additional members of this large family,
utilizing the polymerase chain reaction tech-
nique. These experiments reveal that expression
of the mRNA that encodes these putative olfac-
tory receptors is confined to the sensory neurons
of the olfactory epithelium.
531
Molecular Approaches to Olfaction
We are now engaged in characterizing these re-
ceptors biochemically and in screening for the
specific ligands that activate them. The subcellu-
lar localization of the receptor proteins is being
examined by means of antibodies directed against
conserved regions. These molecular tools should
allow us to address some important questions.
T)oes each of the several million olfactory sensory
cells express a single receptor protein species?
How are the genes that encode these receptors
organized, and how is their expression regulated?
The final step in the transduction of odorant
stimuli is the generation of the intracellular sig-
nal and the firing of an action potential. Special-
ized forms of second messenger-generating en-
zymes and novel ion channels play important
roles in this process. We have identified cDNA
clones encoding three distinct forms of adenylyl
cyclase and are investigating the regulation of
this important enzyme in olfactory tissue. Olfac-
tory neuronal adenylyl cyclase type III has bio-
chemical properties that would be advantageous
for an enzyme involved in sensory transduction.
Electrophysiologic experiments have identified
cyclic nucleotide-gated ion channels in olfac-
tory neurons, and we have recently isolated and
characterized cDNA clones from olfactory tissue
that encode ion channels with properties similar
to those found in the visual system.
The olfactory system is also interesting as a
model for neuron differentiation and develop-
ment. The olfactory neuroepithelium is the only
neuronal tissue in adult mammals that undergoes
continual regeneration. The lifetime of sensory
neurons is approximately 40 days, after which
they are shed from the epithelium and replaced
from a population of neuroblast-like precursor
cells. Moreover, if the nerve leading from the sen-
sory neurons to the olfactory bulb is severed, all
1 0 million receptor cells are rapidly lost and sub-
sequently replaced in a relatively synchronous
fashion. My laboratory is beginning to address the
mechanism of regulation of olfactory neuron-
specific genes.
We have identified a specific DNA sequence
upstream of all the genes known to be expressed
specifically by the olfactory neurons and have ob-
served in olfactory homogenates a binding activ-
ity specific for that sequence. This putative tran-
scriptional regulator might direct the expression
of the entire repertoire of genes involved in olfac-
tory signal transduction and neuronal maturation.
We have recently identified a number of cDNA
clones that encode proteins expressed only in the
mature sensory neurons and are attempting to
elucidate their role in the cell. A novel group of
proteins, those expressed transiently during neu-
ron maturation, may include receptors for neuro-
genic as well as neurotrophic factors. Several of
the genes we have identified appear to encode
membrane-bound or cell surface proteins.
We are continuing to use several techniques to
elucidate the mechanism of signal transduction.
Likewise, the identification of proteins asso-
ciated with the replacement of olfactory neurons
provides the tools to study neural development,
not just in the olfactory system but also in other
areas of the brain. In the future we will focus on
the molecular components that underlie the
complex mechanisms of signal transduction, sig-
nal processing, and the formation of neural
connections.
332
The Molecular Basis of Hereditary Diseases
of the Kidney
Stephen T. Reeders, M.D. — Associate Investigator
Dr. Reeders is also Associate Professor of Internal Medicine and Genetics at Yale University School of
Medicine. He attended Cambridge University with the intention of majoring in physics, but, realizing that
developments in molecular biology were providing the basis for new approaches to the study of human
disease, he switched to medicine and continued to study at Oxford University. After qualifying in
medicine, he sought clinical training in intensive care, cardiology, nephrology, and neurology. Then, with
Sir David Weatherall at Oxford, he began to use molecular genetic techniques to study human disease,
with emphasis on hereditary diseases of the kidney, diseases that heretofore had received little attention
from geneticists.
CRITICAL for normal functioning of the kid-
ney is the integrity of the glomerular base-
ment membrane (GBM) , a complex extracellular
structure that forms one of the main barriers be-
tween the blood and the urine. The GBM is com-
posed of several proteins, including five related
but subtly different collagens that interact to
form a chicken-wire mesh holding the membrane
together. One of the interests of our laboratory is
Goodpasture syndrome, an autoimmune disorder
in which, for unknown reasons, autoantibodies
are suddenly targeted at the collagen components
of basement membrane in the lungs and kidneys.
In the kidney, these autoantibodies produce a
sudden and devastating severe inflammation,
which frequently leads to acute renal failure, irre-
versible unless treated. The nephritis is often ac-
companied by autoimmune lung damage, mani-
fested by bleeding into the alveoli.
Previous studies have shown that the probable
target of Goodpasture autoantibodies is the a 3
chain of basement membrane collagen. To under-
stand the pathogenesis more clearly, we under-
took to isolate and purify the collagen chain to
study its structure. Because the protein is present
in very small amounts and is accompanied by four
similar proteins, purifying it has proved difficult.
We have therefore isolated, cloned, and se-
quenced the gene for the a3 chain of basement
membrane collagen and have used the sequence
information to predict the primary structure and
compare this protein with other basement mem-
brane collagens.
In collaboration with Billy Hudson (Kansas
City), we used knowledge of the primary struc-
ture to identify several potential antibody-
binding sites (epitopes) in the a3 molecule. We
synthesized short peptides and used them to test
the binding of some of these sites, which we local-
ized to within a small region at the carboxyl ter-
minus. At least one of the peptides has very high
affinity for Goodpasture antibodies and adsorbs
them from patients' serum. Knowledge of the
epitope structure should enable us to develop a
means of selectively adsorbing Goodpasture anti-
bodies, opening possibilities for a new treatment
modality. In addition, this information may pro-
vide clues to the development of autoimmunity
in this disorder.
Mariko Mariyama, a Howard Hughes associate
in our laboratory, has recently isolated a novel
basement membrane collagen, a4. The compari-
son of the structure of this molecule with the
known basement membrane collagen allows the
evolution of these molecules to be inferred. Two
classes of the molecule exist in all species from
roundworm to humans, suggesting a functional
divergence. The expression of a3 and a4 occurs
in a limited number of tissues, and they are re-
gionally localized within these tissues. In kidney,
for example, they are found in the glomerulus
but not in the tubular or vascular endothelial
basement membranes. Our gene mapping data
suggest that the expression of a3 and aA is coordi-
nated by transcription of the genes from opposite
DNA strands. A similar arrangement has been
found for the al and a2 genes, homologues of al
and a2.
One of the major projects in our laboratory is a
study of the molecular and cellular pathology of
autosomal dominant polycystic kidney disease
(ADPKD) , one of the commonest causes of kid-
ney failure in humans, affecting at least 1 in
1,000 of the population. The disease is an enor-
mous burden to families and the community,
since the majority of patients develop irreversible
kidney failure in middle life and require dialysis
or transplantation for survival.
Having previously ascertained that the majority
of the inherited mutations in ADPKD lie close to
the tip of the short arm of chromosome 16, we
have isolated and cloned a small segment of DNA
(550,000 base pairs) that includes the mutated
gene. This region turns out to be extremely gene
rich, and we have already isolated 22 genes from
within it. These include novel cyclin A-like and
r«s-like genes, a gene encoding a zinc finger pro-
tein, and a gene having homology to the (8-
335
The Molecular Basis of Hereditary Diseases of the Kidney
subunit of the G protein family. Since we have
not been able to detect any large-scale deletions
or rearrangements affecting any of the 22, we
have begun to examine the sequence of these
genes in detail and to look for mutations that may
affect only one or two nucleotides. This work is
supported by a grant from the National Institutes
of Health.
Our laboratory has been interested in the struc-
ture of telomeres, the ends of chromosomes. We
have shown that banks of repetitive sequence
reminiscent of the sequence of human telomeres
(TTAGGG) are also present at other sites within
the human genome. One of the most interesting
arrays of telomere-like sequence is embedded in
the middle of the long arm of chromosome 2.
Comparison of the chromosome banding pattern
of humans with that of several closely related
apes suggests that this region of the chromosome
contains a point at which two ancestral ape chro-
mosomes fused. Jaap IJdo, a Howard Hughes asso-
ciate, has cloned this fusion point and shown that
it consists of a head-to-head telomere-telomere
fusion.
In situ hybridization of a cloned DNA segment
containing subtelomeric sequences to a human
chromosome 2 (left). The fluorescent probe rec-
ognizes the sequences at the tips of several
chromosomes, including those of the long arm
of chromosome 2 shown here. In addition, the
probe recognizes another sequence (left
arrow) that was buried in the middle of chro-
mosome 2 when chromosomes of ancestral
apes fused to create the human chromosome.
(To this day the great apes have one more chro-
mosome pair than humans.) The point of fu-
sion is very close to a rare fragile site that was
observed in the chromosome 2 used in this
study. The chromosome on the right is shown
with the hybridization signal removed so that
the fragile site, which appears as a gap ( right
arrow ), can be discerned.
From If do, f.W., Baldini, A., Wells, R.A.,
Ward, D.C., and Reeders, S.T. 1992. Genomics
12:835-835.
354
Extracellular Factors Affecting
Neuron Development
Louis F. Reichardt, Ph.D. — Investigator
Dr. Reichardt is also Professor of Physiology and of Biochemistry and Biophysics at the University of
California, San Francisco. He received his Ph.D. degree in biochemistry from Stanford University for work
on control of gene expression by a bacterial virus, the bacteriophage X. Dr. Reichardt entered the field of
neurobiology as a postdoctoral fellow in Paul Patterson 's laboratory at Harvard University, where he
studied factors that regulate the transmitter phenotype of individual neurons. Among his honors are a
McKnight Scholars Award, a Sloan Award, and a Guggenheim Fellowship.
MY laboratory is investigating molecules in
the extracellular environment of neurons
(conducting nerve cells) that direct their devel-
opment in vivo. These include trophic (nutri-
tive) factors, exemplified by nerve growth factor
(NGF), and molecules in the extracellular matrix
or on the surface of cells that serve as substrates
for the growth of axons (long nerve fibers) . Such
molecules help to regulate neuronal survival,
axon growth, and synapse (nerve junction)
formation.
Neurons require contact with targets to survive
during development. Experimentally increasing
or decreasing the volume of target tissue corre-
spondingly increases or decreases neuronal sur-
vival. To explain these target influences, it has
been proposed that target organs synthesize tro-
phic factors required for the survival of the inner-
vating neurons. Defects in the synthesis of these
factors or in the neuron's ability to respond to
them may explain some neurodegenerative
disorders.
The key to understanding trophic factors is to
identify and understand the actions of their re-
ceptors. One of the proteins that functions as a
receptor for the family of NGF-related trophic
factors is called the low-affinity NGF receptor.
Antibodies to this protein have been prepared
and used to show that it is distinct from a second
protein that constitutes a distinct high-affinity re-
ceptor for NGF. Work in the laboratories of David
Kaplan, Mariano Barbacid, and their colleagues
has recently shown that this second protein is the
trk proto-oncogene product, an NGF-tyrosine ki-
nase. The same antibodies have been used to
show that NGF-dependent survival and axon
growth by response neurons do not require the
low-affinity receptor. These antibodies are being
used to examine the role of this class of receptor
in mediating other responses to NGF.
Our laboratory has devoted considerable effort
to identifying molecules that promote the groMT:h
of neuronal processes. We have tried to identify
both the molecules that axons recognize in their
environment and the receptors that neurons use
for binding to these molecules. Our results have
shown that laminin, an adhesive protein, is by far
the most active of the glycoproteins that cells se-
crete into the extracellular matrix. Other pro-
teins that have similar activities include fibronec-
tin, thrombospondin, and vitronectin.
We have also shown that neurons utilize a fam-
ily of receptors, the integrins, to bind to laminin
and other matrix glycoproteins. Distinct recep-
tors appear to mediate neuronal adhesion and
growth cone motility on laminin, fibronectin,
and collagen. In the past two years we have puri-
fied several of these proteins and isolated clones
encoding subunits of the receptors. Specific anti-
bodies to these subunits have been prepared and
are being used to study their regulation . Of partic-
ular interest, we have identified the receptors
used by both peripheral neurons and retinal neu-
rons to interact with laminin, a heterotrimer as-
sembled from three different gene products — A,
Bl, and B2. Josh Sanes and Eva Engvall have
shown that at least two different A-like genes and
two different Bl -like genes exist and are differen-
tially expressed, making it possible to assemble
four different isoforms of laminin, which are dif-
ferentially distributed in embryos. We have
shown that individual integrin heterodimers dis-
tinguish between these isoforms, making it possi-
ble for cells to exhibit different responses to indi-
vidual isoforms. Work in the past year has also
identified candidate receptors for thrombospon-
din and tenascin.
Evidence suggests that the activity of integrin
receptors may modulate the behavior of axons in
vivo. In studies on regulation of a^jSi, the major
laminin receptor in the neuroretina, we have
shown that receptor function can be regulated on
several levels. First, expression of the genes en-
coding the two subunits is regulated. Retinal gan-
glion cells lose responsiveness to laminin when
they contact their synaptic partners in the optic
tectum, and this reflects down-regulation of ex-
pression of the gene. Second, other neurons in
the retina modulate responses to laminin by ex-
hibiting changes in the activity of integrin recep-
tors on their surfaces. These changes occur rap-
idly and can be modulated by external and
internal agents. Yet a third level of regulation ap-
335
Extracellular Factors Affecting Neuron Development
pears to modulate the signals conveyed by inte-
grin receptor binding to laminin and other pro-
teins. These signals can be modulated by trophic
factors, such as NGF.
In addition to proteins secreted into the extra-
cellular environment of cells, neurons also recog-
nize and use for axon extension integral mem-
brane proteins on the surfaces of different cell
types. Recent work in our laboratory has identi-
fied some of the major proteins that neurons use
for axonal growth in the central and peripheral
nervous systems. Thus neurons have been shown
to use both extracellular matrix and cell-cell ad-
hesion molecules for growth of axons on
Schwann cells, the major cell type with which
they interact in peripheral nerves. A similar com-
bination of adhesive interactions promotes
growth of retinal axons on astroglia. The move-
ments of growth cones of retinal ganglion cells
have been directly monitored in vivo by time-
lapse microscopy. Injection of antibodies to indi-
vidual cell adhesion molecules has been shown
to alter dramatically the behavior of these growth
cones.
In studies on interactions of neurons with other
cells in the brain, it became clear that additional
adhesion molecules, not yet identified or charac-
terized, must be important. We have used molec-
ular biological methods to identify two of these
with localized expression patterns in the brain.
One of these, the integrin ag/^n is localized on
axons in many of the major axon tracts of the
brain. Its concentration on axons in these tracts
suggests that it mediates interactions between
these axons. A potential ligand for this integrin
that is also localized in these tracts has been iden-
tified. The second novel adhesion-promoting
molecule is B-cadherin, a Ca^'^-dependent adhe-
sion molecule that is also expressed in the ner-
vous system. It appears to be concentrated in spe-
cific cells in several areas of the brain. Some but
not all neurons are able to interact with it. Both
molecules are also expressed in some areas out-
side the nervous system.
Future work will focus on determining how
these neuronal receptors act to promote growth
cone motility, how they are regulated during de-
velopment, and what their importance is in regu-
lating the grov^T:h of axons during development
and regeneration.
356
Molecular Genetics of RNA Processing
and Behavior
Michael Rosbash, Ph.D. — Investigator
Dr. Rosbash is also Professor of Biology at Brandeis University and Adjunct Professor of Molecular Biology
at Massachusetts General Hospital, Boston. He received his Ph.D. degree in biophysics from the
Massachusetts Institute of Technology and was a postdoctoral fellow at the University of Edinburgh,
where he studied with J. O. Bishop. Dr. Rosbash was a Guggenheim Fellow in Paris, France.
MY laboratory is interested in two fundamen-
tal problems. Our earliest and foremost in-
terest is the molecular genetics of RNA process-
ing. For this subject our principal experimental
system is the budding yeast Saccharomyces cere-
visiae, which is amenable to both genetic and
biochemical attack. Our more recent interest is
the molecular genetics of behavior — in particu-
lar, circadian rhythms. This problem is addressed
in the fruit fly Drosophila melanogaster, be-
cause the organism is amenable to behavioral as
well as biochemical and genetic approaches.
Within the area of RNA processing, we are most
interested in understanding certain aspects of
pre-messenger RNA splicing, the process by
which the undesirable sections of a pre-mRNA
molecule are removed and the remaining
"sense" sections sewn together. Our interests are
primarily focused on the more biological aspects
of the problem. These include how the places in
the molecule to be cut — the two splice sites in
the case of a pre-mRNA with a single intron — are
defined. They also include how splice site
partners, in the case of a pre-mRNA with multiple
introns, are specified.
The latter question is particularly puzzling, be-
cause most of these splice sites appear similar.
Yet there is clearly an order to the process, al-
though the basis for this order is not apparent.
The adjective "biological" is used to distinguish
these issues from more "chemical" consider-
ations, such as how the active site of the splicing
enzyme is formed and how the efficiency and
specificity of the splicing reaction are dictated.
We are addressing these issues of splice site
definition and partner assignment by examining
the interactions between a pre-mRNA substrate
and splicing factors. Although some of the work is
done in vivo — that is, in intact cells where the
interactions are inferred from their conse-
quences— most of our efforts have concentrated
on interactions that take place during in vitro
pre-mRNA splicing in a whole-cell yeast extract.
We have focused particularly on the earliest in-
teractions, those that apparently reflect recogni-
tion of the pre-mRNA substrate by the splicing
machinery.
Our studies indicate that the factor Ul small
nuclear ribonucleoprotein (snRNP) plays a prom-
inent role in these early interactions. Conse-
quently we have expended considerable effort in
characterizing this snRNP and its constituents, as
well as the pre-mRNA-Ul snRNP interaction. Sur-
prisingly, both ends of the intron interact with Ul
snRNP, suggesting that certain aspects of splice
site recognition, if not splice site partner assign-
ment, are already defined early during the
spliceosome assembly process, well before the
cleavage and ligation steps of the actual splicing
process take place. This work is supported by a
grant from the National Institutes of Health.
We are also interested in another biological
aspect of RNA processing — namely, how mRNA
is transported from the nucleus, where it is syn-
thesized, to the cytoplasm, where it is translated
into protein. This transport problem interfaces
with the splicing process, since RNA needs to be
transported to the cytoplasm but usually not be-
fore the splicing is completed. Otherwise, incom-
pletely spliced molecules would be prematurely
transported, which would give rise to untranslat-
able pre-mRNAs in the cytoplasm.
The problem of RNA transport is poorly under-
stood, and even less well understood in yeast than
in mammalian cells. In yeast, however, there is
the possibility of addressing the problem with ge-
netic tools. At present we are localizing pre-
mRNA and splicing factors within the yeast nu-
cleus, in an attempt to define a cytological path
that the RNA may follow in leaving the nucleus.
The goal is to use existing temperature-sensitive
mutants and to uncover new ones, both to study
RNA transport and to define some of the gene
products important for this process.
Rhythms
Our goals in this project are to define the bio-
chemical machinery that underlies the mysteri-
ous yet ubiquitous process of circadian rhythmic-
ity. We are using genetics and biochemistry to
define candidate genes and gene products that
may participate in fundamental aspects of these
rhythms. Our entree into the process is the pe-
riod gene (per) of Drosophila melanogaster.
337
Molecular Genetics of RNA Processing and Behavior
Mutants in this gene, originally identified more
than 20 years ago, have profound effects on circa-
dian rhythms of locomotor activity and of eclo-
sion (emergence of adults from the pupal case) .
Recently we discovered that the per gene prod-
ucts, mRNA as well as protein, undergo circadian
fluctuations in level during the circadian cycle.
These observations and others indicated that there
is a feedback loop in which the per protein nega-
tively affects the level of its own mRNA. Since the
per mRNA also gives rise to the per protein during
translation, this putative feedback loop contains
all of the elements required to define a circadian
clock, including a substantial (and mysterious)
delay between the mRNA and protein accumula-
tion curves. We are studying several aspects of this
loop in an attempt to confirm (or refute) its im-
portance to the circadian clock.
Because no close relatives of the per protein
with a known biochemical function are found in
the database, we cannot say for certain what gene
family it belongs to or what biochemical function
it serves. We suspect, however, that it is a tran-
scription factor, or serves to modulate transcrip-
tion, and that its effect on its own mRNA levels is
quite direct. A current goal is to test this hypothe-
sis and to define the biochemical function of the
per gene product, especially insofar as rhythms
are concerned.
With the support of a grant from the National
Institutes of Health, we are also in the process of
defining and studying several other genes that af-
fect rhythms. Although these genes are in a less
well developed state of examination than the per
gene, some appear interesting and may provide
additional insight into clock mechanisms.
338
Molecular Mechanisms of Transcription,
Regulation, and Development
of the Neuroendocrine System
Michael G. Rosenfeld, M.D. — Investigator
Dr. Rosenfeld is also Professor of Medicine in the Eukaryotic Regulatory Biology Program at the University
of California, San Diego, School of Medicine. He received his undergraduate degree from the Johns Hopkins
University and his medical degree from the University of Rochester. His internship and medical residency
were completed at Barnes Hospital, St. Louis. Before accepting his current position, he received
postdoctoral training at NIH. Dr. Rosenfeld also holds an adjunct position at the Salk Institute.
OVER the past year our central research focus
has been the determination of molecular
mechanisms that induce specific neuroendocrine
phenotypes and the further definition of signal
transduction pathways that lead to regulated pat-
terns of gene expression.
The neuroendocrine system coordinates the
complex pattern of regulation necessary to
achieve the precise temporal, spatial, and homeo-
static patterns of gene expression required by
complex organisms. Development of the central
nervous system and endocrine organs involves
precise patterns of responses to morphogens and
other regulatory signals that ultimately establish
the intricate patterns of neural and endocrine
phenotypes. The cloning and analysis of genes
encoding receptors and cell-specific transcrip-
tion factors have permitted an initial definition of
developmental and regulatory strategies.
The anterior pituitary gland has provided a suit-
able model to investigate the molecular basis for
generating specific cell phenotypes in an organ.
The rat genes for growth hormone and prolactin
(hormone that stimulates and sustains lactation)
exhibit precisely restricted expression in the
cells of origin, somatotrophs and lactotrophs, re-
spectively. We found that prolactin gene expres-
sion is dictated by a distal enhancer and a proxi-
mal region, each containing at least four critical
cell-specific elements. These two domains, each
capable of targeting tissue-specific gene expres-
sion, act synergistically to generate high levels of
prolactin gene expression in transgenic mice.
Similarly, grov^h hormone gene expression in-
volves the action of related cell-specific cis-
active elements. A 33-kDa cell-specific transcrip-
tion factor, referred to as Pit- 1 , was characterized
and its encoding cDNA defined.
Bacterially expressed Pit-1 specifically and
with high affinity binds to prolactin and growth
hormone promoters. Additional elements and
factors are required to achieve the full physiologi-
cal levels and restricted patterns of expression of
the prolactin and growth hormone genes. For ex-
ample, the estrogen receptor synergistically acts
with Pit- 1 in activation of the rat prolactin gene's
distal enhancer.
Pit-1 is a member of a family of regulators that
contain a POU domain consisting of a variant ho-
meodomain and a second, 76-amino acid se-
quence of striking homology. We have found that
the 76-amino acid POU-specific domain func-
tions in high-affinity DNA binding, in conferring
site-specificity, and in protein-protein interac-
tions critical for transcriptional activation by
Pit-1. The major transcription-activating domain
of Pit-1 is a 70-amino acid, N'-terminal, serine,
threonine-rich sequence, distinct from recog-
nized motifs. A detailed structure-function analy-
sis has suggested that the determinants of DNA
binding by the Pit-1 POU-domain protein are dis-
tinct from those of the classical homeodomain
proteins, and that Pit- 1 DNA binding is regulated
by phosphorylation of a specific, conserved resi-
due in the POU homeodomain.
A genetic approach was utilized to determine
the functional role of Pit-1 during organogenesis.
Based on the demonstration of disrupted patterns
of the Pit-1 gene in genetic dwarf mice, it has
been established that this POU-domain protein
acts as a developmental regulator to determine
patterns of commitment, progression, and prolif-
eration of three specific cell types in the anterior
pituitary gland. In the case of lactotrophs and so-
matotrophs, proliferation is restricted in the
Pit- 1 -defective animal cell type, while the thyro-
troph provides an initial example of an estab-
lished cell phenotype that disappears at the time
the initial Pit-1 protein is expressed. These data
indicate that one role of a developmental tran-
scriptional regulator can be survival of an estab-
lished cell type.
A strategy was devised to isolate new members
of the POU-domain gene family. Ten new
members were identified in neural tissues. Two,
referred to as Brn-1 and Brn-2, exhibit virtually
identical patterns of expression in the central
nervous system, though Brn-1 is clearly ex-
pressed in the medullary zone of the kidney
while Brn-2 is not. A third member, Brn-3, is pre-
dominantly expressed in the peripheral nervous
system. Tst l transcripts are present in mamma-
lian brain cells and in testis. Subsequently other
339
Molecular Mechanisms of Transcription, Regulation, and Development
of the Neuroendocrine System
POU domains have been identified, each ex-
pressed in a unique pattern early in development
and during organogenesis.
Most POU-domain genes were widely ex-
pressed in all levels of the neural tube (including
the retina) during early development, and hybrid-
ization in the ventricular (proliferative) zone of
the neuroepithelium is evident for all four tran-
scripts at all levels. The time course of anatomic
restriction in the developing neural tube is
distinct, and the patterns for each gene product
tend to reflect the adult loci of expression. In
addition, reactive transcripts for some of these
genes are also expressed during mammalian
neurogenesis.
We have recently identified many additional
family members in the brain and have obtained
initial evidence that at least some members of the
family can bind to specific elements in distinct
classes of neuronally expressed genes. We find
that neurally expressed POU-domain proteins
can be considered to represent distinct related
families, each of which binds to related, but dis-
tinct, DNA response elements. Putative target
genes have been identified for several of these
factors, and analysis of their function has revealed
both inhibitory and stimulatory domains.
Interactions between diverse transcription fac-
tors generate heterodimers that exert distinct pat-
terns of gene activation. Thus, in the case of the
retinoic acid receptor, we find that a series of cell
type-specific coregulators impose novel hierar-
chies of binding site preferences. Unique posi-
tive and negative transcriptional regulators
impose variable patterns of gene activation, po-
tentially contributing to the refinement of pheno-
typic variance required in the central nervous
system. One class of coregulators has been char-
acterized by expression cloning techniques.
Development of the neuroendocrine system is
also initially regulated by post-transcriptional
strategies. We have provided evidence that a gene
of the neuroendocrine system, the calcitonin/
CGRP (calcitonin gene-related peptide) gene,
contains genomic regions that represent discrete
hormone-encoding domains. The ultimate ex-
pression of these domains is dependent on alter-
native RNA-processing events that differentially
include or exclude specific exons encoding cer-
tain components of polypeptide regulators in the
mature mRNA products. The rat and human calci-
tonin/CGRP genes contain six exons. More than
95 percent of the mature transcripts in thyroid C
cells, encoding calcitonin, are produced by
splicing of the first three to four exons. CGRP
mRNA is the only detectable mature transcript in
rat neuronal tissue, which appears to reflect the
actions of a specific regulatory machinery con-
trolling post-transcriptional RNA splice acceptor
choice. CGRP appears to be an important regula-
tor of blood pressure.
Current investigations continue to explore
these novel aspects of transcriptional and
post-transcriptional regulatory strategies in neu-
roendocrine gene expression. Results may pro-
vide new ways of studying the problem of
organogenesis.
340
Development of the Drosophila Visual System
Gerald M. Rubin, Ph.D. — Investigator
Dr. Rubin is also John D. MacArthur Professor of Genetics at the University of California, Berkeley, and
Adjunct Professor of Biochemistry and Biophysics at the University of California, San Francisco, School
of Medicine. He received his B.S. degree in biology from the Massachusetts Institute of Technology and his
Ph.D. degree in molecular biology from the University of Cambridge. Dr. Rubin's postdoctoral work was
done at Stanford University with David Hogness. He has held faculty positions at Harvard Medical School
and the Carnegie Institution of Washington. Dr. Rubin is a member of the National Academy of Sciences
and counts among his other honors the American Chemical Society Eli lilly Award in Biological Chemistry.
RESEARCH in our laboratory is directed to-
ward studies of differentiation and gene regu-
lation in the developing nervous system. Our ex-
perimental approach involves studying genes
whose mutations disrupt neural development.
During the past year, we have focused our work
on several genes important for the determination
of cell fates in the developing retina of the fruit
fly Drosophila.
Two very different but not exclusive mecha-
nisms can account for the selection of distinct
developmental pathways. First, cells may be pro-
grammed in a lineage-dependent manner by the
asymmetric partitioning of determinants during
cell division. Different developmental pathways
are then selected in the daughter cells in re-
sponse to the different localized determinants. Al-
ternatively, cellular differentiation may occur in
a lineage-independent manner, where the posi-
tion that a cell occupies in a developing field
determines its fate. In this case, diffusible sub-
stances, such as hormones, or interactions with
adjacent cells are the primary determinants of
cellular differentiation. Although the mecha-
nisms used to read and interpret such positional
information are largely unknown, short-range
cellular interactions are thought to be of princi-
pal importance in a wide variety of developmen-
tal phenomena.
The compound eye of Drosophila melanogas-
ter is an attractive system to study the mecha-
nisms underlying lineage-independent develop-
mental decisions, since it consists of a small
number of different cell types that develop in a
lineage-independent manner. The compound eye
is a two-dimensional array of 800 repeating units,
or ommatidia. Each ommatidium contains 8 pho-
toreceptor cells as well as 1 2 nonneuronal acces-
sory cells. Each photoreceptor cell has a distinct
cellular identity, based on both its position
within the ommatidium and its projection pattern
to the optic lobes of the brain. The stereotyped
arrangement of this small number of nerve cells,
together with the dispensability of the visual sys-
tem under laboratory conditions, makes the com-
pound eye an attractive model system to study
genes involved in the specification of nerve cells.
Assembly of ommatidia begins in an initially
unpatterned monolayer of epithelial cells, the
eye imaginal disc. Ommatidial assembly does not
occur synchronously throughout the disc but in-
stead begins at the posterior edge and progresses
anteriorly. Eye discs removed from larvae just
prior to pupariation show a smoothly graded se-
ries of ommatidia at different stages of develop-
ment, covering just over half of the disc. Examina-
tion of individual cells in the forming ommatidia
has shown that the photoreceptors differentiate
in a fixed sequence, beginning with the central
R8 photoreceptor and proceeding pairwise with
R2 and R5, R3 and R4, Rl and R6, and finally R7.
The fate of a cell within a developing ommatid-
ium appears to be governed by the specific com-
bination of signals received by that cell from its
immediate neighbors. We would like to under-
stand how such signals are generated, received,
and interpreted. Our approach has been to study
mutations that specifically disrupt these pro-
cesses, as illustrated by our studies of the seven-
less gene.
The sevenless gene is essential for the develop-
ment of a single type of photoreceptor cell. In the
absence of proper sevenless function, the cells
that would normally become the R7 photorecep-
tors become instead nonneuronal cells. Previous
morphological and genetic analysis has indicated
that the product of the sevenless gene is involved
in reading or interpreting the positional informa-
tion that specifies this particular developmental
pathway. We have isolated and characterized the
sevenless gene. Our data indicate that sevenless
encodes a transmembrane protein with a tyrosine
kinase domain. The structural analogies between
the sevenless protein and certain hormone recep-
tors suggest that developmental pathway selec-
tions dependent on cell-cell interactions may
involve molecular mechanisms similar to physio-
logical or developmental changes induced by
long-range diffusible factors.
To investigate the role of the sevenless protein
341
Development of the Drosophila Visual System
in R7 development, we have examined the pat-
tern of sevenless expression in the developing
retina and have studied the effects of experimen-
tally altering this pattern. By transiently express-
ing the sevenless protein, we have shown that
there is only a brief period during eye develop-
ment when the sevenless protein is required for
t-he formation of the R7 photoreceptor. Our re-
sults are consistent with the proposal that seven-
less directly reads positional information re-
quired to specify the R7 developmental pathway.
A major current goal is to elucidate the intra-
cellular signal transduction pathway that is acti-
vated by stimulation of the sevenless protein.
That is, how does activation of the sevenless tyro-
sine kinase instruct a cell to become an R7 pho-
toreceptor? Defining the biologically relevant
substrates of tyrosine kinase receptors has been a
long-standing and difficult problem. We have
taken a genetic approach toward identifying
genes whose products act downstream of seven-
less, including those that might be direct sub-
strates for the sevenless kinase.
We have utilized two strategies. First, we have
looked for other mutations that give the same phe-
notype as sevenless — transformation of the R7
cell into a nonneuronal cell type. In this way we
isolated the seven-in-absentia (sina) gene.
Function of the sina gene is required only in R7
for correct R7 cell development. The sina pro-
tein, which has a potential metal-binding do-
main, is localized in the nuclei of several omma-
tidial precursor cells, including R7, and sina
expression in R7 appears before R7 overtly be-
gins to differentiate. These data indicate that the
sina gene product is necessary at a stage in the
determination of R7 cell fate when R7 receives
and interprets developmental signals from neigh-
boring cells and possibly acts by regulating gene
expression.
Second, we utilized a crippled sevenless pro-
tein, whose activity is just barely adequate to
specify R7 cell development, to establish a highly
sensitive assay for other components of this signal
transduction pathway. Using this assay we looked
for other genes in which a 50 percent reduction
of the level of their protein products resulted in a
failure to specify R7 cells. In this way we have
identified seven genes that appear to encode
products that act to interpret the signal mediated
by the sevenless receptor.
The putative products of two of these seven
genes have been identified. One encodes a ras
protein. The ras oncogene is implicated in as
many as 30 percent of human tumors. The ras
proteins exist in two different states: an inactive
GDP-bound state and an active GTP-bound state.
The active ras protein transmits a signal by inter-
action with unidentified cellular targets. The
other locus whose product we have identified en-
codes a protein that is homologous to the Sac-
charomyces cerevisiae CDC25 protein, an acti-
vator of GDP-GTP exchange by ras proteins.
These results suggest that the stimulation of ras
protein activity is a key element in the signaling
by sevenless and that this stimulation may be
achieved by activating the exchange of GTP for
bound GDP by the ras protein. The evolutionary
conservation of the ras signaling pathway sug-
gests that studies in Drosophila could provide
clues to the role of ras in oncogenesis and devel-
opmental abnormalities in humans.
342
lebrand Factor Precursor
GPIb
Botrocetin
Heparin Heparin
Collagen Factor VIZI Sulfatide Collagen GPIIb-IIIa
488
474 471 4G4 463
Top: Structure-function relationships of von Willebrand factor.
The factor precursor comprises four types of homologous do-
mains (A-D), repeated two to five times each. The mature sub-
unit consists of 2, 050 amino acids. Binding sites are indicated
for collagen, heparin, factor VIII, platelet glycoprotein lb, bo-
trocetin, sulfatide, and platelet glycoprotein Ilb-IIIa. The loca-
tions of two Arg-Gly- Asp- containing sequences, RGD and
RGBS, are indicated.
Bottom: Mutations, indicated by brackets, in exon 28 of the
gene encoding von Willebrand factor cause von Willebrand
disease types IIA and IIB. The segment of the factor shown in-
cludes amino acid residues 463-921, designated by single-
letter code. The positions of repeated domains D3 and A 1-3 are
indicated. The green zigzag segments from Cys474 to Pro488
and Cys695 to Pro708 indicate regions proposed to interact
directly with platelet glycoprotein lb. One proposed type IIA
mutation (Val551 Phe, orange circle) occurs in the region
of the type IIB mutations.
From Sadler, f.E. 1991. J Biol Chem 266:22777-22780.
The Regulation of Blood Coagulation
J:)
J. Evan Sadler, M.D., Ph.D. — Associate Investigator
Dr. Sadler is also Associate Professor of Medicine and of Biochemistry and Molecular Biophysics at
Washington University School of Medicine, St. Louis. He obtained his undergraduate degree in chemistry
from Princeton University. He then attended Duke University, where he received first his Ph.D. degree in
biochemistry with Robert Hill and then his M.D. degree. Following his internship and residency in
medicine at Duke University Medical Center, Dr. Sadler was a Hematology Fellow in the laboratory of Earl
Davie at the University of Washington, Seattle.
UNDER normal circumstances, blood clots oc-
cur only at sites of vascular injury, and un-
necessary clots are dissolved promptly. Inappro-
priate blood clots can cause devastating illness,
such as stroke and myocardial infarction. Abnor-
mal thrombosis also complicates many com-
mon diseases, including certain cancers and
infections.
In the blood, proteins and small cells called
platelets are required for clot formation. The en-
dothelial cells that line all blood vessels and cir-
culating white blood cells are not, however, pas-
sive bystanders in these reactions, but actively
promote or inhibit clotting. Compounds that are
produced during inflammation modulate these
cellular activities.
We are investigating the structure, function,
regulation, and evolution of proteins that control
blood coagulation. Our goal is to understand how
these opposing tendencies — to stimulate or to in-
hibit clotting — are balanced to achieve normal
hemostasis and prevent dangerous thrombosis.
These studies will increase our knowledge of the
interaction between blood coagulation and in-
flammation and may provide a foundation for the
design of new therapies for thrombotic disorders.
Studies of thrombomodulin, thrombin, and tissue
factor are supported by a grant from the National
Institutes of Health.
von Willebrand Factor
and von Willebrand Disease
The von Willebrand factor (vWF) is a blood
protein that is made by endothelial cells and is
required for normal platelet function. vWF also
binds to and stabilizes blood coagulation factor
VIII, the factor that is deficient in classical hemo-
philia. The structure of vWF was determined indi-
rectly by cDNA cloning: vWF contains 12 re-
peated domains that belong to four families of
ancestral sequences. Hereditary deficiency of
vWF, or von Willebrand disease, is the most com-
mon genetic bleeding disorder of humans. Mild
abnormalities of vWF function can be detected in
nearly 1 percent of the population.
We determined the structure of the human vWF
gene and also of a related pseudogene that has
diverged recently from the authentic vWF gene.
This allowed us to investigate von Willebrand
disease at the level of gene sequence. We charac-
terized deletions of the vWF gene that cause se-
vere von Willebrand disease in five unrelated pa-
tients. These particular patients treat transfused
vWF as a foreign protein and make inhibitory anti-
bodies to it. Deletions in the vWF gene predis-
pose to the formation of such antibodies.
These studies were extended to include pa-
tients with variants of von Willebrand disease
who make a defective vWF molecule. Among
more than 30 unrelated patients, 1 1 difi'erent
mutations were characterized within a single
exon of the vWF gene. Six of these mutations are
within a small vWF domain that appears to modu-
late the affinity of vWF for platelets; of these six
mutations, five cause a paradoxical increase in
binding and the sixth causes a decrease in bind-
ing. The remaining five mutations are within an
adjacent domain of the protein and cause loss of
function, either by impairing vWF biosynthesis or
increasing vWF degradation. The association of
severe bleeding with both increased and de-
creased function illustrates the importance of bal-
anced vWF function for normal hemostasis.
A recently described variant of von Willebrand
disease, recognized in a patient from Normandy,
France, is characterized by defective binding of
vWF to blood coagulation factor VIII. In such pa-
tients factor VIII is unstable, and this results in a
secondary factor VIII deficiency that mimics clas-
sical hemophilia. Among several unrelated af-
fected families, three diff^erent mutations were
identified in the factor Vlll-binding site of vWF.
The corresponding recombinant vWF proteins ex-
hibited the same defect in factor VIII binding as
natural vWF Normandy, confirming that these
mutations cause the disease. The genetic defects
of these and other such patients provide insight
into structure-function relationships of vWF and
may suggest new therapeutic strategies to inhibit
or augment vWF function.
345
The Regulation of Blood Coagulation
Thrombomodulin and Thrombin
Thrombomodulin is a protein of the endothe-
lial cell surface that binds thrombin, a blood clot-
ting en2yme. Because of its effects on thrombin
activity, thrombomodulin is an essential natural
anticoagulant. Several inflammatory mediators
decrease the expression of thrombomodulin by
endothelial cells. Understanding this process
may help us understand the abnormal blood coag-
ulation that accompanies much human disease.
We have cloned and expressed variants of recom-
binant human thrombomodulin in a variety of
cultured cell lines. These cells have been used to
define the structural requirements for thrombo-
modulin cofactor activity.
Thrombin is a protease enzyme that is required
to form blood clots from fibrinogen. Thrombin
also can inhibit clotting by first binding to throm-
bomodulin and then digesting certain other
blood coagulation factors. Disruption of throm-
bin's normal balance between promoting and in-
hibiting clotting can, in principle, cause either
bleeding or thrombosis. We have constructed
mutant forms of thrombin w^ith predominantly an-
ticoagulant or procoagulant activities. Such mu-
tant thrombins help to define the structural basis
for the different activities of thrombin. In addi-
tion, they provide reagents to test the physiologi-
cal importance of specific thrombin activities,
and mutant thrombins may also be found to
have therapeutic procoagulant or anticoagulant
properties.
Tissue Factor
Tissue factor, a cell surface protein that is
found on many cells that do not normally contact
the blood, is the most important physiological
initiator of blood coagulation. When blood ves-
sels are damaged, tissue factor is exposed to and
binds to blood coagulation factor VII. The factor
Vll-tissue factor complex then initiates a cascade
of reactions that cause blood to clot. Both mono-
cytes and endothelial cells express tissue factor
activity in response to many stimuli, and this con-
tributes to the abnormal thrombosis that accom-
panies systemic infections.
We have isolated cDNA clones for human tissue
factor and localized the gene to chromosome 1 .
These clones have been used to study the regula-
tion of tissue factor in endothelial cells. Tumor
necrosis factor, a protein made during inflamma-
tion, causes a dramatic but transient increase in
tissue factor activity. This appears to be the result
of activating the gene, which is normally silent in
endothelium. A second level of control can be
employed to amplify this response. Tissue factor
mRNA normally is degraded very rapidly, but it is
stabilized by some agents that stimulate endothe-
lial cells, and this may contribute to the induc-
tion of tissue factor activity during inflammation.
We are currently studying the structural ba-
sis for this regulation of tissue factor mRNA
degradation.
The systems we are investigating provide abun-
dant opportunities to answer biological ques-
tions concerning the regulation of blood coagula-
tion and to approach fundamental questions
related to signal transduction, gene expression,
and protein structure-function relationships.
These studies may illustrate how several proteins
can be coordinately regulated to promote blood
clotting reactions on the vascular endothelium
during inflammation. We will continue to ex-
plore the mechanisms by which vWF, thrombo-
modulin, thrombin, and tissue factor are regu-
lated and will extend this work to other
endothelial cell proteins that can promote or in-
hibit thrombosis.
346
Molecular Mechanism of Transmembrane Signal
Transduction by G Protein-coupled Receptors
Thomas p. Sakmar, M.D. — Assistant Investigator
Dr. Sakmar is also Assistant Professor at the Rockefeller University. He received his A.B. degree in
chemistry and his M.D. degree from the University of Chicago. He completed a medical residency at
Massachusetts General Hospital, Boston, and conducted postdoctoral research in the laboratory
of H. Gobind Khorana at the Massachusetts Institute of Technology.
IN our laboratory the vertebrate visual proteins
rhodopsin and transducin serve as a model sys-
tem for structure-function studies on the molecu-
lar mechanism of transmembrane signaling.
These visual proteins are members of a superfam-
ily of related G proteins (guanine-binding regula-
tory proteins) and G protein-coupled receptors.
Light-activated rhodopsin catalyzes guanine nu-
cleotide exchange by transducin, v^^hich ulti-
mately leads to a change in membrane cation
conductance and a neural signal. Our approach is
to reconstitute heterologously expressed rho-
dopsin and transducin in defined in vitro systems
and to use biochemical and biophysical methods
to probe site-directed mutants.
Our current interests include structure-
function relationships in rhodopsin. For exam-
ple, we are studying the ground state structure of
the receptor, the interactions between specific
amino acid residues and the 1 l-c?s-retinal chro-
mophore that control spectral properties and
photochemistry, the mechanism by which a pho-
tochemical signal is transmitted from the core of
the receptor to the surface, and the specific do-
mains on the cytoplasmic surface that bind and
activate transducin.
We employ a multifaceted approach, including
the use of a variety of complementary spectro-
scopic techniques. For example, the structure
and environment of the retinal chromophore in
rhodopsin and its photointermediates can be stud-
ied with resonance Raman spectroscopy. In a con-
tinuation of collaborative work with Steven Lin
and Richard Mathies, we obtained microprobe
resonance Raman spectra of solutions containing
only microgram quantities of mutant pigments.
The results confirmed and supplemented our
earlier observations concerning the role of a spe-
cific carboxylate group in rhodopsin that acts to
stabilize the positive charge of the protonated
Schiff base chromophore linkage. A model of the
chromophore binding pocket of rhodopsin was
proposed and is being used to direct further stud-
ies into the mechanism of wavelength regulation
by visual pigments.
We are also interested in identifying specific
domains of rhodopsin and transducin involved in
binding and activation. Flash photolysis studies
of site-directed rhodopsin mutants had previ-
ously shown that at least the second and third
cytoplasmic loops of rhodopsin are involved in
activation of bound transducin. Some cytoplas-
mic mutations prevent transducin binding as
well. Recently we have developed a spectrofluo-
rimetric method designed to allow simultaneous
illumination and measurements of rhodopsin-
transducin interactions by intrinsic fluorescence.
Rhodopsin-catalyzed binding of GTP, or a GTP
analogue to transducin, results in a large increase
in its intrinsic fluorescence. Mixtures of transdu-
cin and rhodopsin can be assayed by this method
to determine the kinetic rate constants of their
interaction and to evaluate the specific effects of
mutations. A series of site-directed mutants of
rhodopsin with alterations in their cytoplasmic
domains have been studied. The results may be
relevant to other seven-transmembrane helix re-
ceptors that couple to G proteins and play roles
in cellular physiology, growth, development,
and differentiation in the nervous system.
347
i
I
1
Molecular Genetics of Development
in Drosophila
Shigeru Sakonju, Ph.D. — Assistant Investigator
Dr. Sakonju is also Assistant Professor of Human Genetics at the University of Utah School of Medicine. He
received a B.A. degree from Columbia Union College and a Ph.D. degree in biology from the Johns Hopkins
University, having developed his doctoral thesis in the Department of Embryology at the Carnegie
Institution of Washington, Baltimore, with Donald Brown. He was a Helen Hay Whitney Postdoctoral
Fellow with E. B. Lewis at the California Institute of Technology and at Stanford University
with David Hogness.
DURING the development of organisms, the
fertihzed egg undergoes many divisions to
produce a mukicellular body. In the fruit fly Dro-
sophila melanogaster, the body is made up of
several fused segments in the head, 3 thoracic
segments with wings and legs, and 10 abdominal
segments, each showing unique characteristics.
This basic pattern of body segments, invariant
from generation to generation, is dictated by a
genetic blueprint within the organism's own
genome.
The characteristics, or identity, of each body
segment are determined by the activities of the
so-called homeotic genes. When these do not
function properly, a body segment or group of
segments transforms to take on the characteristics
of another segment. Thus homeotic genes can be
thought of as master regulatory switches that trig-
ger the genetic circuits necessary to form normal
body patterns. Genes similar to the homeotic
genes of the fruit fly are found in other organisms,
including humans. The goal in my laboratory is to
understand how the homeotic genes accomplish
their task at the molecular level.
In Drosophila, three homeotic genes — Ultra-
bithorax (JJbx), abdominal- A (abd-A), and
Abdominal-B (Abd-E) — are responsible for de-
termining the characteristics of two thoracic and
nine abdominal segments. These genes are lo-
cated in a chromosomal region called the bitho-
rax complex (BX-C) . Flies carrying mutations in
any one of the three BX-C genes show characteris-
tic transformations of body segments. By observ-
ing which segments are transformed in these mu-
tants, we know that Ubx is required in two
thoracic and eight abdominal segments, abd-A in
the second through eighth abdominal segments,
and Abd-B in the fifth through ninth abdominal
segments.
To learn what homeotic proteins do in the cell,
we have focused on the proteins encoded by Ubx
and abd-A. Many homeotic proteins, including
these, have been shown to bind DNA in vitro,
suggesting that they act by binding to transcrip-
tional signals of other genes to regulate expres-
sion. Whether this is how they act in living organ-
isms has not been directly shown. Therefore we
have chosen to study the regulation of a potential
target gene of Ubx and abd-A proteins, called An-
tennapedia (Antp), as a paradigm for the mecha-
nism of homeotic protein actions.
We have shown that both Ubx and abd-A pro-
teins bind to a number of sites on the DNA seg-
ment that contains the signals necessary for tran-
scription of the Antp gene. We have further
shown, by creating mutations in the binding sites
and assaying their effects, that this binding is es-
sential for repressing the Antp gene expression in
the embryo. Our study therefore provides direct
evidence that homeotic proteins do turn on or oflF
transcription of their target genes.
What are the target, or downstream, genes regu-
lated by homeotic proteins? Little is known about
potential target genes, even though many are
thought to exist. To identify such targets, we have
utilized a method, called the enhancer trap, of
detecting genes with expression patterns that
suggest homeotic gene regulation. A number of
candidates have been isolated, and we are testing
them for further evidence of being the down-
stream genes. We have shown that the expression
of several candidate genes is abnormal in the mu-
tant embryos that do not produce Ubx protein,
suggesting that the protein does regulate this
gene.
We are also interested in determining how the
homeotic genes with a relatively small number of
encoded proteins can specify many unique body
segments. One answer lies in the fact that these
genes are expressed in different but overlapping
sets of segments. For example, of the three ho-
meotic proteins from the BX-C, only Ubx protein
is detected in the second and third thoracic seg-
ments; both Ubx and abd-A proteins are present in
the second through fourth abdominal segments;
and all three BX-C proteins are found in the fifth
through eighth abdominal segments.
These three combinations would of course de-
fine only three segment identities if all cells
within segments were expressing the same combi-
nation of the homeotic genes. In fact, cells within
a segment do not express the same combination.
349
Molecular Genetics of Development in Drosophila
We find that some cells within a given abdominal
segment express abd-A protein exclusively or pre-
dominantly; other cells, Ubx protein; and still
other cells, Abd-B protein. This kind of "mosaic"
expression can, in theory, specify an unlimited
number of segmental identities. We have there-
fore asked the following question: Is the mosaic
-expression necessary for correct specification of
these identities?
To answer this question, we altered flies geneti-
cally so that homeotic proteins could be ex-
pressed at will. When Abd-B protein is expressed
in all cells, thoracic and abdominal segments ex-
hibit some characteristics of the eighth abdomi-
nal segment. When both Ubx and Abd-B proteins
are expressed simultaneously, thoracic segments
transform to the first abdominal segment, which
is specified by Ubx, but abdominal segments re-
main largely unaltered. Therefore, surprisingly,
we do not observe completely nonsensical iden-
tities, as might have been expected from this ab-
normal situation, suggesting that mosaic expres-
sion is not required, at least at a gross level, for
correct specification of segments.
350
Generating a Repertoire of Antigen-Specific
Receptors
W9
David G. Schatz, Ph.D. — Assistant Investigator
Dr. Schatz is also Assistant Professor of Immunobiology at Yale University School of Medicine. He received
undergraduate degrees in molecular biophysics and biochemistry from Yale University and in philosophy
and politics from Oxford University. His Ph.D. degree and postdoctoral training were done with David
Baltimore at the Massachusetts Institute of Technology and the Whitehead Institute.
CELLS of the immune system act in concert to
protect against infectious agents and trans-
formed (malignant) cells. At the heart of this pro-
tective system are the antigen receptor molecules
found on B and T lymphocytes: the immunoglob-
ulin (Ig) and the T cell receptor (TCR). Each
lymphocyte expresses a distinct receptor mole-
cule that confers on the cell a unique antigen
specificity. The millions of different genes
needed to encode these receptors are assembled
from component gene segments by a site-specific
process known as V(D)J recombination — so
named for the V (variable) , D (diversity) , and J
(joining) gene segments used in the reaction.
V(D)J recombination is critical for the develop-
ment of B and T lymphocytes and is the only site-
specific recombination process know^n to occur
in vertebrates. We are interested in two funda-
mental questions concerning such recombina-
tion: What is the biochemical mechanism of the
reaction, and what molecular mechanisms regu-
late the reaction during lymphoid development?
This recombination reaction has been inten-
sively studied since its discovery in 1976, yet lit-
tle was known by the late 1980s about the enzy-
matic machinery (recombinase) that carried it
out. Particularly frustrating was the inability to
identify the gene or genes encoding the V(D)J
recombinase, despite a detailed understanding of
the substrates and products. While working with
David Baltimore, I developed a novel genetic ap-
proach to the identification of these genes. Using
a highly sensitive assay for V(D)J recombinase ac-
tivity, I was able to demonstrate that gene transfer
(transfection of genomic DNA) could activate the
V(D)J recombinase in nonlymphoid cells, a sur-
prising result implying that a single genetic locus
was sufficient for the activation.
Marjorie Oettinger and I then isolated this ge-
netic locus and quickly encountered a second
surprise. The locus contained not one but rwo
genes, which functioned together to activate the
V(D)J recombinase. Indeed, a mixture of these
recombination-activating genes, called RAG-1
and RAG- 2, was thousands of times more potent
than was either gene alone. We then went on to
demonstrate that the two genes are only coex-
pressed in developing lymphocytes — in exactly
those cells that are assembling Ig and TCR genes.
This and a variety of other data suggest that RAG- 1
and RAG- 2 encode the critical, lymphoid-specific
components of the V(D)J recombinase. Interest-
ingly, we found that RAG- 1 , but apparently not
RAG- 2, is transcribed in neurons in the central
nervous system, raising the possibility that RAG- 1
has important roles in processes other than clas-
sic V(D)J recombination.
A central goal in our current research is to un-
derstand the enzymatic mechanism of V(D)J re-
combination, an elusive goal thus far, largely be-
cause efforts to reconstitute the reaction in a
cell-free system have been unsuccessful. As a first
step toward this goal, we are studying the bio-
chemical and enzymatic properties of the RAG- 1
and RAG-2 proteins. Our initial focus is on devel-
oping the necessary reagents — in particular,
highly purified preparations of the proteins and
antibodies that specifically recognize them.
We will use these tools to examine the bio-
chemical properties of the RAG proteins, asking
whether they exhibit the activities expected of
proteins involved in recombination (e.g., topo-
isomerase, endonuclease, exonuclease, or ligase
activities) . We are also interested in determining
if the RAG proteins bind to DNA, either nonspecif-
ically or by interacting with elements of the
V(D)J "recombination signal sequence" [the
DNA element required for V(D)J recombination].
In addition, we hope to ascertain whether the
RAG-1 and RAG-2 proteins interact with one an-
other or with other proteins and how these inter-
actions vary during lymphoid development and
among different cell types.
Our second major focus is to understand the
molecular mechanisms that regulate V(D)J re-
combination during lymphoid development. The
assembly of Ig and TCR genes is a complex and
highly ordered process regulated in part at the
level of expression of the V(D)J recombinase. Us-
ing RAG-1 and RAG-2 as indicators, we are study-
ing when and how the recombinase is turned on
and then off again as B and T cells develop. Ex-
periments performed in collaboration with Craig
351
Generating a Repertoire of Antigen-Specific Receptors
Thompson (HHMI, University of Michigan) and
Larry Turka suggest that signals transduced
through membrane-bound TCR molecules may
play a role in mediating the termination of
V(D)J recombinase expression during T cell
development.
V(D)J recombination is also regulated at the
ievel of the availability, or accessibility, of the
gene segments to the V(D)J recombinase . Accessi-
bility in turn appears to be mediated by, or paral-
leled by, transcription of the unrearranged
("germline") gene segments. Analysis of germ-
line transcription during B cell development has
yielded important insights into the relationship
between transcription and accessibility but has
left unaddressed important questions concerning
T cell development. We are interested both in
how a stem cell (a cell with the potential to de-
velop into multiple cell types) becomes commit-
ted to the T cell lineage and how the different
sublineages of T cells are established. Since the
regulation of V(D)J recombination is interwoven
with these developmental decisions, we are ex-
amining the structure and developmental profile
of germline TCR gene transcripts. We hope to de-
termine the sequence of molecular events that
lead to the assembly of TCR genes and to ask
whether defects in this process contribute to the
pathogenesis of immunological disease, particu-
larly autoimmunity.
352
Intracellular Protein Transport
Randy W. Schekman, Ph.D. — Investigator
Dr. Schekman is also Professor of Biochemistry and Molecular Biology at the University of California,
Berkeley, and Adjunct Professor of Biochemistry and Biophysics at the University of California, San
Francisco. As a graduate student, he studied the enzymology ofDNA replication with Arthur Kornberg at
Stanford University. His current interest in cellular membranes developed during a postdoctoral period
with S. J. Singer at the University of California, San Diego. At Berkeley, he developed a genetic approach
to the study of eukaryotic membrane trafftc. Among his awards is the American Society for Microbiology
Eli Lilly Award in Microbiology and Immunology. Dr. Schekman was recently elected to the National
Academy of Sciences.
RESEARCH in our laboratory is devoted to mo-
lecular description of the processes of poly-
peptide translocation from the cytosol into the
endoplasmic reticulum (ER) and of vesicular
transport among organelles of the secretory
pathway.
Genetic and Biochemical Dissection
of the Secretory Process
A genetic approach to the study of eukaryotic
protein transport involved the isolation of condi-
tional mutants. We isolated a series of secretory
{sec) mutants in the yeast Saccharomyces cerevi-
siae that are temperature-sensitive for cell sur-
face growth, division, and secretion. Most of the
mutants accumulate secretory proteins in an in-
tracellular pool that can be released when cells
are returned to a permissive temperature. More
than 30 gene products have been implicated in
the process of delivering membrane and secre-
tory proteins to the cell surface.
A combined genetic and cytologic evaluation
of the sec mutants has allowed a description of
the secretory pathway. Protein transport in yeast
appears to be mediated by the same organelles
and proteins that operate in mammalian cells. Mo-
lecular cloning analysis of SEC genes has re-
vealed striking structural and functional homol-
ogy with corresponding mammalian genes.
We have developed biochemical assays that
measure the early events of polypeptide translo-
cation into the ER and of vesicle-mediated pro-
tein transport from the ER to the Golgi apparatus.
The cell-free reactions represent physiologically
meaningful processes. Extracts prepared from
mutant cells reproduce the temperature-sensitive
defects observed in vivo. In favorable circum-
stances the mutant defects are repaired by addi-
tion of a protein fraction obtained from wild-type
yeast, and such restoration of transport activity
may be used to purify functional forms of SEC
gene products.
Protein Translocation Across Membranes
Protein translocation into the lumen of the en-
doplasmic reticulum represents the initial step in
assembly of the eukaryotic cell surface. This pro-
cess has been reconstituted with detergent-solubi-
lized membrane proteins and purified cytosolic
proteins, yet the mechanism of polypeptide pen-
etration is unclear. We have isolated mutants that
are defective in translocation, using a genetic se-
lection that requires secretory polypeptides to be
retained in the cytosol. The work on these mu-
tants is supported by a grant from the National
Institutes of Health and will not be described
here.
An independent line of investigation concerns
the mechanism of protein translocation from the
cytosol into the vacuole. The vacuole contains an
array of hydrolytic enzymes and is believed to
play an important role in the degradation of cyto-
solic proteins and intracellular membranes. The
mechanism for importing substrates into the vacu-
ole has not been evaluated.
Last year we reported a novel pathway for the
localization and degradation of fructose 1,6-bis-
phosphatase (FBPase), a key regulatory enzyme
of gluconeogenesis. FBPase is localized to the cy-
tosol when cells are grown on a poor carbon
source, such as ethanol. When cells are trans-
ferred to glucose, FBPase is degraded in a process
called catabolite inactivation, which depends on
active vacuolar proteases. In a protease-deficient
strain, FBPase enters the vacuole and remains in-
tact. Import into the vacuole was shown to de-
pend on protein synthesis during the period of
transfer to glucose medium. In addition, vacuolar
import requires the transfer of a protein, possibly
an import receptor, via the secretory pathway.
The mechanism of this new import pathway is
being pursued by the isolation of mutants defec-
tive in the degradation of FBPase. Thus far a large
number of genes have been identified that are
required for import of FBPase into the vacuole.
Preliminary evidence suggests that the import
353
Intracellular Protein Transport
may be mediated by an independent organelle,
the peroxisome. Mutations that block peroxiso-
mal assembly also block FBPase degradation. Fur-
thermore, a peroxisomal enzyme, acyl-CoA thio-
lase, is subject to the same glucose-stimulated,
vacuolar-dependent degradation process. One
possible explanation of these results is that
FBPase may be imported into or somehow asso-
ciated with peroxisomes and then localized to
the vacuole by autophagic uptake of the
peroxisome.
Vesicle Transport Early in the Secretory
Pathway
Subsequent stages in the secretory pathway in-
volve protein sorting and transport from the en-
doplasmic reticulum to the Golgi apparatus and
from there to the cell surface. Genes required for
each of these steps are being evaluated by molec-
ular cloning and by development of cell-free reac-
tions that measure individual steps in the trans-
port process. An assay that depends on Sec
proteins has been reconstituted in vitro. Yeast
a-factor precursor is translocated into the ER lu-
men of gently lysed yeast spheroplasts. In the
presence of soluble proteins and ATP, the precur-
sor is transferred to the Golgi apparatus. This sys-
tem allows the purification and functional charac-
terization of Sec proteins.
Transfer of secretory proteins from the ER to
the Golgi apparatus is mediated by small vesicle
carriers. In this limb of the pathway, defective sec
mutants fall into two categories: class I, mutant
cells that accumulate ER tubules at a restrictive
temperature {SEC 12, -13, -16, and -23); and
class II, mutant cells that also accumulate several
thousand 60-nm vesicles (SEC17, -18, and -22).
Genetic epistasis tests indicate that class I
genes must execute their function prior to class II
genes. This implies that class I products partici-
pate in the production of the 60-nm vesicles that
are consumed, by fusion with the Golgi appara-
tus, through the action of class II gene products.
Genetic interactions among members of class I
and II genes suggest that the Sec proteins in each
group act in a complex, or at least in a concerted
manner, to perform their respective roles in vesi-
cle budding or fusion.
Transport of the a-factor precursor in vitro is
mediated by diffusible vesicles. Transport vesi-
cles contain a core-glycosylated precursor and
are physically separable from donor ER and target
Golgi membranes. Budding of vesicles from the
ER requires a crude cytosol fraction, ATP,
Secl2p, Secl3p, Sec23p, GTP, and a ras-like
GTP-binding protein, Sarlp. Fusion of the vesi-
cles to the Golgi compartment is measured by
conversion of the precursor to a more highly gly-
cosylated form. Enriched transport vesicles target
to the Golgi compartment and then fuse in dis-
tinct subreactions that require cytosol, Ca^"^, ATP,
and only a subset of Sec proteins.
To allow the purification of functional Sec23p,
we have developed an assay based on restoration
of transport in sec23 lysates by wild-type protein
fractions. The purified protein has been isolated
in two forms: a Sec23p monomer, and a hetero-
oligomer that consists of Sec23p and a 105-kDa
protein (Sec24p) that is also required for vesicle
budding. The SEC 2 3 and -24 genes have both
been cloned, and though the sequences confirm
the observed molecular weights of the isolated
polypeptides, no homology to known proteins
was found.
Localization of Sec proteins in yeast has met
with only limited success. Yeast morphologic
analysis is limited by the small cell size and the
difficulty of specimen preservation. Fortunately
the functional and structural conservation of Sec
proteins allows their localization in mammalian
cells. The first and most favorable example is
Sec23p, where antibodies directed against the
yeast protein cross-react with the mammalian ho-
mologue. Immunolocalization studies reveal a
striking enrichment of Sec23p in the cytoplasmic
pocket that separates the transitional ER and the
cis face of the Golgi complex. This location is
completely consistent with the role proposed for
yeast Sec23p and further confirms that the secre-
tory pathway is fundamentally conserved across
the broad spectrum of the eukaryotic kingdom.
354
Development and Function of the Synapse
Richard H. Scheller, Ph.D. — Associate Investigator
Dr. Scheller is also Associate Professor of Molecular and Cellular Physiology and Associate Professor of
Biological Sciences (by courtesy) at Stanford University. He received his B.S. degree from the University
of Wisconsin- Madison and his Ph.D. degree in chemistry from the California Institute of Technology.
Dr. Scheller was a postdoctoral student with Richard Axel and Eric Kandel at Columbia University.
THE nervous system is composed of large num-
bers of unique cells that communicate with
each other via the regulated release of chemical
neurotransmitters. These synaptic interactions
govern animal behavior, and modulation of the
efficacy of synaptic communication is thought to
underlie learning and memory. We are interested
in understanding the molecular mechanisms of
synaptic formation during development and re-
generation in the peripheral nervous system after
nerve injury. It is also our goal to contribute to an
understanding of how the nerve terminal func-
tions in the regulation of neurotransmitter
release.
Processing and Packaging of Neuropeptides
Many synapses release two types of chemical
messengers: fast-acting, or classical transmitters,
and slower-acting messengers, or neuromodula-
tors. Most of the various chemicals used as mes-
sengers in the brain are neuropeptides. These
molecules are synthesized as larger precursors
that are processed to smaller active peptides. One
interesting neuropeptide precursor is expressed
in an identified set of neurons, the bag cells, in
the marine snail Aplysia. When these neurons
fire, they release a set of neuropeptides derived
from a single precursor. These peptides act on
neurons and peripheral tissues to regulate egg
laying, a stereotyped behavior.
Interestingly, the peptides produced on the
egg-laying hormone (ELH) precursor are pack-
aged in two types of vesicles. These vesicles con-
tain different sets of peptides and are differen-
tially localized within the neurons. We are
interested in understanding how the peptides ini-
tially synthesized on a single precursor are sorted
into different vesicles. We are also interested in
understanding the physiological significance of
the differential packaging and localization.
When the ELH precursor is transfected into
mammalian pituitary tumor cells (AtT-20 cells),
ELH is packaged with the endogenous hormone.
The amino-terminal region of the precursor is de-
graded within the secretory cell, probably in the
lysosomes. Thus the AtT-20 cells, like the bag
cells, differentially route the two regions of the
ELH prohormone. Mutating the first cleavage site
from a set of four basic residues to two basic resi-
dues results in constitutive secretion of the
amino-terminal region of the precursor, not intra-
cellular degradation. These results suggest that
the subcellular location of the first endoproteoly-
tic processing event is critical in determining the
routing of the processing intermediates.
Mechanisms of Synaptic Transmission
When the action potential travels down the
nerve and enters a release zone, changes in the
membrane potential open channels that allow cal-
cium to enter the cell. The calcium promotes
transmitter release and membrane fusion. The
membrane then recycles, forming new vesicles,
which are then replenished with chemical trans-
mitter. This cycle might be considered the funda-
mental process that underlies nervous system
function, yet little is known about the molecular
mechanisms involved. In an attempt to define the
molecular mechanisms that regulate membrane
flow in the nerve, our laboratory and others have
begun to characterize the proteins associated
with the critical organelle in the process, the syn-
aptic vesicle. For these studies, we use mamma-
lian brain and the electric organs of marine rays.
These electric organs have a concentration of syn-
apses approximately 100-fold higher than that of
skeletal muscle. In addition, these synapses are
homogeneous; they all use the neurotransmitter
acetylcholine.
Purified synaptic vesicles contain about 20-50
protein bands when fractionated on acrylamide
gels. Genes encoding many of these proteins have
been characterized and the primary sequence of
the molecules determined. Some of the proteins
show interesting homologies to other molecules,
and others are turning out to have counterparts in
yeast where genetic studies of membrane traffick-
ing have provided insight into the secretory pro-
cess. It has also become apparent that many of the
synaptic vesicle proteins are members of small
gene families. Individual members of these gene
families are differentially expressed through the
355
Development and Function of the Synapse
brain, resulting in a variety of combinations of
these molecules on different vesicles.
Molecular fractionation and immunoprecipita-
tion techniques suggest that several of the synap-
tic vesicle proteins interact to form a large multi-
meric complex. This complex contains a
previously uncharacterized 35-kDa protein that
is a major substrate for casein kinase. We have
purified this protein and isolated cDNAs encod-
ing the molecule. The protein is unique in the
database and is predicted to be anchored in the
membrane by a hydrophobic carboxyl terminus
and to have an amino-terminal region oriented
toward the cytoplasm. Like the synaptic vesicle
proteins, this molecule is a member of a small
gene family. Experiments are directed tow^ard un-
derstanding the localization and function of this
protein and the other synaptic vesicle proteins.
Synapse Development
Motor neurons in the spinal cord send axons to
muscle fibers throughout the body. When axons
contact muscle fibers, a highly ordered structure
consisting of a presynaptic nerve terminal and a
postsynaptic site develops. The presynaptic ter-
minal comprises an active zone rich in synaptic
vesicles containing neurotransmitter. The post-
synaptic element is made up of a membrane rich
in receptors for the neurotransmitter and an in-
dentation in the membrane called the junctional
fold. An extracellular matrix, or basal lamina,
surrounds the muscle fiber, including the space
between the nerve and muscle.
One of the key events in the development of
the neuromuscular junction is the redistribution
of neurotransmitter receptors that occurs when
nerve contacts muscle. Initially receptors for the
neurotransmitter, in this case acetylcholine, are
randomly distributed on the muscle fiber. When
the nerve contacts muscle, neurotransmitter re-
ceptors aggregate under the nerve terminal in an
appropriate position to detect the chemicals re-
leased during synaptic transmission.
Agrin, a component of the extracellular matrix,
causes acetylcholine receptors to cluster when
added to muscle fibers growing in culture. We
have isolated recombinant DNA clones encoding
agrin molecules and, through an analysis of the
nucleotide sequence, have defined the primary
amino acid sequence of the molecule. When we
compare the predicted agrin sequence with the
proteins in the data bank, two types of similarities
are revealed. The first is to a class of molecules
that inhibit proteases and the second to a protein
motif called EGF (epidermal growth factor) re-
peats. The gene is expressed in embryonic motor
neurons at the time they are first contacting mus-
cle fibers. Two regions of the agrin gene are alter-
nately spliced and may produce up to eight forms
of the molecule.
Expression of agrin encoding cDNAs in CHO
and COS cell lines results in the association of the
protein with the surface of the transfected cells,
probably through assembly into an extracellular
matrix. Coculture of agrin-expressing cells with
primary muscle fibers or a C2 myoblast cell line
results in aggregation of acetylcholine receptors
at sites of contact between the transfected cell
and the muscle fibers. Only the forms of the pro-
tein containing an eight-amino acid sequence,
which is the product of alternate RNA splicing,
are capable of causing clusters on S27 cells, a
mutant C2 line lacking proteoglycans. These data
suggest that agrin may cluster receptors via two
mechanisms, one of which is proteoglycan de-
pendent. Another possibility is that the eight
amino acids provide a binding site that results in a
high-affinity interaction, overriding the need for
the proteoglycan component.
Since agrin is stably maintained in the synaptic
basal lamina after nerve or muscle damage, it may
also play a role in regeneration events. Under-
standing the mechanisms of peripheral synapse
regeneration may lead to procedures that could
aid in central nervous system regeneration.
356
Molecular Pathogenicity Studies
of Enteric Bacteria
Gary K. Schoolnik, M.D. — Associate Investigator
Dr. Schoolnik is also Associate Professor of Medicine and of Microbiology and Immunology at Stanford
University School of Medicine. He received his M.D. degree from the University of Washington. He was an
intern, resident, and chief resident in internal medicine at Massachusetts General Hospital, a fellow in
infectious diseases with King Holmes and Thomas Buchanan at the University of Washington, and a
postdoctoral fellow with Emil Gotschlich at the Rockefeller University. He founded the Division of
Geographic Medicine at Stanford University and has established a research center for the study of
infectious diseases in southern Mexico.
BACTERIAL, viral, and parasitic infections of
the gastrointestinal tract cause an estimated
500 million illnesses and 5 million deaths each
year among children living in the developing
countries. The principal mission of our labora-
tory is to discover how these infectious agents
cause disease, how they are spread, and how this
information can lead to new tactics for preven-
tion and control. This effort has entailed work in
two settings: molecular studies in our laboratory
at Stanford University, and epidemiological in-
vestigations at a field laboratory in southeastern
Mexico, where infections of this kind are
common.
In the first setting, the unit of analysis is the
organism itself. We try to determine how it at-
taches to, invades, and damages human cells. This
anthropomorphic orientation views the disease-
causing capacity of an infectious agent as the cen-
tral research issue. In the second setting, the unit
of analysis is a household or a village. In this con-
text we seek to understand how the organism is
transmitted within the community, what its reser-
voirs are, and how it manages to survive as a via-
ble entity in the real world.
This ecological orientation seeks to understand
how the organism adapts to different environmen-
tal habitats. Within the context of a Third World
village, these habitats include contaminated food
and well water, sewage, and the gastrointestinal
tracts of people and animals. By using biochemi-
cal and genetic tools to study infectious agents as
they inhabit and move among these different envi-
ronmental niches, we have begun to understand
the underlying molecular mechanisms for this re-
markable capacity. This in turn is beginning to
lead to new strategies for the control of these dis-
eases through vaccination and epidemiological
interventions. Examples of work in progress are
described below.
Enteropathogenic Escherichia coli
Enteropathogenic E. coli (EPEC) are a common
cause of infantile diarrhea in Third World chil-
dren. When biopsies of the small intestine are
performed, colonies of EPEC are found attached
to the underlying epithelia. It is evident that the
bacteria interact not only with the host cells to
which they are bound but also with each other.
Beneath these adherent colonies, structural
changes in the host cell also occur, indicating
that EPEC have altered the absorptive surface of
the intestinal cell. It is now clear that structural
changes of this kind are directly responsible for
the diarrheal syndrome.
From these studies we have learned that EPEC's
pathogenic strategy consists of at least three dis-
tinct steps, which probably occur in the follow-
ing order: the coalescence of individual bacteria
into infectious units composed of several to
hundreds of organisms (an event that probably
occurs in the small intestine, soon after ingestion
of the bacteria); attachment of these infectious
units to the surface of intestinal epithelial cells;
and, following close contact between the bacte-
ria and the epithelial cells, the organism's partial
penetration of the cell, in association with con-
comitant rearrangement of its cytoskeleton.
Each of these steps is performed by a distinct
surface structure of the bacteria. First, the bacte-
ria are organized into infectious units through the
production of rope-like appendages that emanate
from the organism's surface. Termed "bundle-
forming pili," these appendages create a network
within which the bacteria become enmeshed.
Then the bacteria bind to human intestinal epithe-
lial cells through the activity of rod-like filaments
that project from the organism like spines from a
porcupine. These filaments bind fibronectin mol-
ecules that are located around the periphery of
each intestinal epithelial cell. Finally, an outer
membrane protein of the organism interacts with
integrin-like molecules of the epithelial cell, an
event that triggers changes in the architecture of
the cell's cytoskeleton.
The production of at least one of these surface
structures is controlled by physiochemical sig-
nals that operate in the intestinal lumen, where
they serve as unique signatures of that habitat.
Thus, when under the influence of these signals,
357
Molecular Pathogenicity Studies of Enteric Bacteria
EPEC produce bundle-forming pili and coalesce
as infectious units. In contrast, when in inani-
mate environmental reservoirs such as water or
sewage, their production is repressed and they
exist as single, well-separated organisms. This
phenomenon, together with the findings de-
scribed above, shows that EPEC exemplify two
features of bacterial pathogenesis: first, that
pathogenicity is often the consequence of several
distinct virulence determinants of the organism
acting together and, second, that the expression
of these determinants may be regulated by local
features of the organism's environment.
We are now attempting to use these findings to
design an EPEC vaccine for the prevention of diar-
rheal illnesses in children. We hope to delete
from the EPEC chromosome the gene that causes
functional abnormalities of the host cell while
retaining the genes that direct the cell attachment
capacity and the production of bundle-forming
pili. A genetically modified strain of this kind
should be attenuated with respect to its viru-
lence, yet fully capable of colonizing the small
intestine, where it would be expected to stimu-
late a protective immune response.
Using Attenuated Strains of the Typhoid
Fever Bacillus as Vaccines
Fully virulent strains of Salmonella typhi
cause typhoid fever, whereas an attenuated strain
of the same species, administered as a living oral
vaccine, is now FDA approved and widely used
for the prevention of typhoid fever. Attenuated
strains of this kind have been proposed as vehi-
cles that might be able to carry other vaccine sub-
stances— termed "passenger antigens" — de-
rived from a variety of microorganisms. Such a
vaccine would stimulate immunity not only to S.
typhi, but to other infectious agents as well. We
have been studying this system for the delivery of
T cell epitopes.
For the purposes of this study, T cell epitopes
are considered to be localized regions of a pro-
tein that stimulate a cellular immune response
and are required for the elimination of intracellu-
lar pathogens, including some bacteria and many
viruses and protozoan parasites. We have found
that T cell epitopes can be effectively processed
and presented to the immune system when ex-
pressed in the center of the flagellin protein of S.
typhi. Flagellin is the protein building block of a
whip-like extracellular filament that acts like a
motor to propel the bacterium through liquids.
Because these vaccine strains would express a fla-
gellin protein that would also contain a foreign T
cell epitope, they are referred to as "chimeric"
flagellins.
We have taken advantage of recent information
about how epitopes of this kind are processed by
cellular components of the immune system. It is
now clear that one common pathway would en-
tail the presence of the chimeric flagellin in the
endosome of a macrophage. In that specialized
compartment of the cell, it would be digested by
endosomal proteolytic enzymes, resulting in the
release of the epitope as a small peptide prior to
its presentation to other components of the im-
mune system. A T cell epitope was placed in the
flagellin protein flanked at either end by amino
acids known to be sites of cleavage by endosomal
proteolytic enzymes. This has yielded a five- to
eight-fold increase in the magnitude of the result-
ing immune response.
Further modifications have entailed placing
the expression of these chimeric flagellins under
the control of the heat-shock protein promoter,
which is known to be activated in Salmonella
growing in macrophages. Other modifications in-
clude the use of multiple tandem copies of these
epitopes and the use of flagellins that are secreted
as soluble proteins into the macrophage endo-
some. Taken together, these studies have demon-
strated that it is possible to use attenuated strains
of S. typhi to deliver T cell epitopes and that the
effectiveness of this delivery system can be en-
hanced by the design features described above.
We are now working in collaboration with sci-
entists at the National Institutes of Health to place
T cell epitopes of a human immunodeficiency
virus (HIV) protein in Salmonella flagellins,
with a view to preparing useful vaccines for the
prevention of AIDS.
358
HEp-2 cells infected with Escherichia coli express the inv gene from the invasive bacterial pathogen
Yersinia enterocolitica. Data were recorded from a computer-controlled confocal laser microscope
and processed on a graphics display terminal. The HEp-2 cells (green) were stained with the
actin-speciflc stain FITC-phallicidin, while the invasin- bearing bacteria (red) were visualized by
indirect immunofluorescence using antisera against E. coli outer-membrane proteins. Areas of
extensive overlap appear yellow. Results indicate that actin, a major component of the cytoskele-
ton that controls cell shape and movement, accumulates in the region of entering bacteria.
From Young, V.B., Falkow, S., and Schoolnik, G.K. 75>i)2. J Cell Biol 116:197-207, by copyright
permission of the Rockefeller University Press.
359
Stereo pairs showing a three-dimensional view of in situ hybridization signals to two neighboring
genes. DNA from the white gene (seen here in green ) and from the Notch gene ( red ) were hybrid-
ized to the giant chromosomes from Drosophila salivary glands, fluorescently labeled, and visual-
ized by wide-field three-dimensional optical sectioning microscopy. Top picture shows these sig-
nals overlaid on the chromosome (blue ); bottom, the signals alone. The structure of the two genes
is seen to be distinctly different. Scale bar = 1 fim.
Research and photograph by Susan Parmelee in the laboratories of fohn Sedat and David
Agard.
360
Three-Dimensional Structure of Eukaryotic
Chromosomes
John W. Sedat, Ph.D. — Investigator
Dr. Sedat is also Professor of Biochemistry and Biophysics at the University of California, San Francisco.
He received his Ph.D. degree in biology from the California Institute of Technology. His postdoctoral work
with Fred Sanger was done at the Medical Research Council in Cambridge, England. Before joining the
faculty at UCSF, Dr. Sedat was a research associate at Yale University.
THE three-dimensional structure of chromo-
somes, both in the nucleus and during cell
division, remains a major unsolved problem in
biology. Our laboratory, in collaboration with
that of David Agard (HHMI, University of Califor-
nia, San Francisco) , is investigating chromosome
structure from the perspective of several inter-
locking questions: 1) What is the architecture of
the chromosome in the intact diploid nucleus?
How does the three-dimensional structure
change as a function of development, or progres-
sion through the cell cycle? 2) What is the archi-
tecture of a given gene in the nucleus? Do the
structural attributes reflect the detailed molecu-
lar information? 3) How do interphase chromo-
somes condense to form the intricate mitotic
structure at cell division?
The fruit fly Drosophila melanogaster, well-
known for its genetics, development, and bio-
chemistry, was chosen as a model biological sys-
tem. Although the initial emphasis is structural,
molecular genetics and biochemistry provide
functional correlations.
The UCSF three-dimensional optical micro-
scope has been developed to the point that data at
several wavelengths can be routinely collected,
even as a function of time (four-dimensional mi-
croscopy) , and can be used without computer ex-
perience. Still, we continue to perfect and en-
hance the instrumentation. We increased the
time resolution for data collection and greatly
improved the image quality of the four-
dimensional data, permitting analysis of much in-
formation on biological structures. We continue
to write software, with extensive mathematical
analysis, to correct systematic image acquisition
problems, to display results in a variety of for-
mats, and to model and analyze, often quantita-
tively, the complex three-dimensional data. We
have started to develop a computer-based meth-
odology to extract and analyze quantitatively the
large- and small-scale motion of chromosomes or
structure within the nucleus.
Four-Dimensional Optical Microscopy
We have continued to study the structure of the
cellular nucleus in living Drosophila embryos.
Nuclei were labeled by microinjection of fluores-
cent histones, or other chromosomal proteins.
Nuclear and chromosomal structures were fol-
lowed throughout the cell cycle during embry-
onic development. In addition to discerning
structural changes, we can now infer function.
Topoisomerase II — A Key Nuclear Protein
Our studies include an effort to understand the
role of various proteins in the organization and
dynamics of chromosomes. We have therefore
studied the distribution and dynamics of the DNA
strand-passing (unknotting) enzyme topoisomer-
ase II. High-resolution three-dimensional imag-
ing of Drosophila embryonic chromosomes
shows a heterogeneous distribution of topoiso-
merase II along the chromosome. At metaphase
and anaphase, the enzyme can be clearly seen to
be situated adjacent to the chromosome. This
suggests that its localization may be linked to its
activity: the enzyme may concentrate at sites of
chromosome condensation and/or segregation.
These data argue against a purely structural role
for the enzyme.
We have studied the dynamics of localization
by injecting fluorescently labeled antibodies
against the enzyme, or the labeled enzyme itself,
into live embryos and then imaging them by our
three-dimensional microscopy as a function of
time. The resulting time-lapse movies have
shown that the concentration of nuclear topoiso-
merase II changes dramatically throughout the
cell cycle. The highest levels occur in late inter-
phase, the lowest levels in telophase. A very dy-
namic fibrillar complex is evident during inter-
phase. Experiments that will disclose the
functional relevance of this structure are in
progress.
A Molecular Dissection
of the Nuclear Periphery
Recently we showed that the lamin proteins of
the nuclear envelope (NE) form a highly discon-
tinuous network in somatic interphase nuclei.
Several obvious questions arise. First, where are
361
Three-Dimensional Structure of Eukaryotic Chromosomes
the other known components of the nuclear pe-
riphery (pore complexes, chromatin) relative to
this network? Second, what, if anything, occupies
these large, lamin-empty regions? Third, how are
these structures assembled as the NE re-forms
during telophase?
Chromatin in the nuclear periphery displays an
^mteresting structural paradox in that a large frac-
tion appears to be aligned beneath the lamin net-
work, but with very little contacting lamins di-
rectly. The majority of it seems to be at a distance
of about 0.2 /xm. This result is consistent with
much indirect evidence for a strong interaction
between chromatin and the nuclear lamina, but
strongly suggests that a direct physical contact is
not involved.
We injected lamins and lamin-specific mono-
clonal antibody Fab fragments, both fluores-
cently labeled, into early Drosophila embryos to
study four-dimensional lamin-NE dynamics. In
these experiments the embryos develop normally
and hatch on time. We observe a highly discon-
tinuous lamin network in vivo, with interlamin
fiber spacings at least as large as those observed in
fixed samples. From these experiments, new
four-dimensional data, spanning prophase to
metaphase in the cell cycle, show a surprisingly
complex series of lamin structural rearrange-
ments. Lamins do not completely disassemble
and disperse at the onset of mitosis, but remain
well localized with complex structural dynamics
until well into the mitotic process. Further lamin
structural changes take place at anaphase and
telophase.
These studies emphasize that lamins appear es-
sential for the nuclear structural reorganizations
that take place at all points of the cell cycle. If,
however, we inject fluorescently labeled inter-
phase lamins, a very different picture results.
Arrested nuclear structures leading to chromo-
somal/nuclear aggregates are seen. These studies
suggest that structural/functional assays will be
required for proper interpretation of the
biochemistry.
Three-Dimensional in situ Hybridization
We are continuing our study of nuclear organi-
zation, using three-dimensional fluorescence in
situ hybridization. The studies, described in our
previous report, were performed in Drosophila
embryos, primarily at the histone gene locus, a
500-kb, tandemly repeated gene cluster. We have
extended our techniques to other whole-mount
tissues from Drosophila, including developing
imaginal tissues (which will later form the adult
fly). We have also improved our hybridization
protocols and can now detect chromosomal
probes as small as 1 2 kb with high signal-to-noise
resolution. Using these techniques, we are in-
vestigating the arrangement of chromosomes
throughout the cell cycle and as a function of
development.
A focus of our studies is the question of homolo-
gous chromosome pairing (one chromosome
from the male parent, the other from the female) .
This is of particular interest in Drosophila biol-
ogy because genetic evidence has indicated that
different alleles at certain homologous loci can
influence one another, implying that the loci
communicate in some fashion. Such phenomena
have been termed transvections, or more gener-
ally, "trans-sensing effects." By determining the
nuclear positions of the histone gene cluster in
developing embryos, we have shown that the ho-
mologous chromosomes bearing this locus are for
the most part spatially distinct, or unpaired,
throughout most of early embryonic develop-
ment, but that a transition occurs just prior to
cellularization, a distinct time point in develop-
ment, after which the locus is seen to be paired at
high frequency.
We are currently extending this analysis to
other genetic loci, to determine whether this un-
paired/paired transition occurs simultaneously at
all chromosomal positions. We have preliminary
evidence that loci more distant from the centro-
mere than the histone locus may show different
pairing behavior in early embryos. We are particu-
larly interested in carrying out this type of analy-
sis for loci that are known to have transvection or
trans-sensing effects.
This general methodology has a number of po-
tential applications to problems of cell lineage,
neural architecture, and pattern formation in de-
velopment. We are pursuing some of these inter-
ests in collaboration with other laboratories.
362
A Molecular Basis of Familial Hypertrophic
Cardiomyopathy
Jonathan G. Seidman, Ph.D. — Investigator
Dr. Seidman is also Professor of Genetics at Harvard Medical School. He received his undergraduate degree
from Harvard University and his Ph.D. degree from the University of Wisconsin-Madison, where he
studied with William McClain. His postdoctoral studies were carried out in Philip Leder's laboratory at
the National Institute of Child Health and Human Development.
FAMILIAL hypertrophic cardiomyopathy (FHC)
is a heart muscle disorder with an autosomal
dominant pattern of inheritance. The disease is
characterized clinically by unexplained myocar-
dial hypertrophy and variable symptomatology
that can include syncope, arrhythmias, conges-
tive heart failure, and sudden death. Diagnosis in
young people is particularly important. The inci-
dence of sudden death appears higher in this
group and can occur without warning.
Indeed, hypertrophic cardiomyopathy is one of
the most common autopsy findings among young
athletes who die suddenly. The large majority of
these were undiagnosed previously. Diagnosis in
this age group may be particularly difficult, since
the diagnostic clinical and echocardiographic cri-
teria may not be manifest until adulthood.
Because mutations in the cardiac myosin
heavy-chain (MHC) genes were implicated as the
cause of FHC in two families, we decided to un-
dertake a direct analysis of these genes in affected
individuals from other families. During the past
year we spent considerable effort attempting to
find in 24 unrelated families the cardiac MHC
mutations that cause FHC.
We had previously demonstrated that most
FHC mutations are missense or point mutations in
the MHC genes. We have successfully employed a
variety of techniques in the detection of missense
mutations within genes. Most of these are based
upon amplification of genomic DNA sequences
and analyses of individual exons. Application of
these approaches to the study of FHC mutations
was more difficult because the ^ MHC polypep-
tide is encoded in 40 exons, and hence 40 inde-
pendent analyses are required to examine the en-
tire gene. Furthermore, FHC is an autosomal
dominant disorder, and affected individuals are
heterozygous, bearing one mutated and one nor-
mal gene. Genomic analyses may fail to detect
deletions of entire exons or mutations that al-
tered gene splicing because of the presence of
one normal gene.
Access to messenger RNA in which intronic se-
quences have been excised would overcome
these limitations and allow analysis of coding re-
gions in a more rapid and convenient manner.
Although MHC mRNAs are abundant in the heart,
expression elsewhere is low and restricted to se-
lected fibers in slow-twitch skeletal muscle. Nor-
mal and mutant MHC sequences were detected
in RNA transcripts from peripheral lympho-
cytes and lymphocyte cell lines transformed by
Epstein-Barr virus (EBV) . This finding permitted
examination of (8 MHC mRNA, even though car-
diac tissue was not available.
Seven different ^ cardiac MHC mutations were
found among 24 unrelated FHC probands. Four
mutations were identified in two or more fami-
lies. One of these, the Arg453Cys mutation, prob-
ably occurred independently in families B and E,
because only family B also contains a hybrid a/jS
cardiac MHC gene on the same chromosome.
Whether other mutations that are shared by appar-
ently unrelated individuals arose independently
within mutational hot spots or represent a
founder mutation is uncertain. Characterization
of other families should also elucidate whether
the mutations that can cause the FHC phenotype
are restricted in number.
The identification of seven different mutations
in a disease with significant morbidity and pre-
mature death suggests that many of these are of
relatively recent origin in human evolution.
Since FHC mutations are not likely to provide a
selective advantage, but have not been lost in the
population, they probably reflect a high inci-
dence of new mutational events in the (8 cardiac
MHC gene.
The natural history of FHC is quite variable,
and diagnostic tests have been unable to identify
those with a more serious prognosis or those at
risk for sudden, unexpected death. To determine
whether particular ft cardiac MHC gene muta-
tions correlate with clinical outcome, we com-
pared several indices with genotype. Data from
families with the same mutation are pooled.
Disease-related deaths were infrequent in fami-
lies with the Val606Met mutations as compared
with the Arg249Gln, Arg403Gln, or Arg453Cys
mutations. While the incidence of disease-related
deaths in individuals with the Arg249Gln muta-
363
A Molecular Basis of Familial Hypertrophic Cardiomyopathy
tion is similar to that associated witli other muta-
tions, the average age at death is significantly
older for affected individuals within this family.
To assess cumulative survival of affected indi-
viduals with respect to age, we compared sur-
vival curves of FHC families with five different
mutations for which sufficient numbers of af-
fected individuals (alive or deceased) were avail-
able. These analyses confirmed that persons with
the Val606Met mutation survive longer than
those with the Arg453Cys or Arg403Glu
mutations.
The Arg249Glu mutation appears to produce
an intermediate phenotype. Survival in these indi-
viduals is better than in those with the Arg453Cys
or the Arg403Glu mutation. While survival ap-
pears shorter in individuals with the Arg249Glu
mutation than in those with the Val606Met muta-
tion, this difference is not statistically significant.
Individuals with the Arg453Cys mutation (with
or without the hybrid gene) and those with the
Arg403Cys mutation have similar life expectan-
cies, dying prematurely.
The seven different mutations are clustered in
the globular head of the polypeptide, and we
postulate that the defective myosins made by
these genes poison myosin function by impairing
physiologic interactions with other contractile el-
ements. Six of seven missense mutations affect
the charge of the altered residue. Perhaps the fact
that the Val606Met mutation does not alter the
net charge of the polypeptide accounts for the
better survival of affected individuals.
Identification of multiple disease-causing mu-
tations implies that mutational events within the
iS cardiac MHC gene are not uncommon and that a
number of FHC-causing cardiac MHC mutations
have occurred during the course of human evolu-
tion. A high frequency of myosin mutation may
explain the relatively high incidence of sporadic
(10-20 percent) hypertrophic cardiomyopathy.
The value of identifying FHC mutations was fur-
ther demonstrated by analysis of a large family
affected by mutation Arg249Glu. Related adult
family members were clinically evaluated for
FHC, and blood samples were obtained for inde-
pendent genetic diagnosis. The clinical and ge-
netic diagnoses were in complete concordance.
Thirteen children (ages 2-20) were also evalu-
ated. Statistically, half should have been affected,
but only one child had clinically demonstrable
FHC. Genotype analysis of these children re-
vealed six who inherited the mutant MHC gene.
These data underscored the insensitivity of
clinical diagnostic criteria for FHC in children
and young adults. Genetically based diagnoses of
FHC permit preclinical diagnosis and should fa-
cilitate prenatal diagnosis. Furthermore, the abil-
ity to make a preclinical diagnosis in families
makes possible longitudinal studies of disease de-
velopment and interventional trials.
364
I
Time course of the calcium concentration in \
the dendrite of a CAl hippocampal pyramidal
neuron in response to synaptic stimulation as a
function of time, running horizontally. The
concentration along the dendrite, running ver-
tically, is shown in false color, with the resting
level in blue and the highest concentration in
red.
Research of Richard Adams and Terrence
Sejnowski.
366
Computational Neurobiology of Sensory
Representations
TerrenceJ. Sejnowski, Ph.D. — Investigator
Dr. Sejnowski is also Professor at the Salk Institute for Biological Studies and Professor of Biology and
Neuroscience at the University of California, San Diego. He received his B.S. degree in physics from Case
Western University and his M.A. and Ph.D. degrees in physics from Princeton University. He was a
postdoctoral fellow with Alan Gelperin in the Biology Department at Princeton and with Stephen Kuffler
at Harvard Medical School, where he studied mechanisms of synaptic transmission. Dr. Sejnowski was a
member of the faculty of the Biophysics Department at the Johns Hopkins University before moving to
San Diego. He and Patricia Churchland have recently written The Computational Brain, a book on
computational neuroscience.
WE do not yet understand how the nervous
system enables us to recognize objects, to
learn new skills, and to plan actions. The discov-
ery that single neurons in the visual system can be
highly selective in responding to visual stimuli
led to the view that the perception of complex
objects could be directly linked to the activity of
individual neurons. This possibility raises a num-
ber of questions, such as what degree of influ-
ence a single neuron can have on behavior and
whether there are enough neurons in the brain to
account for the large number of objects that can
be perceived.
An alternative possibility relies on populations
of neurons to represent perceptual states. On this
account, the information essential to the repre-
sentation of an object is distributed over a large
population of neurons. It is difficult to imagine
how a pattern of activity in a large number of
neurons distributed widely throughout the brain
could be used to recognize an object and serve as
the input for motor actions. Computer models
incorporating cellular information from single-
cell recordings and constrained by psychological
measurements on performance can help to orga-
nize these data and provide a conceptual frame-
work for understanding distributed represen-
tations. Such models are being used to explore
how the visual cortex represents the three-
dimensional world, how this representation may
arise during development, and how the informa-
tion coded by these neurons might be used to
coordinate actions such as eye movements.
The perception of depth depends upon a num-
ber of visual cues, but only one of them relies on
the slight positional shift that occurs between the
diff^erent viewpoints of the two eyes, called the
image disparity. Neurons in the first cortical stage
of vision are sensitive to disparity, and the devel-
opment of this sensitivity is dependent on binocu-
lar vision during a critical period.
In the adult visual cortex, neurons are ob-
served to be either dominated by input from one
eye (monocular cells) or relatively balanced with
input from both eyes (binocular cells) . Further-
more, binocular cells tend to be stimulated maxi-
mally when images are in exact correspondence
in both eyes, thus preferring zero disparity, and
relatively monocular cells tend to prefer nonzero
disparities. We have simulated the development
of a layer of cortical cells receiving inputs from
both eyes and show how such a relationship be-
tween ocularity and disparity might arise.
The key feature of our model is the use of
correlations of activity both within each eye and
between the eyes. We assume that two retinal
cells close to each other will have more corre-
lated activity than two cells far apart. Corre-
sponding points in the two eyes will also tend to
be correlated, since they will look, on average, at
the same point in space. However, the correla-
tion between the eyes will also be spread out by
convergent and divergent eye movement. We can
simulate visual development in two stages: prena-
tal, when the two retinae have essentially inde-
pendent activities, and postnatal, when the eyes
are open and have correlated activities. By vary-
ing the amount of development that occurs in the
model before eye opening, we can show that a
mixture of monocular and binocular cells arises
with the observed relationship to disparity.
Disparity provides information about the rela-
tive positions of objects in space, but this cue is
insufficient to recover the absolute distance of
objects from the viewer. However, the distance of
an object from the viewer can be computed by
combining relative depth cues with other infor-
mation, such as eye position. We have developed
a network model to explore how the vergence of
the two eyes (angle between the two lines of
sight) and the binocular disparity could be com-
bined to represent the distance to an object.
Single neurons have a wide range of disparity
tuning curves that are broad and overlapping.
Such a distributed representation of disparity was
used in a network model to encode the inputs.
The network was trained to transform disparity
and vergence input information by projections
367
Computational Neurobiology of Sensory Representations
through a layer of hidden units to an output layer
that represented the perceived egocentric depth
of the object, as determined by psychophysical
measurements. The disparity tuning curves of the
hidden units were similar to those of the input
units, and varying the vergence did not change
the shape of the tuning curves; however, the ver-
-^'gence did modulate their amplitude.
Similar "gain fields" for conjugate eye move-
ments have been observed in the posterior pari-
etal cortex, a region of the brain that is essential
for our internal representation of external space.
The predictions of this model can be tested by
recording single-unit activity in the cerebral cor-
tex of awake and behaving monkeys, and several
laboratories are pursuing these experiments. Pre-
liminary results support the model.
Neurons in the early stages of visual processing
in the cerebral cortex are organized in retinoto-
pic maps. Thus visual features are arranged in a
system of coordinates that is based on the posi-
tion of features in the visual field of the retina
rather than on the absolute position of features in
space. Psychological experiments provide fur-
ther evidence that simple visual features such as
orientation and direction of motion are organized
according to retinal coordinates. At later stages of
visual processing, the receptive fields of neurons
become very large; and in the posterior parietal
cortex, containing areas important for sensory-
motor coordination, the visual responses of neu-
rons are modulated by both eye and head posi-
tion. A previous model of the parietal cortex
showed that the modulation of the neurons ob-
served there is consistent with a distributed spa-
tial transformation from retinal to spatial coordi-
nates. Our model of the transformation from
disparity to distance by vergence modulation can
be considered a generalization of this model to
include the third dimension of space.
All these models assume that the responses of
neurons in the early stages of visual processing in
cerebral cortex depend only on retinal informa-
tion and not on the direction of gaze. Several labo-
ratories have now reported that eye position does
in fact modulate the visual response of many neu-
rons in early stages of visual processing. Further-
more, this modulation appears to be qualitatively
similar to that previously reported for neurons in
the parietal cortex. These new findings suggest
that transformations from retinal to spatial repre-
sentations could be initiated much earlier than
previously thought.
We have used network models to study the
consequences of incremental spatial transforma-
tions in a feedforward hierarchy of cortical maps.
Our model shows that it is possible for visual fea-
tures to be encoded in spatial coordinates already
at very early stages of visual processing. We call
this new type of spatial map a retinospatiotopic
representation and are exploring its counterin-
tuitive properties. The model makes several sur-
prising predictions that we are testing with per-
ceptual experiments on human observers.
The primate visual system is very good at com-
plex motion-processing tasks such as tracking a
moving object against a textured background
under a variety of luminance conditions. In order
to track a moving object, the visual system must
integrate many local motion estimates from many
neurons, each with limited spatial receptive
fields. No single neuron has the information
needed to estimate the velocity of the object.
We have developed a simple model for motion
processing in the visual areas of cortex that spe-
cialize in representing motion. The model as-
sumes two pools of filters at each location on the
visual field: one pool computes estimates of mo-
tion in a local region of the visual field, while the
other estimates the relevance or reliability of
each local motion estimate, based on the estimate
itself and on additional information from the
visual scene. Outputs from the second pool can
"gate" the outputs from the first pool through a
gain-control mechanism, before the local motion
estimates are integrated to form more-global esti-
mates. The proposed mechanism of gain control
is consistent with measured responses of cortical
cells under conditions of interfering motion of
transparent stimuli.
These models provide representations of ob-
jects in space that are highly distributed. We also
want to understand how these distributed repre-
sentations can be used to direct the motor system
to orient toward these objects. For example, mo-
tion estimates can be used to direct the eyes to
track moving objects, and distance estimates can
be used to guide hand movements to reach out for
objects. We are developing models of motor sys-
tems in the brain that will complement these
models of sensory processing. The models of mo-
tor control are based on networks of neurons that
include feedback connections, which makes
them highly dynamic. New principles of neural
processing may emerge as more-detailed dynami-
cal properties of neurons are incorporated into
these models.
368
Adenovirus as a Model for Control
of Gene Expression
Thomas E. Shenk, Ph.D. — Investigator
Dr. Shenk is also Elkins Professor of Molecular Biology at Princeton University and Adjunct Professor of
Biochemistry at the Robert Wood Johnson Medical School, University of Medicine and Dentistry of New
Jersey. He received his Ph.D. degree in microbiology from Rutgers University for studies with Victor Stollar,
and his postdoctoral training with Paul Berg at Stanford University. Before coming to Princeton, he was
Assistant Professor of Microbiology at the University of Connecticut Health Center and then Professor of
Microbiology at the State University of New York School of Medicine at Stony Brook. Dr. Shenk counts
among his honors the Eli Lilly Award in Microbiology from the American Society for Microbiology and an
American Cancer Society Professorship.
ADENOVIRUSES are widespread, and humans
are first infected when quite young. Gener-
ally the infection results in cold-like symptoms
and resolves without complication. Some human
adenoviruses, however, induce a variety of be-
nign and malignant tumors if injected into a rat or
hamster. Since these viruses contain DNAand are
tumorigenic under certain conditions, they are
classified as DNA tumor viruses.
Adenoviruses can be propagated easily in cul-
tured cells. When human cells are infected, the
approximately 30 viral genes are expressed, the
viral chromosome is replicated, and individual
DNA molecules are packaged into protein shells
to produce virus progeny. Since viral genes are
expressed at high levels compared with most cel-
lular genes, and since this expression is tightly
regulated, the adenovirus is a useful probe for
studying the control of gene expression.
During the past year, much of our effort has
focused on transcriptional control of viral gene
expression. The first viral gene to be expressed
after infection of a cell is the ElA gene, which
encodes a protein that activates expression of ad-
ditional viral genes. The ElA protein appears to
activate transcription (copying of genetic infor-
mation) through several physiologically distinct
mechanisms. One of these involves a cellular
transcription factor that we have termed YY- 1
We first identified the binding site for YY- 1 in
the P5 transcriptional control region of adeno-
associated virus, a defective virus that depends on
a variety of adenovirus gene products for its repli-
cation. The ElA protein activates expression of
the P5 control region, and the critical sequence
element required for activation is a DNA segment
constituting the binding site for YY-1 . To investi-
gate its function, this site was inserted upstream
of several heterologous promoters, and it re-
pressed their activity. The repression was re-
lieved in the presence of ElA protein. In fact, the
protein also activated transcription through the
YY-1 binding site. The combination of these two
effects, relief of repression and activation, in-
duced transcription by a factor of 1,000 in some
test genes. Thus the combination of the YY-1
binding site and ElA protein formed a powerful
biological on/off switch.
In order to study the YY- 1 factor, we prepared
some from cultured human cells and determined
a short amino acid sequence from the purified
protein. This sequence was used to design a short
probe DNA, which enabled us to identify and iso-
late a cDNA clone encoding the protein. Se-
quence analysis of the clone revealed that YY- 1 is
a 4l4-amino acid protein with a zinc finger
DNA-binding motif. Protein was expressed from
the clone and shown to bind specifically to the
YY-1 recognition site.
The binding activity of YY- 1 was altered by fus-
ing it to the DNA-binding domain of the yeast
GAL4 protein. This approach is widely used to
study transcription factors, since it provides the
opportunity to direct binding of the factor under
study in a highly specific fashion to a test gene
construct that contains the yeast GAL4 DNA-
binding site. By redirecting the factor to bind to a
novel site, it was possible to study the activity of
the fusion protein in cells that contain high levels
of endogenous YY-1 . As anticipated, the YY-1 fu-
sion protein repressed transcription of the test
gene in the absence of the ElA protein, and the
repression was relieved by the ElA protein. Muta-
tional analysis of the fusion protein has demon-
strated that the YY- 1 zinc finger domain is respon-
sible for its ability to repress transcription.
Since the ElA protein would not be expected
to cause the YY-1 fusion protein to detach from
the GAL4 DNA-binding site, it appears likely that
YY- 1 remains bound to the control region but is
somehow altered in the presence of the ElA pro-
tein so that transcription is not repressed. Several
experiments indicate that the ElA protein can
bind directly to YY-1. For example, if the two
proteins are mixed, they sediment as a complex
in a sucrose gradient. Furthermore, radioactively
labeled ElA can bind to YY-1 that has been sepa-
rated from a complex mixture of proteins by elec-
trophoresis and bound to a membrane filter.
369
Adenovirus as a Model for Control of Gene Expression
We are currently mapping the domains on ElA
and YY-1 through which they interact. The loca-
tion of these domains might provide insight to
the mechanism by which the viral activator alters
YY-1 function. We are also searching for cellular
proteins that might interact with the domain of
YY-1 responsible for repression.
^ The P5 promoter of the adeno-associated virus
contains two binding sites for YY- 1 . The site that
was originally studied is centered about 60 base
pairs upstream of the transcriptional initiation
site, and it is responsible for the repression and
ElA-mediated activation discussed above. The
second site is centered at the P5 transcriptional
control region. Its location suggested that YY-1
might serve an initiator function in addition to its
repression function. Short DNA sequences
surrounding several transcriptional start sites
have been shown to be capable of starting tran-
scription and have been termed initiator ele-
ments. These sequences can direct RNA polymer-
ase to initiate transcription at the correct start site
in the absence of any other binding sites for
known transcription factors. When binding sites
for additional factors are added, the initiator se-
quences become much more efficient.
The binding site for YY- 1 proved to behave as
an initiator element. Furthermore, by studying its
activity in cell-free extracts and using antibodies
specific for YY- 1 , we were able to show that YY- 1
protein is required for the initiator activity dis-
played by its binding site. Work in progress to
determine whether YY- 1 interacts with other pro-
teins in the initiation complex should elucidate
its role in transcription.
It is intriguing that YY- 1 (Yin and Yang factor
1 ) is able to exert opposite effects when it binds
to different locations within a promoter. In the
adeno-associated virus promoter, it represses
transcription when bound upstream of the start
site, and it contributes in a positive sense to the
initiation event when bound at the start site. Our
longer term goal will be to understand how the
factor can mediate these two different activities.
Finally, it is important to note that in addition
to activating transcription through YY-1 and
other cellular transcription factors, the ElA pro-
tein can oncogenically transform cells. Almost
certainly, at least part of its oncogenic activity
results from its ability to bind to proteins such as
the cellular retinoblastoma protein, a recessive
oncogene product whose proper function is re-
quired for appropriate growth regulation of cells.
It is possible, however, that interactions of ElA
protein with transcriptional regulatory proteins
also contribute to oncogenesis. Work is in
progress to determine whether alteration of
YY-1 function can play a role in cellular
transformation.
370
Growth Control of Myeloid Cells
Charles J. Sherr, M.D., Ph.D. — Investigator
Dr. Sherr is also a member of the Department of Tumor Cell Biology at St. Jude Children 's Research
Hospital and Adjunct Professor of Biochemistry at the University of Tennessee College of Medicine,
Memphis. He received his medical degree and his Ph.D. degree in immunology from New York University
School of Medicine, where he studied with Jonathan Uhr. After a pathology residency at Bellevue Hospital
Center, New York, he joined George Todaro's laboratory at the National Cancer Institute, where he began
studies on retroviral oncogenes. After 10 years on the staff of the NCI, Dr. Sherr relocated to St. Jude
Children 's Research Hospital.
EACH day humans produce billions of blood
cells, which enter the circulation from their
sites of origin in the bone marrow. The majority
are red cells (erythrocytes) , which transport oxy-
gen, and the remainder are white cells (leuko-
cytes), which play a vital role in preventing in-
fection by bacteria, viruses, and other parasites.
Different classes of white cells carry out special-
ized functions: macrophages and granulocytes in-
gest and kill microorganisms, and lymphocytes
recognize foreign antigens and produce antibod-
ies to combat them.
The process of blood cell production (hema-
topoiesis) is regulated by a group of protein
growth factors, termed colony-stimulating fac-
tors (CSFs) or interleukins. These factors stimu-
late the precursors of mature white cells to form
colonies in agar composed of differentiated
blood cell elements and v/ere named for the types
of colonies they produced. For example, M-CSF
(or CSF-1) specifically induces macrophage colo-
nies, G-CSF promotes granulocyte development,
and GM-CSF stimulates the growth and differen-
tiation of both types of cells. CSFs, now produced
in quantity through genetic engineering tech-
niques, have become part of the clinical arma-
mentarium and are efficacious in extrinsically
regulating blood cell production and in height-
ening host defense against infection.
Signal Transduction by the CSF-1 Receptor
The actions of CSFs in supporting cell prolifera-
tion and survival are mediated through their bind-
ing to specific receptors expressed on the sur-
faces of their target cells. The macrophage CSF-1
receptor (CSF-1 R) consists of an extracellular
growth factor-binding portion, joined through a
single membrane-spanning segment to an intra-
cellular enzymatic domain. Binding of CSF-1 to
its receptor on the outside of the cell triggers the
activity of the intracellular enzymatic moiety — a
kinase — inducing it to add phosphate molecules
to other proteins. These phosphorylation events
modify the biochemical behavior of multiple tar-
get proteins, some of which relay signals to the
cell nucleus that alter gene expression, DNA syn-
thesis, and cell division.
The ability of many cell types to respond to
CSF- 1 is simply determined by whether they ex-
press its receptor. By introducing the gene en-
coding CSF-IR into naive cells, we can sensitize
them to the stimulatory effects of the growth fac-
tor. Such manipulations have allowed us to study
genetically engineered receptor variants for their
capacity to transduce signals for cell growth or
differentiation.
One strategy is to alter different portions of the
intracellular domain of CSF-IR in order to pin-
point structural motifs that determine its interac-
tion with its "downstream" targets. A conse-
quence has been the development of receptors
that are impaired in transducing signals through
certain pathways but not others. In particular in-
stances the introduction of other complementing
genes into cells that express partially defective
receptor mutants has reconstituted full receptor
activity, thereby providing genetic evidence for
functional relationships between different gene
products in signaling pathways.
Such experiments have revealed that the com-
binatorial actions of target proteins that bind to,
or are phosphorylated by, CSF- 1 R can in part de-
termine the specificity of the biological response
in different cell types, thereby influencing deci-
sions governing cell proliferation, survival, and
fate.
Oncogenic Potential of CSF-IR
CSF- 1 R is encoded by the FMS proto-oncogene
and can be converted by mutations to an "onco-
protein" capable of inducing tumors. Certain
mutations in the extracellular domain of CSF-IR
can mimic the action of CSF-1 and activate the
receptor kinase in the complete absence of the
growth factor. The sustained and unregulated sig-
nals for cell growth that arise from this class of
mutant receptors might naturally contribute to
malignancies involving macrophages or their
bone marrow progenitor cells.
We have recently used a prospective genetic
371
Growth Control of Myeloid Cells
approach to identify sites within CSF-IR that,
when mutated, can endow the receptor with on-
cogenic activity. By randomly mutagenizing seg-
ments of FMS and screening "libraries" of mu-
tated genes for their ability to induce cell
transformation, we identified several sites in the
receptor where "activating mutations" occur.
-With such information in hand, it is now possible
to search for the presence of similar genetic le-
sions in the FMS genes of myeloid leukemia cells
and so determine whether FMS mutations play an
etiologic role in such diseases.
Role of CSF-1 in Cell Cycle Progression
After proliferating macrophages complete cell
division (mitosis, or M phase), they enter an 8- to
10-hour gap phase (Gi), during which they pre-
pare to replicate their chromosomal DNA. The
ensuing period of DNA synthesis (S phase) lasts
for seven hours, and once DNA replication is
complete, the cells enter a second, shorter gap
phase (G2) before dividing again and redistribut-
ing copies of duplicated chromosomes to each
daughter cell. CSF-1 is only required throughout
Gi for cells to enter S phase, and once DNA syn-
thesis begins, macrophages can complete cell di-
vision in the absence of the growth factor. On the
other hand, the requirement for persistent CSF-
IR-mediated signals throughout the entire Gj
interval implies that the expression of growth
factor-responsive genes must be temporally regu-
lated over an 8- to 10-hour period.
Genetic data accumulated through studies of
yeasts indicate that cell division-cycle genes
called Gi cyclins act to prepare cells for DNA syn-
thesis. We recently isolated a novel class of "D-
type" cyclins from mammalian cells, at least two
of which are differentially regulated by CSF-1
during the Gj interval of the macrophage cell cy-
cle. Related genes are expressed in other cell lin-
eages, where their expression is governed by dif-
ferent growth factors. Our idea is that these
cyclins control progression through the Gj inter-
val in mammalian cells by mechanistically link-
ing early steps in growth factor-mediated signal
transduction with the timing of the cell cycle
clock. As might be expected, perturbations in the
regulation of these cyclins occur in specific types
of tumor cells and thus appear to contribute to
malignancy.
372
The Role of Second Messengers
in Ion Channel Regulation
Steven A. Siegelbaum, Ph.D. — Associate Investigator
Dr. Siegelbaum is also Associate Professor of Pharmacology in the Center for Neurobiology and Behavior,
Columbia University College of Physicians and Surgeons. He received his A.B. degree in biochemical
sciences from Harvard College and his Ph.D. degree in pharmacology from Yale University, studying the
role of calcium in cardiac electrical activity. He then did postdoctoral research with David Colquhoun at
University College, London, and with Philippe Ascher at the Ecole Normale Superieure in Paris, where he
studied the nicotinic acetylcholine receptor ion channel, before joining the faculty of Columbia University.
He has received the Herbert J. Kayden Award in Biomedical Science of the New York Academy of Sciences.
THE electrical activity of nerve and muscle
cells is regulated by the actions of hormones,
neurotransmitters, and sensory stimuli such as
light, odors, and pressure. Regulation of neuro-
nal activity often depends on the production of
intracellular second messengers — small metabo-
lites such as cyclic AMP, cyclic GMP, and various
products of phospholipid metabolism. These sec-
ond messengers then act to alter the function of
ion channels, the membrane proteins that govern
the electrical signaling of cells. Previous research
in our laboratory focused on the role that regula-
tion of ion channel function by second messen-
gers plays in learning and memory. Recently we
have become interested in the role of second mes-
sengers in olfactory signal transduction, the sub-
ject summarized below.
The second messenger that was first found to
play a role in regulating electrical activity was
cyclic AMP. From initial studies by Earl Suther-
land and his colleagues, a number of hormones
and neurotransmitters have been shown to act by
elevating cAMP concentrations in cells. Later
studies by Edwin Krebs and his colleagues
showed that most of the effects of cAMP were due
to the activation of a cAMP-dependent protein ki-
nase (cAMP-PK), which phosphorylates many
types of proteins. Over the past several years it
has become clear that neurotransmitters can alter
the activity of ion channels by causing the produc-
tion of cAMP, leading to the activation of cAMP-
PK, which can then directly phosphorylate ion
channels. In general, cAMP-dependent actions
are relatively slow; they require several seconds
to produce changes in electrical activity because
of the relatively slow rates of protein phosphory-
lation and dephosphorylation. Thus phosphory-
lation-dependent second messenger actions are
generally not well suited to mediating rapid neu-
ronal signaling that occurs during fast synaptic
transmission or sensory processing.
Role of Second Messengers in Olfactory
Signal Transduction
Olfactory signal transduction has provided neu-
robiologists with an intriguing puzzle on several
levels. First, how does the olfactory system recog-
nize and discriminate among thousands of differ-
ent odors? Are there a limited number of recep-
tors that each bind many hundreds of odorants, or
are there hundreds of receptors that are each spe-
cific for a single or a few different odorants? Sec-
ond, how does the binding of an odorant to its
receptor generate an electrical signal in the olfac-
tory neuron? Third, how does our olfactory sys-
tem enable us initially to detect odors at very low
concentrations and yet become insensitive to the
same stimuli after several minutes? In one sense
this form of adaptation is the simplest form of
learning: our olfactory system "learns" to ignore
a certain stimulus.
Recently some of the puzzles associated with
olfaction began to be solved. Linda Buck and
Richard Axel (HHMI, Columbia University Col-
lege of Physicians and Surgeons) identified a sur-
prisingly large gene family that may code for
hundreds of distinct odorant receptors in rat ol-
factory neurons, providing for the requisite speci-
ficity. The discrepancy between the slow time
course of most cAMP-mediated responses and the
more rapid olfactory signaling was resolved when
Tadashi Nakamura and Geoffrey Gold reported an
ion channel in toad olfactory neurons that was
directly activated by cAMP. Thus the activation or
gating of the channel did not depend on the rela-
tively slow processes of phosphorylation and de-
phosphorylation, but rather seemed to be due to
the direct binding of cAMP to the channel. This
channel was activated equally well by cAMP or
cGMP.
Properties of a Cyclic Nucleotide-gated
Channel
The first gene for a cyclic nucleotide-gated
(CNG) channel to be cloned came from photore-
ceptors, where the channel participates in photo-
transduction and is selectively activated by
cGMP. More recently several laboratories have
cloned the genes for olfactory neuron CNG chan-
nels from several species. In collaboration with
373
The Role of Second Messengers in Ion Channel Regulation
Richard Axel's group, we have been studying the
properties of a CNG channel cloned from olfac-
tory neurons of catfish. Our goal is to understand
how the binding of cyclic nucleotides leads to
channel activation and how the channel itself
may participate in adaptation.
The gene for the catfish olfactory channel is
"highly homologous to olfactory channel genes
from other species and to the gene for the photo-
receptor channel. The channel is expressed selec-
tively in olfactory neurons, supporting its role in
olfactory signal transduction. Like the rat olfac-
tory CNG channel, the catfish channel is also acti-
vated directly by both cAMP and cGMP. Unlike
the rat channel, however, the catfish channel
does not discriminate between cAMP and cGMP,
suggesting a structural difference in the cyclic
nucleotide-binding sites. Using genetic engineer-
ing, we are trying to define the structural bases
for these differences.
We have also used single-channel recording to
measure the unitary currents that flow through an
open CNG channel. We find that the probability
of a channel being open increases with rising cy-
clic nucleotide concentration. When the cloned
channel is expressed in frog oocytes, it requires
relatively high concentrations of cAMP or cGMP
(around 50 micromolar) to become activated.
Surprisingly, the same channel studied in its na-
tive environment, the catfish olfactory neuron,
requires 20-fold lower cyclic nucleotide concen-
trations for activation. This difference in cyclic
nucleotide sensitivity could mean that the oo-
cytes fail to process the channel correctly (for
example, by not phosphorylating it properly) or
that they lack some important channel-regulating
protein. Defining the factors responsible for the
discrepancy between the cloned and native chan-
nels is an important goal.
In other work, we are focusing on the role of
the channel in odor adaptation. When the olfac-
tory CNG channel is activated, it allows calcium
ions to enter the olfactory neuron from the out-
side medium. In photoreceptors, a similar influx
of calcium underlies visual adaptation. We find
that intracellular calcium greatly reduces the re-
sponse of the olfactory channel to cyclic nucleo-
tides. This effect appears to result from a shift in
the dose-response curve for channel activation to
higher concentrations of cyclic nucleotides. The
inhibitory effect of calcium occurs at physiologi-
cal calcium levels. Moreover, the effect is not due
to a direct action of calcium on the channel but
rather appears to involve an intermediate regula-
tory protein that is loosely associated with the
channel. Thus calcium acts as a negative feedback
regulator of olfactory responses.
Thus the olfactory system provides a useful
model for studying neuronal signal transduction
and neuronal plasticity. Studies on the molecular
bases of these phenomena should provide us with
insight into many of the basic mechanisms con-
trolling nerve cell behavior.
374
Chemistry of Cellular Regulation
Paul B. Sigler, M.D., Ph.D. — Investigator
Dr. Sigler is also Professor of Molecular Biophysics and Biochemistry at Yale University. He studied
chemistry at Princeton University and received his M.D. degree from Columbia University. He then spent
two years as a house officer in the Department of Medicine at Columbia-Presbyterian Medical Center,
New York. He began his work on crystallography with David Davies at NIH. He studied as a Helen Hay
Whitney Fellow at the MRC Laboratory of Molecular Biology in Cambridge, England, where he received
his Ph.D. degree in biochemistry. Before accepting his present position. Dr. Sigler was Professor of
Biochemistry and Molecular Biology at the University of Chicago. Dr. Sigler was recently elected to the
National Academy of Sciences.
MY colleagues and I continue to pursue our
interest in the molecular mechanism of reg-
ulation at the level of transcription and trans-
membrane signaling. Our approach is to use
high-resolution x-ray crystallography to deter-
mine the structure of the relevant macromole-
cules and their assemblies. We infer chemical
mechanisms from these structures and test them
with biochemical and physicochemical experi-
ments and with directed mutagenesis.
Transcriptional Regulation
In the area of transcriptional control, our pri-
mary focus remains the mechanism by which
DNA-binding regulatory proteins are targeted to
their responsive genes. Our earliest and best-
studied experimental system has been the alloste-
ric regulation of the trp repressor and the forma-
tion of the trp repressor-operator complex.
Having defined the stereochemistry of the
protein-DNA interface, we are now extending the
study (partly in collaboration with Julian Sturte-
vant, Yale University) to include an analysis of
the thermodynamic parameters underlying the
trp repressor's ability to bind selectively to the
trp operator.
Crystallographic structural analyses have been
extended to the mechanism by which steroid re-
ceptors bind selectively to their DNA targets. The
first complex in this series (in collaboration with
Len Freedman and Keith Yamamoto, University
of California, San Francisco) shows the DNA-
binding domain in complex with its idealized tar-
get, the symmetrical glucocorticoid response ele-
ment. It shows how the two zinc fingers of the
domain interact with each other to form a unified
globular domain and how the domain makes
chemical contacts with the bases that identify the
"half sites" of the symmetrical response element.
It also shows a surprisingly novel mechanism by
which the receptor recognizes the invariant
three-base pair spacing between the half sites.
Interactions with the DNA phosphates stabilize
a change in the protein's structure that causes it
to form a firm dimer. In so doing, the DNA-bind-
ing surfaces of each subunit are placed in perfect
registry with the complementary surfaces of the
DNA target's half sites. Thus the DNA itself helps
form a dimer that discriminates between DNA se-
quences that have the correct spacings between
their half sites and those that do not. These same
DNA-induced conformational changes are also
likely to potentiate functional contacts between
the receptor and other components of the tran-
scription initiation assembly.
Work is under way to examine the arrangement
by which the glucocorticoid receptor contacts
other transcription factors on "composite" regu-
latory sites. We are also examining the way other
members of this steroid receptor superfamily in-
teract with their respective response elements.
Only through a comparative study can we build a
reliable picture of the targeting mechanism.
Most recently we have solved (in collaboration
with Laimonis Laimins, HHMI, University of Chi-
cago) the structure of the complex between the
E2 transcription factor of bovine papilloma virus
and the DNA sequence (enhancer) to which it
binds. E2 is involved in replication of the viral
genome and controls expression of the genes re-
sponsible for transformation of cells infected
with this oncogenic virus. The structure has been
refined at an unprecedented level of detail (1.7
A) and exhibits a barrel-like dimeric architecture
that has never been seen before in any protein
structure, let alone in a DNA-binding domain.
The DNA is severely bent over this barrel as it
forms specific contacts with the protein. These
results are now being correlated with genetic in-
formation produced by others.
Several other specific DNA complexes of tran-
scriptional regulatory proteins are in earlier
stages of crystallographic analysis. A particularly
interesting one is the arg repressor, which regu-
lates the expression of arginine biosynthetic
genes. It is unique because it is a hexamer of
identical subunits. Its target is also unusual, in
that it is two symmetrical 1 8-base pair sequences
whose symmetry axes are separated by 2 1 base
375
Chemistry of Cellular Regulation
pairs, or exactly two helical turns of DNA. Our
studies (in collaboration with Werner Maas, New
York University) support the inference that the
operator must be sharply bent as it wraps around
the hexamer. In view of the distortion that this
protein can introduce into DNA, it is interesting
that the arg repressor has been implicated as a
-binding "scaffold" in the site-specific recombina-
tion process that resolves fused ColEl plasmids.
The architecture of this protein-DNA complex
should give insight into the DNA topology in-
volved in site-specific recombination.
Transmembrane Signaling
In the area of transmembrane signaling, we
have recently solved (in collaboration with Jeff
Browning of Biogen) the structure of the human
secretory phospholipase A2 that is associated with
both acute and chronic inflammation. This study
contrasts the free enzyme with the enzyme in a
complex with a transition-state analogue and
shows that the mechanism of this enzyme con-
forms to what we have proposed earlier for inter-
facial catalysis. This structure could provide a
firm stereochemical platform for anti-inflamma-
tory drug design.
In order to function, this class of enzymes must
attach itself firmly to the surface of the mem-
brane. Others have shown that the attachment of
the enzyme to the membrane surface is primarily
due to electrostatic forces. Our calculations (in
collaboration with Barry Honig, Columbia Uni-
versity) of the charge-potential distribution for a
large series of phospholipase A2 structures ac-
count nicely for this effect.
Recently our attention has turned to the trans-
membrane signal cascade that employs seven-
helical transmembrane receptors, G proteins,
and target enzymes involved in either the synthe-
sis or cleavage of phosphodiester bonds (e.g.,
adenyl cyclase, phospholipase C, cGMP phospho-
diesterase) . Because of the immediate availability
of large amounts of pure protein, we have chosen
to work first on the bovine visual receptor sys-
tem. Our first goal is to solve a recently obtained
well-ordered crystal form of T^^ • OTP, the solu-
ble transducer of the activation signal. We hope
to extend this work to include complexes of
Tg„ with the activated receptor (rhodopsin)
and its target molecule, the 7-subunit cGMP
phosphodiesterase .
376
The Mitochondrial Genome of Trypanosomes
Larry Simpson, Ph.D. — Investigator
Dr. Simpson is also Professor of Cell and Molecular Biology at the University of California, Los Angeles.
He received his B.A. degree in biology from Princeton University and his Ph.D. degree in molecular
parasitology with William Trager at the Rockefeller University. His postdoctoral training was with Maurice
Steinert at the Free University of Bruxelles.
THE kinetoplastids represent a large group of
parasitic protozoa that are tiie causal agents
for a variety of human and animal diseases, in-
cluding African sleeping sickness, Chagas' dis-
ease, and leishmaniasis. There are no vaccines for
any of these, and chemotherapies are either non-
existent or inadequate. As cells, the kinetoplast-
ids represent one of the most ancient lineages of
the eukaryotic kingdom and thereby have many
novel physical and biochemical features. Some
species, such as Crithidia, are parasitic in only a
single invertebrate host. Others, such as Leish-
mania, Trypanosoma, and Pfoytomonas, are par-
asitic in both an invertebrate and a vertebrate (or
plant) host.
Kinetoplast DNA
The mitochondrial genome of these cells,
which is known as kinetoplast DNA, consists of a
single giant network containing approximately
10,000 minicircles and 20-50 maxicircles, all
linked together by catenation. The maxicircles
comprise a subset of mitochondrial genes that en-
codes two small rRNAs, three subunits of cy-
tochrome oxidase, cytochrome b, four subunits
of NADH dehydrogenase, and three yet unidenti-
fied proteins. No tRNAs appear to be encoded by
the mitochondrial genome and therefore must be
imported into the organelle by some yet to be
defined mechanism. The function of the minicir-
cle was unknown until recently.
RNA Editing
RNA editing is a post-transcriptional process in
which uridine (U) residues are inserted and de-
leted from coding regions of the primary tran-
scripts of several maxicircle "cryptogenes." The
extent of editing varies from a few U residues to
hundreds, at hundreds of sites throughout the en-
tire mRNA (pan-editing) . The function of editing
is to create translatable mRNAs encoding mito-
chondrial proteins.
The extent of editing of specific genes varies
from species to species. For example, the NADH
dehydrogenase subunit 7 mRNA is internal and
5'-edited in Leishmania tarentolae but pan-
edited in Trypanosoma brucei. We have re-
cently shown that at least one G-rich intergenic
region in both species encodes a transcript that is
pan-edited to produce ribosomal protein SI 2 for
the mitochondrial ribosome. It is likely that five
other G-rich regions represent additional pan-
edited cryptogenes, which would bring the total
mitochondrial structural gene content in these
cells to 17, of which 12 are cryptogenes.
Mechanism of RNA Editing
We discovered in 1990 that maxicircles also
encode another class of RNAs — the guide RNAs
(gRNAs) — which contain the necessary se-
quence information for the edited genes. These
are small RNAs, which can form perfect duplex
hybrids with edited mRNAs, both within the
edited region and 3' of the edited region, pro-
vided G-U base pairs are allowed. The gRNAs also
have nonencoded 3' oligo-[U] tails ranging in
length from 5 to 28 nucleotides. We then discov-
ered that gRNAs are also encoded in the minicir-
cles, finally providing a genetic role for these
enigmatic molecules.
We have proposed two models for the involve-
ment of gRNAs in editing. In both models the ini-
tial interaction of the gRNA and the mRNA is the
formation of an anchor duplex just downstream
of the pre-edited region. The "enzyme cascade"
model invokes an endonuclease that cleaves at
the first mismatch, a terminal uridylyl transferase
that adds a U to the 3'-hydroxyl, and an RNA ligase
joining the two mRNA fragments. In the "transes-
terification" model, the added U's are derived
from the oligo-[U] tail of the gRNA by means of
two successive transesterifications. In both mod-
els the gRNA provides an internal guide sequence
to specify the precise addition and deletion of U's
at specific sites.
A major goal is to obtain an in vitro system in
which the entire editing process occurs in an ac-
curate manner, thereby allowing a biochemical
dissection and reconstitution of the underlying
enzymatic machinery. We have recently shown
that at least the initial step of RNA editing — the
formation of gRNA-mRNA chimeric molecules —
577
The Mitochondrial Genome of Trypanosomes
can be performed in vitro by incubation of syn-
thetic RNAs with a mitochondrial extract and that
this process is anchor-dependent.
The work on RNA editing was supported by a
grant from the National Institutes of Health.
Evolutionary Considerations
' The existence of split genes is not novel, but
the existence of genes in which the RNA product
of one gene contains information for the correc-
tion of coding sequences within transcripts of the
other gene is novel. If the transesterification
model proves to be correct, this would suggest
that RNA editing is on the same evolutionary
pathway as RNA splicing and may in fact repre-
sent a primitive type of trans-splicing with partial
integration. The trypanosome type of RNA editing
has not yet been reported in other organisms, but
other types of modifications of the sequences of
coding RNAs have been observed in organisms as
diverse as plants and humans. C to U changes at
specific sites occur in transcripts of several hu-
man genes and also in transcripts from many plant
mitochondrial and chloroplast genes. The deter-
mination of site specificity for these multiple
transitions is completely unknown.
The evolution of RNA editing in the trypano-
somes is interesting in itself. To investigate it, we
plan to examine representatives of more-primi-
tive kinetoplastid lineages.
The Kinetoplast Genome as a Target
for Disease Diagnosis
We have shown that the kinetoplast DNA mini-
circle molecule of Trypanosoma cruzi, the
causal agent of Chagas' disease, is an appropriate
multicopy target for detection and strain classifi-
cation of the parasite in patients, animals, or in-
sects. The minicircle of T. cruzi consists of four
conserved regions and four variable regions.
There are multiple minicircle sequence classes in
a network, and minicircles in different strains are
very polymorphic. Primers to the conserved re-
gion were used for polymerase chain reaction
(PGR) amplification of minicircle fragments
from either the conserved region or the variable
region. This method is being developed into a
diagnostic procedure to detect small numbers of
parasites in blood of chronically ill patients and
to classify the strain of the parasite.
We recently developed a method to recover
blood from patients and preserve total DNA with-
out refrigeration. The DNA is then cleaved with a
chemical nuclease to release linearized minicir-
cles, and the fragments are amplified with spe-
cific primer sets. We are attempting to expand
this diagnostic procedure by developing a multi-
plex PCR-based assay for the detection of multi-
ple blood-borne viral and parasitic disease
agents.
The kinetoplast DNA minicircle has also
proved to be an appropriate target for the diagno-
sis of other pathogenic kinetoplastid infections,
such as those caused by Leishmania species.
Grants from the World Health Organization
and the Rockefeller Foundation supported this
diagnostic work.
RNA Editing as Possible Target
for Intervention
Whenever a parasite has a biochemical path-
way that is unique to the parasite and not found in
the human host, there is a potential for selective
chemotherapy. As RNA editing is dissected bio-
chemically, we plan to examine the possibility of
inhibiting components of the editing machinery
in the parasites without affecting the host. This
may open up a new direction for selective che-
motherapy of the many trypanosome-caused
diseases.
378
Regulation of Gene Activity During
B Cell Development
Harinder Singh, Ph.D. — Assistant Investigator
Dr. Singh is also Assistant Professor in the Department of Molecular Genetics and Cell Biology at the
University of Chicago. He received his Ph.D. degree in biochemistry, molecular biology, and cell biology
with Lawrence Dumas at Northwestern University. His postdoctoral research was done with Phillip Sharp
at the Massachusetts Institute of Technology, as a Jane Coffin Childs fellow. He remained at MIT as a
research associate until his move to Chicago.
MY research interests are focused on the
analysis of transcriptional regulatory cir-
cuits that turn genes on or off during the growth
and differentiation of B lymphocytes, cells that
produce antibodies. The B cell lineage is a very
useful model for exploring the molecular basis of
differential gene activity in mammalian develop-
ment. We are seeking answers to the following
questions. What is the nature of the genetic ele-
ment (s) linked to a target gene that controls its
transcriptional activity? What is the nature of the
regulatory protein (s) that recognizes this genetic
element? How upon binding near its target gene
does the regulatory protein modulate the activity
of the enzyme complex that transcribes the gene?
What is the mechanism by which a growth or de-
velopmental signal is transduced by the regula-
tory protein (s) to effect gene activity? How does
the structure of the chromatin within which the
target gene is packaged influence the function of
the regulatory protein (s)?
The heavy- and light-chain genes encoding the
immunoglobulin (Ig) molecule are selectively
transcribed in B cells. These genes are assembled
from gene segments through an ordered series of
somatic recombination events that occur in a de-
veloping B cell. The heavy-chain gene locus is the
first to undergo recombination and transcrip-
tional activation, thereby defining the pre-B cell
developmental state. Subsequently one of two
light-chain gene loci is recombined and ex-
pressed, resulting in the development of a mature
B cell. Ig genes contain multiple cis-acting tran-
scriptional regulatory elements that restrict their
expression to appropriate stages in the B lineage.
Previous work by various research groups has
resulted in the identification, characterization,
and cloning of a B cell-specific regulatory pro-
tein, Oct-2. This protein recognizes the octanu-
cleotide sequence ATTTGCAT, which confers B
cell specificity to Ig gene promoters. The same
sequence motif is also a functional component of
the heavy-chain and K-light-chain gene en-
hancers. Oct-2 can activate transcription of a re-
porter gene linked to an Ig promoter in a non-B
cell. Thus Oct-2 appears to be both necessary and
sufficient for regulating the activity of Ig pro-
moters. The deduced amino acid sequence of
Oct-2 reveals a region of similarity that is shared
with three other regulatory proteins: Pit- 1 , Oct- 1 ,
and unc-86. This region is termed the POU box
and includes a subdomain related to the
homeobox.
The levels of the Oct-2 protein are regulated
during B cell differentiation. The Oct-2 protein is
expressed at low levels in pre-B cells. Oct-2 lev-
els are elevated 5- to 10-fold upon pre-B cell
differentiation, and higher Oct-2 protein levels
correlate with activation of the Ig /c-light-chain
gene locus. Increased expression of Oct-2 in
pre-B cells is induced by signaling with the B cell
mitogen, bacterial lipopolysaccharide, as well as
the lymphokine interleukin-1 (IL-1). Transform-
ing growth factor-/8, an inhibitor of k gene induc-
tion in pre-B cells, blocks the up-regulation of
Oct-2 but not the activation of NF-kB. The latter is
another regulatory protein that has been previ-
ously implicated in controlling the activity of the
K locus. We have proposed a model in which the
concerted action of increased levels of Oct-2 and
activated NF-kB controls the proper stage-specific
expression of the k locus. To test genetically the
function of Oct-2 in regulating immunoglobulin
gene transcription, we have used gene targeting
to disrupt sequentially the two copies of the
Oct-2 gene in a B cell. The analysis of the mutant
and wild-type cells is currently under way.
Expression of the Oct-2 gene has been shown
to be regulated at the level of transcription dur-
ing pre-B cell differentiation. By studying the
control of Oct-2 expression, regulatory proteins
interacting with Oct-2 in the genetic hierarchy
underlying B cell difi'erentiation can be identi-
fied and isolated. A promoter controlling tran-
scription of the Oct-2 gene has been identified.
This region is being analyzed and will enable the
characterization of transcription factors that
regulate Oct-2 expression as well as B cell
development.
B cells need to activate the expression of two
other genes, mb- 1 and B29, to function as anti-
gen-recognizing cells. These genes encode mem-
brane proteins that associate with the antibody
379
Regulation of Gene Activity During B Cell Development
molecule (antigen receptor) and are required for
the expression of the receptor on the cell surface.
The mb-l gene is activated early in B cell ontog-
eny, continues to be expressed in mature B cells,
but is turned olF in plasma cells. We have identi-
fied the mb-l promoter and shown that it con-
tains a regulatory domain that functions in a cell-
=-«ype and stage-specific manner. A 25-base pair
element within this domain is necessary and suf-
ficient for activity. This element is recognized by
a novel transcriptional activator termed BLyF,
whose distribution correlates with the pattern of
expression of mb-l in B lineage cells. Thus BLyF
may play an important role in B cell develop-
ment, in part by regulating the activity of the
mb-l gene.
380
Regulation of Gene Expression
in Developing Lymphocytes
Stephen T. Smale, Ph.D. — Assistant Investigator
Dr. Smale is also Assistant Professor of Microbiology and Immunology and a member of the Molecular
Biology Institute at the University of California School of Medicine, los Angeles. He received his Ph.D.
degree in biochemistry from the University of California, Berkeley, where he studied with Robert Tjian.
Dr. Smale 's postdoctoral research was done with David Baltimore at the Whitehead Institute,
Massachusetts Institute of Technology.
.1
HEMATOPOIESIS refers to a complex develop-
mental process through which pluripotent
stem cells in mammalian fetal liver or adult bone
marrow give rise to several types of terminally
differentiated blood cells. The immune system in-
corporates many of these cell types, which in-
clude B and T lymphocytes, granulocytes, mono-
cytes, and killer cells, to protect the host from
infection by a variety of means. When defects
arise in the hematopoietic pathway, the effects
are often severe. In some cases, precursor cells
cannot develop to maturity, resulting in immuno-
deficiency. In other cases, uncontrolled prolifera-
tion of developing cells results in leukemias and
lymphomas. To understand the basis of these de-
fects, we must explain the regulation of hema-
topoiesis at the molecular level.
The approach used in our laboratory to study
the regulation of the hematopoietic pathway is to
identify proteins that directly activate or inacti-
vate genes expressed at specific stages of B and T
lymphocyte development. Lymphocytes play a
central role in the immune response by mediat-
ing the recognition of foreign and infectious mat-
ter. As B and T cells mature, a wide variety of
genes are turned on and off in a specific temporal
pattern. An analysis of the mechanisms by which
these genes are precisely controlled is a good
starting point for studying the regulation of
hematopoiesis.
Our laboratory currently focuses on the regula-
tion of the gene encoding terminal deoxynucleo-
tidyltransferase (TdT) . This gene is turned on for
only a short time during both B and T cell devel-
opment. The TdT protein is a template-indepen-
dent DNA polymerase that appears to play a role
in generating diversity within the antibody and T
cell receptor molecules.
We chose to analyze the TdT gene because of
its expression patterns in normal and leukemic
cells. In normal cells TdT expression is unusual
because it is found in both early B and T cells,
suggesting that it may be regulated by transcrip-
tion factors common to these two related but dis-
tinct lineages. Moreover, TdT is expressed at high
levels in acute lymphocytic leukemias (ALLs) and
at lower levels in many acute myeloid leukemias
(AMLs). TdT expression in AMLs is particularly
intriguing because it does not occur in normal
myeloid cells or myeloid precursors. Therefore
the abnormal expression of TdT in AMLs suggests
that the protein or proteins that deregulate TdT
may also play a role in leukemogenesis.
To study the mechanisms of regulating TdT ex-
pression, we have analyzed the promoter, the
transcriptional control region surrounding the
start site for TdT RNA synthesis. Our analysis fo-
cuses on two aspects of promoter function: 1) the
DNA sequence elements and proteins responsible
for directing TdT expression specifically in lym-
phoid cells and 2) the unusual architecture of the
TdT promoter region, which is fundamentally
different from that of most other mammalian pro-
moters that have been studied in detail. Our pro-
gress in these two areas is described below.
Regulated Expression of the TdT Gene
To understand the specific activation of the
TdT gene in B and T lymphocytes, we are
currently focusing on a DNA sequence element
located 60 nucleotides upstream from the TdT
transcription start site. We have shown that this
element of approximately 15 base pairs, called
D', is essential for efficient promoter activity in
lymphocytes. We have identified two different
classes of proteins that interact with the D' re-
gion, one or both of which may be important for
TdT transcription. We identified the first protein,
called LyF- 1 , by looking for protein-DNA interac-
tions at the D' site in protein extracts derived
from lymphocytes. We have characterized this
novel protein after purifying it from a murine
thymoma cell line. LyF-1 is a 50-kDa protein and
was found to be highly enriched in lymphoid
cells relative to a variety of nonlymphoid cells.
Moreover, LyF- 1 was found to bind to two differ-
ent sites in the TdT promoter as well as to the
promoters for four other lymphocyte-specific
genes.
The second class of proteins that bind to the
TdT D' site are those related to the ets family of
DNA-binding proteins. Mammalian ets-1 was first
581
Regulation of Gene Expression in Developing Lymphocytes
identified as the cellular homologue to a retro-
viral oncogene and was shown only recently to
bind to specific DNA sequence elements. As with
most genes expressing DNA-binding proteins, the
ets- 1 gene is related to several other genes found
in the mammalian genome, all of which encode
proteins that bind to similar DNA sequence ele-
ments. We have expressed in bacteria three dif-
ferent members of the ets family — ets-1, fli-1,
and PU.l — and have found that all three bind
tightly to the TdT D' site. Interestingly, all three
of these proteins are expressed at high levels in
certain subsets of hematopoietic cells, and both
ets-1 and fli-1 are expressed in most, if not all,
cell types that express the TdT gene.
We have determined that LyF-1 is not ets-1 or
fli-1 and is unlikely to be a member of the ets
family of proteins. Therefore our current goal is
to determine which of these proteins — LyF-1,
ets- 1 , fli-1 , or a combination of LyF- 1 and an ets
member — is required to activate TdT transcrip-
tion by interacting with the D' element. Those
proteins required for TdT activation are likely to
be important for regulating lymphocyte develop-
ment and will become candidates for involve-
ment in the deregulation of TdT expression in
ALL and AML. This work is supported by a grant
from the National Institutes of Health.
Unusual Design of the TdT Promoter
The TdT promoter is unusual in that it does not
contain a TATA box, which is a common DNA se-
quence element found in most promoters that
have been characterized in detail. The TATA ele-
ment is known to carry out several important
functions in promoters, including 1) recruitment
of the RNA polymerase to the transcription start
site region, 2) determination of the site's loca-
tion, and 3) determination of the direction of
transcription from the promoter. To understand
the regulation of the TdT promoter, we are inquir-
ing into how these functions are carried out in the
absence of a TATA box. This question is espe-
cially important for an understanding of tran-
scriptional control during lymphocyte develop-
ment, since it appears that the promoters of most
genes expressed during this process do not con-
tain TATA boxes.
Previously we found that in place of the TATA
element, which is typically located 30 nucleo-
tides upstream of the transcription start site, the
TdT promoter contains a distinct element that in-
stead overlaps the start site. This element, which
we call an initiator (Inr) , is like the TATA element
in its importance for promoter function and also
in pinpointing the RNA start site to a specific
nucleotide.
We are now comparing further the activities
and mechanisms of action of TATA boxes and Inr
elements. Currently our data suggest that the
mechanisms for TATA and Inr-mediated tran-
scription follow very similar steps prior to the
actual onset of RNA synthesis.
We would also like to understand the contribu-
tions of TATA and Inr elements toward determin-
ing the start site location and the direction of
transcription within a promoter. Some promoters
contain both TATA and Inr elements. Although
both elements can influence start site localiza-
tion and transcriptional directionality, the TATA
element appears to be much more powerful than
an Inr. Our results, however, also have led us to
challenge the belief that the orientation of a TATA
box within a promoter alone determines the di-
rection of transcription. Instead, our data suggest
that the direction of transcription within a simple
promoter is determined by the location of a TATA
box or Inr element relative to a specific activator
element.
We are continuing to define the rules by which
a promoter determines the direction of RNA syn-
thesis in order to understand further the design of
more-complex promoters, including those that
contain, like the TdT promoter, important se-
quence elements both upstream and downstream
of the transcription start site.
382
Developmental Genetics
Philippe M. Soriano, Ph.D., D.Sc. — Assistant Investigator
Dr. Soriano is also Assistant Professor at the Institute for Molecular Genetics and Department of Cell
Biology, Baylor College of Medicine. He obtained his Ph.D. degree in biochemistry and his D.Sc. degree
from the University of Paris. He did postdoctoral research in France, and then with Rudolf Jaenisch in
Germany and at the Whitehead Institute for Biomedical Research of the Massachusetts Institute of
Technology, before joining the faculty at Baylor. Dr. Soriano is a Pew Scholar in the biomedical sciences.
THE major aim of our research is to extend the
understanding of early development of the
mouse. Our approach is to create mice that are
mutant in genes critical for this process. Two
techniques are being used: random mutagenesis,
which should discover unknown genes, and tar-
geted mutagenesis in previously characterized
genes for which no mutation is known.
The random mutagenesis project is designed to
identify developmentally regulated and critical
genes in the embryo. Identification of the mu-
tated gene in many of the classical mouse mu-
tations has been difficult because there is no
convenient tag by which to clone the muta-
tion. Insertion mutagenesis in transgenic mice,
wherein a gene is disrupted by introducing a for-
eign fragment of DNA into the germline, is attrac-
tive, since the transgene can serve as a tag. How-
ever, only 1 out of 20 transgenic strains displays
an overt phenotype, so the approach is laborious
and time consuming. We have circumvented this
problem by preselecting for mutations in embry-
onic stem (ES) cells, which are then introduced
into embryos to colonize the germline.
The selection procedure we have used is
termed "promoter trapping." In this method, a
reporter gene is placed downstream of a splice
acceptor, and the construct is then introduced
into ES cells. Expression of the reporter gene can
only originate from a flanking cellular promoter.
Therefore transgenic mice derived from ES cells
selected for such events can be used both to trace
the activity of the tagged gene, by expression of
the reporter gene, and to mutate the gene. The
reporter gene we have used encodes a fusion pro-
tein with both (8-galactosidase (;8-gal) and neo-
mycin phosphotransferase (neo) activity. This al-
lows one to select directly for promoter trap
events, since neo confers resistance to the drug
G418, and to follow the activity of the trapped
gene by incubating the embryos in a dye that
turns cells blue if |8-gal is present.
Thirty-one transgenic lines have now been gen-
erated using a retroviral promoter trap vector and
have been examined both for patterns of expres-
sion and for phenotype. We are particularly inter-
ested in strains that exhibit restricted patterns of
expression at gastrulation, a critical stage of de-
velopment in the early mouse embryo. Twelve
strains carry a recessive lethal mutation due to the
promoter trap insertion, and one induces steril-
ity. This demonstrates that the method is very
valuable for isolating developmental mutants.
Further analysis of one of the embryonic lethal
strains suggests a mutation in a transcription fac-
tor. The other strains, which do not display a phe-
notype associated with the promoter trap event,
may reflect mutations in nonessential genes. This
first aspect of our research is supported by the
National Institutes of Health.
It is also possible to make specific mutations by
targeting genes believed to play critical roles in
development, based on their previous character-
ization. This approach relies on the ability to in-
troduce into ES cells mutant gene constructs that
will recombine homologously with the normal
gene, at the correct chromosomal location, there-
fore creating a mutation in the gene of interest.
Our initial efforts focused on the gene encoding
c-src, the first proto-oncogene identified.
Src is a protein-tyrosine kinase, broadly ex-
pressed but at particularly high levels in neurons
and in platelets. The src gene was knocked out in
ES cells, and the mutation was transmitted
through the germline. Surprisingly, animals ho-
mozygous for the mutation do not die at an early
stage, but develop osteopetrosis, a disease char-
acterized by an impaired function of the osteo-
clasts, the cells that normally resorb bone. As a
result of the mutation, the mutant mice fail to
undergo normal tooth eruption and have to be
maintained on a soft food diet. We have been able
to show that the defect lies in the osteoclasts
proper, rather than in accessory cells such as os-
teoblasts that condition osteoclast activity. How-
ever, no defect is found in cells or tissues where
src is most highly expressed.
To explain the lack of a more severe pheno-
type, we have tested the hypothesis that another
5rc-related kinase may be compensating for loss
of src in the mutant mice. Mice mutant in the
closely related c-yes and fyn genes have therefore
been generated by homologous recombination in
ES cells. Neither of these mutations results in an
383
Developmental Genetics
overt phenotype. However, further analysis of the
fyn mutation has revealed that thymocytes are un-
able to proliferate in response to stimulation
through the T cell receptor, demonstrating that
this kinase is essential for normal thymocyte
function. Surprisingly, peripheral T cells can re-
spond through the T cell receptor, implying that
-Other kinase (s) may be involved. This mutation
may have further effects on the repertoire of T
cells, and further studies are under way.
To test the concept that 5rc-related kinases
compensate for one another, we have crossbred
the mutant mice and examined progeny for dou-
ble mutant genotypes. The vast majority of src-
yes and src-fyn double mutants die at birth —
supporting the compensation hypothesis. The
nature of the defect in the double mutants is
under further investigation. This second aspect of
our research has been supported in part by the
National Institutes of Health.
Expression pattern of the ^galactosidase (^-gal) gene in blastocysts of a 'promoter trap" strain
of transgenic mice, showing labeling (blue) in both the inner cell mass and the outlying trophec-
toderm. By this technique, mice derived from selected embryonic stem cells can be used both to
trace the activity of a ^-gal-tagged gene and to mutate the strain.
Research of Glenn Friedrich and Philippe Soriano.
384
Understanding How Eggs Work
Allan C. Spradling, Ph.D. — Investigator
Dr. Spradling is also a staff member of the Department of Embryology at the Carnegie Institution of
Washington in Baltimore and Adjunct Professor of Biology and Microbiology at the fohns Hopkins
University. He earned his B.A. degree in physics from the University of Chicago and his Ph.D. in cell biology
from the Massachusetts Institute of Technology. His postdoctoral study was done at Indiana University
with Anthony Mahowald. Dr. Spradling is a member of the National Academy of Sciences and has received
many honors for his work.
ALTHOUGH the union of egg and sperm initi-
ates the complex processes that ultimately
result in a new animal, the roles played by the
two cells in embryonic development are by no
means equivalent. Even nonspecific treatments,
such as pricking with a needle, will stimulate
many types of eggs to develop in the absence of
sperm. Without an egg, however, neither a sperm
cell nor any other can even begin the processes
that lead to an embryo and ultimately an adult. In
fact, eggs have undergone extensive and intricate
preparations that allow them to support and di-
rect embryogenesis. If the mystery of a new life
could be condensed into a single cell, that cell
would surely be an egg.
Eggs Pose Exceptional Biological Problems
Egg cells are very different in structure and in
biological capacity from other cells. Most
chicken cells weigh less than one millionth of an
ounce, whereas a single chicken egg makes a nice
breakfast. Eggs not only differ in size but in many
other important ways. For example, the egg's
genes function differently from those of other
cells; so extensive are these differences that the
chromosomes appear in the microscope quite
unlike those of other tissues. In eggs the products
of many genes accumulate and are stored in spe-
cial forms so that they can be utilized at precisely
the right times during embryonic development.
Indeed, storage of material is one of the reasons
many eggs are enormous.
A great deal remains to be learned about egg
structure. How are gene products specially
stored, and what are their specific functions later
in development? Is each product located in a par-
ticular place? What are the mechanisms that al-
low these materials to be produced and stock-
piled in the appropriate manner? The sheer
complexity of an egg, with its tens of thousands
of specific, highly organized components, has
until recently blocked all attempts to describe
egg structure in molecular detail, much less to
understand the logic that allows this structure to
develop into an even more intricate organism.
However, over the past 10 years or so, powerful
new methods in molecular biology, such as gene
cloning, gene transfer, and fruit fly genetics, have
begun to unravel the fascinating secrets stored
within an eggshell.
Using Genetics to Study Eggs
It is now possible to study structures as com-
plex as eggs because techniques of both genetics
and molecular biology allow us to take eggs apart
gene by gene. Each component of an egg is speci-
fied by a gene carried on the chromosomes. Much
research on eggs utilizes the fruit fly Drosophila,
since far more has been revealed about its genes
during 70 years of genetic studies than about
those of any other complex organism, including
humans. A genetic mutation in the fly inactivates
one gene among the 20,000 to 50,000, and if that
gene specifies an egg component, then the mu-
tant fly will produce defective eggs.
Frequently, however, the eggs can still par-
tially function. For example, in a particular mu-
tant, development might begin normally, but
then stop because of a lack of stored food. By
studying what goes wrong with the mutant eggs,
biologists can deduce the normal function of a
particular gene's product. In this case, one might
conclude that the gene was involved in produc-
ing the stored food. To understand eggs, it will be
necessary, at a minimum, to create mutations and
carry out such studies on all the fruit fly genes,
one by one.
If researchers were limited to looking at egg
defects in a microscope, progress would still be
slow. Many of the defective eggs would not con-
tain any detectable problems; they just wouldn't
work. It is at this point that the ability to clone the
gene molecularly becomes essential to further ad-
vancement. Cloning is simply a way of purifying
an individual gene so that its DNA structure can
be determined in the laboratory.
There are two major benefits. First, the struc-
ture of the gene's product, a specific protein that
actually becomes part of the egg, can now be de-
termined by the process of DNA sequencing. Sec-
ond, specific antibodies can be prepared that
only bind to the product of that particular gene.
385
Understanding How Eggs Work
When these antibodies are used, eggs that previ-
ously appeared normal in the microscope can
suddenly be seen to have specific defects. Such
molecular biological studies allow us to under-
stand in much greater detail vv^here in an egg par-
ticular gene products are located, where they go
as development begins, and what role they are
4ikely to play in the process.
Transporting Materials Into Eggs
One project of our research group concerns
how the massive amount of material found in an
egg actually gets there. In the fruit fly most of the
egg contents are piped in from 15 nearby cells
called nurse cells. Four of these cells are attached
directly to the egg by special "pipes" made of the
same material as the outer surface of cells, and
the remaining 1 1 are attached to other nurse cells
by similar pipes. Thus some of the material in an
egg has been shipped through as many as four
nurse cells on the way to its destination. Nurse
cells not only control somehow the rate and direc-
tion of transport, but even select certain products
for priority delivery, moving them well ahead of
the majority of the egg's contents.
To learn more about the equipment flies use to
make their eggs, we identified some mutant
strains in which either the connections between
nurse cells and egg did not form properly, or ma-
terial did not flow very well. We have now stud-
ied one mutant in some detail. When mother flies
lack this gene, they are able to attach on average
only four nurse cells to their eggs. As a result the
eggs usually do not grow properly and are unable
to function. Rarely, a developing egg becomes
connected to as many as 11 nurse cells, and a few
of these eggs can develop, so the affected females
are not completely sterile.
We determined the structure of the protein that
was missing in the mutant, and were surprised to
learn that it was very similar to one found in large
amounts just under the surface of human and
other mammalian red blood cells. This protein,
called "adducin," has been studied for some time
by Vann Bennett (HHMI, Duke University). It has
been found at various levels in many types of cell,
where it is thought to strengthen cell membranes
by acting as a kind of skeleton supporting the
floppy lipid bilayer. Red cell adducin is thought
to act with other membrane skeleton proteins to
mold red cells into their characteristic concave
disk shape and to lend the physical properties
that let them pass through small capillaries. As
with many human proteins, however, the actual
function is uncertain because the effects of re-
moving a red cell's adducin are unknown.
At present we are attempting to learn in detail
what adducin does to help nurse cells and eggs
hook up properly. The pipes form during cell di-
vision. Normal daughter cells separate when a re-
gion of membrane clamps down as though con-
stricted by a knot from outside. Eventually the
two cells are choked off completely. During the
last few divisions that give rise to the nurse cells
and egg, these constrictions simply stop short,
leaving a small interconnecting region of cell
membrane. This junction subsequently forms
into a pipe. In the adducin mutants, many of the
constrictions are unable to stop, so cells that
should have built a connecting pipe become com-
pletely separated. The pipes that do form in the
mutant are much weaker looking than normal,
may not transport materials very well, and appear
susceptible to breakage.
Understanding how adducin contributes to
fruit fly egg production may provide insight into
the details of how it works in humans. The mu-
tant flies also have several other abnormalities
besides their inability to connect nurse cells and
the egg properly, indicating that the function of
adducin can be studied in a variety of different
situations. Experimenters can now make what-
ever changes in the protein's structure they wish,
introduce a modified gene encoding the altered
protein back into the mutant strain of fruit fly,
and see if the modified protein will "rescue"
some or all of the fly's problems or create new
ones. This approach can be applied to the many
different gene products that make up a fruit fly egg.
586
Structural Studies of Regulatory Proteins
Stephen R. Sprang, Ph.D. — Associate Investigator
Dr. Sprang is also Associate Professor of Biochemistry at the University of Texas Southwestern Medical
Center at Dallas. He received a B.S. degree from California State University at Los Angeles, and a Ph.D.
degree from the University of Wisconsin-Madison, where Muttaiya Sundaralingam was his advisor. His
postdoctoral training was at the University of Alberta, Edmonton, with Robert Fletterick. Again with Dr.
Fletterick, he was an assistant research biochemist at the University of California, San Francisco,
before assuming his present position.
CELLS communicate with each other by se-
creting growth factors, cytokines, or hor-
mones into the extracellular space, and by re-
sponding to factors produced by other cells.
These chemical messengers act by changing pat-
terns of gene expression within target cells,
thereby altering the cells' metabolic or develop-
mental program. Among the major unsolved ques-
tions in biology is how cells specifically recog-
nize these chemical messengers and how the
messages are transduced within cells.
We are using x-ray crystallographic techniques
to define the molecular nature of the interaction
between specific growth factors and their cog-
nate receptors. Receptors are themselves protein
molecules distributed on the outer surface of the
plasma membrane surrounding the target cell.
The goal of our research is to learn how receptor
molecules specifically recognize growth factors,
and how, if at all, the receptors alter their struc-
ture and chemical properties as a consequence of
their interaction with the factors.
Tumor Necrosis Factors and Fibroblast
Growth Factors
Tumor necrosis factors (TNF) are produced by
cells of the immune system. Macrophages — the
white blood cells responsible for engulfing cel-
lular debris — produce large quantities of TNF-a
(also known as cachectin) when stimulated by
toxins carried on the surface of bacteria, as oc-
curs in an infection. The avid attachment of this
cytokine to receptors present on a variety of cells
in the body triggers a series of events that mediate
inflammation, endotoxic shock, and the wasting
phenomenon, cachexia, from which the cytokine
takes one of its names.
A related cytokine, TNF-|8, is produced by T
lymphocytes and has many properties in common
with TNF-a, including the ability to bind to the
same receptors. Both molecules are toxic to many
types of tumor cells.
Two different types of TNF receptor have been
found which, surprisingly, bear little amino acid
sequence identity to each other. Both receptors,
however, contain a structural "motif" composed
of four inexact copies of a repeat rich in the sul-
fur-containing amino acid cysteine. Preliminary
data from our own and other laboratories suggest
that each trimer of TNF-a interacts with three re-
ceptor molecules. We hope to learn how these
distantly related molecules interact with the
same receptors and how this interaction might
mediate a cellular response.
The three-dimensional atomic structures of
TNF-a, and most recently of TNF-/?, have been
determined by Michael Eck, a graduate student in
the laboratory (and now a Research Associate
with Don Wiley and Stephen Harrison at HHMI,
Harvard University). Despite the differences in
amino acid composition and sequence, TNF-a
and TNF-/3 have the same three-dimensional
structure and assemble to form trimers composed
of three identical protein subunits packed about
a threefold axis of symmetry. However, because
of the differences in the amino acid composition
of the two cytokines, the surface of the TNF-a
trimer is chemically quite different from that of
the TNF-/3 trimer. We hope to determine how the
same receptor can recognize these different mo-
lecular surfaces with equal affinity.
More importantly from a pharmacological per-
spective, we would like to be able to design TNF
molecules that interact with one, but not the
other receptor, or receptor antagonists that inter-
act with only one of the two TNF species. Toward
this goal, we are now attempting to determine the
structure of theTNF-receptor complexes. The lab-
oratory has been successful in growing crystals of
the TNF-binding fragment of one of the two re-
ceptors, and crystallographic studies are now
under way. We are also attempting to produce
crystals of the complex formed between TNF-a
and one of the receptors in order to study directly
the interaction between the two molecules.
Last year we reported the three-dimensional
structure of basic fibroblast growth factor
(bFGF), one of a group of seven structurally re-
lated proteins that promote cell division. This
factor is present in the space between cells, par-
ticularly in endothelial tissue, where it stimulates
the movements of cells to sites of tissue genera-
587
Structural Studies of Regulatory Proteins
tion or repair. Our work has lately focused on the
interaction between bFGF and heparin, a com-
plex, negatively charged sugar polymer that coats
endothelial cells and is required to induce the
mitogenic response that bFGF elicits. In a collabo-
ration with Phillip Barr of Chiron, Inc., the labo-
ratory is also attempting to determine the struc-
ture of the receptor for bFGF.
The signals generated by the engagement of ex-
tracellular messenger molecules with their re-
ceptors are, in many cases, transmitted across the
plasma membrane to members of the G protein
family. For example, when epinephrine binds to
the /3-adrenergic receptor, an intracellular pro-
tein called Gsa is induced to bind the nucleotide
GTP and discard a pair of regulatory subunits.
The GTP-bound Gsa is then an activator of a
membrane-bound enzyme that catalyzes the syn-
thesis of the intracellular messenger molecule
cyclic AMP. In collaboration with Alfred Gilman
(University of Texas Southwestern Medical
Center at Dallas), we are undertaking crystallo-
graphic studies of Gsa and the related protein
Gia to learn how these proteins may interact with
other components of the signal transduction sys-
tem. Research Associate David Coleman has ob-
tained small crystals of the complex between Gia
and a nonhydrolyzable GTP analogue, and we
hope soon to resolve the structure of this fascinat-
ing molecule.
Glycogen Phosphorylase
A second focus of the laboratory has been to
understand how a certain class of complex biolog-
ical catalysts, called allosteric enzymes, regulate
their own activity. An example of such an en-
zyme, studied in our laboratory, is glycogen
phosphorylase. This molecule catalyzes the
breakdown of a storage carbohydrate called gly-
cogen into sugar units that can be used directly by
the body to fuel muscle contraction or to main-
tain constant levels of glucose in the blood. The
activity of this enzyme increases when it binds to
"cellular messenger" molecules such as adeno-
sine monophosphate, which signals an energy
deficit in the cell, and decreases when it binds
glucose, which signals an energy surplus. Glyco-
gen phosphorylase can be chemically modified
(phosphorylated) by other enzymes in response
to hormonal signals (epinephrine), which also
increases the catalytic activity of this enzyme.
Our goal is to understand the molecular mechan-
ics of the process by which catalytic machines
such as phosphorylase can alter their activity in
response to the cellular messengers.
We have recently determined the three-dimen-
sional structure of active glycogen phosphorylase
with the activator AMP bound to a regulatory site
in the molecule. In comparing this structure with
the inactive conformation of the molecule deter-
mined in Robert Fletterick's laboratory (Univer-
sity of California, San Francisco) , we have learned
how the three-dimensional structure of the mole-
cule is altered in such a way as to increase the
affinity of the enzyme for glycogen.
388
Insulin and the Islets of Langerhans
Donald F. Steiner, M.D. — Senior Investigator
Dr. Steiner is also A. N. Pritzker Distinguished Service Professor of Biochemistry and Molecular Biology
and of Medicine at the University of Chicago Pritzker School of Medicine. He received his M.D. degree at
the University of Chicago. His interest in insulin developed during postdoctoral training with Robert
Williams at the University of Washington School of Medicine. After joining the faculty at the University of
Chicago, he studied insulin action in the liver and, later, insulin biosynthesis. This work led to his
discovery of proinsulin and preproinsulin. Dr. Steiner has received many honors, including the Lilly and
Gairdner Awards, the Wolf Prize in Medicine, and several honorary degrees.
INSULIN is essential for normal growth and uti-
lization of food. Diabetes, a disease due to
insulin deficiency or defects in its action, is char-
acterized by high blood sugar and such complica-
tions as blindness, heart disease, stroke, and in-
creased susceptibility to infections. It affects 2-3
percent of people in developed countries.
Diabetes can be controlled through various
combinations of diet, oral hypoglycemic agents,
and/or insulin injections, depending on the type
and severity of the disease. Such therapies, how-
ever, are often less than fully satisfactory because
they may only retard the development of compli-
cations. A better understanding of how insulin is
formed and secreted in a regulated manner into
the bloodstream and how it acts on tissue recep-
tors to control metabolism and growth are vitally
important to the development of new therapeutic
approaches.
Insulin is only made in the islets of Langerhans
— small clusters of cells dispersed throughout
the pancreas. Specialized islet cells also secrete
other hormones that influence metabolism, in-
cluding glucagon, somatostatin, amylin (or islet
amyloid polypeptide), and pancreatic polypep-
tide. The islet hormones, like other regulatory
peptides in the body, are derived from larger pro-
teins called prepro hormones.
These precursors contain additional portions
that may guide them along special intracellular
pathways, where they are concentrated into stor-
age vesicles and processed into their biologically
active forms. The contents of these vesicles are
then released into the bloodstream in varying
proportions to meet physiological requirements.
One goal of our research is to learn more about
how newly formed prohormones are separated
from other proteins in the cell, concentrated into
secretory granules, and processed into active hor-
mones by highly specialized enzymes before be-
ing secreted.
Insulin secretion from |8-cells in response to
elevated plasma glucose is a complex electro-
chemical process resembling the transmission of
nerve impulses or the contractions of the heart.
In the insulin-producing |8-cells, a specialized
sensor mechanism couples the metabolism of
glucose to ion channels in the plasma membrane.
These channels, upon membrane depolarization,
allow the selective entry of calcium into the cell,
triggering the release of secretory granule con-
tents. Certain oral hypoglycemic agents used to
treat diabetes — the sulfonylurea drugs — appear
to stimulate insulin secretion by inhibiting a spe-
cialized potassium channel in the ;8-cell mem-
brane, which then initiates electrical depolariza-
tion of the cell. We are trying to learn more about
the structure of this and other important ion
channels in insulin-producing cells to under-
stand both their normal functions and their possi-
ble malfunction in some forms of diabetes.
We also are studying mutations that affect insu-
lin or proinsulin structure. Some of these occur
in families and are associated with mild diabetes
as a result of the synthesis of abnormal insulin
molecules with greatly reduced biological activ-
ity. Other mutations in the insulin gene primarily
affect the conversion of proinsulin into insulin,
leading to elevated proinsulin in the circulation.
Insulin acts on tissues by binding to a large and
complex protein receptor that is present on the
surface of most cells, activating a tyrosine kinase
that alters many intracellular processes through a
cascade of intracellular phosphorylations. Insu-
lin binding to the receptor also leads to the up-
take and degradation of the hormone in the liver
and other tissues. This process, known as recep-
tor-mediated endocytosis, plays an important
role by rapidly removing insulin from the circu-
lation. By studying inherited defects in proinsu-
lin and insulin receptor molecules, we are learn-
ing more about the normal processes of islet
hormone production and action and how their
derangement can lead to disease.
Prohormone-converting Enzymes
We have recently identified two cDNAs, which
we call PC2 and PC3, in neuroendocrine cells.
These encode proteases having catalytic domains
similar to that of Kex2, a yeast prohormone-
589
Insulin and the Islets of Langerhans
processing enzyme. PC2 is more abundant in islet
(S-cells, while PC 3 predominates in the anterior
pituitary. Expression of PC2 in Xenopus oocytes
gives rise to a 68-kDa protein that is active on
substrates having characteristic sequences for
prohormone cleavage recognition (i.e., Lys-Arg
or Arg-Arg) . The proteolytic activity requires cal-
cium ions and is highest at pH 5.5. Other expres-
sion studies carried out in collaboration with
Gary Thomas and his co-workers at the Vollum
Institute indicate that PC 2 and PC 3 separately or
together can appropriately process both proopio-
melanocortin or proinsulin. Studies are now in
progress on the genes encoding these proteases
and their evolutionary origins, as well as on the
synthesis, sorting, and activation of PC2 and/or
PC3 in various neuroendocrine cells.
Regulation of Insulin Secretion
We have recently isolated and characterized
cDNA and genomic clones encoding several volt-
age-dependent K"^ channel isoforms expressed in
human islets or insulin-producing tumors. These
channels are related to the Shaker family of Dro-
sophila K"^ channels. Their electrophysiological
characteristics have been examined by voltage
clamp recordings of oocytes injected with syn-
thetic mRNA. Efforts are also under way to iden-
tify and characterize an ATP-dependent potas-
sium channel believed to play a key role in
initiating /S-cell depolarization in response to
glucose or sulfonylureas.
Insulin Receptor Studies
Studies nearing completion have revealed an
interesting and potentially important aspect of
insulin proreceptor processing, namely that un-
cleaved proreceptors having exon 1 1 (which en-
codes 1 2 amino acids near the carboxyl terminus
of the a-subunit) are fully functional, while those
lacking this exon (due to alternative splicing of
the mRNA) are markedly reduced in their afi&nity
for insulin. Since most cells express receptors
without exon 1 1 , failure to cleave the prorecep-
tor (as occurs in a family we have studied that has
mutant proreceptors that cannot be processed)
leads to severe insulin-resistant diabetes. The
liver, kidney, islet ;8-cells, and placenta are the
only tissues that express insulin receptors with
exon 1 1 , suggesting that alternative splicing may
be physiologically relevant.
Biosynthesis of Islet Amyloid
Polypeptide (lAPP)
LAPP, or amylin, a recently discovered product
of the (8-cell, is a peptide related to CGRP (calci-
tonin gene-related peptide) that is found in amy-
loid deposits in the islets of elderly diabetics and
may play a role in impairing |8-cell function. We
have characterized the gene and cDNAs encoding
the precursor of lAPP in humans and several other
mammalian species. We are currently studying a
transgenic mouse model to identify factors that
may contribute to amyloid deposition (in collabo-
ration with Niles Fox of the Lilly Research
Laboratories) .
Evolution of Insulin and Insulin-like
Growth Factors
We have used the polymerase chain reaction to
identify genes encoding insulin or the closely re-
lated insulin-like growth factors IGF-I and -II in
primitive vertebrates. The identification of a hy-
brid insulin/IGF molecule in amphioxus, a pro-
tochordate, suggests that the insulin-like growth
factors diverged from an ancestral preproinsulin-
like protein in the very earliest stages of verte-
brate evolution (about 600 million years ago).
Studies nearing completion on the identification
and structure of insulin and IGF receptors in
these lower forms also lend support to this model
for the origin of the insulin-like growth factors as
unique vertebrate growth regulators.
390
Autoantibody Probes for Mammalian
Gene Expression
Joan A. Steitz, Ph.D. — Investigator
Dr. Steitz is also Professor of Molecular Biophysics and Biochemistry at Yale University School of Medicine.
She received her Ph.D. degree in biochemistry and molecular biology ( with James Watson ) from Harvard
University and did postdoctoral work at the Medical Research Council Laboratory of Molecular Biology
(with Frances Crick) in Cambridge, England. Her many honors include the Passano Foundation Young
Scientist Award, the Eli Lilly Award in Biological Chemistry, the U.S. Steel Award in Molecular Biology,
the National Medal of Science, the Dickson Prize for Science, the Warren Triennial Prize (shared with
Thomas Cech), and the Christopher Columbus Discovery Award. Dr. Steitz is a member of the National
Academy of Sciences.
KNOWLEDGE gained from basic research in
the biomedical sciences sometimes pro-
vides answers to key questions in clinical medi-
cine. Occasionally, however, the path is reversed
and clinical studies provide information or mate-
rials that help to unravel basic biological pro-
cesses. An example of this is our use of sera from
human patients to determine the roles of previ-
ously mysterious small particles in normal cells.
Particles called small nuclear ribonucleoproteins
(snRNPs: pronounced "snurps") are found in the
nucleus of the cells of humans and other higher
organisms. Each snRNP is a tight cluster of one or
more proteins with a small RNA molecule.
SnRNPs are abundant in virtually all human cells
and are remarkably similar among various mam-
malian species, suggesting that the particles must
play important cellular roles.
Systemic lupus erythematosus (SLE) is one of a
number of diseases in which the immune system
mistakenly makes antibodies against the body's
own molecules. Curiously, molecules that are
very abundant in cells and highly conserved in
evolution, such as DNA, are the most common
targets of autoimmunity. Thus SLE patients often
make autoantibodies against snRNPs.
Using SLE patients' antibodies to probe both
the structures and functions of snRNPs, we have
investigated the roles of various kinds of snRNPs
in gene expression. These investigations began in
1979 when studies by Michael Lerner (then an
M.D./Ph.D. student; now HHMI, Yale University
School of Medicine) led to the hypothesis that the
most abundant snRNP in mammalian cells, called
the Ul snRNP, might be involved in RNA splic-
ing, an early step in gene expression. In the mak-
ing of a gene's product, its information coded in
DNA is transcribed into an RNA copy called pre-
messenger RNA, which is then "processed" into
mRNA to direct the synthesis of a protein. The
DNA and the pre-mRNA contain segments called
exons, which code for the gene's product, and
segments called introns, which are intermittent
noncoding regions. Before leaving the cell nu-
cleus as mRNA, the pre-mRNA is cut, the exons
are spliced together, and the introns discarded.
The individual exons must be precisely joined in
the order they originally had in the gene. Some-
times differences in the way exons are spliced
can lead to anomalous protein products in
various tissues.
Evidence that snRNPs play central roles in pre-
mRNA splicing has been obtained in several types
of experiments, including use of autoantibodies
from SLE patients to inhibit splicing in active cell
extracts. We now know that participation of the
most abundant snRNPs in mammalian cells (the
Ul , U2, U5, and U4/U6 particles) is essential and
that splicing requires assistance from the snRNP
proteins as well as their RNA molecules. SnRNPs
recognize the splice junctions and the so-called
intron branch point (where an unusual RNA
structure is formed as an intermediate in splic-
ing) and then assemble to align the exon ends so
that precise splicing can occur. In this sense,
snRNPs are much like the ribosomal subunits
(also containing both RNA and protein) that as-
semble onto an mRNA to translate it into protein.
Current efforts in splicing are directed at un-
derstanding exactly how RNA-RNA interactions in
the active splicing body (called the spliceosome)
contribute to catalysis. Here, we are using two
types of crosslinking approaches to identify con-
tacts between the pre-mRNA molecule and either
proteins or snRNAs in the assembled splice-
osome. Novel crosslinks are now being analyzed
that suggest how a cut-off exon may be held in the
spliceosome for subsequent ligation. Also evolv-
ing is an increased understanding of how the
spliceosome is related to some "self-splicing" in-
trons, which are removed without proteins or
other factors.
Mammalian cells also contain many minor
snRNPs that are closely related to the splicing
snRNPs. One is the U7, which is only 1/1 ,000 as
abundant as the splicing snRNPs. We have re-
cently demonstrated that it participates in form-
ing the 3' ends of histone mRNAs by using base-
pairing to recognize a specific sequence in the
391
Autoantibody Probes for Mammalian Gene Expression
pre-mRNA just downstream of the cut site. Other
related low- abundance snRNPs containing Ull
and Ul 2 RNA interact to form a two-snRNP com-
plex, but their function is unknown.
A new autoantibody directed against the Ull
snRNP has recently been found in the serum of a
patient with scleroderma. Also in this group of
-related snRNPs, which probably all function in
some aspect of mRNA maturation, are viral
snRNPs. For instance, in marmoset cells infected
by Herpesvirus saimiri, which causes malignant
transformation of the T cells, the viral genome
encodes five small RNAs. Current results suggest
that these viral snRNPs may act to slow the degra-
dation of mRNAs coding for oncoproteins or
growth factors, thereby enhancing the process of
cellular transformation.
Another type of patient autoantibody is di-
rected against a different class of small RNPs lo-
calized in the nucleolus, where ribosomal RNA
processing and assembly occur. The most abun-
dant nucleolar snRNP (containing U3 RNA) is es-
sential for the first step of ribosomal RNA process-
ing. Recent analyses have revealed how U3 binds
to the ribosomal pre-RNA, making specific base
pairs. Separate studies are dissecting the signals
that dictate the delivery of the U3 and related
snRNPs to their nucleolar site of action.
Yet another autoantibody type precipitates
EBERs, two small RNAs specified by Epstein-Barr
virus (EBV), the causative agent of infectious
mononucleosis, also implicated in several human
cancers. Since EBERs are among the few viral
products that are expressed in EBV-transformed
cells, they must be important to the induction or
maintenance of the transformed state. A highly
abundant, highly conserved cell protein that
binds the EBERs appears to reside in the cell's
nucleolus. We hope that its further characteriza-
tion will lead to an elucidation of EBER function.
Thus autoantibodies are potent probes for de-
ciphering some of the fundamental reactions oc-
curring in all mammalian cells, those involved in
gene expression. Characterization of new cellu-
lar particles like snRNPs is significant for future
studies of basic cellular processes and their alter-
ation by disease. Furthermore, our research has
provided new ways of diagnosing patient autoan-
tibodies, which are helpful in the diagnosis and
treatment of diseases like SLE.
592
Structural Studies of Protein-Nucleic Acid
Interactions
Thomas A. Steitz, Ph.D. — Investigator
Dr. Steitz is also Professor of Molecular Biophysics and Biochemistry and of Chemistry at Yale University.
He received a B.A. degree in chemistry from Lawrence College in Appleton, Wisconsin, and a Ph.D. degree
in molecular biology and biochemistry from Harvard University, with W. N. Lipscomb. After a
postdoctoral year at Harvard, he moved to the MRC Laboratory of Molecular Biology in Cambridge,
England, as a fane Coffin Childs fellow, with D. M. Blow. He next Joined the Yale faculty, where he has
remained, except for sabbatical work with K. Weber in Gottingen, West Germany, with A. Klug at
Cambridge, and with f. Abelson at the California Institute of Technology. He has received the Pfizer Prize
of the American Chemical Society and is a member of the National Academy of Sciences.
OUR general long-term goal is to determine
the detailed molecular mechanisms by
which those proteins and nucleic acids that are
involved in the central dogma of molecular biol-
ogy (DNA replication, transcription, and transla-
tion) achieve their biological functions. Virtually
all aspects of the maintenance and expression of
information stored in the genome involve inter-
actions between proteins and nucleic acids. We
are seeking to provide a structural and chemical
basis for these fundamental processes.
Synthetase-tRNA Complex
Enzymes called aminoacyl-tRNA synthetases
translate the genetic code by attaching the
correct amino acid to a tRNA containing the ap-
propriate anticodon. Of significant current inter-
est is how these synthetases can accurately distin-
guish among the 60 or so similar tRNA molecules.
Furthermore, how does RNA recognition by a
protein differ from DNA recognition? Finally, can
the structure of the present-day synthetase-tRNA
complex provide any insights into the evolution
of this central process and the evolution of the
genetic code itself?
We have determined the crystal structure of
glutaminyl-tRNA synthetase (GlnRS), a 64,000-
molecular-weight monomeric protein, com-
plexed with tRNA^'" and ATP. The GlnRS consists
of four domains arranged to give an elongated
molecule that interacts with the inside of the L-
shaped tRNA from its anticodon to its acceptor
end. GlnRS specifically recognizes the correct
tRNA by interactions with the three bases of the
anticodon and with base pairs of the amino acid
acceptor stem of the tRNA. The three bases of the
anticodon are unstacked and splayed out; each
base binds into a separate recognition pocket on
the enzyme. The extensive hydrogen-bonding in-
teractions between the protein and the anticodon
bases make the enzyme specific for the two gluta-
mine anticodons (UUG and CUG) but none of the
other 62 possible anticodons. The structural
bases of Gln-tRNA synthetase recognition are
currently being pursued by determining the
structures of mutant tRNAs complexed with the
enzyme.
The protein domain that contains the active
site has a structure similar to that of the homolo-
gous domains of the tyrosyl- and methionyl-tRNA
synthetases and by amino acid sequence similar-
ity is homologous to another 10 of the 20 synthe-
tases. Synthetases for 10 amino acids belong to a
second unrelated class of synthetases. This work
is supported by a grant from the National Insti-
tutes of Health.
Regulation of Gene Expression
In Escherichia colt a reduction in glucose con-
centration results in a rise in the levels of a sec-
ond messenger molecule, cAMP, and subse-
quently to an increase in the proteins that
metabolize other sugars. This is achieved because
cAMP binds to the catabolite gene activator pro-
tein (CAP), which in turn binds to specific se-
quences at transcription start sites, activating the
transcription of the catabolite genes. We wish to
know how the binding of cAMP promotes the se-
quence-specific DNA binding of CAP and how
this binding then activates the transcribing en-
zyme RNA polymerase.
We have now determined the structure of CAP
cocrystallized with both a 30-bp DNA fragment
and cAMP. The earlier CAP • cAMP structure had
shown each subunit of this dimer to consist of
two domains, the larger of which binds cAMP.
The two small domains are seen to bind DNA,
with the helix-turn-helix interacting in the major
groove as anticipated. Strikingly, in the first com-
plex the DNA is bent, with an overall bend of
about 90° . In both complexes most of the bend is
achieved by two large kinks of about 43° each.
The relationships between this CAP-induced DNA
bend and transcription activation are presently
being pursued by attempts to crystallize CAP with
polymerase and DNA.
393
Structural Studies of Protein-Nucleic Acid Interactions
Replication of DNA
E. co// DNA polymerase I functions primarily in
the repair of DNA but is homologous to polymer-
ases involved in replication. We have determined
the structure of the Klenow fragment, a portion
of Pol I that retains the polymerase and a 3'- to
5'-editing exonuclease activity. We have shown
that a larger structural domain, which has a cleft
sufficient in size to bind duplex DNA, contains
the active site for the polymerase reaction,
whereas a smaller domain has the active site for
the exonuclease activity. Using site-directed mu-
tagenesis, we have made an enzyme devoid of the
editing exonuclease activity and determined its
structure. We have grown two crystal forms of
this protein complexed with a small DNA sub-
strate. A high-resolution structure of one crystal
form shows a single-stranded tetranucleotide
bound to the exonuclease active site and 1 1 base
pairs of duplex DNA bound to a cleft that runs at
right angles to the major cleft. There are changes
in the structure of the polymerase domain, in-
cluding the movement of a thumb-like structure.
To access the polymerase active site in the cleft, it
appears that the DNA will have to bend by 90°.
These structures begin to address the issues of
how these two active sites work together on the
same DNA substrate and how they both function
to enhance the DNA-copying fidelity of this and
other polymerases. This work is supported by a
grant from the American Cancer Society.
Genetic Recombination
We have recently determined the crystal struc-
tures of two proteins that are involved in genetic
recombination. One protein, called resolvase,
catalyzes a site-specific recombination between
two duplex DNAs of identical sequence. Resolv-
ase is the product of a transposable element (a
jumping gene) that can move throughout the E.
coli population spreading drug resistance genes.
This protein can bind to a specific duplex DNA
sequence, align two DNA segments having the
same sequence, cleave the two DNA duplexes,
rearrange the duplexes, and re-ligate them, re-
sulting in a recombinational event. We have de-
termined the structure of the catalytic domain of
this enzyme at 2. 5 -A resolution. This structure
helps to explain the phenotypes of many resolv-
ase mutant proteins. This structure and that of the
intact protein determined at 3-A resolution pro-
vide ideas for a possible recombination mecha-
nism. We have recently cocrystallized this pro-
tein with a 31 -bp fragment containing the
recombination site, whose structure should pro-
vide further clues to the mechanism of this
reaction.
E. coli recK protein is essential for general re-
combination in E. coli. Using the energy of ATP
hydrolysis, recK protein promotes the pairing of
homologous duplex DNAs in preparation for re-
combination. The structure of rech. protein has
been refined at 2. 3 -A resolution. The subunit
forms a helical filament in the crystal very similar
to that formed on DNA and thus enables us to
understand the many mutant recK proteins made
during the past decade and relate its structure to
its functions in nucleotide binding, DNA binding,
and the SOS response. Our goal is to understand
how ATP hydrolysis and the homologous pairing
of DNA are coupled. We have now produced
good crystals of the next enzyme in the pathway
of recombination, ruvC, which cleaves the DNA
recombination intermediate called the Holliday
structure. The work on resolvase and the recK
protein is supported by a grant from the National
Institutes of Health.
HIV Proteins
During the past year we have determined the
structure of human immunodeficiency virus
(HIV) reverse transcriptase complexed with a
nonnucleotide inhibitor that shows promise as an
anti-AIDS drug. The enzyme is a heterodimer of a
66-kDa polypeptide containing a polymerase and
RNase H domain and a 5 1 -kDa polypeptide con-
taining only the polymerase domain. This di-
meric polymerase shows an asymmetric structure
with one active-site cleft running from the p66
domain to the RNase H domain. The p5 1 subunit
has no cleft. The similarities between the cata-
lytic domains of HIV reverse transcriptase and
Klenow fragment suggest that all polymerases
have the same catalytic site. We have been suc-
cessful in diffusing other nonnucleotide inhibi-
tors into the crystal and find that they bind into a
deep pocket near to, but not overlapping with,
the polymerase catalytic site. We expect that de-
sign of additional inhibitors based on the struc-
ture will be possible.
Recently a peptide fragment of the transactiva-
tor protein Tat has been cocrystallized with a
fragment of the RNA site to which it binds, TAR.
This work is supported in part by a grant from the
National Institutes of Health.
394
A representation of the crystal structure o/ Escherichia coli tecA protein. This
protein catalyzes DMA strand exchange in general genetic recombination.
Shown are 12 subunits of the crystal in a-carbon backbone representation,
with adjacent subunits alternately light and dark blue. A bound ADP molecule
is represented in orange.
Figure by Randall Story and Thomas Steitz from research reported in Story,
R.M., Weber, I.T., and Steitz, T.A. 1992. Nature 555:318-525.
395
Visualization of photoreceptor axons in the developing visual system o/Drosophila. The axon
bundles emerge from each ommatidium of the developing retina and project through an epithelial
tube, the optic stalk, to specific retinotopic positions in the brain. Guidance of these axons relies
largely on positional information.
Research of Hermann S teller.
396
Pattern Formation and Neuronal Cell Recognition
in the Drosophila Visual System
Hermann Steller, Ph.D. — Assistant Investigator
Dr. Steller is also Associate Professor of Neurobiology at the Massachusetts Institute of Technology and
Adjunct Assistant Neurobiologist at Massachusetts General Hospital, Boston. He was born in the Federal
Republic of Germany and received a Diplom in biology from the Johann- Wolf gang- Goethe University,
Frankfurt. His graduate studies were done with Vincenzo Pirrotta at the European Molecular Biology
Laboratory and with Ekkehard Bautz at Heidelberg University. His postdoctoral work was done with
Gerald Rubin in the Department of Biochemistry at the University of California, Berkeley. Dr. Steller is
currently also a Searle Scholar and a Pew Fellow in the Biomedical Sciences.
THE overall objective of our research is to un-
derstand how functional neuronal circuits
are established and maintained during develop-
ment. Our current work is focused on three major
areas.
Axon Guidance and Neuronal
Cell Recognition
We are studying two different stages of visual
system development to investigate the cellular
and molecular mechanisms by which axons find
and recognize their proper synaptic partners. The
optic nerve of the Drosophila larva is a simple,
well-described model system. Connectivity de-
fects of the larval optic nerve can be rapidly and
reliably detected in mutant embryos by staining
with neuron-specific antibodies. In addition, a
simple behavioral test, larval phototaxis, pro-
vides an assay for functional connections of the
larval optic nerve. This permits systematic
screening for mutants with abnormal axonal pro-
jections, which can be subsequently analyzed in
detail with respect to defects in axonal guidance,
target recognition, and synapse formation.
We have previously identified a gene, discon-
nected {disco), which is required for establish-
ing stable connections between the larval optic
nerve and its target cells in the developing brain.
We have cloned the disco gene and determined
its structure, nucleotide sequence, and pattern of
expression. These studies suggest that disco en-
codes a transcription factor with autoregulatory
properties. Consistent with such a function we
have recently found that disco protein has se-
quence-specific DNA-binding activity in vitro
and that two high-affinity binding sites are lo-
cated very close to the disco transcription unit.
Ectopic expression of disco protein under an in-
ducible promoter in transgenic flies results in se-
vere developmental defects and embryonic lethal-
ity. These defects include a drastic reduction of
the axon scaffold and connectivity defects in both
the peripheral and central nervous systems. We
are now testing the idea that disco regulates the
expression of cell adhesion and/or cell recogni-
tion molecules that are required for the establish-
ment of stable connections between the larval
optic nerve and its target cells in the brain.
More recently we have begun to study axon
guidance and neuronal cell recognition in the
adult visual system. The compound eye of Dro-
sophila consists of approximately 800 repeating
units called ommatidia. Each ommatidium con-
tains eight photoreceptor neurons, which repre-
sent three major cell types that project to differ-
ent target cells in the optic ganglia. The major
class of photoreceptors, Rl-6, establishes synap-
tic connections with neurons in the first optic
ganglion, the lamina. Photoreceptor axons from
R7 and R8 project deeper into the brain to differ-
ent regions of the second ganglion, the medulla.
Early during visual system development, all eight
photoreceptor axons from each ommatidium
grow as a bundle to specific retinotopic positions
in the developing brain. The growth cones of
these axons have to navigate over a long distance
and make a number of highly specific choices.
We would like to understand what signals
guide axons to their proper destinations and how
these signals are generated, received, and inter-
preted. To address these questions, we have
screened for mutations that perturb the projec-
tion pattern of photoreceptor cells at very early
developmental stages, when axons enter the
brain. We have found a number of mutants with
severely abnormal patterns of axon ingrowth. The
developmental and genetic characterization of
this material is in progress.
Role of Innervation for Neurogenesis
and Survival of Target Cells
It has been noticed for many years that synaptic
input can have a profound influence on the fate
and differentiation of target cells. Cell death in
the absence of incoming projections is a dramatic
example of how innervation can affect develop-
mental decisions, and many neurological dis-
orders are thought to arise from defective interac-
tions between neurons and their targets. In
397
Pattern Formation and Neuronal Cell Recognition in the Drosophila
Visual System
Drosophila the proper development of the aduh
optic gangha, the central nervous system portion
of the visual system, depends on innervation from
the eye. In the absence of retinal innervation,
adult flies entirely lack the first optic ganglion,
the lamina, which receives direct synaptic input
from the outer photoreceptor cells Rl-6.
^' We have found that the birth of lamina neurons
is controlled by innervation from the developing
eye. The arrival of photoreceptor axons in the
brain induces a wave of mitotic activity that pro-
duces the lamina neurons. These results suggest a
novel mechanism for matching the number of tar-
get neurons in the first optic ganglion to the num-
ber of incoming photoreceptor axons, and they
explain how developmental synchrony between
the Drosophila retina and first optic ganglion is
achieved. We are now using several different ap-
proaches to elucidate the detailed cellular and
molecular mechanisms underlying this process.
While the importance of retinal innervation on
the development of the adult optic ganglia of
Drosophila is well documented, little is known
about retrograde effects of the brain on photore-
ceptor cells in the compound eye. We have re-
cently discovered the first evidence for the exis-
tence of such retrograde effects in the
Drosophila visual system. Although photorecep-
tor cells develop normally in the absence of con-
nections to the optic ganglia, we have found that
their continued survival requires these connec-
tions. This situation is reminiscent of trophic in-
teractions that are commonly found in
vertebrates.
Genetic Control of Cell Death
Apoptosis, the deliberate and orderly removal
of cells by natural death, is a prominent feature of
normal development throughout the animal
kingdom. In many organisms, a large number of
cells die in the absence of obvious external in-
sults. For example, in vertebrates neurogenesis
produces about twice as many neurons as are
needed in the mature nervous system, and approx-
imately half of these neurons are eliminated by
cell death. We are interested in isolating genes
that are required for the initiation or execution of
cell death in Drosophila. We have found that the
ultrastructural characteristics of cell deaths seen
in the Drosophila embryo are strikingly similar
to apoptotic deaths described in other systems.
We have developed techniques utilizing the vital
dyes acridine orange and nile blue that permit the
rapid and reliable visualization of apoptotic cells
in live embryos, and we have used these methods
to screen for cell death-defective mutants. We
have identified a complex genetic locus on the
third chromosome that is required for either the
commitment to or the execution of a cell death
program. We have cloned the DNA encompassing
this locus, and we expect that its molecular char-
acterization will provide insight into the bio-
chemical mechanisms underlying cell death in
Drosophila, and possibly other organisms as
well.
398
Molecular Genetics of Nematode Development
and Behavior
Paul W. Sternberg, Ph.D. — Associate Investigator
Dr. Sternberg is also Associate Professor of Biology at the California Institute of Technology and Adjunct
Assistant Professor of Anatomy and Cell Biology at the University of Southern California School of
Medicine, los Angeles. He received a B.A. degree in biology and mathematics from Hampshire College and
a Ph.D. degree in biology from the Massachusetts Institute of Technology for work with Robert Horvitz.
He did postdoctoral research in yeast molecular genetics with Ira Herskowitz at the University of
California, San Francisco. Dr. Sternberg is also a Presidential Young Investigator.
USING the nematode Caenorhabditis ele-
gans, our laboratory takes a molecular genet-
ics approach to basic questions in developmental
biology and neurogenetics: What are the molecu-
lar mechanisms by which cells interact to estab-
lish a spatial pattern of cell types? What is the
genetic and cellular basis for morphogenesis?
What establishes the asymmetry of individual
cells? How are the instructions for innate behav-
ior encoded in the genome? Our major strategy is
to identify mutations that make cells or animals
misbehave and then to study the functions of the
genes defined by these mutations, using a combi-
nation of molecular cloning and genetic analysis.
A second strategy is to clone nematode homo-
logues of genes identified in mammals and then
to elucidate the functions of those genes in
nematodes.
In this past year we focused on the develop-
ment and function of the C. elegans male spic-
ules— innervated structures crucial to successful
mating. Each of the two spicules comprises nine
cells: two sensory neurons, one motoneuron, and
six supporting cells. By studying spicule develop-
ment, we have identified a new example of in-
duction during nematode development. In the de-
veloping male, either of two cells signals spicule
precursor cells to generate particular sets of spic-
ule cells. This inductive signaling process re-
quires the tin- 3 growth factor, the let- 2 3 tyrosine
kinase, and the let-60 ras genes that we cloned
over the last two years, tin- 3 encodes an induc-
tive signal for the hermaphrodite vulva, and it is
likely that it acts as an inductive signal for proper
spicule development as well. Thus we have
found that a cascade of proto-oncogenes specifies
cell fates in several aspects of nematode develop-
ment. In addition to this inductive signal, at least
three other signals are also necessary for the
correct specification of spicule precursor cells.
Because C. elegans hermaphrodites are inter-
nally self-fertilizing — each animal producing
both sperm and ova — male mating and thus spic-
ule function is dispensable. Thus mutant strains
defective in male mating can be easily propa-
gated and the mating process studied. We have
used a simple behavioral test — the ability of
males to sire progeny — to isolate mutants that are
unable to mate. Some mutant males have obvious
defects in the development of male-specific
structures. Others, called Cod (for copulation
defective) , are anatomically normal yet defective
in mating behavior.
By studying the Cod mutants, we hope to eluci-
date how genes control each step in male mating
behavior. We have isolated a set of mutants, have
characterized the mating defect of each strain,
and have begun placing these mutations on the
genetic map. Most of the mutants analyzed are
defective at only a single step in the mating pro-
cess. These steps include 1) attraction to her-
maphrodites, 2) maintaining contact with her-
maphrodites, 3) location of the vulva, 4)
insertion of spicules, and 5) transfer of sperm.
For example, a mutant male defective in step 4
will locate the hermaphrodite vulva but fail to
insert his spicule. Having mutants blocked at de-
fined steps will allow us to identify genes neces-
sary to specify this innate behavior.
To identify the cells responsible for each step
in mating behavior, we kill individual cells with a
laser microbeam and observe the consequences.
For example, the spicule motoneuron and both
spicule sensory neurons are necessary for spicule
insertion. One of our G protein a-subunit genes is
expressed in one of these neurons, suggesting
that it might serve to regulate spicule insertion.
Another G protein, homologous to human G,,, is
expressed in male diagonal muscles, required for
initial steps in mating. We have begun to test
whether any of the Cod mutants are defective in
these G protein genes.
The establishment of cellular asymmetry is a
fundamental aspect of cell regulation. We have
begun to study this problem in the context of the
2° vulval precursor cell. The lin-18 gene (cell
lineage gene number 18) is necessary for the
asymmetry of the 2° vulval precursor cell. We
have mapped the lin-18 gene to a manageable
region of the X chromosome. We have found that
the 2° cells will orient posteriorly in the animal,
unless they receive a signal from the developing
gonad to orient anteriorly. This signal is distinct
399
Molecular Genetics of Nematode Development and Behavior
from the inductive signal that instructs the 2°
cells to be 2° and not other types of cells. The
lin-18 gene must be involved in the response to
this signal.
In another project, we have been analyzing the
function of the nematode homologues of signal
transduction molecules identified in mammals.
-We have cloned a nematode gene similar to the
raf-1 protein kinase as well as four nematode
genes similar to mammalian G protein a-subunit
genes. We have placed these on the C. elegans
genomic map. To identify the cells expressing
these genes, we have constructed chimeric genes
consisting of their transcriptional control regions
linked to a reporter gene. We then introduced the
engineered genes back into the nematode by mi-
croinjection of DNA to create transgenic nema-
todes. All four G protein genes are expressed pri-
marily in neurons, including neurons in the male
tail. We plan to engineer poisonous forms of
these genes to disrupt their function in intact
animals.
A G protein a-subunit gene expressed in a sensory neuron of a male spicule o/ Caenorhabditis
elegans. The cell body is seen as a large round blue spot, and the dendrite as blue inside the yellow
spicule.
Research of Paul Sternberg; photograph by Jane Mendel.
400
Why Do We Drink Coffee and Tea?
Charles F. Stevens, M.D., Ph.D. — Investigator
Dr. Stevens is also Professor of Molecular Neurobiology at the Salk Institute for Biological Studies and
Adjunct Professor of Pharmacology and of Neuroscience at the University of California School of Medicine,
San Diego. He received his B.A. degree in psychology at Harvard University, his M.D. degree at Yale
University, and his Ph.D. degree in biophysics at the Rockefeller University for studies with Keffer Hartline.
He was a member of the faculties at the University of Washington Medical School and at Yale Medical
School before joining the Salk Institute. Dr. Stevens is a member of the National Academy of Sciences.
COFFEE and tea are widely used in many soci-
eties as mild stimulants. How do they work?
Recent research in our laboratory provides an
answer.
One of the more remarkable features of the
brain is its ability to modify its own neuronal
characteristics, a phenomenon known as neuro-
modulation. Neurons can cause other neurons to
alter their properties, and thus can change the
computations carried out by the modified neuro-
nal circuits.
Neuromodulation occurs in many ways. One of
the most important is the regulation of synaptic
strength — the force and effectiveness with which
one neuron transmits signals to another. Some
changes in synaptic strength are very long lasting
and are thought to underlie the storage of memo-
ries. Other modifications of synaptic strength oc-
cur more rapidly and constitute a moment-to-
moment tuning up of circuit function. A major
goal of our laboratory has been to elucidate the
mechanisms through which synaptic strength is
neuromodulated. In the course of this work, we
have gained insights into the actions of caffeine
and theophylline, the stimulants in coffee and
tea, respectively.
The chemical adenosine is released at synapses
together with neurotransmitters and is present in
the fluids bathing neurons. Neurons display on
their surface several different types of receptors
for adenosine, which couple adenosine binding
to second messenger cascades (such as the one
involving cAMP) . Adenosine is a very potent regu-
lator of synaptic strength: concentrations above
the usual levels result in large decreases in
strength. Here is the interesting part: methylxan-
thines, a family of chemicals that includes caf-
feine and theophylline, bind to adenosine recep-
tors and block their uptake of adenosine.
Some "fake" agonists that bind to a receptor
will activate it as if they were natural agonists, but
the methylxanthines do not do this. They occupy
the receptor's adenosine binding site and thereby
prevent adenosine from doing so. Rather than
mimic the action of adenosine, the methylxan-
thines act as if taking it away. Thus they increase
synaptic strength, indicating that the normal
brain levels of adenosine are sufficient to pro-
duce and maintain a partial decrease in synaptic
strength.
We examined the effect of adenosine itself, and
of adenosine receptor blockers, on synaptic
transmission in brain slices from the hippocam-
pal region (specifically from dentate) of rats. Us-
ing the whole-cell recording method, a tech-
nique that permits high-resolution detection of
currents that flow as a result of synaptic activa-
tion, we were able to determine the specific con-
sequences of adenosine and methylxanthine
action.
To interpret the effects of these drugs, one
needs to know certain features of normal synaptic
transmission. A synapse contains a number of mi-
croscopic membrane-bounded spheres, known as
synaptic vesicles, whose contents are released
from the cell by a membrane fusion event (exo-
cytosis). Each of these vesicles contains a unit
amount of neurotransmitter, such as glutamate,
that binds to special receptors in the target neu-
ron surface membrane and produces a signal. The
strength of a synapse therefore depends jointly on
the number of vesicles whose contents are re-
leased and on the size of the response produced
by one vesicleful of the neurotransmitter. Be-
cause the size of the synaptic response must be
integral — a multiple of that produced by a single
vesicle — this single-vesicle response is called a
quantum or quantal response. The quantal size
depends, in turn, on the number of transmitter
molecules contained in a vesicle and on the num-
ber and responsiveness of the receptors displayed
by the target neuron.
Only a certain number of vesicles are available,
of course, to release their contained neurotrans-
mitter. This number is usually denoted by N.
What is interesting and significant about the neu-
rotransmitter release process is that it is probabi-
listic. That is, only a fraction of the N vesicles
available to release their contents do so, and that
number is determined the same way one would
calculate the number of heads in A'' coin flips,
with a probability p for heads on any flip. Thus
401
Why Do We Drink Coffee and Tea?
the strength of a synapse is determined by TV, p,
and the quantai size. There are techniques for
measuring the quantai size directly, and a kind of
statistical computation, known as quantai analy-
sis, can estimate TV and p.
We have carried out this procedure in the dentate
neurons with either added adenosine or adenosine
receptor antagonists. The quantai responded size
was unaffected by any of the drugs. Quantai analysis,
however, reveals a pronounced effect on the release
probability: adenosine decreases and methylxan-
thines increase the chances that a vesicle will un-
dergo exocytosis and release its neurotransmitter
when a nerve impulse arrives at the synapse.
Thus methylxanthines antagonize baseline
adenosine effects and strengthen synapses. Why
does this act as a stimulant? About two-thirds of
the brain's synapses are excitatory and the other
third are inhibitory. Many drugs shift the balance
between excitation and inhibition and thereby
have a net stimulatory or depressive effect on the
brain's activity. Barbiturates, alcohol, and benzo-
diazipines (e.g., Valium) all selectively increase
the strength of inhibitory synapses and shift the
balance from excitation toward inhibition. We
find that the effects of adenosine and adenosine
antagonists on synaptic transmission are limited
to excitatory synapses. Apparently the inhibitory
synapses lack adenosine receptors or the second
messenger cascades that are necessary for the ex-
pression of their action.
In summary, coffee and tea act (as many have
discovered independently) as antialcohol and an-
tibarbiturate agents, in that they shift the brain's
excitation/inhibition balance in the excitatory di-
rection. They do this by antagonizing the effects
of background levels of adenosine on synaptic
strength. Their precise mechanism of action is to
increase the probability of release of neurotrans-
mitter molecules from synaptic vesicles.
402
Morphogen Gradients and the Control of Body
Pattern in Drosophila
Gary Struhl, Ph.D. — Associate Investigator
Dr. Struhl is also Associate Professor of Genetics and Development at Columbia University College of
Physicians and Surgeons. He received B.S. and M.S. degrees in biology from the Massachusetts Institute of
Technology and his Ph.D. degree in genetics from the University of Cambridge, England. His graduate
studies were carried out with Peter Lawrence at Cambridge. Before joining the faculty at Columbia, he
conducted molecular and genetic studies on Drosophila in the laboratory of Tom Maniatis at Harvard
University. In addition to several fellowships, he received the McKnight Neuroscience Development Award.
SINCE the birth of embryology as an experi-
mental discipline, it has been apparent that
the development of cell and body patterns de-
pends on robust and complex systems of spatial
information. Yet, until recently, we had little
idea of the physical nature of such systems or the
mechanisms by which they are generated or
interpreted. Considerable attention has been
given to the possibility that gradients of "form-
generating" substances, or morphogens, might
provide this information. For example, a local-
ized diffusible morphogen could generate a gra-
dient that would provide a series of concentra-
tion thresholds, each determining distinct
cellular behaviors (e.g., the development of par-
ticular pattern elements) as a function of distance
from the source.
Current research in this laboratory is directed
toward identifying potential morphogen gra-
dients and determining both how they arise and
how they control pattern. Over the past 20-30
years, a variety of embryologic and genetic exper-
iments have suggested that such gradients play a
key role in controlling cell and body patterns in
insects. Indeed, Christiane Niisslein-Volhard and
her colleagues have identified several such mor-
phogens involved in specifying the basic body
plan of the fruit fly Drosophila.
One of these determinants, the protein product
of the gene bicoid (bed) , is clearly expressed as a
gradient that peaks at the anterior pole of the
early embryo and controls anterior body pattern
(head and thorax). Another, nanos (nos), is re-
quired for generating posterior body pattern (ab-
domen), though its properties are less well un-
derstood. Our immediate goal during the past
year or so has been to determine how these two
systems work. As we describe below, the con-
trolled expression of a single regulatory protein,
hunchback (hb), has proved to be the key to un-
derstanding both molecular mechanisms.
rior pole to a boundary halfway down the body.
This event depends critically on the bed morpho-
gen. Moreover, it is essential for the generation of
a large portion of anterior pattern, including all
three thoracic segments.
In an initial series of experiments, we found
that bed behaves as a transcriptional regulator,
capable of directly binding the hb gene and acti-
vating transcription in a remarkably sensitive,
concentration-dependent fashion. Hence, when
the concentration of bed protein exceeds a criti-
cal threshold, it binds a series of adjacent sites in
the DNA immediately upstream of the hb pro-
moter and activates transcription; however, when
the concentration of bed protein falls beneath
this threshold, binding does not occur and the hb
gene remains silent. The concentration gradient
of bed protein therefore determines where the hb
gene is transcribed by positioning this threshold
along the anteroposterior axis.
In the process of examining how the bed gra-
dient controls hb gene expression, we created a
series of hybrid genes that contain only some of
the binding sites for bed protein normally present
upstream of the hb promoter. These hybrid
genes, like hb itself, are activated in anterior por-
tions of the body under the control of the bed
protein gradient. However, their domains of ex-
pression are abbreviated relative to that of hb,
extending only a quarter or a third of the way
down the body. These and other experiments
suggest that the bed gradient can trigger several
spatially distinct responses, the posterior bound-
ary of each depending on the affinity of bed pro-
tein for a given target gene.
Recently several new genes have been identi-
fied that appear to be expressed in a series of
overlapping anterior domains. These genes are
required for generating particular portions of
head and thoracic pattern and may constitute
other direct targets of the bed gradient.
Anterior Pattern: Transcriptional
Activation of hb by bed
Early in embryogenesis, the hb gene is acti-
vated in a broad domain extending from the ante-
Posterior Pattern: Translation Repression
of hb by nos
Transcriptional activation of the hb gene by
bed is only one of two sources of hb mRNA in the
403
Morphogen Gradients and the Control of Body Pattern in Drosophila
early embryo. The second source is maternal hb
mRNA synthesized during oogenesis and depos-
ited uniformly throughout the egg. These tran-
scripts are translated only after fertilization, and
their translation is repressed in more-posterior
portions of the egg, owing to nos activity emanat-
ing from the posterior pole.
We recently found that the ability of nos to
repress hb translation depends on the presence of
two copies of a short cis-acting regulatory se-
quence in the 3' noncoding portion of the hb
mRNA. As in the case of the DNA sites mediating
the binding and regulation of hb by bed protein,
the number and quality of the target sites in hb
mRNA can determine the pattern of expression of
hb protein.
However, in contrast to bed, which has many
other signaling roles, nos appears to have no role
other than to block posterior expression of hb
protein. Indeed, we have been able to show that
nos activity is completely dispensable if the
translation of maternally derived hb mRNAs is
blocked by other means (such as mutation) . Thus
the control of posterior body pattern may depend
solely on the repression of maternal hb tran-
scripts by nos.
Posterior Body Pattern Is Controlled
by hb Protein Acting
as a Classical Gradient Morphogen
The bed and nos determinants have opposite
effects on hb protein expression. The bed gene
triggers the synthesis of hb transcripts, and hence
protein, anteriorly; nos blocks the translation of
both maternal and zygotic hb transcripts posteri-
orly. Together the actions of bed and nos gener-
ate a graded pattern of hb protein in which the
concentration declines from uniformly high lev-
els in the anterior half of the body to undetect-
able levels in the posterior half.
As described above, our analysis of the interac-
tion between nos and hb indicates that nos itself
can be rendered dispensable, provided that ma-
ternal hb transcripts can be inactivated by other
means. In these unusual embryos, hb protein is
still expressed differentially along the anteropos-
terior axis, owing to zygotic activation of the hb
gene by bed. This finding, taken together with the
critical role nos normally plays in repressing the
translation of hb mRNAs, suggests that the distri-
bution of hb protein may be the critical determi-
nant of posterior body pattern.
We tested this possibility by generating em-
bryos in which the differential expression of hb
protein along the anteroposterior axis has been
systematically altered while all other known sig-
naling systems are eliminated or held constant. By
examining the expression of the subordinate reg-
ulatory genes Kriippel, knirps, and giant, each of
which is normally responsible for controlling a
particular subdomain of thoracic and abdominal
pattern, we have been able to show that the gra-
dient of hb protein provides a series of distinct
concentration thresholds that govern where these
genes are expressed.
Thus hb appears to control posterior body pat-
tern by acting as a classical gradient morphogen.
In this regard, it functions as the posterior coun-
terpart to the anterior morphogen bed.
404
Secretory Pathways in Neurons
Thomas C. Siidhof, M.D. — Investigator
Dr. Siidhof is also Professor of Molecular Genetics at the University of Texas Southwestern Medical Center
at Dallas. He received his M.D. degree and his doctorate from the Georgia Augusta University of Gottingen,
FRG. He obtained postdoctoral training first with Victor Whittaker at the Max Planck Institute for
Biophysical Chemistry, Gottingen, and then with Michael Brown and Joseph Goldstein in Dallas.
NEURONS communicate with one another by
means of chemical signals. The known com-
munication pathways between neurons are of sev-
eral kinds:
• Fast, point-to-point transmission of signals
between neurons occurs at the synapse.
• Long-lasting modulatory signals that often
reach many cells are transmitted outside of syn-
apses by neuropeptides and other mediators.
• Short-range diffuse signals are probably
spread by lipophilic messengers, such as nitric
oxide or arachidonic acid.
Of these pathways, signaling between neurons
at the synapse is quantitatively the major form of
cell-to-cell communication in the central ner-
vous system. Synapses are abundant in the ner-
vous system, and their activity provides the basis
of brain function. However, the slow communica-
tion pathways are clearly an essential counterpart
to the fast point-to-point signals. The coexistence
of different signaling pathways in the same neu-
ron increases the complexity of the neuronal
networks. Brain function will clearly not be
understood until we gain insight into the molecu-
lar mechanisms that govern these signaling
pathways.
Work in our laboratory addresses the question
of how nerve cells send out chemical signals. We
are concentrating on the synapse as the most
abundant signaling pathway. Here the chemical
signals, the neurotransmitters, are prepackaged
in unique cellular organelles called synaptic vesi-
cles and released from the presynaptic neuron by
secretion. This secretion is achieved by exocyto-
sis, the fusion of synaptic vesicles with the synap-
tic cell membrane. After fusion the empty synap-
tic vesicles are quickly re-endocytosed and
refilled with neurotransmitter. They become
competent for secretion again in a short time, al-
lowing the neurons to fire rapidly.
We have taken two avenues to the exploration
of the molecular basis of signal transmission at
the synapse. The first approach has been to study
synaptic vesicles and their components as the
central organelle in neurotransmitter release.
The second approach consists of a characteriza-
tion of the presynaptic plasma membrane as the
point of signal release.
Characterizing the molecular components of
synaptic vesicles constitutes a long-term project
that we are largely carrying out in collaboration
with Reinhard Jahn (HHMI, Yale University).
This project has led to the molecular characteriza-
tion of more than 10 synaptic vesicle proteins,
which together account for approximately one-
third to one-half of the total vesicle protein by
mass. The goal of this project is twofold.
First, we would like to achieve a complete de-
scription of the synaptic vesicle in molecular
terms. This is not only a precondition to any even-
tual understanding of how the synaptic vesicle
pathway works, but would also provide the first
molecular anatomy of an organelle.
Second, we would like to explore the functions
of each vesicle protein in neurotransmitter re-
lease, using biochemical and genetic techniques.
This part of the project has progressed to the
point that interesting biochemical properties of
several vesicle proteins have been elucidated.
Moreover, the feasibility of mouse genetics to
probe the functions of these proteins has been
demonstrated. For example, we have recently
shown that a synaptic vesicle protein named syn-
aptotagmin binds Ca^"^ and phospholipids in a
ternary complex at physiologic Ca^"^ concentra-
tions. This result suggests a function for synapto-
tagmin in synaptic vesicle fusion, a possibility
now being explored in transgenic mice.
Another approach we are pursuing to elucidate
synapse function is the study of the presynaptic
plasma membrane. The membrane serves two ba-
sic functions. It binds synaptic vesicles and fuses
with them in a Ca^"^-dependent manner, thereby
releasing neurotransmitter; and it contacts the
postsynaptic site and aligns pre- and postsynaptic
membranes with each other. Both functions
are probably performed by specific protein
components.
To identify a component of the presynaptic
plasma membrane that may be involved in its
functions, we have studied the receptor for a neu-
rotoxin called a-latrotoxin. This protein, derived
from venom of the black widow spider, binds spe-
405
Secretory Pathways in Neurons
cifically to the presynaptic membrane and causes
massive neurotransmitter release. Purification
and amino acid sequencing of the receptor for
a-latrotoxin led to the discovery of a new family
of neuron-specific cell surface proteins that we
named neurexins. Their structure and localiza-
tion suggest that they may be synaptic recogni-
tion molecules.
Surprisingly, we also found that the a-latrotoxin
receptor interacts with a synaptic vesicle protein,
synaptotagmin, whose Ca -binding properties
have already implicated it in exocytosis. These
results raise the possibility that the neurexins
may have a role in aligning the synapse both
extra- and intracellularly. Further studies of the
neurexins should allow us to gain new insights
into how synapses are formed and neurotransmit-
ters released. Results should be relevant for un-
derstanding brain function under normal and
pathologic conditions.
406
Transcription Factors in Cell Growth
and Kidney Differentiation
Vikas P. Sukhatme, M.D., Ph.D. — Assistant Investigator
Dr. Sukhatme is also Associate Professor of Medicine and of Molecular Genetics and Cell Biology at the
University of Chicago. He received his Ph.D. degree in theoretical physics from the Massachusetts Institute
of Technology. He then received his M.D. degree from Harvard Medical School. After residency and clinical
fellowship training at the Massachusetts General Hospital, Boston, he completed his postdoctoral work at
Stanford University.
MY laboratory has been engaged in cloning
and characterizing mammalian genes that
code for transcription factors. One focus is on
signal transduction, specifically on mitogenic sig-
naling, and another aims at identifying transcrip-
tion factors that control kidney development.
These two interests have recently intersected, as
delineated below.
The Egr Family of Immediate-Early
Transcription Factors
Extracellular "signals" in the form of neuro-
transmitters, growth factors, hormones, and ma-
trix are known to control cellular phenotype.
These agents lead to the generation of second
messenger signals in the plasma membrane and
cytosol. In turn, these biochemical events modu-
late the expression of so-called immediate-early
genes (lEGs) , whose induction does not require
de novo protein synthesis. Several years ago, we
and others identified several lEGs in the context
of a mitogenic response, and more specifically in
the transition of a cell out of a quiescent state
(Go) into Gi . Of particular interest to us has been
a subset of lEGs that encode transcription factors
(proteins that bind DNA and regulate gene tran-
scription), since they might couple short-term
responses in the form of second messenger events
to long-term changes in gene expression instru-
mental in altering phenotype.
The best characterized of these immediate-
early transcription factor genes include members
of the Fos family. c-Fos, identified as the cellular
homologue of the \-Fos oncogene present in two
viruses that cause osteosarcomas, was discovered
in 1984 to be an lEG in serum-stimulated fibro-
blasts. However, it was not until four years later
that Fos was shown to be part of the AP- 1 tran-
scription factor complex, composed of Fos-Jun
heterodimers and other less well characterized
proteins. c-Jun was likewise first identified as the
cellular homologue of the \-Jun transforming
gene. It was suspected to be a transcription factor
through homology to part of the yeast GCN4
protein.
In 1 987 we discovered (concurrently with sev-
eral other laboratories) the Egr family of lEGs.
The best-characterized gene in this family is
Egr-1 (early growth response gene-1). Egr-1
(also known as Zif-268, Tis-8, NGFI-A, and
Krox-24) was isolated as a serum-inducible lEG
in quiescent fibroblasts (Gq-Gj transition), utiliz-
ing a differential screening protocol. The gene is
induced by mitogen stimulation in every mam-
malian cell type tested, including B cells; T cells;
kidney mesangial, glomerular, and tubular epi-
thelial cells; hepatocytes; and vascular smooth
muscle and endothelial cells. It is also induced by
nerve growth factor in PCI 2 pheochromocytoma
cells, a physiological context in which mitotic
cells convert to a nonmitotic state.
The cDNA structure predicts a protein whose
carboxyl terminus contains three zinc fingers of
the Cys2-HiS2 type, first identified in the Xenopus
transcription factor TFIIIA. This prediction has
recently been verified by Carl Pabo (HHMI, Mas-
sachusetts Institute of Technology) and his col-
leagues through analysis of the crystal structure
of the zinc finger domain cocrystallized with its
target DNA sequence GCGGGGGCG.
We have been interested in identifying the
events, from cell surface to nucleus, that modu-
late Egr l expression. Although identification of
such "upstream" or "proximal" events can ei-
ther be attempted in the context of mitogenesis
or in other situations in which Egr- 1 mRNA levels
change, most of our work has been restricted to
cell proliferation studies in fibroblasts. It has
been found that multiple kinases regulate Egr- 1
expression. For example, activation of the PDGF
(platelet-derived growth factor) receptor or the
EGF (epidermal growth factor) receptor by their
cognate ligands leads to Egr-1 induction.
v-Src has been shown to regulate Egr-1 inde-
pendent of protein kinase C. More recently, we
have found that v-Raf, a serine-threonine kinase,
whose activation results from the convergence of
diverse cell surface signals, leads to Egr-1 induc-
tion. Furthermore, a dominant negative mutant of
v-Raf will ablate the v-Src induction of Egr-1 , sug-
gesting that v-Src stimulates Egr- 1 via v-Raf. Even
if Egr-1 served as nothing more than a target
407
Transcription Factors in Cell Growth and Kidney Differentiation
marker, studies like these are helping to define
the circuitry among second messenger cascades.
Additional studies currently under way are
aimed at defining structure-function relation-
ships of the Egr-1 protein. The protein contains
several activation domains and a small modular
repressor region whose sequence has been highly
conserved in evolution. Nuclear localization is
dependent upon a bipartite sequence, and DNA
binding is determined by the three zinc fingers.
A major challenge ahead is to define a func-
tional role for the Egr l protein. Attempts at
abrogating Egr- 1 activity may prove useful in de-
fining a phenotype. In this regard, the use of anti-
sense methodology — either as oligonucleotides
or in the form of stable inducible vectors — is
under investigation. The latter technique, when
applied to c-Fos, has resulted in a remarkable and
reversible inhibition of cell growth. Another ap-
proach is also available to inhibit Egr- 1 activity:
nature has designed its own inhibitor of Egr-1
function in the form of a tumor-suppressor gene,
one whose absence is likely to lead to Wilms' tu-
mor (see below).
Transcription Factors in Kidney
Development
A second recent focus of our laboratory is on
trying to define molecular events that character-
ize the development of the kidney. It is well
known that metanephric mesenchymal (blaste-
mal) cells convert into epithelial cells over a 4- to
5 -day period (in the rat or mouse) in response to
invasion of the ureteric bud. Little is known at the
molecular and cellular level of the events that
transpire during this conversion process. Our aim
is to define a hierarchy of transcriptional regula-
tors whose expression is modulated during
nephrogenesis. Toward this end, several studies
are under way.
One strategy is to identify zinc finger transcrip-
tional regulators expressed in the kidney during
differentiation. Using a so-called H/C-link probe
directed against a region often conserved in zinc
finger proteins, we have isolated from a kidney
library a dozen cDNA clones encoding zinc finger
proteins. These are now being characterized by
their developmental profile (Northern blots and
in situ hybridization) and by limited sequence
analysis. One of these clones hybridizes at high
stringency to the candidate Wilms' tumor anti-
oncogene (WTl) identified recently by reverse
genetic means in the laboratories of David Hous-
man (Massachusetts Institute of Technology)
and Gail Bruns (Harvard University). Another
clone has identified a zinc finger protein that
is expressed very early in human kidney
development.
In a second approach, work is in progress to set
up stem cell cultures from embryonic kidney for
the development of cell lines arrested at different
stages of differentiation. If these cells can be
maintained and then made to differentiate in cul-
ture under appropriate conditions, they will be
invaluable for studying the early molecular and
cellular events in nephrogenesis.
Third, we are pursuing an interesting connec-
tion between Egr-1 and WTl . WTl is a zinc finger
protein; three of its four zinc fingers show a 60-
70 percent similarity to the three zinc fingers of
Egr l. Indeed, WTl and Egr-l bind to a common
sequence, as shown by the work of Frank
Rauscher (Wistar Institute). We have recently
shown (in collaboration with Dr. Rauscher) that
WTl acts as a transcriptional repressor, whereas
Egr- 1 is a transcriptional activator. These findings
may have exciting biological implications, be-
cause several binding sites exist for the Egr-1/
Wilms' tumor proteins in the promoter sequence
of the insulin-like growth factor II (IGF-II).
IGF-II levels are known to be high in Wilms'
tumor, and these levels fall during development.
Thus the possibility exists that IGF-II is a target
for the repressive action of the Wilms' tumor pro-
tein and for positive regulation by Egr- 1 . If this
model holds, it will be a way to explain the find-
ings of aberrant regulation of IGF-II in Wilms'
tumor and will also provide major insight into the
mesenchymal cell-to-epithelial cell conversion
in the kidney.
408
Structure and Function of Voltage-Dependent
Calcium Channels
Tsutomu Tanabe, Ph.D. — Assistant Investigator
Dr. Tanabe is also Assistant Professor of Cellular and Molecular Physiology at Yale University School of
Medicine. He received his B.A., M.A., and Ph.D. degrees from Kyoto University in Kyoto, Japan. He began
studies on the structure and function of the receptors and ion channels in excitable membrane when he
was a graduate student in Shosaku Numa's laboratory. After he received his doctorate, he continued to
conduct research in Dr. Numa's laboratory as a faculty member before coming to Yale.
VOLTAGE-dependent calcium channels play
important roles in the regulation of a variety
of cellular functions, including membrane excit-
ability, muscle contraction, synaptic transmis-
sion, and secretion. At least four types of calcium
channels have been distinguished by their elec-
trophysiological and pharmacological properties.
Recently, molecular biological studies, com-
bined with electrophysiology, have provided evi-
dence that this diversity of calcium channels
derives largely from differences in their pore-
forming ttj-subunit and that the other subunits
associated with can modify channel function.
Furthermore, the diversity of the several subunits
was shown to originate not only from differences
in the genes encoding them but also from alterna-
tive splicing of their RNAs.
Several types of calcium channels are known to
be coexpressed in single cells, and the cells ap-
parently use them for different purposes. We are
interested in the structure-function relationships
of calcium channels and the molecular basis of
one type's specialization.
Muscular dysgenesis (mdg) is a fatal autosomal
recessive mutation of mice. It eliminates excita-
tion-contraction (E-C) coupling and dihydropyri-
dine (DHP) -sensitive calcium current of the slow
L-type from skeletal muscle. Analysis of genomic
DNA and skeletal muscle RNA indicates that the
mdg mutation is associated with alterations of the
structural gene for the skeletal muscle DHP re-
ceptor. Injection of an expression plasmid carry-
ing the cDNA of the receptor restores both E-C
coupling and skeletal L-type calcium current,
suggesting a dual role of this molecule.
The restored coupling resembles that of nor-
mal skeletal muscle, which does not require
entry of extracellular calcium ions. By contrast,
injection into dysgenic myotubes of an expres-
sion plasmid carrying cDNA of the cardiac DHP
receptor produces rapid, cardiac-like L-type
current and cardiac-type E-C coupling, which
does require calcium ion entry.
To investigate the molecular basis for these dif-
ferences in calcium currents and E-C coupling,
we expressed various cDNAs of the chimeric DHP
receptor in dysgenic myotubes, with the follow-
ing results. Expression of cDNAs encoding chi-
meras with regions of the skeletal muscle DHP
receptor replacing one or more of the corre-
sponding large, putative cytoplasmic regions of
the cardiac DHP receptor showed that the region
linking repeats II and III is a major determinant of
skeletal muscle-type E-C coupling.
Expression of cDNAs encoding chimeras in
which repeats of the skeletal muscle DHP recep-
tor are replaced by corresponding repeats from
the cardiac receptor showed that repeat I deter-
mines whether the chimeric calcium channel ac-
tivation will be slow (skeletal muscle-like) or
rapid (cardiac-like).
We are also focusing on the drug-binding sites of
calcium channel molecules, including those of
channel antagonist (L-type channel blocker), and
the mechanism of calcium-dependent inactivation.
409
The Molecular Biology of Liver Regeneration
Rebecca A. Taub, M.D. — Associate Investigator
Dr. Taub is also Assistant Professor of Human Genetics at the University of Pennsylvania School of
Medicine. She received her B.A. degree in biochemistry from Yale University, attended Yale University
School of Medicine, and completed residency training in internal medicine at Yale. She was a postdoctoral
fellow with Philip Leder at Harvard Medical School, where she later joined the Department of Internal
Medicine as Assistant Professor before moving to Philadelphia.
THE liver has unusual properties of regenera-
tion. When the two largest lobes of the liver
are removed, leaving the smaller lobes com-
pletely intact, the remaining cells can grow and
divide until the liver regains its former size,
whereupon growth ceases. Many growth factors
have been implicated in regulating this process,
but the mechanisms remain poorly understood.
After many years of study the same interesting
questions about liver regeneration remain: What
makes the liver start regenerating? What regulates
liver growth during regeneration, allowing the
liver to maintain its normal architecture? What
makes the liver stop regenerating when it has re-
gained its initial size?
Understanding liver regeneration will help ex-
plain how the liver responds to toxic damage or
infections like hepatitis. Additionally, because
increasing numbers of liver transplants are being
performed and successful transplants require
liver regeneration, it is essential to understand
the biological and molecular bases for liver cell
growth.
My colleagues and I are interested in determin-
ing what genes are expressed during liver regen-
eration and how their protein products are in-
volved in regulating the process. It is important
to determine if these genes are identical to those
that regulate the growth of all cells. We are study-
ing liver cell growth in two systems: a continu-
ously growing liver cell line that responds to in-
sulin as a growth factor and regenerating liver
tissue from rats.
In our early studies, we identified more than
40 novel genes that are rapidly expressed in re-
sponse to growth factors in liver cells. Although
many of the genes are expressed in other growing
cells and seem to be part of the general growth
response, some of the genes are specific to grow-
ing liver. Many of these genes encode proteins
that function in the cell nucleus, possibly regu-
lating the cascade of gene expression occurring
when cells grow. Some of these genes have al-
tered expression in cancer cells, contributing to
the aberrant proliferation.
Our studies have focused on understanding the
actions of the proteins encoded by some of these
genes and their potential roles in regulation of
liver regeneration. Because we isolated so many
novel genes, it was important to establish criteria
for determining which of the encoded proteins
are likely to have important regulatory roles in
liver regeneration. We decided to explore further
the exact roles of novel proteins falling into four
functional categories.
The first category includes proteins that regu-
late the expression of genes. Because so many
genes need to be turned on for liver regeneration
to proceed, proteins that regulate gene expres-
sion are likely to be important.
The second category includes proteins that are
secreted from cells and may function to regulate
the growth of surrounding cells. The liver must
maintain its cellular architecture during regener-
ation, and because it is made up of many different
cell types, intercellular communications must
exist during regeneration. Secreted proteins
could be involved in mediating such intercellular
signals.
The third category includes proteins whose ex-
pression is specific to regenerating liver and not
other growing cells. We found several in this cate-
gory, a few of which are highly expressed in re-
generating liver. These proteins could have im-
portant functions in liver-specific growth.
The fourth category includes several genes
whose expression is nicely induced in regenerat-
ing liver but abnormally high in the liver tumor
cell line that grows in response to insulin. These
genes could be functioning as oncogenes in the
liver tumor cell line.
One of the novel genes that we isolated, RL/
IF-1 (regenerating liver inhibitory factor), en-
codes a protein that inhibits the activity of certain
gene-transactivating proteins in the NF-kB/RcI
family. We examined the relative activity of
NF-/cB/Rel proteins during liver regeneration and
found to our surprise that although most of the
proteins in this family remain in an inhibited or
inactive form, one protein, PHF-1, becomes dra-
matically active in its ability to interact with gene
sequences within minutes after hepatectomy.
The activation of PHF-1 is the earliest change we
411
The Molecular Biology of Liver Regeneration
have observed in the liver posthepatectomy, and
our findings suggest that PHF- 1 could have a role
as an initiating signal in liver regeneration.
LRF- 1 (liver regeneration factor) is one of the
proteins that we have studied that is highly ex-
pressed in regenerating liver and functions as a
DNA-binding protein, controlling the expression
of target genes. It falls into the category of so-
called leucine zipper proteins, in which a stretch
of amino acids containing evenly spaced leucine
residues allows one molecule of LRF- 1 to interact
with another molecule of LRF-1 or of a related
protein. LRF-1 activates or inhibits the expression
of target genes in a singular manner, implying
that LRP-1 has a unique role in regulating events
in the regenerative process. As the relative level
of LRF- 1 complexes increases posthepatectomy,
activation of the many liver-specific genes oc-
curs, resulting in maintenance of normal liver
function and metabolic homeostasis during
regeneration.
Another goal of our research (supported in part
by research funds from other agencies) is to un-
derstand the involvement of insulin in the regula-
tion of hepatic growth. It is well known that dia-
betics have poor healing capacity, and early
studies have shown that the livers of diabetic ani-
mals show poor regenerative capacity. Because
we are studying a liver cell line that is growth-
regulated by insulin, we have been able to com-
pare the insulin-regulated growth response in
these cells with the response during liver regen-
eration. Differences in expression of more than
1 0 of the novel genes we have identified suggest
that insulin, if it is an important growth factor
during liver regeneration, must act several hours
after the hepatectomy and is not an initiating fac-
tor. Additionally, in examining the encoded func-
tion of several genes that are aberrantly regulated
in the hepatic cell line, we are learning more
about the specific signaling pathways involved in
hepatic growth.
412
Protein-Tyrosine Phosphatases and the Control
of Lymphocyte Activation
Matthew L. Thomas, Ph.D. — Assistant Investigator
Dr. Thomas is also Associate Professor of Pathology and Assistant Professor of Molecular Microbiology at
Washington University School of Medicine, St. Louis. He received his Ph.D. degree from the University of
Utah after completing his thesis research at Harvard Medical School. His postdoctoral training was done
with Alan Williams at Oxford University and with Ian Trowbridge at the Salk Institute. Prior to his present
appointment. Dr. Thomas was an Established Investigator of the American Heart Association.
THE molecular mechanisms by which cells re-
spond to their environment are a central
theme in many areas of medical research. Our
interests are in understanding how signals re-
ceived by proteins of leukocyte surface mem-
branes result in changes in a wide variety of cel-
lular processes. In particular, we are interested in
lymphocyte activation.
Lymphocytes, by virtue of their ability to recog-
nize an infinite variety of foreign antigens, play a
key role in effecting and regulating an immune
response. Upon binding antigen, lymphocytes
undergo profound biochemical changes to en-
gage the cellular machinery required for clonal
expansion and to produce molecules needed to
fight infection. A key regulatory mechanism by
which many different cell types control the signal
transduction process is phosphorylation and/or
dephosphorylation of distinct protein-tyrosine
residues of specific substrates. This process is
controlled by families of enzymes that either add
phosphate (protein-tyrosine kinases) or remove
phosphate (protein-tyrosine phosphatases). The
research interests of my laboratory are centered
on how protein-tyrosine phosphatases control
lymphocyte activation.
The protein-tyrosine phosphatase family can
be divided into two main branches: transmem-
brane and intracellular. The leukocyte-common
antigen CD45 is a major transmembrane protein-
tyrosine phosphatase of lymphocytes and is ex-
pressed by all nucleated cells of hematopoietic
origin. To study the function of this molecule, we
generated nontransformed T cell clones deficient
in the expression of CD45. The deficiency re-
sulted in the cells' inability to respond to antigen.
CD45-deficient cells still respond to other prolif-
erative signals, such as stimulation by growth fac-
tors or plant lectin mitogens or direct activation
of protein kinase C. However, these cells cannot
proliferate or produce cytokines in response to T
cell antigen receptor stimulus, and their ability to
cytolyze target cells is also impaired. Therefore
the protein-tyrosine phosphatase CD45 is re-
quired for antigen-induced T cell activation.
We are investigating the molecular mechanism
of antigen-induced CD45-controlled activation
by examining the differences in tyrosine phos-
phorylation among the CD4 5 -expressing and
-nonexpressing cell lines. We have observed in-
creased tyrosine phosphorylation of members of
the Src-tyrosine kinase family of proteins in the
CD45-deficient cells, indicating that they are po-
tential substrates for CD45. Three members of
the Src family, p56'^\ p59'^, and ip62^^\ are ex-
pressed by T cells, and recent studies have indi-
cated that kinase activity for all three is decreased
in the CD45-deficient T cells. Src family
members are negatively regulated by phosphory-
lation of a carboxyl-terminal tyrosine residue.
CD45 may serve to dephosphorylate the car-
boxyl-terminal tyrosine site and thus function to
activate members of the Src family. The inability
of antigen to activate CD45-deficient T cells im-
plies that activation of Src family members is criti-
cal to antigen-induced T cell activation and that
CD45 is the phosphatase important in initiating
this process. This work is supported in part by a
grant from the National Institutes of Health.
Immortalization of CD45-deficient T cells by
fusion with a transformed thymoma cell line has
permitted the development of a model system in
which function can be reconstituted by cDNA
transfection. We are currently analyzing regions
of functional importance in the CD45 molecule
by reconstituting the cells with altered CD45
cDNAs. These experiments should allow precise
definition of the regions of the molecule impor-
tant in regulating T cell activation.
To analyze further how protein-tyrosine phos-
phatases control lymphocyte activation, we have
isolated multiple cDNAs that encode protein-
tyrosine phosphatases expressed by leukocytes.
LRP is a transmembrane phosphatase whose exte-
rior domain is predicted to be a highly glycosy-
lated, elongated rod. This type of structure is also
found at the amino terminus of CD45's exterior
domain and may provide a novel means of regu-
lating the phosphatase activity. To understand its
function and regulation, we have developed a
monoclonal antibody that recognizes the native
protein. Use of this antibody has allowed us to
413
Protein-Tyrosine Phosphatases and the Control of Lymphocyte Activation
examine the expression of the protein by var-
ious leukocyte populations, the extent of post-
translational modification, and the ability of the
antibody to induce changes in lymphocyte activa-
tion and proliferation.
PEP and SHP are both intracellular protein-
tyrosine phosphatases expressed primarily by
leukocytes. PEP contains a large carboxyl-termi-
nal extension following the phosphatase domain.
This region is unusual when compared with other
intracellular phosphatases, in that it contains se-
quences indicative of nuclear localization and
rapid protein turnover. By tagging the protein
with a sequence that can be traced, we have ob-
tained preliminary results indicating that PEP
does indeed localize to the nucleus. It is possible,
therefore, that PEP is involved in regulating gene
transcription or other nuclear functions.
SHP contains two src-homology region-2
(SH2) domains linked in tandem immediately
amino terminal to the PTPase catalytic domain.
SH2 domains are found in proteins involved in
transducing mitogenic signals, such as all the
nonreceptor protein-tyrosine kinases, the p21"*
GTPase-activating protein, and the 7-isoform of
the phosphatidylinositol-specific phospholipase
C. Functionally, SH2 domains bind phosphory-
lated tyrosine residues. Thus a protein that con-
tains SH2 domains serves to amplify and direct
the biochemical pathway of a response that is in-
duced by an initial increase in tyrosine phosphor-
ylation. Recent results indicate that the SHP SH2
domains are capable of binding a low-molecular-
weight phosphotyrosine protein. The existence
of a phosphatase that contains SH2 domains sug-
gests that SHP may serve to modulate kinase-in-
duced signals. We are examining the plausibility
of this model by identifying the specific proteins
with which SHP interacts.
Through an understanding of how protein-
tyrosine phosphatases effect lymphocyte activa-
tion, we hope to gain a more thorough knowledge
of the biochemical steps involved in control-
ling and regulating an immune response. The
functional characterization of multiple protein-
tyrosine phosphatases expressed by lymphocytes
has allowed us to define steps in which members
of this family function in lymphocyte activation.
414
Molecular Regulation of Lymphoid Cell Growth
and Development
Craig B. Thompson, M.D. — Associate Investigator
Dr. Thompson is also Associate Professor of Internal Medicine and of Microbiology and Immunology at the
University of Michigan Medical School. He received his undergraduate degree from Dartmouth College
and his medical degree from the University of Pennsylvania. Following an internship and residency at the
Peter Bent Brigham Hospital, Boston, he spent eight years as a research medical officer in the United States
Navy. During this time, he conducted his research at Boston University, the Uniformed Services University
of the Health Sciences, the Fred Hutchinson Cancer Research Center, and the Naval Medical Research
Institute.
THE lymphoid immune system comprises two
major cell types: the T cell, which identifies
and destroys cells expressing foreign proteins;
and the B cell, which secretes antibodies that
bind to foreign substances, targeting them for
elimination. The central role of the system in the
natural resistance to infectious diseases is demon-
strated by the infections encountered by patients
with immunologic deficiencies. Many of the seri-
ous infectious and neoplastic complications asso-
ciated with the acquired immune deficiency syn-
drome (AIDS) are the result of depletion of the
helper T cells.
A better understanding of the molecular
mechanisms associated with generation of both B
and T cells during development would aid in our
ability to understand and treat various immunode-
ficiencies. The goals of our laboratory are to un-
derstand the molecular events associated with
the development of the lymphoid immune sys-
tem and to define the mechanisms by which the
functions of these cells are controlled.
T cells are divided into two major subsets:
helper T cells, which produce the lymphokines
that regulate immune responses, and cytotoxic T
cells, which can kill cells expressing foreign pro-
teins. The cooperation of these two cell types is
needed for the immune system to reject a foreign
cell or a cell bearing foreign proteins. Initiation
of a helper T cell response requires activation of
the T cell receptor by foreign histocompatibility
genes. Recent evidence suggests, however, that T
cell receptor activation alone does not lead the
helper cell to produce sufficient amounts of lym-
phokines to initiate an immune response. Addi-
tional co-stimulatory signals are required.
Previously we showed that the CD 2 8 receptor
expressed on helper T cells serves as a surface
component of a signal transduction pathway that
can enhance T cell lymphokine production. In
vitro, interaction of CD28 with its natural ligand
B7, expressed on activated B cells or macro-
phages, can act as a co-stimulus to such enhance-
ment. Now we have evidence that CD28 activa-
tion of primary T cells is also a required
co-stimulatory event in the initiation of a cell-
mediated immune response.
We have investigated the role of the CD 2 8
pathway in allogenic responses by using a soluble
CD28 receptor homologue termed CTLA-4-Ig,
produced by our collaborator Peter Linsley. This
protein is a recombinant that displays an affinity
for the CD28 ligand, B7, approximately 20-fold
higher than does cell-bound CD28. Therefore it
acts as a competitive inhibitor of CD28 engage-
ment. In vitro, CTLA-4-Ig was found to be able to
inhibit completely the ability of cells from one
rat strain to respond against cells from another.
Based on these data, the ability of CTLA-4-Ig to
prevent the induction of organ graft rejection was
tested in a cardiac transplant model. Trans-
planted hearts in control animals receiving no im-
munotherapy are rejected within one week of
transplantation. In contrast, when activation of
the CD28 pathway is blocked by daily administra-
tion of CTLA-4-Ig, rejection of the cardiac trans-
plant is prevented.
These data support the hypothesis that helper T
cell co-stimulation by the CD28-ligand, B7, is a
required event in the initiation of a T cell-
mediated immune response. Furthermore, the
work suggests that preventing B7 activation of a T
cell may not only help to prevent transplant re-
jection but may also help to decrease the severity
of autoimmune diseases such as rheumatoid ar-
thritis. Work to address this issue is being
planned.
The major role of the B cell immune system is
to generate the approximately 10 million diff'er-
ent antibody molecules needed to protect the
body from foreign substances. B cells derive their
name from the bursa of Fabricius, a developmen-
tal organ in birds that is required for B cell matura-
tion. Mammals lack this organ but are still able to
generate a B cell immune system. Over the past
several years, our laboratory has investigated the
role of the bursa in B cell development in the
chicken. We are attempting to characterize dif-
415
Molecular Regulation of Lymphoid Cell Growth and Development
ferences in the generation of B cells in mammals
and birds to shed light on mammalian B cell
production.
Our studies have helped to demonstrate that
the primary molecular mechanism by which
chickens generate a wide variety of antibody mol-
ecules is different from that used by mammals.
Antibody diversity in the chicken is achieved by a
process known as gene conversion, which occurs
during B cell development in the bursa of Fabri-
cius. Gene conversion requires that the immature
B cell migrate to the bursa, and that the mature B
cell, having undergone conversion, emigrate to
the peripheral lymphoid organs. Understanding
how this migration is controlled in a regulated
fashion should lead to insights into how cells
in a multicellular organism migrate during
development.
416
The Molecular Basis of Metamorphosis
Carls. Thummel, Ph.D. — Assistant Investigator
Dr. Thummel is also Assistant Professor of Human Genetics at the University of Utah School of Medicine.
He obtained his undergraduate degree in biology from Colgate University and his Ph.D. degree in
biochemistry, working with Robert Tjian, at the University of California, Berkeley. He received
postdoctoral training in the laboratory of David Hogness at Stanford University.
THE fruit fly Drosophila melanogaster pro-
vides an ideal model system for studying the
development of eukaryotes. Three-quarters of a
century of biological, physiological, and genetic
experiments, combined v^^ith recent intensive mo-
lecular studies, has led to a greater understanding
of its development than that of any other higher
organism.
Halfway through the fly's life cycle, a pulse of
the steroid hormone ecdysone triggers a dramatic
morphological transformation, from the rela-
tively immobile feeding larva to a highly motile,
reproductively active adult fly. We are studying
the molecular basis of the ecdysone-induced reg-
ulatory mechanisms that allow metamorphosis to
proceed.
When the larva begins to undergo metamorpho-
sis, its salivary glands contain giant polytene
chromosomes that can be visualized by light mi-
croscopy. These 500-fold overreplicated, inter-
phase chromosomes lie in register beside one an-
other. A characteristic banding pattern along the
length of the polytene chromosomes allows any
gene of interest to be located precisely. Regions
of the genome that are undergoing transcription
are often represented by large areas of decon-
densed chromatin that can be seen as puffs. Thus
the transcriptional activity of specific genes at
specific times can be followed in these chromo-
somes by observing the appearance and disap-
pearance of puff^s during development.
Approximately 10 puffs can be distinguished
when the salivary gland chromosomes first be-
come large enough to see. These puff's remain
until the end of the larval phase, when the burst
of ecdysone triggers a dramatic change in the
puffing pattern. Approximately six puffs are in-
duced directly by the steroid hormone. These
early ecdysone-inducible puffs appear to encode
regulatory proteins that repress their own expres-
sion and induce the formation of over 100 late
puffs. This second wave of puffs is believed to
encode the proteins responsible for initiating
metamorphosis.
By isolating and characterizing the ecdysone-
inducible genes that lie within the early puffs, we
hope to learn how these genes are induced by the
hormone and how their encoded proteins might
function in a regulatory capacity. In a broader
sense, this project provides a model system for
characterizing the role of steroid hormones in reg-
ulating gene expression and for addressing the
question of how gene hierarchies are controlled
during development.
Our studies have focused on E74, an ecdysone-
inducible gene that is located within the large
early puff at position 74EF in the polytene chro-
mosomes. This unusually complex gene encodes
three nested mRNAs that derive from unique start
sites but share a common 3' end. The distal pro-
moter directs the synthesis of a 60-kb primary
transcript that is spliced to form the 6-kb E74A
mRNA. Two other promoters, located 40 kb
downstream from the E74A promoter, direct the
synthesis of 4.8- and 5.1-kb E74B mRNAs. Al-
though the E74A and E74B mRNAs are distinct
from one another by virtue of their unique 5'
exons, the majority of these mRNAs are identical,
derived from a common set of three 3' exons. This
nested arrangement of the E74 transcripts leads
to the synthesis of two related E74 proteins that
have unique amino-terminal domains joined to a
common carboxyl-terminal domain.
The sequence of the carboxyl terminus of the E74
proteins is very similar to a portion of the protein
encoded by the ets oncogene. This 85-amino acid
ETS domain defines a family of proteins and has been
shown to function as a site-specific DNA-binding do-
main that recognizes a purine-rich DNA sequence.
Studies of proteins related to oncogenes, such as
E74, may help us learn more about how the normal
counterparts of these disease genes function during
development.
By using antibody detection techniques to lo-
calize the E74A protein bound to the giant poly-
tene chromosomes, we have identified approxi-
mately 70 binding sites, most of which
correspond to late ecdysone-inducible puffs.
Based on this observation, we predict that at least
one function for the E74A protein is to activate
late gene expression. In support of this predic-
tion, many late puffs are either reduced or absent
417
The Molecular Basis of Metamorphosis
in the polytene chromosomes of mutant animals
that do not express E74A protein. Our long-term
goal is to identify some of these late genes in
order to determine what role they might play
during metamorphosis and whether they are di-
rectly regulated by E74A.
The precise timing of the genetic response to
efcdysone can be clearly seen in the pattern of
puffs that arise at the onset of metamorphosis.
Our characterization of E74 transcription has
provided insights into how timing can be built
into a genetic regulatory hierarchy. Ecdysone di-
rectly activates the E74A promoter, resulting in a
dramatic induction of the 6-kb mRNA. This tran-
script, however, does not appear in the cyto-
plasm until one hour after promoter activation.
The delay corresponds quite closely to the time it
takes for RNA polymerase to traverse the 60-kb
transcription unit, indicating that the length of
the .£"74^1 unit functions as a timer to delay signifi-
cantly the appearance of the encoded mRNA. The
unusual length of the E74A primary transcript
sets it apart from most transcription units in Dro-
sopfoila, which are only slightly longer than the
final processed mRNA.
Ecdysone also directly activates the 20-kb
E74B transcription unit. In agreement with its
primary transcript length, mature E74B mRNA
appears between 1 5 and 30 minutes after ecdy-
sone addition. Thus the structure of the E74 gene
dictates an invariant order of appearance of its
transcripts in response to ecdysone.
The earlier appearance of E74B mRNA is en-
hanced by its activation at an approximately 25-
fold lower ecdysone concentration than E74A.
E74B is further distinguished from E74A by its
repression at a significantly higher ecdysone con-
centration than that required for its induction,
close to the concentration required for E74A ac-
tivation. These regulatory properties lead to an
ecdysone-induced switch in E74 expression,
with an initial burst of E74B transcription fol-
lowed by a burst of E74A transcription. These
studies provide a means to translate the profile of
a steroid hormone pulse into different amounts
and times of regulatory gene expression that, in
turn, could direct different developmental re-
sponses in a temporally and spatially regulated
manner.
To date, E74 has provided a valuable paradigm
for our molecular characterization of the ecdy-
sone regulatory hierarchy. We have recently be-
gun to extend our studies to include three other
early genes, all of which, like E74, are unusually
long and encode multiple DNA-binding proteins.
Characterization of the temporal regulation of
these early genes, like that described above for
E74, has confirmed that both promoter sensitiv-
ity to ecdysone and primary transcript length con-
tribute to the timing of early gene activation.
In addition, these studies have allowed us to
divide the early transcription units into two
classes. One class of transcripts is activated by a
low ecdysone concentration, like E74B, and ap-
pears to play earlier roles in the regulatory hierar-
chy. This class includes transcripts encoding the
ecdysone receptor, needed for initiating the ge-
netic response to the hormone. The second class,
typified by E74A, is activated by a higher hor-
mone concentration and appears to play a later
role in the hierarchy.
In addition, analysis of the effects of early gene
mutations on both early and late gene expression
should provide valuable clues regarding func-
tional regulatory interactions within the hierar-
chy. The predictions that arise from these genetic
studies can then be tested at the molecular level.
An additional long-term goal v/ill be to isolate
more early and late ecdysone-inducible genes to
extend our understanding of this complex devel-
opmental process.
418
The Regulation of Mammalian Development
Shirley M. Tilghman, Ph.D. — Investigator
Dr. Tilghman is also Howard A. Prior Professor of the Life Sciences in the Molecular Biology Department at
Princeton University and Adjunct Professor of Biochemistry at the University of Medicine and Dentistry
of New Jersey, Robert Wood Johnson Medical School. She obtained a B.Sc. degree at Queen's University in
Kingston, Ontario, Canada. Following two years in Sierra Leone, West Africa, where she taught secondary
school, she attended graduate school at Temple University in Philadelphia, where she received her Ph.D.
degree in biochemistry. Her postdoctoral work was done with Philip Leder at NLH Before joining the
faculty at Princeton, Dr. Tilghman held positions at Temple University and the Institute for Cancer
Research, Philadelphia.
ORDERLY development of the mammalian
embryo requires the appropriate activation
and subsequent modulation of genes in a spatial
and temporal manner. For the vast majority of
genes, both the mother's and father's copies are
activated and modulated identically, but for a
small class of genes, only the mother's or father's
copy is expressed. Such genes are parentally im-
printed. That is, during the process that generates
eggs or sperm, these genes are marked in such a
way that the resulting embryo can distinguish the
parental origin and express it accordingly. Our
laboratory is studying a locus that encodes at least
two imprinted genes on the distal end of mouse
chromosome 7.
One, the insulin-like growth factor II gene
(Igf2) encodes a fetal-specific growth factor that
is exclusively expressed from the paternal chro-
mosome. The other gene is HI 9, which encodes
an RNA, evolutionarily conserved, that is only
found in high abundance during fetal develop-
ment in tissues originating in endoderm and me-
soderm. Unlike Igf2, HI 9 is exclusively ex-
pressed from the maternal chromosome. These
genes lie in tandem about 75 kilobases (kb) of
DNA apart and are expressed in a very similar
manner during mouse embryogenesis.
We are investigating the activation and role of
these two differentially imprinted genes. The
function of the H19 RNA is unknown. Its pattern
of sequence conservation in mammals is reminis-
cent of other functional RNAs, such as those asso-
ciated with telomerases and RNase P. We are us-
ing both genetic and biochemical approaches to
understand its role during development.
Over the past 50 years a large number of muta-
tions have been described that affect all aspects of
mouse physiology. However, molecular access to
the genes, which would allow us to identify those
of developmental importance, has been difficult
because the mouse genome is so large. Advances
in DNA analysis and cloning methods have effec-
tively reduced the barriers to studying these
genes at the molecular level. We have generated a
yeast artificial chromosome library that now con-
tains 2.5 copies of the mouse genome in over
28,000 yeast strains, each of which harbors a seg-
ment of mouse genomic DNA averaging 275 kb.
The library has been constructed to serve the
mouse genome community, and over 60 labo-
ratories worldwide have so far screened it suc-
cessfully. This work is supported by a grant from
the National Institutes of Health.
We have used the mouse Fused (Fu) locus to
test the utility of the library for isolating large
chromosomal DNA segments. Fu, a dominant mu-
tation on mouse chromosome 17, generates
kinky tails in heterozygotes and early embryonic
lethality in homozygotes. The lethality is asso-
ciated with overgrowth of neuroectoderm and
duplications of the body axis. To localize Fu pre-
cisely, 1,000 progeny were generated by back-
crossing mice carrying the Kinky allele of Fu
with a distantly related wild mouse, Mus spretus.
Because these mice are genetically very distinct,
it is easy to follow the segregation of genes in
their progeny.
The offspring were scored for seven molecular
markers that map in a small interval around Fu,
and a high-density genetic map gives us molecu-
lar landmarks every 100-200 kb of DNA. One
marker, a pseudogene of the a-globin gene fam-
ily, cannot be separated from Fu in this cross,
suggesting that it is very close to the gene —
within 100-200 kb. We have used this close
proximity to isolate approximately 650 kb of
DNA around the marker. By comparing the segre-
gation of this DNA with that of Fu, we should be
able to pinpoint Fu's location.
A similar approach has been adopted for the
piebald (5) locus on mouse chromosome 14.
Mice carrying the original 5 mutation have spot-
ted coats, a result of the absence of melanocytes
in genetically specified regions of the midsec-
tion. Mice carrying more-severe s mutations are
almost entirely white. In addition, they develop
419
The Regulation of Mammalian Development
megacolon, as a result of the absence of enteric
ganglia in the gut. Both melanocytes and enteric
ganglia develop from the neural crest, a migrat-
ing cell population that contributes to many cell
types. By studying the s gene, we hope to gain
insight into the genes that control the birth, mi-
gration, and maturation of these cells.
A high-density map around s is under way, us-
ing mice carrying the original 5 mutation and an-
other wild strain, Mus castaneus. In addition,
the molecular studies are greatly aided by the
many different mutant alleles of 5 created over
the past 40 years at the Oak Ridge National Labo-
ratory by William and Leane Russell. These al-
leles, many of which are deletions, will be invalu-
able in mapping the region on chromosome 14
that contains the 5 gene.
To gain insight into the nature of the 5 mutation,
we are also comparing the behavior of neural crest
cells in s-bearing and normal mice. For this pur-
pose we have exploited a transgenic mouse strain
carrying a (S-galactosidase gene that is expressed in
early neural crest cells and their derivatives. To
study later stages in melanocyte development, we
are utilizing an antibody to the cell surface protein
c-kit, which is expressed in these cells. We will
thus identify the stage in neural crest development
that the s mutation affects.
420
Mechanisms of Gene Regulation in Animal Cells
Robert Tjian, Ph.D. — Investigator
Dr. Tjian is also Professor of Molecular and Cell Biology at the University of California, Berkeley, and
Adjunct Professor of Biochemistry and Biophysics at the University of California, San Francisco. He
received the Ph.D. degree in biochemistry and molecular biology from Harvard University. Following this
he was a Junior Fellow of the Harvard Society of Fellows while a resident at Cold Spring Harbor Laboratory
and later a staff investigator there before moving to Berkeley. His honors include the Monsanto Molecular
Biology Award of the National Academy of Sciences and the Pfizer Award for Enzymology. Dr. Tjian was
recently elected to the National Academy of Sciences.
THE main research interest of our laboratory is
the mechanism by which the genetic infor-
mation stored in DNA molecules is retrieved in a
controlled and orderly fashion during the process
called transcription, which leads to the produc-
tion of specific proteins in animal cells. We have
taken a biochemical approach to the problem of
gene control and have devised various means of
isolating the cellular components responsible for
transcription and of reconstructing this complex
reaction in the test tube. In this way we are study-
ing how specific genes are turned on and off dur-
ing cell growth and development of eukaryotic
organisms. The mechanisms that govern the acti-
vation of genes are of fundamental importance in
understanding the normal metabolic processes
that maintain and perpetuate living cells, as well
as in deciphering cellular and genetic disorders.
Biochemical Analysis of Cancer Genes
A living cell contains hundreds of thousands of
protein molecules, each carrying out its allotted
function. However, if either the production or
action of these molecules is altered, severe mal-
function can result, such as uncontrolled growth
leading to cancer. Thus certain key molecules in
the milieu of a normal cell's constituents have
the potential to cause tumors when their func-
tion is disrupted. Such molecules are called
oncoproteins.
Our group previously isolated from human
cells a family of rare proteins that have subse-
quently been shown to be oncogenic, encoded by
thejun and fos genes. These regulatory proteins
are normally involved in controlling the action of
many other genes, but when their activities are
perverted — for example, by viruses — they can
lead to the production of cancer-causing cells.
Recent advances in the study of the nuclear on-
cogene ^wn reveal that the ability of its protein to
activate transcription is regulated by a cell-rype-
specific inhibitor that interacts with a unique
portion of the molecule, rendering it less potent.
It is anticipated that the isolation and character-
ization of this specific negative regulator of jun
will help unravel the molecular signaling path-
ways responsible for transducing information
from the outside of the cell into the nucleus,
where gene expression is controlled. Moreover,
such inhibitors of jun activity may also be found
to represent new members of the anti-oncogenic
or tumor-suppressor family of biological regula-
tory molecules.
Studies of Trans-activating Proteins
That Regulate Gene Expression
A major hurdle has been the development of
biochemical techniques that allow the purifica-
tion of certain rare and fragile regulatory pro-
teins. Through use of DNA-affiniry chromatogra-
phy procedures pioneered in this laboratory, it is
now possible to isolate such transcription pro-
teins and, in turn, to clone molecularly the genes
that encode them. The ability to proliferate this
biologically important class of genes provides a
powerful approach toward understanding their
structure and function. In the past two years the
laboratory has isolated and characterized some
10 different genes that are directly responsible
for the tissue-selective, temporally programmed,
and basal-level control of gene expression in ani-
mal cells.
In addition, these recent studies are beginning
to reveal new concepts regarding the surprisingly
modular construction of the derivative proteins
as well as their unusual plasticity and functional
flexibility. Most importantly, specific structural
motifs that lie within these proteins have been
recognized as carrying out distinct functions.
These findings provide the theoretical basis for
analysis of other as yet undiscovered regulatory
factors and will greatly aid our ability to decipher
their mechanisms of action.
How Promoter-Specific Regulators
Trigger Transcription
One of the remaining fundamental mysteries is
the mode of action by which sequence-specific
421
Mechanisms of Gene Regulation in Animal Cells
DNA-binding proteins, such as the prototype hu-
man factor Spl, direct transcriptional interac-
tions. To address this critical issue, our group re-
cently fractionated and isolated the multiple
components necessary to reconstitute transcrip-
tion. In the process of dissecting the general tran-
scriptional apparatus, we discovered two previ-
ously undetected components that serve as the
functional bridge between upstream trans-activa-
tors and the initiation complex. These novel fac-
tors appear to be part of the missing link that di-
rects promoter-selective transcription in animal
cells, and it is likely that they will be members of
a diverse and essential class of regulatory
proteins.
Indeed, this past year has seen significant pro-
gress in the purification and characterization of
transcriptional "coactivators" and TATA-binding
protein (TBP) -associated factors (TAFs). In partic-
ular, several TAFs and coactivators have been pu-
rified from both Drosophila and human cells. In
addition, the genes encoding this novel class of
transcription factors have recently been cloned.
The structure and function of these multisubunit
complexes should prove to be very revealing.
One of the most exciting and unexpected find-
ings this year was the discovery that the TBP, a
general transcription factor thought to be only
responsible for RNA polymerase II transcription
of mRNA, is also an integral subunit of a TBP-TAF
complex responsible for recognition of the RNA
polymerase I promoter. Most interestingly, the
RNA polymerase I complex carries out all the
functions ascribed to the species-specific tran-
scription factor SLl , and the subunit composition
reveals the presence of TBP and three novel TAFs,
apparently adapted uniquely to direct RNA poly-
merase I transcription. This surprising finding
provides a unifying mechanism of transcription
initiation.
Transcription of Developmentally
Regulated Genes
One of our long-term interests is the mecha-
nisms underlying regulation and expression in
the development of higher organisms. We have
begun to address this issue in two ways: first, by
initiating a series of in vitro experiments aimed
at dissecting the transcriptional regulation of
Drosophila genes, including the alcohol dehy-
drogenase, Ultrabithorax, Antennapedia, dopa
decarboxylase, and hunchback genes. A major
advance was the development of in vitro tran-
scription reactions from staged Drosophila em-
bryos that accurately initiate RNA synthesis and
recapitulate the temporal program of transcrip-
tion displayed by these tissue-specific and devel-
opmentally regulated genes.
A second approach has been to investigate the
regulatory mechanism of RNA polymerase initia-
tion factors in vivo. Various systems have been
adapted to introduce altered genes back into cells
or whole organisms to study their patterns of
expression.
These in vitro and in vivo studies have re-
cently led to two exciting results. First, a negative
regulator of a developmentally important gene
has been identified by direct biochemical means,
and its mode of operation can now be dissected.
The finding of specific transcriptional repressors
is of particular importance because it is thought
that an interplay of positive activators and nega-
tive regulators is seminal to the spatially re-
stricted patterns of expression observed during
embryogenesis.
Also arising from Drosophila studies are the
identification and subsequent biochemical char-
acterization of a transcription factor that appears
to govern the expression of genes in cells of the
central nervous system. The gene encoding this
neurogenic-specific activator has recently been
isolated, and its structure is expected to reveal
interesting information. Advantages in the use of
fruit flies include the ability to probe the develop-
mental and tissue-specific function of this neuro-
genic regulator in a rapid and highly informative
manner not readily applicable to mammalian
cells. These studies are expected to yield new
insights concerning the tissue-specific distribu-
tion and temporal timing of expression during
development.
422
Studies on T Lymphocytes and Mammalian Memory
Susumu Tonegawa, Ph.D. — Investigator
Dr. Tonegawa is also Professor of Biology at the Massachusetts Institute of Technology. He received a B.S.
degree in chemistry from Kyoto University in Kyoto, Japan, and a Ph.D. degree in biology from the
University of California, San Diego. His postgraduate training and research were at UCSD in the
laboratory of Masaki Hayashi and at the Salk Institute with Renato Dulbecco. Dr. Tonegawa was a
member of the Basel Institute for Immunology in Basel, Switzerland, before joining the Department of
Biology and Center for Cancer Research at MIT. He was awarded the Nobel Prize for physiology or
medicine in 1987.
T lymphocytes play a pivotal role in the body's
defense against a variety of infectious agents
and malignant tumors as well as in the rejection
of grafted foreign tissues. Our laboratory contin-
ues to study the development and functions of T
lymphocytes, with a particular focus on more re-
cently discovered and less-characterized yb T
cells. In addition, outwork on mammalian mem-
ory has progressed significantly during the past
year.
Our study on yb T cells could be divided into
three categories: development and selection,
specificities, and function. As for the early devel-
opment of these cells, one major issue that we
continued to pursue is the mechanism by which
the yb lineage segregates from the a.0 lineage
from common progenitor cells. Following our
earlier observations that differential activation of
the transcriptional silencer associated with the y
genes of the T cell receptor (TCR) plays a pivotal
role in the cell lineage segregation, we character-
ized the silencer DNA elements in detail and
identified proteins that bind to these elements.
We intend to clone the genes encoding the
silencer-binding proteins and produce mice with
mutations in these genes. Analysis of the mice
should be highly informative in the dissection of
the lineage segregation mechanism.
a/? T cells are known to undergo in the thymus
a critical maturation step called positive selec-
tion. This depends on an appropriate interaction
between the TCRs and a self determinant en-
coded in part by the genes of the major histocom-
patibility complex (MHC). Following our earlier
study indicating that epithelium-associated yb T
cells (called s-IEL and vut-IEL) undergo a similar
positive selection, we have now obtained evi-
dence of positive selection for circulating yb T
cells. Thus the development of a transgenic yb T
cell clone (KN6) specific for an MHC class I prod-
uct (encoded by the T2^ gene) was shown to be
blocked at an immature state in a mouse geneti-
cally deficient in the expression of MHC class I
gene products.
We also studied T cell development by exploit-
ing the embryonic stem (ES) cell gene-targeting
method. We produced four types of mutant mice
relevant for the analysis of T cell development
and functions. The first and second types of mice
have mutations in the TCRa and TCRjS genes, re-
spectively, and are deficient in mature T cells
but not in mature yb T cells. The third type of
mouse has a TCR6 gene mutation and is deficient
in yb T cells but not in afi T cells. The fourth type
was produced by mutating the RAG-1 gene,
whose product is required for somatic rearrange-
ment of immunoglobulin and TCR genes. Neither
mature T cells nor B cells are present in the
RAG-1 mutant mice.
The initial analysis of these mice indicated that
there is no major interaction between the devel-
opment of and yb T cells and that the muta-
tions in the various genes block lymphocyte de-
velopment at distinct stages. Detailed analysis of
the immature lymphocytes accumulating in these
mutant mice is expected to augment our under-
standing of lymphocyte development.
Specificities of yb T cells were studied by char-
acterizing the self-reactive yb T cell subset
(V4C1) on the one hand, and by testing our hy-
pothesis that TL class I molecules have evolved to
present peptides to yb TCRs. In the V4C1 subset
study, we produced three types of monoclonal
antibody (mAb), each of which blocks self-
reactivity of these TCRs in vitro. These mAbs rec-
ognize 1) a determinant specifically present on
the 75 TCR utilized as the immunogen (called
clonotypic), 2) determinants present on all
V74C71 -expressing TCR, or 3) determinants pres-
ent on the receptor for vitronectin, a family of
cell adhesion molecules previously implicated as
coreceptors for this T cell subset. These mAbs
will be extremely useful in the elucidation of the
ligand for the self-reactive yb T cell subset, as
well as for the analysis of its development and
tissue localization.
We are studying the general peptide-present-
ing role of TL class I molecules by producing
mouse mutants lacking a cluster of these mole-
cules. We have thus far identified ES cell clones
423
Studies on T Lymphocytes and Mammalian Memory
devoid of a large number of TL class I genes. Also
being pursued is biochemical characterization of
a TL class I molecule encoded by the gene 73'',
whose product has been shown to be expressed
specifically on the surface of gut epithelial cells
with which a 76 T cell subset, i-IEL, is associated.
We intend to purify the putative peptides bound
to the T3^ molecule and subject them to microse-
quencing. The sequences may lead to the identi-
fication of the antigenic proteins.
We have pursued functions of 76 T cells by im-
munizing KN6 transgenic mice with the ligand-
bearing C57BL/6J spleen cells and following the
proliferation of the transgenic cells and the disap-
pearance of the injected spleen cells from the
host spleen. These studies indicate that KN6 76 T
cells are capable of responding to the allogenic
cells by cytotoxicity.
Functions of 76 T cells have also been studied
by following the antibody response of afi T-defi-
cient mutant mice to thymus-dependent and
-independent antigens. So far these mice respond
only to the thymus-independent antigen. A possi-
ble role of 76 T cells in this type of antibody re-
sponse is being studied using 76 T-deficient mu-
tant mice. The T- or 76 T-deficient mice have
also been infected with listeria, mycobacteria, or
malaria. Resistance and/or immunity against
these infectious agents will be determined. We
are also studying the response of the mutant mice
to skin grafts and injected tumor cells.
In addition to the studies on 76 T cells, we
found, using the multigene transfection tech-
nique, that the afi TCR-CD3 complex can be ex-
pressed on the surface of nonlymphoid cells
without either the CD37 or CD35 subunit. This
suggests the intriguing possibility that these mul-
tiple forms of TCR-CD3 complex are utilized by
normal T cells for difi'erential purposes. The hy-
pothesis is being tested by producing mouse mu-
tants that are defective either in the CD37 or
CD36 gene.
We are also interested in studying how infor-
mation is stored and retrieved in the brain. The
approach that we have taken is to investigate the
biochemistry, physiology, and behavior of mice
mutant for genes thought to be involved in synap-
tic plasticity. Using homologous recombination,
we have disrupted the a-subunit of the calcium
calmodulin kinase II (CaMKII) gene in embry-
onic stem cells and have used these cells to gener-
ate a mouse strain lacking the gene. The a-sub-
unit of CaMKII is neural specific and comprises
most of the CaMKII holoenzyme in postnatal hip-
pocampus and forebrain. Peptides that emulate
either the calmodulin-binding domain or the in-
hibitory domain of this kinase seem to block the
induction of long-term potentiation (LTP) . Mice
lacking the a-subunit develop normally and are
viable.
Gross neuroanatomical studies did not detect
any abnormalities. We studied the electrophysiol-
ogy of the CAl fields of the mutant hippocampus
in collaboration with the laboratory of Charles
Stevens (HHMI, Salk Institute). In normal animals
we were able to induce LTP in more than 90 per-
cent (n = 12; 3 animals) of all slices. However,
we could stably potentiate less than 1 0 percent of
slices from mutant animals (« = 17; 5 animals).
Furthermore, the unpotentiated excitatory post-
synaptic potentials of normal and mutant animals
were identical, suggesting that synaptic transmis-
sion was not impaired in the mutant animals.
We also studied the mutant mice in difl'erent
versions of the Morris water maze, in collabora-
tion with Jeanne Wehner's laboratory. The results
indicate that the mutant mice, but not their nor-
mal litter mates, are specifically impaired in tasks
that demand the use of configured spatial infor-
mation. Our analysis of the mice mutant for the a
CaMKII demonstrates the involvement of this ki-
nase with LTP and with configural learning.
Past evaluations of the involvement of LTP on
learning and memory have mainly depended on
the potential role that the NMDA (7V-methyl-D-
aspartate) receptor plays on both phenomena. Un-
fortunately, NMDA blockers seem to impair both
learning and performance, confounding the in-
terpretation of those experiments. However, our
results suggest that performance measures unre-
lated to learning were not responsible for the
learning impairment observed in our mutant
animals.
A key aspect of our studies has been the appar-
ent specificity of the phenotype: the develop-
ment and neuroanatomy of the mutant mice are
apparently normal, as well as nonpotentiated syn-
aptic transmission. Furthermore, behavioral anal-
ysis of the mutant animals argues for a selective
impairment on hippocampal-dependent learn-
ing. We are continuing to test the Hebbian hy-
pothesis by making mutants for other compo-
nents of LTP and for genes that might affect
the physiology of specific regions involved in
learning.
424
dIV in ftplysia sensory neuron: processes vs« cell bodii 29.2
Before 5~HT 19» after 50 mM 5-HT 49s after B-HT
Concentrations of the intracellular messenger cAMP (cyclic adenosine 3' ,5' monophosphate)
within a single neuron from the sea snail Aplysia californica. 77?^ cAMP is detected by its ability to
affect an enzyme, cAMP- dependent protein kinase. This enzyme was produced by recombinant
DNA technology, labeled with fluorescent dyes, and injected into the neuron, which was then
grown in tissue culture. The cAMP-sensitive signal from the enzyme was imaged by confocal
fluorescence microscopy, a technique that can isolate one plane of focus within a live specimen.
The concentrations of cAMP are denoted by a rainbow of colors, with blues and reds representing
the lowest and highest levels, respectively.
The upper left panel shows the neuron before stimulation. The upper middle panel shows the
same cell shortly after addition of 5 hydroxytryptamine ( 5-HT, also known as serotonin ) to the
medium. Although applied to the entire cell, 5 HT, an important neurotransmitter, elevates cAMP
to a much greater extent in the fine dendrites. Over the next 100 seconds (upper right, lower left
panels ), the cAMP shows some spread to the main cell body, but the dendrites continue to have the
highest cAMP levels. The cAMP response begins to decay even while the 5-HT is still present (lower
middle panel ) and returns to baseline when it is removed (lower right). These images are the first
direct visualization of local generation of c AMP in the fine outgrowths and its diffusion toward
the cell body and nucleus. It is known to be a key controller of both short- and long-term plasticity
in many neurons.
Research of Brian Bacskai, Benny Hochner, Martyn Mahaut-Smith, Stephen Adams, Bong-kiun
Kaang, Eric Kandel, and Roger Tsien.
426
Molecular Engineering Applied to Cell Biology
and Neurobiology
Roger Y. Tsien, Ph.D. — Investigator
Dr. Tsien is also Professor of Pharmacology and of Chemistry at the University of California School of
Medicine, San Diego. His undergraduate degree was from Harvard College, in chemistry and physics, but
it was at the University of Cambridge, England, while obtaining a Ph.D. degree in physiology, that he
was "introduced to the potential synergism between organic chemistry and cell biology. " After a
postdoctoral fellowship at Gonville and Caius College, Cambridge, Dr. Tsien became a faculty member at
the University of California, Berkeley. Seven years later his laboratory moved to the University of
California, San Diego. His recent honors include the Passano Foundation Young Scientist Award, the
Spencer Award in Neurobiology from Columbia University, and the Bowditch Lectureship of the American
Physiological Society.
THE overall goal of my laboratory is to gain a
better understanding of information process-
ing both inside individual living cells and in net-
works of neurons. Our preferred approach is
through the rational design, synthesis, and use of
new molecules to detect and manipulate intra-
cellular biochemical signals, usually by optical
means such as fluorescence readout or photo-
chemical release of messenger substances. For ex-
ample, we have created fluorescent dye mole-
cules that detect calcium ions (Ca^"^) with great
specificity and sensitivity, so that while the cells
are living and performing their normal functions,
we can image Ca^"*^ levels inside cells with a spa-
tial resolution of a micron or so and a temporal
resolution of a fraction of a second. These dyes
have found wide application in cell biology,
since a rise in intracellular Ca^"*^ levels is one of
the commoner mechanisms by which cell mem-
branes control biochemical events inside the
ceil, such as muscle contraction, synaptic trans-
mission, glandular secretion, enzyme activation,
embryonic fertilization, and growth stimulation.
The detection of intracellular signals such as
Ca^"^ is doubly important. It should help in trac-
ing the complex biochemistries involved in such
signaling, and it affords a nondestructive way to
watch the activity of many individual cells simul-
taneously. The latter ability is particularly rele-
vant to understanding how neural networks pro-
cess information by harnessing many individual
but interconnected neurons in parallel. The domi-
nant established techniques for monitoring
neural activity either listen intensively to a single
neuron at a time or record some smeared-out
average of what thousands, millions, or billions
of cells are doing. If we can continue to improve
the spatial and temporal resolution of present
Ca^"^ imaging, we may succeed in eavesdropping
on conversations within small groups of individu-
ally identified neurons or in taking snapshots of
the instantaneous state of activity of yet larger en-
sembles. Because imaging is inherently good for
following multiple events in parallel, it would be
a major help in analyzing the workings of the
brain, which is still the most awesome and com-
plex molecular assembly known. We recognize
that optical methods, although unsurpassed in
their combination of spatial and temporal resolu-
tion, are best applied to small regions of thin
transparent tissues. For larger volumes of opaque
organs, especially in intact organisms, other
forms of visualization, such as magnetic reso-
nance imaging, are more appropriate, so we are
also seeking to extend our molecular designs to
create suitable non-optical indicators.
A recent example of molecular engineering is
our development of a fluorescent sensor for
cAMP. This important intracellular messenger
plays a crucial role in the actions of a great many
hormones, in the detection of odors and tastes,
and in the mechanisms of learning and memory.
In this case we did not design the sensing mole-
cules from scratch but rather modified the natu-
ral protein that cells normally use to respond to
cAMP. In collaboration with Susan Taylor and her
laboratory, Stephen Adams attached fluorescent
labels on cAMP-dependent protein kinase in such
a way that cAMP not only activates the normal
activity of this enzyme but produces an immedi-
ate optical signal that we can image microscopi-
cally. This labeled protein enables us to visualize
cAMP levels, to show that diff'erent regions of a
single cell can have differing responses to neuro-
transmitter and drug stimulation, and to see that a
subunit of the enzyme can move in and out of the
nucleus as the cAMP level rises and falls. While it
is in the nucleus, it is ideally placed to modify
gene expression.
A particularly dramatic example of the dy-
namics of cAMP signaling comes from sensory
neurons of the marine mollusk Aplysia califor-
nica. These neurons have been extensively stud-
ied by Eric Kandel (HHMI, Columbia University) ,
James Schwartz, and their collaborators as models
for both short- and long-term neuronal plasticity.
427
Molecular Engineering Applied to Cell Biology and Neurobiology
In collaboration with Dr. Kandel's group, Brian
Bacskai and Martyn Mahaut-Smith have injected
the labeled protein kinase into the neurons, ei-
ther in culture or in intact ganglia, and imaged
the nucleus, the surrounding cytoplasm of the
cell body, and the peripheral outgrowths of the
cell. Bath application of the relevant neurotrans-
mitter, 5-hydroxytryptamine, produces rapid in-
creases in cAMP with remarkable spatial gra-
dients— high in the peripheral outgrovvths yet
only slightly elevated in the central cell body.
Optical sections through the nucleus show that
it tends to exclude the holoenzyme (injected into
the cytoplasm) as long as cAMP concentrations
remain at basal levels. Prolonged elevation of
cAMP and dissociation of the holoenzyme causes
gradual translocation of the catalytic subunit into
the nucleus over tens of minutes. The observed
gradient puts high cAMP where it is most needed
for short-term plasticit}', at the distal processes
where the presynaptic terminals would be in
vivo. Only strong or repeated stimulations would
be able to raise the cAMP concentration in the
cell body sufficiently to release the catalytic sub-
unit to diffuse into the nucleus, phosphorylate
transcription factors, and cause longer-term
changes in gene expression. There is some evi-
dence that cAMP changes in the mammalian brain
may also be important in comparable forms of
plasticity, so we are trying to extend our studies
to the appropriate mammalian neurons.
Eventually we hope to extend optical methods
to detect macromolecular biochemical signals
such as protein phosphorylation or gene tran-
scription. These events currently are assayed by
grinding up millions of cells, so that time resolu-
tion is limited and differences between individ-
ual cells or subregions are impossible to discern.
Our experience with imaging ionic messengers
and cAMP suggests that cells have considerable
individuality and complex behavior patterns.
These were missed with destructive population
assays, which might be somewhat analogous to
studying human psychology on the basis only of
anonymous nationwide averages in which the
respondents are executed after each poll. We
therefore seek continuous, nondestructive read-
out from single cells. Approaches currently
under development (by Julie Matheson and Gre-
gor Zlokamik, respectively) include microinjec-
tion of peptides whose fluorescence is altered by
phosphorylation and development of membrane-
permeant fluorogenic substrates for reporter en-
zymes whose nucleotide sequences can be fused
to genes or promoter sequences of interest.
A complementary approach is to perturb intra-
cellular signals in a controlled manner to see how
the cell or tissue responds. Many important intra-
cellular messengers such as cAMP or inositol
phosphates contain one or more phosphate
groups that prevent permeability through mem-
branes. A general method for making membrane-
permeant analogues has not been available but
would be highly useful for artificial stimulation
of the putative transduction pathways to see what
physiological functions result, especially re-
sponses that cannot be assayed in microinjected
or permeabilized cells. Recently Carsten Schultz
has discovered such a method, esterification of
the organophosphate anions with acetoxymethyl
groups, which increases the potency of extracel-
lularly applied cAMP analogues by about 100-
fold and yields the first membrane-permeant de-
rivatives of inositol phosphates yet reported. This
methodology should help us to find new func-
tions for these ubiquitous messengers.
Our projects encompass a wide range of disci-
plines, including organic synthesis, theoretical
and experimental optical spectroscopy and pho-
tochemistry, protein chemistry, computerized
microscopy and image processing, cell biology,
and neurobiology.
The work of Stephen Adams, Martyn Mahaut-
Smith, Julie Matheson, Carsten Schultz, and Gre-
gor Zlokarnik in my laboratory was supported by
grants from the National Institutes of Health.
428
Genetic Defects in the Metabolic Pathways
Interconnecting the Urea
and Tricarboxylic Acid Cycles
David L. Valle, M.D. — Investigator
Dr. Valle is also Professor of Pediatrics, Medicine, Molecular Biology and Genetics, and Biology at the
Johns Hopkins University School of Medicine. He received both his undergraduate degree in zoology and
his medical degree from Duke University. His internship and residency in pediatrics were completed at
the Johns Hopkins Hospital. His postdoctoral research in metabolism was done at NIH.
KJMAN biochemical genetics has been a
ruitful area of study since its beginning
with the work of Sir Archibald Garrod early in this
century. Inherited defects in our body's chemis-
try or, as Garrod called them, inborn errors of
metabolism, are intrinsically interesting and
serve as important models for all genetic diseases.
My colleagues and I have been involved in the
study of several aspects of these disorders,
including clinical diagnosis, biochemical charac-
terization, delineation of pathophysiologic mech-
anisms, development of new therapeutic ap-
proaches, and molecular studies of the involved
genes.
We have focused on disorders of amino acid
metabolism, particularly those involving two
fundamentally important areas of metabolism:
the urea cycle, which is involved in the conver-
sion of excess nitrogen from a toxic to a nontoxic,
readily excreted form; and the tricarboxylic acid
cycle, an essential component of energy metabo-
lism. Recently we have extended these interests
to include inborn errors in the biogenesis of the
peroxisome, a ubiquitous, subcellular organelle
that contains about 40 enzymes important in a
variety of anabolic and catabolic processes.
One of the amino acid disorders that we are
studying intensively is an inborn error of orni-
thine metabolism known as gyrate atrophy of the
choroid and retina (GA). This progressive, blind-
ing chorioretinal degeneration with associated
cataract formation is inherited as an autosomal
recessive trait. The primary biochemical defect is
deficiency of the enzyme ornithine-6-aminotrans-
ferase (OAT), which results in an approximate
10-fold accumulation of ornithine in all bodily
fluids.
Despite the systemic nature of the metabolic
abnormality in GA, the clinical phenotype is lim-
ited to the eye. Thus GA is one of a very few iso-
lated, inherited retinal degenerations for which a
primary biochemical defect is known. In an ex-
tensive molecular analysis of the OAT genes of 85
probands from GA families around the world, we
have detected 34 OAT mutations. Other investi-
gators have added another 20, and together these
54 OAT mutations account for 128 (75 percent)
of the possible 170 mutant alleles in our patient
population.
This compilation of OAT mutations allows one
to determine their consequences on the steady-
state levels of OAT mRNA and on the structure
and function of OAT protein in the patients' cul-
tured skin fibroblasts or when expressed in a het-
erologous system, Chinese hamster ovary cells,
which lack endogenous OAT mRNA and protein.
We find that more than 80 percent of the mutant
alleles produce normal amounts of normally
sized OAT mRNA. A small fraction (approxi-
mately 10 percent) of mutant alleles, all with
point mutations that truncate the open reading
frame in the penultimate exon or earlier, have
markedly reduced levels of OAT mRNA. In con-
trast to their mRNA phenotype, approximately 80
percent of the OAT mutant alleles, including at
least 13 missense mutations, yield little or no de-
tectable OAT antigen. Thus destabilization of the
protein is the most common consequence of
these mutations. However, two missense alleles,
Rl 80T and Rl 54L, inactivate OAT function with-
out reducing OAT antigen. We speculate that the
involved residues may play a role in the active
site of OAT. Studies are now in progress to pro-
duce the large quantities of OAT necessary for
x-ray crystallography to determine directly the
consequences of these mutations in OAT struc-
ture and function. A portion of this work on gy-
rate atrophy is supported by a grant from the Na-
tional Institutes of Health.
The OAT-catalyzed reaction is an essential step
in the metabolic pathway that interconnects the
urea and tricarboxylic acid cycles and, as might
be predicted, is subject to complex regulation. In
liver, the regulation of OAT expression is coordi-
nated with other urea cycle-related enzymes. We
have identified a sequence motif in the 5'-flank-
ing region of OAT that is also present in the pro-
moters of several other urea cycle enzymes and
have obtained evidence that this motif is a cis-
acting element involved in the regulation of these
genes. Surprisingly, localization of OAT expres-
sion by in situ hybridization and immunohisto-
429
Genetic Defects in the Metabolic Pathways Interconnecting the Urea
and Tricarboxylic Acid Cycles
chemistry reveals an additional complication:
OAT expression is limited to one zone of the he-
patic lobule, namely a small population of hepa-
tocytes surrounding the central vein, whereas
most other urea cycle-related enzymes are in the
periportal region. This zonal expression persists
even when OAT activity is induced 40 -fold by
alterations in dietary protein. We hope to identify
cis- and trans-acting elements that mediate this
aspect of OAT expression. Furthermore, to un-
derstand coordinated aspects of OAT regulation
better, we are cloning the genes for other en-
zymes that are metabolically related to OAT. We
used complementation in Saccharomyces cere-
visiae mutants to clone the human cDNA for
pyrroline-5-carboxylate reductase, the enzyme
that catalyzes the conversion of the product of
the OAT reaction to proline. We have now cloned
and mapped the structural gene for this enzyme
and are beginning a comparison of the promoter
regions of the reductase to that of OAT.
Our interest in GA has stimulated us to identify
other genes that may be involved in inherited reti-
nal degenerations. We have proceeded along two
lines of investigation: 1) cloning genes important
for photoreceptor function and 2) using posi-
tional cloning to identify candidate genes from a
region of the genome known to harbor genes for
several retinal degenerations (Xpl 1 .2-Xpll.3).
As part of the former strategy, we have cloned the
cDNA and cloned and mapped the structural gene
for recoverin. Recoverin is a calcium-binding
protein whose expression is limited to the pho-
toreceptor. When intraphotoreceptor calcium
falls, recoverin stimulates retinal guanylate cy-
clase, so that photoreceptor cGMP concentra-
tions return to high, dark-adapted levels. We are
beginning to examine the possible role of re-
coverin in a variety of retinal degenerations. In
our positional cloning studies of the Xpl 1.2 re-
gion of the human genome, we have assembled
yeast artificial chromosome (YAC) contigs cover-
ing most of this region and have utilized one of
the YACs as a probe to screen a human retinal
cDNA library. We have cloned at least five cDNAs,
all of which map back to the correct Xpl 1.2 re-
gion. These are being sequenced and will be used
as probes in Northern blots of patient samples to
test for their possible involvement in these
disorders.
We have also begun an investigation of inborn
errors of peroxisome biogenesis and function.
Zellweger syndrome, a neurodevelopmental dis-
order fatal in infancy, is the disease paradigm.
Cells and tissues from these patients exhibit defi-
ciency of virtually all peroxisomal enzymes and
lack normal-appearing peroxisomes. We have
cloned the genes for two peroxisomal membrane
proteins, the 70-kDa peroxisomal membrane
protein (PMP70) and the 35-kDa protein
(PMP35). PMP70 isamemberof theATP-binding
cassette (ABC) transporter protein family that
also includes the mammalian multiple-drug resis-
tance protein (MDR) and the CFTR protein in-
volved in cystic fibrosis. We have cloned the en-
tire PMP70 cDNA, determined its sequence, and
used it to clone, map, and characterize the
PMP70 gene. In collaboration with Hugo Moser,
we are analyzing the possible role of PMP70 in
Zellweger syndrome. We have identified three
PMP70 mutant alleles, and our results suggest
that PMP70 mutations account for one of the
Zellweger complementation groups. We also
have determined the complete sequence of hu-
man PMP35 and identified one mutant allele in
another Zellweger complementation group. We
now are focusing on expression systems to test
directly the functional consequences of these
mutations on peroxisomal biogenesis.
450
Human Molecular Genetics in Two X-linked Diseases
Stephen T. Warren, Ph.D. — Associate Investigator
Dr. Warren is also Associate Professor of Biochemistry and of Pediatrics at Emory University School of
Medicine. He received his Ph.D. degree in genetics from Michigan State University. Prior to joining the
faculty at Emory, he did postdoctoral research with Richard Davidson at the University of Illinois School
of Medicine in Chicago.
MY laboratory is involved in human molecu-
lar genetics and is especially interested in
the identification of genes responsible for ge-
netic disease. Our efforts currently focus on X-
linked disease, particularly the fragile X syn-
drome and Emery-Dreifuss muscular dystrophy.
Fragile X Syndrome
Fragile X syndrome is the commonest form of
inherited mental retardation and one of the most
prevalent genetic diseases known, affecting ap-
proximately 1 per 1,000 persons worldwide. As
the name implies, fragile X syndrome is asso-
ciated with a fragile chromosomal site, which has
been localized to band position Xq27.3. Fragile
sites are heritable loci that form cytologically evi-
dent gaps within chromosomes under specific
biochemical induction. Although such sites are
numerous throughout the human genome, the
fragile X site is the only one associated with a
disease.
Fragile X syndrome is unusual among mamma-
lian genetic disorders in that 20 percent of the
males with a fragile X chromosome are not af-
fected while approximately 30 percent of carrier
females show some degree of mental impairment.
The fragile X mutation is less frequently pene-
trant (i.e., resulting in mental retardation) among
the siblings of these normal carrier males (called
transmitting males), and penetrance increases
with each generation from a transmitting male
until it reaches so-called Mendelian ratios, where
half of the male children of a carrier female are
affected, typical of an X-linked gene. One excep-
tion is that daughters of transmitting males are
never affected, but their children may be. This
confusing hereditary pattern, unique among ge-
netically studied organisms, has been referred to
as the Sherman paradox, in reference to Steph-
anie Sherman's description of it.
Our work over the past year has not only identi-
fied the gene responsible for fragile X syndrome
but has also uncovered an unusual mutation
whose behavior explains the Sherman paradox.
Working with an international group of collabora-
tors, including Thomas Caskey (HHMI, Baylor
College of Medicine), David Nelson (Baylor),
and Ben Oostra (Erasmus University, the Nether-
lands) , we identified yeast artificial chromosome
(YAC) clones, previously developed in my labora-
tory, that mapped near chromosome breakpoints
involving the fragile X site. Using cloned DNA
derived from one of these YACs, we identified a
cDNA encoded by a gene that the translocation
breakpoints had interrupted. This gene, termed
fragile X mental retardation 1 {FMR-1), pro-
duces a 4.4-kb message expressed at high levels
in the brain and testes. Male fragile X patients
have macro-orchidism, or enlarged testes, in ad-
dition to mental retardation.
Within the FMR-1 mRNA is an unusual repeat
of the trinucleotide CGG. In normal individuals,
there are most frequently 29 repeats, though this
can vary between 6 and 52. Among transmitting
males and most normal carrier females, there are
between 52 and 200 CGG repeats. Among men-
tally retarded patients, the codon repeats up to
1,300 times and is markedly unstable in mitotic
cells. Importantly, when the repeat expands
beyond 250, it spontaneously methylates the
nearby DNA, turning off the FMR-1 gene. Re-
moval of the gene product by hypermethylation
in response to the massive augmentation of CGG
repeats is believed to be the mechanism of fragile
X syndrome.
Work with our colleagues in Houston revealed
an apparent relationship between the number of
repeats in a normal carrier female and the proba-
bility of having a mentally retarded son. In gen-
eral, the smaller the abnormal repeat, the lower
the risk of expansion to the full fragile X muta-
tion in an offspring. Above a threshold of approxi-
mately 200 repeats, the fragile X chromosome
when passed down always undergoes expansion
to the full mutation. Hence carrier mothers with
200 repeats have a 50 percent risk of having a
retarded son, while those with only 70 repeats
have a 9 percent risk. This explains the paradox
of penetrance in fragile X syndrome: the sequen-
tial increase in the CGG repeat with each genera-
tion imparts a concomitant increase in risk of
having an affected child.
451
Human Molecular Genetics in Two X-linked Diseases
We are focusing on three major efforts related
to the FMR-1 gene. The first is to determine the
mechanism of expansion of the CGG repeat by
both the direct sequence analysis of the fragile X
mutation and the introduction of long synthetic
CGG repeats into mammalian cells and, subse-
quently, into mice. Second, we are attempting to
understand the normal function of the FMR-1
gene product. Toward this goal, we have recently
cloned and sequenced the homologous gene
from the mouse and are in the process of cloning
the yeast and the nematode genes. Finally, we are
excited about finding a number of other human
genes, distinct from FMR-1, that similarly contain
long CGG repeats. These genes, whose functions
are presently not understood, may share a similar
functional utilization of the CGG repeat; may un-
dergo similar mutational changes, perhaps lead-
ing to diseases exhibiting unusual genetic pat-
terns; and may represent other, autosomal fragile
sites.
Emery-Dreifuss Muscular Dystrophy
Emery-Dreifuss muscular dystrophy (EDMD) is
the most frequent X-linked muscular dystrophy
following the Duchenne and Becker types. It is a
slowly progressive disease that usually leaves af-
fected males ambulatory until middle age. Heart
muscle involvement is frequent, sometimes re-
sulting in early sudden death due to heart block.
If the disease is identified early, such a death can
be prevented by pacemaker implantation.
We have performed genetic linkage studies in
two large families and have localized the EDMD
gene to the terminal band of the X chromosome
(band Xq28) just distal from the fragile X site.
We can now place the gene within an approxi-
mate 2,000 kb of DNA. Using selective cDNA li-
braries and DNA of this region cloned into cos-
mids or YACs, we are mapping muscle-expressed
genes within Xq28. This entails use of a somatic
cell hybrid mapping panel containing six distinct
Xq28 fragments. Any genes that map within the
region believed to contain the EDMD gene will be
used to search for mutations in patients.
Another outcome of our genetic mapping stud-
ies has been the identification of young, asymp-
tomatic males who will eventually suffer from
EDMD. Cardiac function of these males is being
carefully followed, and pacemaker implantation
is performed when warranted prior to full heart
block. Thus we have virtually eliminated sudden
death due to EDMD in these two families.
In situ hybridization detecting human DNA
within the metaphase chromosomes of a so-
matic cell hybrid ( micro21D), which contains
a human-rodent translocation between the
centric fragile X chromosome (yellow) and a
rodent chromosome arm ( red). The hybrid was
constructed under conditions favoring re-
arrangements specific for the fragile X site. In-
deed, the human translocation breakpoint in
this hybrid is within the CGG repeat of the frag-
ile X mutation. Somatic cell hybrids with trans-
locations of this nature proved instrumental in
cloning the fragile X site and its associated gene
(FMR-1).
Research of Stephen Warren.
432
The MyoD Gene Family: A Nodal Point During
Specification of Muscle Cell Lineage
Harold M. Weintraub, M.D., Ph.D. — Investigator
Dr. Weintraub is also a Full Member in the Division of Basic Sciences at the Fred Hutchinson Cancer
Research Center and Affiliate Professor of Pathology and Zoology at the University of Washington, Seattle.
He received his M.D.-Ph.D. degree from the University of Pennsylvania School of Medicine and completed
his postdoctoral studies at the Medical Research Council Laboratory of Molecular Biology in Cambridge,
England. Before joining the staff at the Hutchinson Center, Dr. Weintraub was in the Department of
Biochemical Sciences at Princeton. He is a member of the National Academy of Sciences and the American
Academy of Arts and Sciences. Among his many honors are the Eli Lilly Award and the Richard Lounsbery
Award from the National Academy of Sciences.
THE MyoD gene converts many differentiated
cell types into muscle. MyoD is a member of
the protein family characterized by the basic
helix-loop-helix, a 68-amino acid domain in
MyoD that is necessary and sufficient for myo-
genesis. MyoD binds cooperatively to muscle-
specific enhancers and activates transcription.
The helix-loop-helix motif is responsible for di-
merization, and, depending on its dimerization
partner, MyoD activity can be controlled.
MyoD senses and integrates many facets of the
cell state. The gene is expressed only in skeletal
muscle cells and their precursors; in nonmuscle
cells it is repressed by specific genes. MyoD acti-
vates its own transcription, perhaps stabilizing
commitment to myogenesis. Despite this seem-
ingly overwhelming evidence that MyoD is cru-
cial for myogenesis in vertebrates, recent experi-
ments with Michael Krause and with Andrew Fire
show that zygotic deletions of MyoD in worms
result in embryos that retain the capacity to acti-
vate muscle cell differentiation.
Muscle-Specific Transcriptional
Activation by MyoD
Our laboratory has focused on the mechanism
by which MyoD activates transcription. Previous
experiments showed that when the 13-amino
acid basic region of the ubiquitously expressed
basic helix-loop-helix gene E12 replaces the
corresponding basic region of MyoD, the result-
ing MyoD-E12Basic chimeric protein can bind
specifically to muscle-specific enhancers in vitro
and form dimers with El 2, but cannot activate a
cotransfected reporter gene or convert lOTVi
cells to muscle. Back mutation of this chimeric
protein (with the corresponding residues in
MyoD) reestablishes activation. A specific ala-
nine is involved in increasing DNA binding, and a
specific threonine is required for activation.
A reporter gene containing MyoD-binding sites
located downstream from the transcription start
site was used to show that MyoD-E12Basic can
bind in vivo and thereby inhibit expression of the
reporter. In vivo binding is also supported by the
fact that the addition of the "constitutive" VPl6
activation domain to MyoD-El 2Basic restores full
trans-activation potential. The normal MyoD acti-
vation domain maps within the amino-terminal
53 residues and can be replaced functionally by
the activation domain of VP16.
The activity of the MyoD activation domain is
dramatically elevated when deletions are made
almost anywhere in the rest of the MyoD mole-
cule, suggesting that the activation domain in
MyoD is usually masked. Surprisingly, MyoD-
E12Basic can activate transcription in CVl and
B78 cells (but not in IOT1/2 or 3T3 cells), sug-
gesting that the activation function of the basic
domain requires a specific factor present in CVl
and B78 cells. We propose that the masked MyoD
activation domain requires, in order to function,
the participation of another factor that recog-
nizes the basic region.
By replacing the MyoD basic region and the
adjacent four-residue junction region with helix
1 into the corresponding region of El 2, we have
recently shown that this small section of MyoD is
sufficient for myogenesis. Our work suggests that
only three residues, A114, Tn,, and K124, are
uniquely critical for "recognition factor" func-
tion and subsequent activation of myogenic gene
transcription.
Control of MyoD Activity
A variety of transforming agents, including a
variety of growth factors, the oncogenes src, ras,
fos,jun,fps, erbA, myc, and El A, and such chem-
ical agents as butyrate and phorbol esters, inhibit
myogenic differentiation. Most of these reagents
can inactivate the expressed MyoD protein; in ad-
dition, several (such as ras and fos) also inhibit
MyoD transcription. Whether this is a secondary
433
The MyoD Gene Family: A Nodal Point During Specification
of Muscle Cell Lineage
effect due to an inhibition of the autoactivation
function of MyoD protein or a more direct inhibi-
tion of MyoD transcription remains to be deter-
mined. Rhabdomyosarcoma cells (derived from
tumors of patients who harbor a genetic predis-
position to myogenic tumors) differentiate
poorly but express MyoD, suggesting that loss of
anti-oncogene activity at the rhabdomyosarcoma
locus can also impinge on MyoD action. The spe-
cific pathway by which each of these oncogenes,
anti-oncogenes, and growth factors inhibits myo-
genesis provides a potential clue to how MyoD
might integrate information coming from many
aspects of cellular function.
Recently, in collaboration with the laboratory of
Inder Verma, we found that the leucine zipper re-
gion of the jun oncogene actually binds to the he-
lix-loop-helix region of MyoD, both in vivo and in
vitro. Similarly, assays for a recognition factor for
MyoD activation show that such a factor, which is
missing in rhabdomyosarcoma cell lines, can be
provided in trans by fusion with 1 OTVi cells. Possi-
bly failure to activate myogenesis leads to increased
proliferation and then secondary effects that give
rise to rhabdomyosarcomas.
Activation of MyoD During Development
We are studying developmental activation of
MyoD in mice, worms, and frogs. In both mice
and worms, deletional analysis has identified
regulatory sequences upstream of the MyoD
gene that are important for correct developmen-
tal activation of MyoD. Current efforts focus on
identifying trans-acting elements that integrate
with these sequences. In worms, several mater-
nal-effect mutants have been isolated by Jim
Priess and his colleagues. These mutant em-
bryos produce excess muscle from the wrong
lineage. It is possible that these mutants define
elements involved in the segregation of myo-
genic potential to specific cells during early
cleavage stages. In apparent contrast to worms,
frogs seem to activate MyoD in all cells of the
blastoderm; however, expression is stabilized
only in those presumptive mesodermal cells
that become induced by vegetal inducing fac-
tors such as activin. Frogs also contain maternal
MyoD mRNA, which, however, seems not to be
crucial for subsequent myogenesis, as its de-
struction with anti-sense DNA results in normal
muscle gene activation.
454
Structural and Functional Studies of the T Cell
Antigen Receptor
Arthur Weiss, M.D., Ph.D. — Associate Investigator
Dr. Weiss is also Ephraim P. Engleman Distinguished Professor of Rheumatology and Associate Professor
of Medicine and of Microbiology and Immunology at the University of California, San Francisco. He
received his undergraduate education at the Johns Hopkins University and was an M.D./Ph.D. student at
the University of Chicago, where he studied immunology in the laboratory of Frank Fitch. He did
postdoctoral work with Jean- Charles Cerottini and K. Theodore Brunner at the Swiss Institute for
Experimental Research, Lausanne. After an internship and residency in internal medicine at UCSF, he
became a postdoctoral fellow in rheumatology with John Stobo.
THE immune system has evolved to provide an
organism with a flexible and dynamic mecha-
nism to respond specifically to a wide variety of
antigens. During the initiation of an immune re-
sponse, antigen must not only be recognized by
antigen-specific lymphocytes, but this recogni-
tion event must lead to cellular activation. T and
B lymphocytes comprise the antigen-specific
components of the cellular immune system. The
activation of T lymphocytes is critical to most im-
mune responses, since it permits these cells to
exert their potent regulatory or effector activi-
ties. During activation, relatively quiescent cells
undergo complex changes involving cell differ-
entiation and proliferation.
Following exposure to antigen, activation of T
lymphocytes is limited to only those cells ex-
pressing antigen-specific receptors. Activation is
a consequence of ligand-receptor interactions
that occur at the interface of the T cell and an
antigen-presenting cell. These interactions initi-
ate intracellular biochemical events within the T
cell that culminate in cellular responses. Our
goal is to understand how cell surface molecules
on the T cell, and in particular the T cell antigen
receptor (TCR), initiate T cell activation.
Although it is clear that a number of different
cell surface molecules on the T lymphocyte and
the antigen-presenting cell may participate in the
complex cell-cell interaction that occurs during
antigen presentation, the TCR must play a promi-
nent role. Here the familiar lock and key analogy
is appropriate. Antigen is the ligand (key) for a
particular set of clonally distributed receptors
(locks) on T lymphocytes. Antigen often repre-
sents a protein fragment that is physically asso-
ciated with a molecule of the major histocompati-
bility complex (MHC).
The TCR is an extraordinarily complex struc-
ture. It consists of an a//3-chain disulfide-linked
heterodimer (Ti) derived from immunoglobulin-
like genes that is noncovalently associated with
six invariant chains of the CD3 complex and a
f-chain dimer. CD3 consists of four chains (5-, 7-,
and two e-chains) derived from three closely
linked homologous genes located on chromo-
some 1 1 . The f dimer, derived from the products
of two homologous genes on chromosome 1 , may
represent a homodimer (f^ or heterodimer con-
sisting of fry (77 is an alternatively spiced form of
f) or (7, a homologous protein, is also a com-
ponent of the IgfFc receptor on mast cells and
basophils) . Ti is the ligand-binding subunit of the
TCR, since it contains all the information needed
to recognize antigen and MHC specificities. CD3
and f have been thought to play some role in
transducing the ligand occupancy state of Ti
across the plasma membrane. Hence the struc-
tural basis for the association of Ti, CD3, and f is
of interest.
Previous studies from our laboratory have dem-
onstrated that coexpression of all of the chains of
the oligomeric TCR — Ti, CD3, and f — on the
plasma membrane is obligatory for efficient TCR
expression. Studies from this laboratory demon-
strated that the structural and functional basis for
the interaction between Ti, CD 3, and f is con-
tained within regions including the transmem-
brane domains of these proteins. Further muta-
tional studies are in progress to understand more
precisely how Ti, CD3, and f interact function-
ally within these domains.
Since the transmembrane regions are responsi-
ble for the Ti-CD3-r association, we have taken
advantage of this information to separate regions
or domains of CD3 and f from Ti. This has been
accomplished by constructing chimeric mole-
cules between other cell surface molecules
linked to the cytoplasmic domain of TCR chains.
A chimeric molecule consisting of the CDS extra-
cellular and transmembrane domains fused to
the f cytoplasmic domain acquired the signal-
transducing capacity of the entire TCR. This find-
ing demonstrates that the cytoplasmic domain of
CD3 f can link the TCR to intracellular signaling
machinery.
These studies are being extended to define the
region of f that interacts with such intracellular
molecules and to identify these molecules. Re-
cently we identified a 70-kDa tyrosine phospho-
435
Structural and Functional Studies of the T Cell Antigen Receptor
protein that associates with the f-chain; we are in
the process of characterizing this protein. We
would also like to understand the function of the
associated CD3 complex within the complex
oligomeric TCR. Moreover the ability to create
such functional chimeric receptors may permit
the creation of novel antiviral or antitumor TCRs
that may be of value in gene therapy.
Stimulation of the TCR initiates cellular activa-
tion by inducing a transmembrane signal that is
manifested as the formation of intracellular bio-
chemical mediators called second messengers,
which can initiate or influence cellular response
pathways. Recent studies demonstrate that the
TCR activates an uncharacterized protein-tyro-
sine kinase (PTK) as the initial event leading to
cellular activation. This PTK appears to be asso-
ciated with the TCR f-chain. The activation of
this PTK results in the tyrosine phosphorylation
of many cellular proteins, one of which is phos-
pholipase C-7I (PLC-7I). The phosphorylation
of PLC-7I activates this enzyme to hydrolyze a
rare membrane lipid, PIP2 (phosphatidylinositol
4,5-bisphosphate). This yields two potent intra-
cellular second messengers (inositol 1,4,5-tris-
phosphate and diacylglycerol) that regulate the
mobilization of intracellular calcium and activa-
tion of the enzyme protein kinase C, respectively.
These latter two events are physiologically im-
portant to subsequent cellular responses.
The mechanism by which the TCR couples to
intracellular signaling pathways is largely unde-
fined, as are many of the components of the sig-
naling pathways themselves. To define and char-
acterize the molecular basis by which the TCR
regulates these pathways, we are using a somatic
cell genetic approach. In work supported by a
grant from the National Institutes of Health, we
have isolated a number of mutants derived from T
cell leukemic lines that are defective in TCR-
mediated activation of the inositol phospholipid
pathway. Unlike the parental cells, none of these
mutants produce lymphokines in response to
TCR stimulation. These mutants define four dis-
tinct gene products other than the TCR chains
that are required for the functional activation of
the inositol phospholipid pathway.
In one mutant, the activation of the tyrosine
kinase pathway is still intact; however, some of
the substrates of the tyrosine kinase pathway are
not phosphorylated, including PLC-7I. The phe-
notype of this mutant suggests that there may be a
coupling protein or second PTK that is deficient.
The defect in another of these mutants can be
attributed to the absence of a cell surface protein,
CD45, with tyrosine phosphatase activity. The
absence of CD45 prevents the TCR from activat-
ing the tyrosine kinase or phosphatidylinositol
inositol pathway. At least one target of the CD45
tyrosine phosphatase is a regulatory phospho-
tyrosine residue in the PTK Ick. Thus a protein-
tyrosine phosphatase regulates the activity of a
PTK. This suggests a complex autoregulatory sys-
tem that we are intensively studying. In the re-
maining two mutants, biochemical studies have
complemented our genetic approach. In neither
of these mutants is the PTK pathway activated.
Preliminary studies suggest that one of these mu-
tants is deficient in a previously identified PTK.
Thus these mutants are proving to be valuable
tools with which to dissect the complexities of
the signal transduction pathways and their rela-
tionships to cellular responses.
T cell activation is a complex process that is
regulated by cell surface molecules. Investiga-
tion of the molecules and events involved in the
activation of T cells should lead to a more com-
plete understanding of T cell biology and a more
rational approach to the manipulation of the im-
mune system. Moreover, through the study of the
activation of T cells, it is likely that insight into
other biological systems involving cell prolifera-
tion and differentiation will emerge.
456
Following the Life History of Lymphocytes
Irving L. Weissman, M.D. — Investigator
Dr. Weissman is also the Karel and Avice Beekhuis Professor of Cancer Biology and Professor of Pathology,
Developmental Biology, and (by courtesy) Biology at Stanford University School of Medicine. He directs
the Program for Molecular and Genetic Medicine and the Immunology Program. He received his M.D.
degree from Stanford and remained to do postdoctoral studies in the Department of Radiology. He also
studied at Oxford with Jim Gowans in 1964 and returned in 1975 for part of a sabbatical year, which he
then completed with Melvin Cohn at the Salk Institute. Dr. Weissman is a member of the National
Academy of Sciences and the American Academy of Arts and Sciences.
LIKE all other blood cells, lymphocytes — the
principal players in immune recognition of
self from nonself — are derived ultimately from
stem cells in the bone marrow. It is both biologi-
cally and clinically important to delineate the de-
cisions these bone marrow precursors make as
they pass through microenvironments that define
the type of lymphocyte (or other blood cell) they
shall become. We have focused on identifying the
earliest cells in mouse and human bone marrow
that have multipotent capacity, the so-called he-
matopoietic (blood-forming) stem cells.
Several years ago we were able to isolate the
hematopoietic stem cell of the mouse. This year
we showed that no other cell type in the bone
marrow has stem cell activity or potential. We
have also demonstrated its full developmental
potential by transferring a single stem cell from
one mouse strain mixed with 1 00 stem cells from
another strain into lethally irradiated mice of the
second strain. Progeny from the single marked
stem cell regularly gave rise to over 100 million
blood cells of all types, including hundreds to
thousands of stem cells.
These thousands of stem cells, derived from the
initially injected single cell, could be retrieved
and transferred to a second generation of irra-
diated animals, all of whom were fully reconsti-
tuted. Thus this stem cell has a remarkable pro-
file of activities, including that of massive
self-renewal.
In the past year we also found that stem cells in
mouse fetuses have the capability of giving rise to
a broader variety of T cells than do stem cells in
the adult bone marrow.* The additional types of
T cells derived from fetal stem cells are those
cells that move from the fetal thymus to the skin
and other epithelial coverings of the body, pre-
sumably to act as sentinels to protect against in-
coming infectious microorganisms.
Most remarkably, in preparation for this added
capacity in fetal life, stem cells apparently start a
T cell developmental "clock" on one T cell re-
ceptor chromosome. We propose that after sev-
eral stem cell divisions, the clock moves past that
part of the chromosome that will be expressed as
receptors for antigens expressed on epithelial T
cells, and concentrates only on other T cell re-
ceptors to fight off infection in other sites of the
body. We believe that the genetic events that set
developmental clocks and then shut them down
at the level of hematopoietic stem cells lie at the
heart of understanding the determination of
choices that cells can make in general, and hope
to develop in the next few years new methods to
investigate the regulators that set the clock and
how they do so in the developmental history of
the mouse embryo.
Several years ago our laboratory developed a
mouse model of human organ function, wherein
human hematopoietic and lymphoid organs, such
as fetal liver, thymus, and bone marrow, could be
implanted in the immunodeficient SCID mouse.*
We found that the human lymphoid microenvi-
ronments implanted in the SCID mouse could
provide the right soil for human T cell lymphoma
growth, while the same primary tumors from pa-
tients will not grow in the SCID mouse in any
other microenvironment. This presents the op-
portunity to study the earliest stages of growth
and malignant progression of human cancers,
lymphomas, and leukemias, if the general princi-
ple holds that their early growth is dependent on
the organ in which they find themselves. This im-
plies that there might exist factors within human
organs that are responsible for the early neoplas-
tic grov^h of cancer cells, some of which may be
tissue specific.
In the past year, we also identified the genes
that encode a Peyer's patch homing receptor —
the molecule that is involved in the traffic of lym-
phocytes from the bloodstream to intestinal lym-
phoid organs such as Peyer's patches, appendix,
and mesenteric lymph nodes. The molecule is a
member of the integrin family of adhesion pro-
teins and uses the combination a^^-j. Lymphomas
that express a^^^ were found to metastasize to
* This work is supported by a grant from the National Insti-
tutes of Health.
457
Following the Life History of Lymphocytes
Peyer's patches as well as to bind to the blood
vessels in the traffic zones of Peyer's patches,
whether the a^fi-j molecules were found to be ex-
pressed on these lymphomas or were transfected
to them by our cDNA clones. Antibodies to either
or jSy could block this selective adhesive event.
We have also developed a new mouse strain by
transgenic technology, deleting all cells that ex-
press the activated killer cell gene granzyme A
(first identified and cloned in this laboratory) by
attaching to that gene a "suicide" gene obtained
from Richard Palmiter (HHMI, University of
Washington). When killer cells are activated in
this mouse strain they begin to express the diph-
theria toxin A chain suicide protein, and die. Un-
expectedly, we have uncovered in this transgenic
mouse strain a profound effect on the life span of
all CDS T cells, and not just killer T cells and
natural killer cells. In the next year, we plan to
utilize these mice to delineate the role of killer
cells in normal and pathological immune reac-
tions in vivo and to define the mechanism by
which non-killer CDS T cells are deleted after
their emigration from the thymus.
438
Function and Regulation of the Cystic Fibrosis
Transmembrane Conductance Regulator
Michael J. Welsh, M.D. — Investigator
Dr. Welsh is also Professor of Internal Medicine and of Physiology and Biophysics at the University of Iowa
College of Medicine, Iowa City. He earned his M.D. degree from the University of Iowa. He completed his
residency at the University of Iowa College of Medicine; held clinical and research fellowships in
pulmonary diseases and cardiovascular research at the University of California, San Francisco; and did
postgraduate research in physiology and cell biology at the University of Texas, Houston. He then returned
to the University of Iowa as a faculty member.
CYSTIC fibrosis (CF) is a common lethal ge-
netic disease involving defective electrolyte
transport by several epithelia. In normal epithelia
of the airways, the intracellular second messen-
ger cAMP regulates chloride (Cl~) channels in
the apical membrane. These channels provide
both a pathway through which Cl~ flows and a
key point for regulation of its movement. When
cAMP increases, the channels open and Cl~ flows
from the cell into the airway lumen, drawing
water with it. Secretion of salt and water is impor-
tant in generating the respiratory tract fluid, a crit-
ical component of the mucocilliary defense
mechanism. In CF airway epithelia, cAMP fails to
open the Cl~ channels. As a result secretion is
defective and the respiratory tract fluid is abnor-
mal. This defect is believed to be the major cause
of morbidity and mortality in CF lung disease.
CF is caused by mutations in the gene encoding
the cystic fibrosis transmembrane conductance
regulator (CFTR). When we expressed the nor-
mal CFTR gene in CF airway epithelial cells, the
channel defect was corrected. That result indi-
cated that CFTR was somehow intimately asso-
ciated with cr channels, but did not reveal its
function. Toward further understanding, we ex-
pressed CFTR in a number of nonepithelial mam-
malian cells lacking both endogenous CFTR and
cAMP-activated CP channels. In every case in
which CFTR was expressed, the cell generated a
unique CI" channel that was activated by cAMP.
Such channels were not observed in cells lacking
the CFTR gene.
The cAMP-regulated CP channels displayed
regulatory and biophysical properties that were
identical to those observed in cells expressing en-
dogenous CFTR, as well as those in the apical
membrane of airway epithelia. We stress apical
membrane because that is where the CF defect is
observed. These results suggested that CFTR itself
might be a CP channel. This conclusion was
quite controversial because CFTR did not resem-
ble any channels previously described. It seemed
instead to resemble a family of proteins that in-
clude membrane pumps.
To test the hypothesis that CFTR is a CP chan
nel, we changed specific amino acids within the
CFTR sequence. When an ion crosses a cell mem-
brane through a channel, it must interact with the
amino acids in that channel. In changing some of
the positively charged amino acids to negatively
charged ones, we changed the anion selectivity
sequence. The channels normally favor CP over
iodide (P), but after two of the amino acids were
mutated in CFTR, the channel favored P over
CP. The ability to change the properties of the
conduction mechanism by altering specific
amino acids provided the most compelling evi-
dence that CFTR is itself a cAMP-regulated CP
channel.
The evidence that CFTR forms a CP channel
and that the electrolyte transport defect in CF is
in the apical membrane suggested that CFTR
would be located in the apical membrane of se-
cretory epithelia. To test that hypothesis, we de-
veloped antibodies to CFTR and localized it with
immunofluorescence confocal microscopy. In
several lines of intestinal epithelial cells that se-
crete CP, we found that CFTR was located in the
apical membrane. That result indicates that CFTR
is in a position where it can directly mediate CP
movement across the membrane.
An increase in the cellular concentration of
cAMP opens the CFTR CP channel. Several of our
studies have shown that CFTR is phosphorylated
and thus regulated by a cAMP-dependent protein
kinase (PKA). (Kinases regulate cell proteins by
attaching a phosphate group.) When we used
cell-free patches of membrane containing CFTR,
we found that addition of the catalytic subunit of
PKA opened the CFTR CP channel.
Moreover, in vitro biochemical studies dis-
closed that PKA phosphorylates CFTR on seven
residues. In vivo studies further showed that PKA
phosphorylates four serine residues located
within a portion of the protein called the R do-
main. Evidence that these reactions are important
for opening the channel came from the observa-
tion that PKA failed to open the channel when the
four serines had been changed to alanines. Fur-
thermore, when the R domain was deleted from
CFTR, the channel no longer required PKA to
open: it was open even without phosphorylation.
The CFFR CP channel also contains stretches
439
Function and Regulation of the Cystic Fibrosis Transmembrane
Conductance Regulator
of amino acids similar to those found in a number
of proteins that utilize ATP. To determine if the
CFTR Cl~ channel utilizes ATP, we first phos-
phory'lated it with PKA and then examined the
effect of ATP on cell-free membrane patches con-
taining CFTR. We found that ATP is required for
the channel to open. The channel appears to re-
quire the hydrolysis of ATP; nonhydrolyzable ana-
logues would not open it. This observation pro-
vides the first example of an ion channel
regulated by ATP hydrolysis. We were surprised
because an ion channel, by its nature, is a passive
structure. The results suggest that the energy
from ATP hydrolysis may be required to open the
channel and that, once open, Cl~ flows through
passively.
These studies have begun to define the func-
tion of CFTR, and in so doing, they begin to ex-
plain why the apical membrane of CF epithelia
are impermeable to CP. They also suggest that
correction of the underlying defect would be a
feasible therapeutic strategy.
In addition to the studies of the CFTR Cl~ chan-
nel, the laboratory is investigating other channels
in epithelial cells. Other CI" channels are not as
well explored as the CFTR channel; but if their
regulation were understood, they might possibly
be utilized to bypass the CFTR CP channel de-
fect. Potassium channels, too, are important, be-
cause they generate the intracellular voltage that
drives Cl~ out of the cell across the apical mem-
brane into the airway lumen. Finally, the labora-
tory has a major focus on the regulation of intra-
cellular calcium. The calcium ion is known to
regulate some CP channels in the airway epithe-
lial cells, and again, a better understanding of this
process might open an approach by which the CF
defect could be circumvented.
CI
ATP
ATP
Model showing the proposed domains of the cystic fibrosis transmem-
brane conductance regulator ( CFTR). Membrane is cross-hatched area.
MSD refers to membrane- spanning domain, NBD to nucleotide-binding
domain, and PKA to cAMP-dependent protein kinase. Mutation of the
CFTR gene is responsible for the defective function of chloride channels in
CF patients.
Adapted from Anderson, M.P., Berger, H.A., Rich, D.P., Gregory, R.J.,
Smith, A.E., and Welsh, M.f. 1991. Cell 67:775-784. Copyright© 1991 by
Cell Press.
440
Identification of the Gene Responsible for
Adenomatous Polyposis
Raymond L. White, Ph.D. — Investigator
Dr. White is also Professor of Human Genetics and Adjunct Professor of Cellular, Viral, and Molecular
Biology at the University of Utah School of Medicine. He received a B.S. degree in microbiology from the
University of Oregon, Eugene, and a Ph.D. degree in microbiology from the Massachusetts Institute of
Technology. He did postdoctoral research with David Hogness at Stanford University. Dr. White has held
various academic appointments at the University of Massachusetts, Worcester, and the University of Utah.
He was recently named the Thomas D. Dee II Professor of Human Genetics at the University of Utah. Dr.
White is a corecipient of the National Health Council's National Medical Research Award and was recently
elected to the National Academy of Sciences.
FORMATION of a polyp in colonic epithelium
is an early event in the development of colon
cancer. In some families a dominantly inherited
mutation causes familial adenomatous polyposis
coli (APC), a condition characterized by large
numbers of polyps and a consequently high risk
for colon cancer among those who inherit the
mutant allele. Because colon carcinoma is the
leading cause of cancer death in the United
States, identification of the gene that causes poly-
posis in APC families has been the goal of investi-
gators seeking to understand cellular mecha-
nisms that can lead to malignancy. This goal was
achieved when the gene APC was isolated in
1991.
Genetic linkage studies, combined with evi-
dence from microscopic analysis of altered chro-
mosomes, had localized the polyposis-causing
mutation in APC families to a specific region on
the long arm of chromosome 5 . Molecular tech-
niques, including physical mapping of large DNA
fragments, were brought into play to define the
region more clearly, because numerous genes be-
sides APC were likely to reside in that portion of
the chromosome.
With molecular probes derived from the large
DNA fragments, small deletions were found in the
DNA of two unrelated patients with APC. These
deletions proved to be the key to isolation of
APC, for they defined a very small region in
which to search. A nearby gene, MCC, which is
often mutated in sporadic colonic tumors, had
been a candidate; but MCC was not interrupted
by either of the small deletions, and the search
continued.
Three other genes were found that did fall
within the 100-kilobase region deleted in both
patients, and each of those was investigated for
the presence of very small ("point") mutations
in the DNA of other persons with APC. One of
these genes was subsequently identified as APC
on the basis of point mutations in several
patients.
Although each of these APC-specific genetic al-
terations had occurred in a different coding re-
gion of the gene, all were of a kind that could be
expected to prevent the generation of a normal
protein product. In one patient the mutation was
new in his family. His parents' chromosomes
carried normal copies of APC, but two of his chil-
dren had inherited his mutant allele.
APC, like NFl (neurofibromatosis type 1 ) , an-
other tumor-associated gene cloned recently, ap-
pears to fall into the growing category of tumor-
suppressor genes. There were no clues, however,
to its physiological function, as had come to light
for NFl with the discovery of similarities in
amino acid sequence between its predicted prod-
uct and proteins known to participate in trans-
duction of growth signals within the cell. The
only structural motif that could be identified was
the presence of "heptad repeats," or series of
seven amino acids in tandem arrays. Heptad re-
peats often suggest that a protein can interact
with similar proteins to form coiled helical
structures.
The putative product of MCC also contains
heptad repeats. It is intriguing to speculate upon
the possibility of interaction between this pro-
tein and the APC product, in view of the fact that
alterations in both genes are known to play im-
portant roles in the pathway to colon cancer.
Some families exhibit an unusually high inci-
dence of colon cancer without showing a pattern
of classical polyposis; that is, affected members
may develop very few polyps. Inherited muta-
tions in the APC gene are being sought in these
families, and colonic tumors in the general popu-
lation are being tested for mutant APC alleles.
Characterization of a variety of APC mutations
may help to explain differences in clinical pro-
files among people with a genetic predisposition
to colon cancer. It may also lead to new informa-
tion about the role of the gene in normal cellular
processes.
For many families with APC, discovery of the
gene has immediate value for presymptomatic,
even prenatal, diagnosis of carrier status. When a
441
Identification of the Gene Responsible for Adenomatous Polyposis
particular mutation has been characterized in a
family, members who carry the disease allele can
be identified directly. Families for whom the mu-
tant allele is as yet unidentified may still benefit
from analysis of genetic linkage between the pu-
tative disease allele and a highly polymorphic
DNA marker system that has been detected within
one of the other genes lying in the APC deletion
region. Early diagnosis at the molecular level will
free unaffected members of these families from
some anxiety and spare them the discomfort and
expense of frequent colonoscopic examinations.
442
Mechanisms of the Biological Activities of
Membrane Glycoproteins
Don C. Wiley, Ph.D. — Investigator
Dr. Wiley is also Professor of Biochemistry and Biophysics at Harvard University and Research Associate in
Medicine at the Laboratory of Molecular Medicine at the Children's Hospital, Boston. He received his Ph.D.
degree in biophysics from Harvard University. He then joined the faculty at Harvard and served as
Assistant and Associate Professor of Biochemistry and Molecular Biology before attaining his present
position. Dr. Wiley is Chairman of the Department of Biochemistry and Molecular Biology. He is a member
of the National Academy of Sciences and a Fellow of the American Academy of Arts and Sciences. Among
his honors is the Louisa Gross Horwitz Prize from Columbia University.
Tcell recognition occurs when cell surface his-
tocompatibility glycoproteins present anti-
gens, processed to small peptides, to an antibody-
like molecule on the T cell receptor. An individ-
ual organism has only a small number of different
histocompatibility molecules (probably less than
a dozen), so that each histocompatibility gly-
coprotein must be able to "present" many,
possibly thousands, of different antigenic pep-
tides to thousands or more distinct T cell re-
ceptors throughout the immunological life of
the individual.
In the past year we have been able to visualize
the conformation of peptides bound to a histo-
compatibility glycoprotein, HIA-B27. The ex-
tended conformation of the peptide appears to be
specified by the HLA-binding site, so that the two
ends of the peptide are bound to specialized re-
gions at the two ends of the binding groove. A few
of the side chains of a 9-mer peptide interact with
pockets in the surface of the HLA molecule,
whose size and chemical composition vary from
allele to allele in the population. We also eluted,
sequenced, and identified 1 1 self-peptides from
HLA-B27. All 11 had arginine, a positively
charged amino acid, at peptide position 2, which
correlates with the x-ray crystallographic finding
that position 2 fits into a deep pocket with a nega-
tively charged glutamic acid at the bottom. (That
residue is polymorphic and changes the specific-
ity of that pocket in other alleles.)
We are also now able to reconstitute class I mol-
ecules from polypeptide chains produced in
Escherichia coli with single peptides and have
crystallized HLA-A2 with a series of peptide anti-
gens from influenza virus and HIV-l (human im-
munodeficiency virus type 1 ) . One complex dif-
fracts beyond 1 .5-A resolution when the crystal is
frozen at - 160°C. (All of the crystals in our labo-
ratory are now frozen to this temperature to pre-
serve crystallographic order and eliminate radia-
tion damage.)
In our studies of class II histocompatibility an-
tigens (a collaboration with Joan Gorga and Jack
Strominger), we now have three crystals under
study: human DRl, human DRl plus a superanti-
gen, and DRl expressed in insect cells and com-
plexed with a single influenza virus peptide. The
complex of a single peptide with a class II mole-
cule was generated by expressing a soluble class
II molecule in cells from insects, which lack an
immune system. Empty DRl molecules were pro-
duced that rapidly and stoichiometrically bound
peptide. The empty molecules were stabilized
against aggregation and sodium dodecyl sulfate-
induced denaturation by addition of peptide, ar-
guing that peptide binding is accompanied by a
conformational change.
Our laboratory is also studying how influenza
virus infects cells. About 10 years ago we deter-
mined the three-dimensional structure of the in-
fluenza virus hemagglutinin (HA), the viral gly-
coprotein responsible for binding the virus to
cells and for fusing the viral membrane to a cellu-
lar membrane to effect infectious entry. Recently
we determined the structure of a series of com-
plexes between the HA and derivatives of sialic
acid, the cellular receptor for influenza virus. We
have synthesized a number of these new ligands
and determined the crystal structure of the com-
plexes to confirm an atomic model for virus-cell
binding that we proposed. In the process a sec-
ond binding site has been located on the HA at an
interface between domains of the molecule,
which, although probably not physiological, may
offer opportunities for the design of a ligand to
stabilize the interface against the conformational
change required for the HA's membrane fusion
activity.
A number of other crystallographic and bio-
chemical studies are under way on influenza C
virus, on a low-pH fusion-active conformation of
the influenza HA, on trypanosome surface anti-
gens, and on the glycoprotein of HIV- 1 in com-
plex with its cellular receptor, CD4.
443
Studies of Blood Cell Formation
David A. Williams, M.D. — Assistant Investigator
Dr. Williams is also Associate Professor and Kipp Investigator of Pediatrics at Indiana University School of
Medicine, Indianapolis. It was here that he received his medical degree, after graduating in biology from
Indiana State University. His postdoctoral training includes a pediatric residency at Children's Hospital
Medical Center, Cincinnati; research fellowships with Richard Mulligan at the Massachusetts Institute of
Technology Center for Cancer Research and the Whitehead Institute for Biomedical Research; and clinical
fellowships in pediatric hematology /oncology at Harvard Medical School, the Children 's Hospital, and
'%t, Dana-Farber Cancer Institute, Boston. He was Assistant Professor of Pediatrics at Harvard Medical School
before assuming his present position.
THE goal of work in our laboratory is to under-
stand the relationship between supporting
cells in the bone marrow cavity and the develop-
ment of blood cells. The production of blood
cells from primitive "stem cells" in the marrow
is dependent on interactions between the latter
cells and certain supporting cells. Understanding
the process of blood cell formation is important
for treatments that utilize bone marrow trans-
plantation, including gene transfer methods to
correct genetic diseases affecting the blood-
forming system.
Studies of the Bone Marrow Environment
The production of enormous numbers of blood
cells in the bone marrow environment and their
delivery to the blood circulation constitute a
highly regulated and complex process. Either ex-
cessive or insufficient production of the primi-
tive stem cells or the more mature daughter cells
is associated with human diseases. Little is
known, however, about how this process is regu-
lated. Our approach to the problem has been to
simplify interactions between stem cells and sup-
porting cells by immortalizing single supporting
cells from the bone marrow and studying these
interactions in detail.
In collaboration with Vikrum Patel at North-
western University, we have defined a protein
and its receptor that appear to be important in
anchoring stem cells to supporting cells. This in-
teraction can be blocked in mice and in vitro
using antibodies to the receptor. The stem cells
that adhere to supporting cells (stromal cells) us-
ing this receptor appear to have a high capacity to
form new stem and daughter cells. The adherence
of stem cells to supporting cells is important in
the survival of stem cells both in vitro and in the
bone marrow cavity. A better understanding of
this interaction may lead to new methods for
growing blood cells and expanding the number
of stem cells in the laboratory.
An important protein in the growth of blood
cells was recently identified in our laboratory as
"Steel factor." This protein is defective in a
mouse mutant (Steel mouse) , which has a form of
aplastic anemia, as well as infertility and skin ab-
normalities. The protein is normally made in two
forms that differ with respect to how it is pre-
sented to blood stem cells in the bone marrow
environment. We are investigating the role of
these two different types of Steel factor
presentation.
Deniz Toksoz, a former HHMI associate, has re-
cently shown that presentation of Steel factor in a
localized fashion leads to survival of immature
human blood cells (termed progenitor cells)
longer in culture than when the presentation is
nonlocalized. This observation, in combination
with the identification of proteins involved in
blood cell adhesion (described above) , may have
important implications in developing a system
for growing bone marrow cells in vitro. Manas
Majumdar, an HHMI associate, is currently study-
ing the role of this protein on blood formation
during development of the mouse.
Studies on the role of the supporting cells in
the bone marrow have also led us to the identifi-
cation and production (by recombinant meth-
ods) of another protein involved in blood forma-
tion, termed IL-11 (interleukin- 11 ) . In recent
work utilizing IL-11, Xunxiang Du has demon-
strated that it has the capacity to help bone
marrow stem cells reestablish blood formation
after intravenous injection (i.e., bone marrow
transplantation) in mice. This protein affects the
growth of several different types of blood cells,
including platelets and neutrophils. We are in-
vestigating the use of this growth factor to pre-
vent some of the severe complications of chemo-
therapy, such as infection and bleeding. This
work is also supported by grants from the Na-
tional Institutes of Health.
Effects of Growth-regulating Proteins
During Blood Development in the
Early Embryo
In the developing mammalian embryo, blood
445
Studies of Blood Cell Formation
formation begins in the yolk sac. Little is known
about the interaction between supporting cells in
this environment and the earliest primitive stem
cells, a very few of which are thought to give rise
to all blood cells in the adult. Our laboratory has
derived cell lines from each part of the yolk sac,
and Merv Yoder, a visiting scientist, is character-
izing the interactions of supporting cells from the
yolk sac with blood stem cells derived from
mouse adult bone marrow, yolk sac, and the em-
bryonic liver.
In addition, Pamela Kooh, an HHMI associate,
is examining the effects of expression of various
growth proteins during fetal development. She is
utilizing embryonic stem cell technology to ex-
amine mouse fetal blood development in vivo
and in vitro.
Gene Transfer and Somatic Gene Therapy
A long-standing focus of research in our labora-
tory has been the use of viral vehicles, or vectors,
to introduce new genetic material into bone
marrow stem cells. These manipulated cells can
then be introduced into a recipient by bone
marrow transplantation, so that the daughter
blood cells contain the new gene. In the future
these methods, called somatic gene therapy, may
provide a means of curing severe genetic
diseases.
Our work has utilized information we have
gained about blood formation to improve the de-
livery of genes into mouse stem cells. Barry Lus-
key, an HHMI associate in our laboratory, has
shown that use of the Steel factor ensures that
100 percent of mice transplanted with manipu-
lated bone marrow can be made to express hu-
man protein. We hope to apply this knowledge
clinically. Children with adenosine deaminase
(ADA) deficiency, a genetic disease involving
bone marrow-derived cells, exhibit severe com-
bined immunodeficiency (SCID) . With a view to
treating this fatal condition, we are collaborating
with David Bodine and Arthur Nienhuis at the Na-
tional Institutes of Health in testing the ap-
proaches outlined above in monkeys, but so far
without success.
Tom Moritz, a postdoctoral fellow in our labo-
ratory, is collaborating with Ronald Hoffman at
Indiana University in investigating the transfer of
the ADA gene in human bone marrow cells that
have been highly enriched for stem cells. At this
point it is unclear whether the use of such stem
cell enrichment will improve the efficiency of
gene transfer into blood stem cells. Gene transfer
efficiency is a critical issue in the successful ex-
tension to humans of gene transfer methods devel-
oped in mice by this and other laboratories.
446
Growth Factor-stimulated Cell Proliferation
Lewis T. Williams, M.D., Ph.D. — Investigator
Dr. Williams is also Professor of Medicine at the University of California, San Francisco. He received his
undergraduate degree from Rice University and his M.D. and Ph.D. degrees from Duke University, where
he studied with Robert Lefkowitz. He then completed a clinical residency in internal medicine and
specialty training in cardiology at Massachusetts General Hospital, Boston. Before joining the faculty at
UCSF, he was Assistant Professor of Medicine at Harvard Medical School. Among his honors is the
Outstanding Young Investigator Award of the American Federation for Clinical Research.
POLYPEPTIDE groAvth factors regulate the pro-
liferation and migration of cells in the devel-
oping tissues of embryonic animals. The actions
of these factors appear to be recapitulated in
adults when damaged or senescent tissues are re-
paired. Our research group is investigating the
action of platelet-derived growth factor (PDGF) .
This potent growth factor for fibroblast and
smooth muscle cells is found in platelets and is
released at sites of tissue injury. PDGF is also pro-
duced by other tissues, including endothelial
cells that line the inner surfaces of blood vessels.
In this context PDGF is likely to play a major role
in stimulating the proliferation of smooth muscle
cells that constitute atherosclerotic plaques. Its
role in vascular proliferation appears to be espe-
cially prominent in the recurrent blockage of cor-
onary arteries that occurs after clinical interven-
tions such as angioplasty or atherectomy, which
are undertaken in an attempt to restore blood
flow through vessels narrowed by atherosclero-
sis. PDGF also plays a role in the growth of some
tumors. At least one monkey sarcoma is caused by
the aberrant production of PDGF, which stimu-
lates the tumor cells to grow in an uncontrolled
fashion.
Like other growth factors, PDGF acts on cells
by first binding to specific receptor sites located
on the cell surface. This interaction of PDGF with
its receptor is transmitted as a signal across the
cell membrane and triggers a series of complex
reactions inside the cell that culminate in DNA
synthesis and cell division. To study the mecha-
nism of signal transmission by the PDGF receptor,
we purified the receptor from mouse cells,
cloned the gene that encodes the mouse receptor
protein, expressed this receptor in cells that nor-
mally lack PDGF receptors, and demonstrated
that this expressed receptor mimics the actions of
native PDGF receptors and mediates all of the
known responses to PDGF.
The PDGF receptor is anchored at the surface
of the cell and is oriented so that approximately
half of the receptor, the PDGF-binding domain, is
located outside the cell, and the other half of the
receptor is located inside the cell. The receptor
appears to consist of several distant regions,
termed domains, that have distinct functions. Us-
ing the cloned gene for the receptor, we can pro-
duce individual domains of the receptor protein
and study the functions of these domains. For ex-
ample, we have produced receptor fragments
that contain the PDGF-binding domain but lack
the other portions of the molecule. To localize
more precisely the portion of the receptor that is
essential for binding PDGF, we are now prepar-
ing even smaller versions of the binding domain
by deleting portions of the molecule. When a
minimal domain for binding is defined, we will
study the three-dimensional structure of this sim-
plified molecule and use this information to de-
sign agents that can block the binding of PDGF to
its receptor. These agents should function as
blockers of the actions of PDGF and will facilitate
the study of the role of PDGF in atherosclerosis
and cancer.
One of the major problems in growth factor
research has been to determine how the portion
of the receptor inside the cell senses that the bind-
ing domain on the outside of the cell has inter-
acted with PDGF. We have recently found that
the transmission of the signal from the outside
domain to the inside of the cell involves two ma-
jor steps. First, when a receptor molecule binds
PDGF, the receptor pairs with another identical
receptor molecule to form a receptor dimer. Each
of the two receptor molecules in the dimeric
complex phosphorylates its partner, thereby
modifying the partner. The phosphorylation reac-
tion results in the addition of a phosphate group
to the partner and is accomplished by a region of
the receptor termed the kinase domain. We have
designed and produced mutant receptors that
have normal PDGF-binding domains but have de-
fective kinase domains. When these mutants are
introduced into cells that have normal receptors,
a dimer (heterodimer) is formed between the
normal and kinase-defective receptors. The nor-
mal and mutant receptors in the heterodimer
complex cannot phosphorylate each other and
cannot transmit the signals required to initiate
cell growth. These experiments prove that forma-
tion of a dimer consisting of two normal recep-
447
Growth Factor-stimulated Cell Proliferation
tors is required for proper functioning and signal
transduction of the receptor. By introducing mu-
tant receptors into specific tissues of animals we
hope to be able to block the function of the nor-
mal receptors and assess the role of PDGF in phys-
iological processes and in disease states.
When the dimerized receptor is phosphory-
lated, the second major step in signal transduc-
tion occurs. The phosphorylated receptor physi-
cally binds to signaling molecules that are inside
the cell. We have recently found that the interac-
tion between the receptor and signaling mole-
cules occurs at the phosphorylation sites on the
receptor. Using information about the structures
of these sites, we are designing ways to disrupt
the interaction between the receptors and the sig-
naling molecules. We are also using the receptors
as a tool for discovering previously unidentified
signaling molecules that mediate that prolifera-
tive response to PDGF. Recently we have identi-
fied one of these molecules, phosphatidylinositol
3-kinase (PI 3-kinase), that appears to play an im-
portant role in PDGF-stimulated proliferation.
Other investigators have found that PI 3-kinase is
also important in the cell transformation caused
by some viruses that cause cancer in animals. Us-
ing the receptor as a probe, we recently purified
the PI 3-kinase and cloned the gene for this signal-
ing molecule. We hope that by studying the inter-
action of the receptor and PI 3-kinase we can un-
derstand an important set of reactions that are
involved in regulating cell growth.
We recently have begun to study the fibroblast
grovvT:h factors (FGF). These factors appear to
play important roles in the earliest stages of em-
bryogenesis and in angiogenesis (the formation
of new blood vessels) . The development of new
vessels can be beneficial (e.g., in offsetting ath-
erosclerotic narrowing of blood vessels in the
heart) or deleterious (e.g., in the formation of
new vessels that supply nutrients to tumors or in
the vascular proliferation that occurs in the eyes
of diabetic patients) . We have identified some of
the FGF receptors and are now examining their
mechanisms of action. This work on FGF recep-
tors is supported by a grant from the National In-
stitutes of Health.
The long-range goal of these studies is to probe
the role of grov^h factors in normal embryonic
development, in tissue repair, and in prolifera-
tive diseases. Using the tools of molecular biol-
ogy, cell biology, and protein chemistry, we and
other research groups are identifying the factors,
receptors, and regulatory molecules involved in
these processes. Studies of the spatial and tem-
poral distribution of the growth factors and re-
ceptors in normal and diseased tissues will pro-
vide insight into the function of these molecules.
By understanding the molecular details of the
protein-protein interactions involved in growth
factor action, it may be possible to devise new
therapeutic strategies to treat proliferative
diseases.
448
Somatic Cell Gene Transfer
James M. Wilson, M.D., Ph.D. — Assistant Investigator
Dr. Wilson is also Associate Professor of Internal Medicine and Biological Chemistry at the University of
Michigan Medical School. He received his undergraduate degree in chemistry from Albion College and his
Ph.D. (biological chemistry) and M.D. degrees from the University of Michigan. He completed a residency
and clinical fellowship at Massachusetts General Hospital, Boston, and conducted postdoctoral research
with Richard Mulligan at the Whitehead Institute, Massachusetts Institute of Technology. Dr. Wilson is
investigating ways to treat genetic diseases by correcting the basic defects.
GENE replacement therapies are being consid-
ered for the treatment of a variety of ac-
quired and inherited diseases. These novel thera-
peutic modalities involve the transfer of genetic
material into somatic tissues of affected individ-
uals. The development of new therapies for car-
diopulmonary diseases, based on gene transfer
into hepatocytes and airway epithelial cells, is a
major focus of my laboratory.
Familial hypercholesterolemia (FH) is an auto-
somal dominant disorder in humans that is an ex-
cellent model for developing gene replacement
therapies of hyperlipidemic states. Patients with
the homozygous deficient form of FH have severe
hypercholesterolemia and suffer life-threatening
coronary artery disease in childhood that is refrac-
tory to conventional medical therapies. The mo-
lecular basis of this disorder is a systemic defi-
ciency in the receptor that degrades low-density
lipoprotein (LDL) , the primary carrier of choles-
terol in the blood. Hepatocytes are the appro-
priate target for gene transfer in this disease,
since the liver is the organ primarily responsible
for LDL metabolism and cholesterol excretion.
We have used an animal model for FH — the
Watanabe heritable hyperlipidemic (WHHL) rab-
bit— to develop different types of liver-directed
gene replacement therapies. One approach is sim-
ilar in concept to the well-described bone
marrow-directed gene therapies. This ex vivo
method involves isolating hepatocytes from a ge-
netically deficient animal, transferring functional
genetic material into the hepatocytes in vitro,
and returning the corrected cells to the affected
recipient. We have used recombinant retrovi-
ruses to transfer a functional gene for the human
LDL receptor stably into hepatocytes derived
from WHHL rabbits. Transplantation of these
cells into WHHL rabbits leads to substantial de-
creases in serum cholesterol for over four
months. A similar approach may be therapeutic in
patients with homozygous FH.
An alternative and less invasive approach is to
deliver a functional LDL receptor gene into the
hepatocytes in vivo. We have developed an in
vivo gene delivery system that is based on inter-
actions with hepatocyte-specific receptors. Using
this approach we can deliver reporter genes spe-
cifically to hepatocytes in vivo and obtain ex-
pression of the recombinant gene in liver for at
least four months. Administration of a gene
transfer substrate containing the LDL receptor
gene into the circulation of WHHL rabbits led to
significant reductions in the level of serum cho-
lesterol. Hepatic overexpression of LDL receptor
by gene transfer can potentially prevent the ath-
erosclerotic disease in FH and other hyperlipid-
emic states. This work is also supported by a grant
from the National Institutes of Health.
Cystic fibrosis (CF) is an inherited disease
characterized by abnormal salt and water trans-
port that leads to pathology within the pancreas
and lung. CF is the most common lethal congeni-
tal disease among Caucasians, with a prevalence
of 1 in 2,000 births. The primary defect in the
lung appears to be the production of thick abnor-
mal mucus that plugs the airways and leads to
infections. The recent discovery of the gene that
causes CF suggests a new therapeutic strategy in
which normal functioning CF genes are directly
introduced into pancreatic or respiratory cells of
affected patients.
As a first step in the development of a genetic
cure for CF, we have attempted to correct the
physiological abnormality in cells from a CF pa-
tient by introducing into them a CF gene that
encodes CFTR (cystic fibrosis transmembrane reg-
ulator), a normal functioning product. Replica-
tion-defective viruses were used to shuttle a nor-
mal CF gene into pancreatic cells from a patient
with CF. Prior to gene transfer, the cells mani-
fested the typical abnormalities associated with
this disease (i.e., decreased transport of salt
across the membrane). Following gene transfer,
the cells regained normal regulation of salt trans-
port. This demonstrates the feasibility of gene
therapy in CF at a cellular level.
Rational development of approaches for recon-
stituting CFTR expression in vivo requires a defi-
nition of endogenous CFTR expression in the
normal human airway as well as an understanding
of the biology of the epithelial cells that line the
449
Somatic Cell Gene Transfer
airway. Using antibody and RNA probes, we have
characterized the distribution of CFTR expres-
sion in the human bronchus. The airway epithe-
lial cells express low levels of CFTR. The predom-
inant site of expression, however, is in the
submucosal glands that are responsible for the
production and delivery of mucus to the airway
lumen. These findings suggest that effective gene
therapies may require genetic reconstitution in
both the epithelium and the submucosal glands.
This work is also supported by the Cystic Fibrosis
Foundation.
We have entered a new phase of the develop-
ment of CF gene therapies in which approaches
for delivering functional CF genes into airway
cells in vivo are being designed and tested. Di-
rect inhalation of the gene transfer substrates
could provide a noninvasive way of delivering
the genes to the appropriate cells. Several viral
and nonviral approaches are being considered.
450
Model for gene therapy of the cystic fibrosis airway: human bronchial
xenograft reconstituted with cells from a CF patient following retro-
viral infection with the ^-galactosidase reporter gene. Darkfield photo-
micrograph shows blue cells following a histochemical stain for
0-galactosidase.
Research and photograph by fohn Engelhardt in the laboratory of
fames Wilson.
Differentiation of human myeloid leukemia cells to eosinophils in the bone marrow of a mouse
with severe combined immunodeficiency disease (SCID).
From Sawyers, C.L., Gishizky, M.L., Quan, S., Golde, D.W., and Witte, O.N. 1992. Blood
79:2089-2098.
452
Normal and Abnormal Lymphocyte
Growth Regulation
Owen N. Witte, M.D. — Investigator
Dr. Witte is also Professor of Microbiology and Molecular Genetics at the University of California, Los
Angeles, holding the President's Chair in Developmental Immunology. He received his B.S. degree in
microbiology from Cornell University and his M.D. degree from Stanford University, where he trained with
Irving Weissman in the Medical Scientist Training Program. Dr. Witte completed postdoctoral training
with David Baltimore at the Massachusetts Institute of Technology before joining the UCLA faculty. His
honors include the Milken Family Medical Foundation Award in Basic Cancer Research and the Rosenthal
Foundation Award of the American Association for Cancer Research.
OUR ability to resist a wide range of infectious
agents depends on the normal function of
the immune system. The humoral portion of this
system is responsible for the production of spe-
cific antibody molecules from B lymphocytes.
Too low a growth rate can result in immunodefi-
ciency; too high a growth rate, in various types of
leukemia or lymphoma. Our laboratory has
concentrated on defining the growth control
mechanisms that regulate the production of B
lymphocytes.
The ABL Oncogene in Murine and
Human Leukemias
The ABL oncogene was first isolated as the ac-
tive genetic element of the Abelson murine leu-
kemia virus. This agent is capable of causing a
wide range of leukemias in mice. The biological
properties of the ABL gene product depend on its
activity as a tyrosine-specific kinase.
The human homologue of the ABL gene has
now been strongly implicated in the pathogene-
sis of a family of human leukemias that harbor a
specific cytogenetic abnormality. This is a chro-
mosome translocation that uses mRNA splicing to
join part of the BCR gene (of unknown function)
from chromosome 22 to part of the ABL gene
from chromosome 9, forming the so-called Phila-
delphia chromosome, or Ph 1 . The tyrosine kinase
activity of the chimeric BCR/ABL gene product
is evoked and strongly correlates with the trans-
formation activity of the protein.
Two different forms of BCR/ABL protein can
occur, depending on the precise position of the
chromosomal breakpoints. In human chronic my-
elogenous leukemia, a larger protein product
called P210 BCR/ABL is produced, and in the
case of Phi -positive acute lymphocytic leuke-
mia, a PI 85 BCR/ABL protein product is com-
monly found. We have undertaken to determine
the precise contribution of BCR sequences to the
tyrosine kinase activity of the ABL segment and to
the malignant potential of the gene product.
A variety of studies have now documented the
specific role of BCR in the activation of the ABL
tyrosine kinase. Site-directed mutagenesis has es-
tablished that the BCR segment is essential for
transformation by the chimeric oncogene and
acts through the tyrosine kinase domain of ABL.
We have recently described a new class of pro-
tein-protein interactions that regulate this activa-
tion. Within the first exon segment of BCR, there
are two strong SH2 (SRC homologous domain 2)
binding regions that can tightly bind to ABL. All
previous SH2 interactions have been mediated by
proteins containing phosphotyrosine. Interest-
ingly, the BCR protein requires phosphoserine
and phosphothreonine for strong binding, but
not phosphotyrosine. BCR represents a new class
of protein-protein interaction domains important
in intracellular signaling.
Further analysis of BCR has shown that it also
represents a new class of protein kinases. The first
exon region of BCR contains a serine/threonine
kinase activity that can be distinguished from the
traditional protein kinase family by several crite-
ria, including its containment of catalytic activity
within a single exon, its reactivity with nucleo-
tide analogues, and the role of essential cysteines
within the nucleotide-binding cleft. Work from
other laboratories shows that BCR also has a
GTPase-activating protein domain at its carboxyl
terminus that regulates the action of small gua-
nine nucleotide-binding proteins. BCR appears
to be a multifunctional protein. Its normal func-
tion remains to be determined.
The normal ABL oncogene products are ex-
pressed in many cell types, but their role in mam-
malian cell physiology is unknown. Gross struc-
tural changes can activate their oncogenic
potential. It has been difficult to identify more-
subtle mutations that might activate ABL, be-
cause the normal gene is toxic to most cell types
when highly expressed. The precise mechanism
of toxicity is not established, but probably relates
to a cell cycle-blocking effect.
Full-length cDNA copies of the cellular ABL
genes are cloned into a retroviral vector that has
been modified to allow amplification in an acute
transfection system. Retroviral particles are pro-
453
Normal and Abnormal Lymphocyte Growth Regulation
duced that can transmit the cellular ABL gene at
high efficiency to a wide variety of cell types.
Using this system, we have been able to select for
transformed clones that harbor new classes of ac-
tivating mutations in ABL.
One new class of mutants shows the loss of se-
quences downstream of the kinase domain. These
narturally selected deletion mutants and molecu-
lar constructs created to mimic their structure are
weak transforming alleles that require a comple-
menting oncogene to induce fibroblast cells to
grow in agar. Their most striking phenotypic
characteristic is the very low activation of the ty-
rosine kinase activity of ABL. This is very different
than all previous transformation alleles of ABL.
It has previously been shown that all forms of
the ABL oncogene can synergize for their biologi-
cal effects with members of the MYC gene family.
We have recently tested the hypothesis that MYC
is not only complementary to ABL transformation,
but essential. Using a series of dominant negative
mutants to inactivate the effective dosage of MYC
in both fibroblast and hematopoietic cells, we
have observed a dramatic decrease in the ability
of BCR/ABL and viral forms of ABL to transform.
The dominant negative mutants have no suppres-
sive effect on other types of oncogenes, including
MOS. Such essential combinations could be im-
portant considerations in devising strategies for
the treatment of certain cancers.
Regulation of Stem and Progenitor
Cell Growth
In collaboration with Naomi Rosenberg of
Tufts Medical School, we have developed a sys-
tem for retroviral infection of murine bone
marrow stem cells with the BCR/ABL oncogene
and reimplantation into syngeneic hosts. This
procedure leads to tumors with the characteris-
tics of human chronic myelogenous leukemia. In-
terestingly, animals infected with the PI 85 BCR/
ABL forms show more-aggressive tumors that
invade nonhematopoietic organs and show
shorter latency than animals infected with the
P210 form. This system should be valuable for
analyzing new therapies directed at the BCR/ABL
oncogene and for defining the grov^h regulation
of primitive hematopoietic stem cells.
To elucidate the mechanisms that lead to dereg-
ulated growth of stem cells, we have developed
an in vitro system to monitor the effects of BCR/
ABL on multipotential stem cells. Marrow
enriched for such cells was infected with P2 1 0
BCR/ABL expressing retrovirus and plated in agar
in very low concentrations of growth factors like
Steel factor and interleukin-3 (IL-3). These low
concentrations are incapable of stimulating col-
ony growth but could synergize with BCR/ABL.
A variety of colony types grew in this assay sys-
tem, including multipotential cells capable of
differentiating to form mast cells, macrophages,
granulocytes, and pre-B cells. Cell lines of these
different sublineages established from such mul-
tipotent colonies express a high level of the
BCR/ABL gene product but are only subtly trans-
formed in their growth factor requirements and
nonmalignant when transferred to syngeneic
animals.
Reliable test models for the growth of human
myeloid leukemias are limited to a small subset of
these leukemias that can be grown into continu-
ous cell lines in vitro. We have recently estab-
lished a reproducible procedure for the growth
of freshly explanted human acute myelogenous
leukemia and blast crisis specimens from chronic
myelogenous leukemia patients. Cells from the
bone marrow of such patients are implanted
under the kidney capsule of the severe combined
immunodeficient (SCID) mouse. At this site the
human cells grow and then migrate to the bone
marrow and peripheral blood in a pattern quite
similar to that of the original human disease.
One long-range goal of our group is to develop
effective in vitro culture techniques for the prop-
agation and enrichment of stem and progenitor
cells for different lineages. Previously we used
the growth stimulatory properties of the BCR/
ABL oncogene in concert with selected bone
marrow-adherent stromal lines to grow clonal
lines of B lymphoid progenitor cells that could
repopulate the B cell lineage of immunodefec-
tive (SCID) mice. By modifying the culture con-
ditions, we have now been able to cultivate such
progenitor cells without the need for co-stimula-
tion by the oncogene. These populations are very
effective in reconstituting the B cell lineage in
vivo.
We have used these progenitor cells to prepare
several cDNA libraries in order to search for new
members of the tyrosine kinase gene family and
other regulatory genes that may be important in B
cell development. A new member of the SRC fam-
ily of tyrosine kinases has been identified that is
specific to the B cell lineage.
454
Translational Regulation
Sandra L. Wolin, M.D., Ph.D. — Assistant Investigator
Dr. Wolin is also Assistant Professor of Cell Biology at Yale University School of Medicine. She received her
undergraduate degree in biochemistry from Princeton University and her M.D. and Ph.D. degrees from
Yale University, where she worked with Joan Steitz. Her postdoctoral research was done with Marc
Kirschner and Peter Walter at the University of California, San Francisco, where she was a fellow of the
Helen Hay Whitney Foundation and a Lucille P. Markey Scholar.
MY laboratory is particularly interested in the
mechanisms that regulate the translation of
messenger RNA (mRNA) into proteins. The infor-
mation in mRNA is translated by a large RNA-
protein complex, the ribosome. It has been
known for some time that the ribosomes do not
move along the mRNA at an even pace — that they
pause at discrete places along the way. Why ribo-
somes pause is not well understood but is
thought to be due either to rare codons or to sec-
ondary mRNA structures.
In certain mRNAs the pausing of ribosomes is
thought to cause them to slip and lose their place,
such that a different protein is now translated.
This slipping is known as ribosome frameshifting.
We have been investigating why ribosomes pause
during translation, and we would like to under-
stand how cells can use ribosome pausing to regu-
late the synthesis of particular proteins.
To address these questions, we use a method
that allows us to determine the distribution of
ribosomes on an mRNA with single-nucleotide
precision. In this way we can obtain a detailed
picture of the translation process. Using this as-
say, we have found that the ribosomes often
pause directly over the initiation codon of the
mRNA. This pausing by fully assembled ribo-
somes appears to represent a slow step in eukar-
yotic protein initiation that had not been previ-
ously detected.
By performing translation reactions in the pres-
ence of inhibitors that block distinct steps in poly-
peptide initiation and then examining the posi-
tion of the resulting paused ribosomes, we have
narrowed down this major slow step to one of two
points in the initiation pathway. We are continu-
ing to characterize this intermediate in transla-
tion initiation and are also investigating the possi-
bility that this slow step in protein synthesis may
be a point at which cells regulate translation of
particular mRNAs.
We are also interested in understanding how
the availability of particular tRNA molecules (in-
termediates in protein synthesis) affects ribo-
some pausing during elongation of the nascent
protein. To study this question, we have prepared
translation extracts in which the tRNA molecules
have been removed. By adding back different
tRNA populations, we can now manipulate the
levels of individual tRNAs. In this way we hope to
determine how different tRNA species contribute
to ribosome pausing and frameshifting.
We are also studying how the attachment of
ribosomes to the endoplasmic reticulum mem-
brane affects their movement along mRNAs en-
coding secreted proteins. The synthesis of these
proteins designed for export outside the cell be-
gins when ribosomes, free in the cytoplasm, initi-
ate translation on the mRNAs. A common feature
of these secretory proteins is the presence of a
signal peptide, usually an amino-terminal exten-
sion of 15-30 amino acids. A signal-recognition
particle (SRP), a small cytoplasmic ribonucleo-
protein, binds to the signal sequence emerging
from the ribosome and arrests elongation tran-
siently. This translational arrest is released after
the ribosome-SRP complex interacts with a spe-
cific component of the endoplasmic reticulum
membrane, the SRP receptor.
Concomitant with the resumption of protein
synthesis, translocation of the nascent polypep-
tide begins. All subsequent translation is carried
out by ribosomes that are attached to the endo-
plasmic reticulum membrane.
We have found that the point at which mem-
brane insertion of the nascent polypeptide begins
is distinct for different proteins, which may re-
flect differences in the size and structures of indi-
vidual signal peptides. It also appears that certain
features of mRNA structure that cause ribosomes
to pause during translation in solution do not al-
ways result in their pausing when attached to
microsomal membranes. Thus it appears likely
that mRNA secondary structure can be altered
by attachment through ribosomes to these
membranes.
We are currently determining whether ribo-
somes, following termination, are able to reini-
tiate translation on microsomal membranes. If so,
it would provide a molecular explanation for the
"circular polysomes" that have long been ob-
served attached to the endoplasmic reticulum.
455
Molecular Genetics and Studies Toward Gene
Therapy for Metabolic Disorders
Savio L. C. Woo, Ph.D. — Investigator
Dr. Woo is also Professor in the Department of Cell Biology and Institute for Molecular Genetics at Baylor
College of Medicine. He obtained his undergraduate degree in chemistry from Loyola College, Montreal,
and his Ph.D. degree in biochemistry from the University of Washington, where he worked with Earl Davie.
His postdoctoral research was done at the University of British Columbia, Vancouver, in the Division of
Neurological Sciences.
A major focus in my laboratory has been the
analysis of human metabolic disorders at the
molecular level. Phenylketonuria (PKU), the dis-
ease under investigation, is an inborn error in
amino acid metabolism that causes severe and
permanent mental retardation in untreated chil-
dren. The condition is caused by defects in the
liver enzyme phenylalanine hydroxylase (PAH)
and is transmitted from the parents to both boys
and girls. In the United States it affects 300-400
newborns a year, and 1 of every 50 individuals is
an asymptomatic carrier of the disease trait. A sec-
ond area under intense investigation is technol-
ogy development for the cure of these genetic
disorders by somatic gene therapy. The diseases
being studied here include PKU and hemophilia
B, which is transmitted from asymptomatic car-
rier mothers to their sons.
Prenatal Diagnosis for Phenylketonuria
Our laboratory has isolated the human PAH
gene by molecular cloning and used the cloned
gene to analyze cellular DNA of normal and af-
fected individuals. Extensive benign variations in
this gene were discovered and used to trace the
transmission of individual PAH genes from the
parents to the children in PKU families. A fetus
inheriting the same PAH genes as an affected sib-
ling will have PKU. This has led to a prenatal
diagnosis procedure for PKU in families with pre-
viously affected children, and the procedure has
been adopted in the United States and other west-
ern European countries.
Prognosis by Gene Analysis
There is a wide range of severity of clinical
symptoms among PKU patients, and their treat-
ment depends on measurement of the phenylala-
nine level in blood, which is often highly vari-
able. The development of an independent
method for the determination of prognosis is im-
portant for proper management. We have ana-
lyzed the mutations in the PAH gene of a large
number of PKU patients who have been closely
supervised for the past 20 years. The severity of
their clinical conditions is primarily dependent
on the inheritance of mutations in their PAH
genes that are either totally or partially defective.
In the future this correlation will allow physi-
cians to prescribe proper medical treatment after
simple gene analysis.
Population Dynamics
A number of these PKU genes have distinctive
distributions in the European continent. One is
very prevalent in eastern Europe, and its fre-
quency decreases in a gradual fashion from east to
west. These results suggested that the mutation
occurred in eastern Europe some time ago and
was then spread throughout the European conti-
nent by migration of people in prehistoric times.
Two other prevalent PKU genes are very frequent
in either northern or southern Europe but less
frequent in the neighboring countries. When sim-
ilar analysis was performed in Israel and China,
independent centers of major PKU mutations
were also discovered. These results strongly sug-
gest that multiple PKU mutations occurred inde-
pendently in various regions of the world and
then spread into neighboring areas in Europe and
Asia.
Somatic Gene Therapy
The other major goal of our laboratory is to ex-
plore the potential for somatic gene therapy of
genetic disorders. The PAH gene has been in-
serted into the genome of an incapacitated virus.
The recombinant viruses are able to transduce
mammalian cells and incorporate themselves
into the genome of the host cells, but they are no
longer able to produce new virus to continue the
infection process. The recombinant viruses were
used to transduce cultured rodent hepatoma cells
and normal liver cells, thereby conferring upon
them the ability to synthesize the corresponding
human enzyme.
These results have led to the development of
hepatocyte transplantation technologies in labo-
ratory animals. A variety of inert substances were
used as support for mouse liver cells prior to
transplantation, but the transplanted hepatocytes
457
Molecular Genetics and Studies Toward Gene Therapy
for Metabolic Disorders
lived for only a few weeks in the recipient ani-
mals. When the hepatocytes were returned to the
liver by direct injection into the portal vein or the
spleen of recipient mice, however, the cells mi-
grated to the liver and incorporated themselves
into the parenchyma. They not only lived for as
long as the mice did but also continued to func-
tion as liver cells in the transplanted animals. Us-
ing liver cells isolated from a mouse strain that is
deficient in hepatic PAH activity, we demon-
strated that enzymatic activity was reconstituted
in these cells after retroviral-mediated gene
transfer. The PKU mouse model will be critically
important to test the efficacy of our hepatic gene
transfer and hepatocyte transplantation protocols
for the correction of PKU in the future.
To develop a larger animal model for somatic
gene therapy, with technologies that may be di-
rectly applied to human patients in the future, we
selected a colony of hemophilic dogs. Methods
were developed to obtain a liver lobe from nor-
mal dogs by partial hepatectomy and to disperse
the hepatocytes into single cells in culture. In-
stead of a few million cells (which we obtained
from the mouse), several billion cells were ob-
tained from a single canine liver lobe. The iso-
lated hepatocytes were transplanted back into the
same animal by direct injection into the spleen.
Alternatively, a catheter was inserted into the
splenic vein with a subcutaneous port to permit
direct external injection of hepatocytes into the
port. It was observed that a billion hepatocytes
can be easily transplanted, and greater than half
of these cells migrate to the liver and survive in
the animal for a minimum of four months. Using a
recombinant retroviral vector containing the hu-
man coagulation factor IX gene to transduce nor-
mal canine hepatocytes, we also observed high
levels of human factor IX protein in the culture
media. With these technological developments,
we shall attempt to correct the genetic deficiency
in hemophilia B dogs by somatic gene therapy.
Prenatal diagnosis of PKU and disease progno-
sis by gene analysis were supported by a grant
from the National Institutes of Health. Somatic
gene therapy research on the PKU mouse and he-
mophilia B dog models is also supported in part
by a grant from the National Institutes of Health.
458
Paracrine Control of Blood Vessel Function
Role of the Endothelins
Masasbi Yanagisawa, M.D., Ph.D. — Associate Investigator
Dr. Yanagisawa is also Associate Professor of Molecular Genetics at the University of Texas Southwestern
Medical Center at Dallas. He received his M.D. and Ph.D. degrees from the University of Tsukuba, Japan,
where he worked with Tomoh Masaki. Before moving to Dallas, he was Assistant Professor of
Pharmacology at Kyoto University, Japan.
DISORDERS of blood vessels, including all
forms of heart attack and stroke, represent
the most frequent disease cause of death in devel-
oped countries. The inner surface of blood vessel
walls is lined with a thin monolayer of flat cells
called vascular endothelium. These cells cover a
total surface area of nearly 700 m^ throughout
the human body. The endothelium is also unique
in that it is the only tissue that has direct physical
contact with circulating blood under healthy
conditions. Nevertheless, until a little over 10
years ago, the endothelial cells were considered
to be merely a "bio-inert dialysis bag." In other
words, they just kept blood flowing smoothly
without unwanted clotting and allowed nutri-
tional components and metabolites to pass freely
between the blood and interstitial space.
Evidence accumulated recently, however, indi-
cates that the endothelium plays much more
complex roles in many different facets of physiol-
ogy and pathology. The endothelial cells can re-
spond to both chemical (circulating and local
hormones) and mechanical (local blood flow and
pressure) information carried in the circulating
blood.
In response to these stimuli, these cells trigger
remodeling and de novo formation of blood ves-
sels by secreting growth factors and migrating
into adjacent tissues. They also regulate leuko-
cyte infiltration and lymphocyte homing by ex-
pressing a variety of cell adhesion molecules.
They integrate blood coagulation, fibrinolysis,
and platelet function by producing various pro-
and anticoagulant factors. They control vascular
permeability by actively transcytosing and metab-
olizing plasma components. Finally, they regu-
late blood pressure and local blood flow by both
activating and inactivating circulating vasoactive
factors and, more importantly, by generating an
array of vasoactive molecules. Thus we now rec-
ognize the vascular endothelium as an active and
dynamic transducer that senses and interprets
blood-borne signals.
Central to these functions of the endothelium
is local communication of the endothelial cells
with vascular smooth muscle cells underneath.
As mentioned above, the endothelial cells send
out an array of physiological and pathological
signals toward the smooth muscle cells by means
of the endothelium-derived vasoactive factors.
Among those are prostacyclin and nitric oxide
(also called endothelin-derived relaxing factor),
both strong vasodilators.
This background led us in 1988 to identify en-
dothelin-1 (ET-1), a novel vasoconstrictor pro-
duced by vascular endothelial cells. ET-1 is a
small peptide consisting of 21 amino acid resi-
dues wound into a rigid structure by two sets of
intrachain disulfide bridges. It is the most potent
vasoactive molecule known, causing a strong and
extremely sustained constriction of all blood ves-
sels both in vitro and in vivo.
Our subsequent studies demonstrated in hu-
mans and other mammals three endothelin-
related genes that encode, besides ET-1, two ad-
ditional isopeptides of the endothelin family
called ET-2 and -3. Endothelin isopeptides are
expressed with distinct distribution patterns in
many mammalian tissues. While the endothelial
cells are the most abundant source of ET- 1 , one or
more of the three isopeptides are expressed
widely in other tissues such as brain, lung, and
kidney.
The production of endothelins is regulated in
both directions. In endothelial cells, it is up-
regulated by various chemical stimuli, including
classical vasoactive hormones (epinephrine, an-
giotensin II, and vasopressin) , products from co-
agulation/platelet activation (thrombin and
transforming growth factor-jS) , factors implicated
in septic shock (bacterial endotoxin, interleukin-
1, and tumor necrosis factor), oxidized low-
density lipoprotein, etc. ET-1 production is also
augmented by mechanical stimuli such as stretch
and fluid-dynamical shear stress. In contrast, cer-
tain vasodilators (nitric oxide and atrial natri-
uretic peptides) attenuate the ET-1 production
via increase of intracellular cGMP levels.
Like many other peptide hormones and neuro-
peptides, endothelins are processed from the
corresponding larger precursor proteins (pre-
pro-proteins). However, biologically active 21-
459
Paracrine Control of Blood Vessel Function: Role of the Endothelins
residue endothelins are produced via a formerly
unknown type of proteolytic processing. Biologi-
cally inactive intermediates of approximately 40
amino acids, called big endothelins, are first cut
out from the prepro-proteins. Their carboxyl-
terminal half is then sheared off to produce the
amino-terminal active peptides.
'This unusual endoproteolytic activation is cata-
lyzed by a novel membrane-bound metallopro-
tease called endothelin-converting enzyme. This
unique enzyme is sensitive to the metallopro-
tease inhibitor phosphoramidon and is distinct
from any other protease know^n. Since big ET-1 is
at least 100-200 times less active than mature
ET-1 in constricting vascular smooth muscle
strips, the converting enzyme is essential in pro-
ducing the potent vasoconstrictive agent. There-
fore the enzyme could also be an important target
for pharmacological intervention in the endothe-
lin system. If one could develop an inhibitor of
endothelin-converting enzyme, it could possibly
be used as a novel class of vasodilatory drug.
Apart from their potent and long-lasting vaso-
constrictor/pressor activities, endothelins pos-
sess a wide spectrum of nonvascular actions in
various tissues. Furthermore, two distinct sub-
types of endothelin receptors (called ET^ and
ETg) with different selectivities to the three iso-
peptides have been pharmacologically demon-
strated and molecularly cloned by us and other
researchers. In keeping with the wide spectrum
of endothelin action, the two receptors are ex-
pressed in a variety of vascular and nonvascular
tissues.
The observation that many endothelin-produc-
ing tissues also express one or more subtypes of
endothelin receptors suggests the importance of
the peptide family as locally acting mediators
(paracrine and/or autocrine) rather than circu-
lating agents. The idea is further supported by
several lines of evidence. For example, clearance
of endothelins from the circulation is extremely
rapid, with an initial half-life of less than 2-3
min; plasma concentration of immunoreactive
ET-1 is well below the threshold concentration
for pharmacological actions; and circulating en-
dothelins are capable of inducing the release of
vasodilator substances such as nitric oxide and
prostacyclin via the receptors on the endothelial
cells, thereby limiting their own vasoconstrictor
actions.
Increased plasma concentrations of FT- 1 have
been reported in patients with various disease
states, including vasospasm and hypertension,
where abnormal vasomotor function is impli-
cated. Moreover, in animal models of certain vas-
cular disorders, such as acute myocardial infarc-
tion, cerebral vasospasm, and acute ischemic
renal failure, treatment with antiendothelin
neutralizing antibody significantly ameliorates
the abnormal vasoconstriction seen in these
conditions.
It is hoped that further insight into the physio-
logical and pathobiological roles of this complex
system of peptidic mediators will be gained in
the near future with the development of endothe-
lin receptor antagonists, inhibitors of endothelin-
converting enzyme, and mice deficient in en-
dothelin/endothelin receptors. This may lead to
a new level of understanding of how cardiovascu-
lar homeostasis is maintained via local communi-
cation between cells of blood vessel walls in
health and disease.
460
Mechanism of Phototransduction in Retinal Rods
and Cones
King- Wat Yau, Ph.D. — Investigator
Dr. Yau is also Professor of Neuroscience at the Johns Hopkins University School of Medicine. He received
an A.B. degree in physics from Princeton University and a Ph.D. degree in neurobiology from Harvard
University. He did postdoctoral research at Stanford University with Denis Baylor and at Cambridge
University, England, with Alan Hodgkin. For six years thereafter, he was on the faculty at the Department
of Physiology and Biophysics of the University of Texas Medical Branch at Galveston. He has received the
Rank Prize in Optoelectronics from the Rank Prize Funds, England.
VISION begins in the rods and cones of the
retina, where light is absorbed and trans-
duced into a neural signal consisting of an elec-
trical hyperpolarization at the photoreceptor
membrane. This signal is relayed to second-order
neurons in the retina through a modulation of the
release of synaptic transmitter at the photorecep-
tor's terminal. In darkness the transmitter is re-
leased at a high rate, and in light the membrane
hyperpolarization reduces the release in a graded
fashion. This modulation of synaptic transmitter
release can lead to a hyperpolarizing or depolar-
izing response to light in a second-order neuron,
depending on the polarity of a given synapse.
The phototransduction process — the way the
hyperpolarizing response to light is generated in
the receptors — is as follows. In darkness an ionic
conductance in the plasma membrane of the re-
ceptor's outer segment (the part of the cell that
contains the visual pigment) is kept open by the
cyclic nucleotide guanosine 3':5'-cyclic mono-
phosphate (cGMP), letting both Na^ and Ca^"^
into the cell. This "dark" current depolarizes the
cell and causes the steady release of synaptic
transmitter described above.
Light activates the following reaction cascade:
light photoisomerization of visual pigment -»•
G protein activation -> cGMP phosphodiesterase
stimulation cGMP hydrolysis. As a result, the
cGMP level falls in the outer segment, causing
the ionic conductance to close and leading se-
quentially to membrane hyperpolarization and
the reduction of synaptic transmitter release. This
phototransduction scheme applies to both rods
and cones, with only quantitative differences be-
tween the two types of receptors.
One consequence of the conductance closure
in the light is that the Ca^^ influx stops. The re-
sulting imbalance between influx and efflux
leads to a decrease in the intracellular free Ca^*
concentration. This Ca^"*^ decrease reduces a tonic
inhibition exerted by Ca^+ on the cGMP-synthe-
sizing enzyme guanylate cyclase and causes an
increase in the synthesis of cGMP in the light.
Thus Ca^"^ mediates a negative feedback control
on the light-activated cGMP hydrolysis, and this
feedback should be a candidate mechanism un-
derlying the well-known phenomenon of back-
ground light adaptation in photoreceptors. In-
deed, we have found that this adaptation
essentially disappears upon removing the feed-
back experimentally by eliminating the Ca^"^ in-
flux and efflux.
The effect of Ca^"^ on rod guanylate cyclase has
been studied by others in biochemical experi-
ments in vitro. This effect is now known to in-
volve recoverin, a novel Ca^^-binding protein
that activates guanylate cyclase at low Ca^^ con-
centrations but loses this ability when Ca^"^ is
bound to it. One drawback of the biochemical
experiments is that unphysiological ionic con-
centrations (e.g., very high Mg^"^) were used to
measure the cyclase activity. We have now stud-
ied this Ca^^ modulation of the cyclase in more
physiological conditions, by recording the
cGMP-activated current from an isolated, open-
ended rod outer segment with a suction pipette
while dialyzing the interior of the outer segment
with different Ca'^^ concentrations.
The Ca^^ effect on the guanylate cyclase could
be derived from the magnitude of the cGMP-
activated current. We have found that the cyclase
activity is very sensitive to the free Ca^^ concen-
tration, with its maximum activity being approxi-
mately halved at 100 nM Ca^^. This is similar to
biochemical measurements. The cyclase depen-
dence on Ca^^ shows a Hill coefficient of approxi-
mately 1.5, which is lower than the value of
around 4 in biochemical measurements. The cy-
clase activity becomes relatively insensitive to
Ca^^ at a GTP concentration of greater than 1 mM,
suggesting that the effect of Ca^"^, acting through
recoverin, may primarily be to reduce the affinity
of the enzyme for its substrate. Further experi-
ments on this problem are in progress.
Another problem we are working on is a molec-
ular characterization of the cGMP-activated con-
ductance mediating phototransduction. The con-
ductance now appears to belong to a family of
cyclic nucleotide-gated channels that includes
the cGMP-gated channels in retinal rods and
cones, as well as the cGMP/cAMP-gated channel
461
Mechanism of PhototransdiicHon in Retinal Rods and Cones
mediating olfactory transduction in olfactory
cilia. These channels show both similarities and
differences. In particular, the rod and olfactory
channels have similar ion-permeation character-
istics, but the olfactory channel shows a 30-fold
higher affinity for cyclic nucleotides. To under-
stand the molecular determinant for this differ-
ence, we, in collaboration with Randall Reed
(HHMI, Johns Hopkins University), have con-
structed chimeras between the rod and the olfac-
tory channels and tested for modifiications in
function. The difference in amino acid residues at
the putative cyclic nucleotide-binding site on
the two channel molecules has little to do with
their difference in affinity for cyclic nucleotides.
Other regions of the molecule must therefore be
involved, possibly through steric or allosteric in-
teractions in the folded molecule. Identification
of these regions is still in progress.
Separately, several independent clones have
been isolated from a human retinal cDNA library
based on structural homology to the published
sequence of the bovine rod channel. One of these
has been successfully expressed in human 293
cells and found to have properties identical to the
bovine rod channel. Thus this clone most proba-
bly encodes for the human rod channel. The
other clones have yet to be expressed function-
ally in a cell line. Based on the presence of a puta-
tive cyclic nucleotide-binding domain as well as
other features in their nucleotide sequences,
however, they probably also encode for cyclic
nucleotide-gated channels. Antipeptide antibod-
ies are being made to identify the locations of the
encoded proteins in the retina. Continuing ef-
forts are also being made to express these clones
functionally. Cloning at the genomic level is be-
ing carried out to provide further clarification.
Part of the above work is supported by a grant
from the National Institutes of Health.
462
Molecular Mechanisms of Ion Channel Function
Gary Yellen, Ph.D. — Assistant Investigator
Dr. Yellen is also Assistant Professor of Neuroscience and Biophysics at the Johns Hopkins University
School of Medicine. He received his undergraduate degree in biochemical sciences from Harvard College
and his Ph.D. degree in physiology from Yale University, where he studied with Charles Stevens. Dr. Yellen
did his postdoctoral research on ion channel physiology as a Life Sciences Research Foundation
postdoctoral fellow at Brandeis University, where he worked with Christopher Miller.
ALL electrical signaling in the nervous system
is controlled by ion channels, a class of
membrane proteins that form pores through the
membrane. Charged ions such as sodium, potas-
sium, and calcium pass through ion channels and
carry an electrical current. The channels them-
selves are regulated, so that the pores are only
open when the proper chemical or electrical sig-
nal is present, and only certain ions can pass
through a particular kind of channel. By under-
standing how channels open and close and how
they are regulated, we define the repertoire of
molecular changes used by neurons when they
signal, sense, and learn.
Ion channels, like other membrane proteins,
have resisted standard biochemical and structural
analysis. Their structure has only recently begun
to be elucidated by a combination of protein
chemistry and molecular biology. On the other
hand, we have detailed knowledge of the func-
tioning of ion channels. Because each ion chan-
nel catalyzes the transport of millions of ions per
second, we can measure electrically the current
carried by just a single-channel protein molecule.
This technique of single-channel recording has
allowed us to make a detailed model for the con-
formational changes between open and closed
states induced by chemical ligands and changes
in voltage, but we still have no knowledge of the
protein structures that underlie these conforma-
tional changes.
My laboratory uses a combination of high-
resolution functional analysis (by single-channel
recording) and direct manipulation of the struc-
ture of the channel protein. Site-directed muta-
genesis allows us to modify any amino acid in a
protein for which we have the cloned genetic ma-
terial. Rather than modifying the protein directly,
we change the DNA sequence and then inject the
modified messenger RNA into immature frog eggs
(oocytes) , which manufacture the modified pro-
tein. This method allows us to test specific the-
ories about which parts of the channel protein are
important for specific functional features.
We have applied this combination of strategies
to voltage-activated potassium channels, which
participate in electrical signaling. By systematic
mutagenesis, we have identified the region of the
potassium channel protein that lines the pore
through which ions cross the membrane. We
found specific amino acid residues in the protein
sequence that control the sensitivity of the chan-
nel to tetraethylammonium, an organic ion that
can block current through the channel. Natural
variation of one of these amino acids explains the
differences in drug sensitivity between potassium
channels in different organs or species. Amino
acids in this region of the protein can also alFect
the rate at which ions are conducted through the
pore. These discoveries put us in a position to
discover the basic mechanisms of ion selectivity
and channel gating at the level of individual
amino acids.
We have also used recording from single potas-
sium channels to demonstrate that one of the
mechanisms by which these channels open and
close is a simple occlusion of the pore by part of
the channel protein. Earlier work established that
the pore could be directly blocked or occluded
by internal organic ions; we have established that
the natural gating occurs by a very similar mecha-
nism involving a tethered blocking particle. The
most direct demonstration of this is that potas-
sium ions passing through the pore from one side
can clear the tethered blocking particle from the
opposite side.
Further work in progress to determine the
structural basis for potassium channel gating in-
cludes introducing chemically reactive cysteine
residues at specific locations in the protein se-
quence. Channel proteins with cysteines inserted
at critical locations should show a specific
change in function when treated with reagents
that modify the cysteine side chain. The reactivity
of these side chains will depend on both their
location in the channel protein and the specific
conformational state of the protein at the time of
reaction.
We have also used the site-directed strategy to
study acetylcholine-activated cation channels,
which convert neurochemical signals into elec-
trical signals at synapses. We have changed amino
465
Molecular Mechanisms of Ion Channel Function
acids in the region of the protein that binds ace-
tylcholine and identified specific residues that
play a critical role in binding and signal transduc-
tion by acetylcholine. These studies are teaching
us more about the molecular basis of drug rec-
ognition and of signal transduction in this
protein.
Dr. Yellen is now Associate Professor of Neu-
robiology at Harvard Medical School.
464
Drosophila Behavior and Neuromuscular
Development
Michael W. Young, Ph.D. — Investigator
Dr. Young is also Professor of Genetics at the Rockefeller University. He received his B.A. degree in biology
from the University of Texas, Austin. Staying on to work at the university with Burke Judd, he earned his
Ph.D degree for genetic and cytological studies o/Drosophila chromosome structure. Dr. Young did
postdoctoral work in biochemistry at Stanford University Medical School with David Hogness. He moved
to Rockefeller as a fellow of the Andre and Bella Meyer Foundation.
A biological clock, composed of a few thou-
sand cells within the mammalian brain, con-
trols timing of daily behaviors such as sleep with
an accuracy of minutes. Chemical and electrical
rhythms have been detected in these mammalian
pacemaker tissues. Still, little is known about the
underlying biochemistry used to calculate time.
The genes and proteins central to biological
timing are beginning to be recovered and charac-
terized in a simpler model organism, the fruit fly
Drosophila. The best-studied gene in the Dro-
sophila clock system has been named per {pe-
riod^. Several mutant forms of the gene have
been recovered that affect the pace of the insect's
clock and certain aspects of cell physiology.
In the per^ mutant, circadian locomotor activ-
ity rhythms have a long period of 30 rather than
24 hours. For the mutant per^, daily cycles have a
shortened, 1 9-hour period. Mutants with no daily
rhythms are designated per^ . Corresponding
changes in cycle time are found for a high-
frequency rhythm — a courtship song produced
in males by pulses of wing beating: an 80-second
song (instead of the normal 55 seconds) for per\
40 seconds for per^, and song arrhythmicity for
per^. Also, the mutations change the period of a
daily oscillation in per transcription, which may
be important for establishing rhythmic behavior.
Finally, for at least some tissues, the mutants ap-
pear to alter conductance of specialized channels
(gap junctions) between cells.
The molecular changes associated with the
mutations have been established. The per^ mu-
tant cannot express a full-length protein. On the
other hand, per^ and per^ make per proteins, but
these are altered by substitution of a single, dif-
ferent amino acid. Comparable changes in cycle
time can also be effected by altering the amount
of per protein the fly produces. For example, mi-
croinjection of a gene that underproduces the per
protein 20-fold induces 40-hour daily rhythms.
From these results it has been suggested that the
per^ and per^ mutations generate, respectively,
hyper- and hypoactive proteins.
In a recent effort to understand how changes in
protein structure can affect per activity, genes
carrying new mutations, produced in vitro, were
reintroduced into the fly by microinjection. It
was found that mutations changing the structure
of a certain segment of about 20 amino acids pre-
dominantly confer short-period (per'-like) rhyth-
micity. Apparently the mutations identify a re-
gion of the per protein that regulates the activity
level in normal flies.
A variety of experiments have demonstrated
that the per protein acts in the nervous system to
control daily and circaminuten rhythms. We have
become interested in tracking the development
of the fly's clock in an effort to determine when it
begins to run, whether it requires signals from
outside the organism to start, and where the first
cells expressing per arise and develop. We have
learned that Drosophila reared in constant envi-
ronmental conditions spontaneously start their
clocks only hours after formation of the embryo.
Evidence of a running clock is first seen after
completion of the embryonic nervous system and
just following cessation of high levels of per ex-
pression in certain neural cells.
Until recently, only the per locus was known to
be essential for production of biological rhythms
in the fruit fly. Genetic screens for rhythm muta-
tions have led to the discovery of additional, in-
dispensable genes. Of special interest is a new
mutation found on the second chromosome {per
maps to the X chromosome). The new mutation,
like per^, renders flies arrhythmic, and in molecu-
lar studies appears to block the circadian rhythm
in per transcription. Thus the newly recognized
gene may be required for per to function.
Development of Skin, Muscle, and Nerve
In the embryo the nervous system and skin are
derived from a common set of cells, the ecto-
derm. Each of these cells must choose a fate, and
in certain Drosophila mutants the choice goes
awry. For one set of mutants known as "neuro-
genic," the capacity to choose skin has been lost
and only nerve is formed. We believe the mutants
have lost the ability to form a set of signals that act
early in development, so we are using the muta-
465
Drosophila Behavior and Neuromuscular Development
tions to mark, isolate, and characterize the prod-
ucts of these genes and putative developmental
signals. Neurogenic mutations have been recov-
ered at seven genetic loci: Notch, big brain,
mastermind, neuralized. Delta, almondex, and
Enhancer of split.
Notch has been characterized more completely
than any other gene in the neurogenic group. It
produces a very large, 2 ,700-amino acid protein,
predominantly composed of an uninterrupted
array of 36 copies of a hormone-like molecule, a
relative of epidermal growth factor. The Notch
protein spans the cellular membrane, with the 36
hormone copies exposed to its neighbors. We
surmise that the neighbors read that signal and, in
return, tell the A'^orc^-bearing cell to come up
with the correct allocations of skin and nerve.
Since the entire string of hormone copies is teth-
ered to the cell's surface, signaling between cells
must be intimate. Only cells that can touch each
other could communicate through such a
protein.
From work with temperature-sensitive muta-
tions, we know that Notch proteins are used to
instruct development throughout embryonic,
larval, and pupal life. Mutations altering the
structures of individual hormone-like elements
of the Notch protein have been examined to de-
termine the role each plays in early and late devel-
opment. Of the 36 hormone repeats, no two are
identical, and changes in different hormone ele-
ments produce different developmental abnor-
malities. This must mean that alternate parts of
the Notch hormone string are read as develop-
ment unfolds. In part these specificities could al-
low a cell to talk to changing neighbors from the
time of cell birth to differentiation into adult
tissue.
What do signals from these genes tell a cell to
do? For several years gene action at Notch, Delta,
big brain, almondex, neuralized, mastermind,
and Enhancer of split have been assumed to stim-
ulate an undifferentiated ectodermal cell to de-
velop as an epidermal cell. The genes are thought
to provide a series of epidermalizing signals dur-
ing cell differentiation, with loss of function gen-
erating a nerve cell. New work in our laboratory
shows this simple picture to be inaccurate. Notch
proteins have now been found in cells giving rise
to embryonic muscle. In Notch mutants, strong
effects on muscle development are seen, with in-
creased numbers of some muscle cell types gen-
erated, probably at the expense of others. Thus
parallel changes in muscle, skin, and nerve devel-
opment take place in Notch mutants. Of most sig-
nificance, comparable effects on muscle develop-
ment are seen with mutations of Delta, big brain,
mastermind, almondex. Enhancer of split, and
neuralized.
We have learned three things from these stud-
ies: 1 ) the developmental fates of many cell types
are switched in neurogenic mutants; 2) the genes
must provide differentiation signals that cells
composing different germ layers can read, with
no apparent overlap in the final developmental
fates; and 3) the genes defined by the neurogenic
mutations probably work together to form a sin-
gle developmental pathway, which generates a
common differentiation signal in all cells af-
fected in the mutants, since any developmental
anomaly caused by loss of one gene in the group
predicts a comparable developmental change
upon loss of any other gene in the series.
466
los/jun
NGF
r
Fos/Jun
binding site
TH
inactii^e
Fos and Jun
protein
levels low
Fos/Jun
binding site
TH
active
Fos/Jun
synthesis
induced
by NGF
lra?/jun?
Fos/Jun
binding site
TH
repressed
Fra/Jun
synthesis
induced
Model for a mechanism of control over a gene that may play a role in Parkinson's disease. The
polypeptide hormone NGF (nerve growth factor) binds to receptors on the surface of nerve cells
( neurons) and induces expression of the gene encoding tyrosine hydroxylase (TH). This enzyme
carries out a critical step in the synthesis of the neurotransmitter dopamine, whose release by
neurons transmits signals to neighboring cells. Thus NGF may indirectly control the neuron's
signaling capacity. In the model, NGF induces synthesis of members of the Fos family of proteins,
which form complexes with fun family proteins. These complexes bind to a regulatory element of
the TH gene and stimulate TH expression. Later, proteins related to Fos — Fra's — are expressed
and repress TH gene activity. A deficiency in dopamine production can lead to Parkinson's dis-
ease and other neurological disorders.
Research of Elena Gizang- Ginsberg and Edward Ziff.
468
Control of Transcription by Transmembrane
Signals
Edward B. Ziff, Ph.D — Investigator
Dr. Ziff is also Professor of Biochemistry at New York University Medical Center. He received his B.A.
degree in chemistry from Columbia University and the Ph.D degree in biochemistry from Princeton
University. He then studied DNA structure with Fred Sanger at the MRC Laboratory of Molecular Biology
in Cambridge. He later conducted research on DNA tumor viruses at the Imperial Cancer Research Fund
Laboratory, London, and in the Department of Molecular Cell Biology at the Rockefeller University. He
later joined New York University Medical Center and began the study of cellular mechanisms that control
proliferation and differentiation.
THE remarkable process of development re-
quires that a fertilized egg undergo many
rounds of cell division with accompanying differ-
entiation to generate specialized cell types that
ultimately form the mature organism. For devel-
opment to proceed normally, just the right num-
ber and types of cells must be available at each
stage. It follows that any cell's decision to divide
and/or differentiate must be carefully regulated.
These two critical processes, proliferation and
differentiation, are also controlled in the adult —
for example, during the maintenance of tissues
and in wound healing.
The decision of a cell to divide or to express a
specialized function is often determined by sig-
nals from its environment. Prominent among the
agents that convey such signals are the growth
factors, which are polypeptides synthesized and
secreted by cells. Our laboratory studies the mo-
lecular mechanisms by which growth factors and
other transmembrane signaling agents exert their
effects on cell proliferation and differentiation.
The transforming genes of DNA tumor viruses,
such as adenovirus, often modify these programs
and are therefore useful for dissecting the growth
regulatory pathways.
Growth factors transmit signals to cells by bind-
ing to specific receptor proteins, which span the
cell's plasma membrane and induce second mes-
senger signals in the cytoplasm. The latter signals
have a multitude of targets, some in the cyto-
plasm and some in the nucleus. Although individ-
ual growth factors may exert profound changes
on cells, some effects, such as the induction of
cell proliferation, may require the combined ac-
tions of more than one growth factor. When the
signal pathways are inappropriately activated,
cells may lose control of growth and form a tu-
mor. When the pathways are blocked, essential
cell types may degenerate and die. It follows that
errors in signaling can result in diverse diseases.
In the case of the nervous system, transmem-
brane signals induced by neurotransmitters regu-
late the properties of neurons. These small mole-
cules, released by neurons at synapses, bind to
receptors on postsynaptic target cells. Neuro-
transmitter stimulation of target neurons is a criti-
cal step in the rapid transmission of nerve im-
pulses, but it can also more slowly regulate the
activities of specific genes, allowing the nervous
system to modify its properties in response to its
environment. Such modification may underlie
the processes of neural adaptation and memory.
Our laboratory has identified a group of imme-
diate early-response genes that are rapidly in-
duced by growth factor stimulation and appear to
be primary targets in the nucleus for the growth
factor-induced signals. Our work focuses on the
c-fos gene, which is very rapidly induced by a
wide range of transmembrane signals. This gene
encodes a protein, c-Fos, which is a member of a
family of transcription factors that bind to spe-
cific sites in the regulatory regions of other genes
and thereby control their activity. In this manner
c-Fos acts as an intermediary for the conversion of
short-term transmembrane signals into longer-
term changes in the cell. These studies are sup-
ported by the American Cancer Society.
We are particularly concerned with the role of
c-fos in programs of neuronal differentiation in-
duced by nerve growth factor (NGF). In vivo,
NGF is required for the differentiation and main-
tenance of peripheral neurons. Expression of c-
fos appears to be a first step in the activation of a
multistage gene expression program induced by
NGF that can culminate in cell division or the
induction of terminally differentiated functions.
In our model neuronal cells, we have shown
that neurotransmitters as well as growth factors
can also induce the expression of c-Fos. It is ap-
parent from this and other studies that c-Fos has a
critical role in the adult nervous system, not just
in neural development.
We have shown that c-Fos induced by NGF can
cooperate with a second protein factor, c-Jun, to
induce the gene for tyrosine hydroxylase (TH),
an enzyme that catalyzes a critical step in the pro-
duction of neurotransmitters in the catechol-
amine family. The combined c-Fos and c-Jun form
a heterodimer that binds to a TH gene regulatory
469
Control of Transcription by Transmembrane Signals
element and induces expression. This participa-
tion of c-Fos in TH control may allow a neuron to
coordinate the production of catecholamine neu-
rotransmitters with the activity of the neuron
during its function in the nervous system. Indeed,
stimuli provided by light or smell may use this
pathway to control neuronal activity.
The c-Fos protein belongs to a family of pro-
teins including FosB and Fral and Fra2, all of
which may bind to c-Jun or to members of the Jun
protein family. The resulting complexes in turn
may all bind to the same DNA element. The dif-
ferent c-Fos family members differ in their pat-
terns of expression following cell stimulation,
as well as in their structures outside the DNA-
binding domain. This suggests that they may regu-
late gene activity differentially when complexed
with a DNA element such as the TH gene regula-
tory element.
Recent studies in our laboratory indicate that
the TH gene is repressed by a mechanism in
which the c-Fos protein, an activator, is replaced
by a different c-Fos family member, which serves
as a repressor. Indeed, other Fos family members,
including FosB, become the predominant species
as c-Fos levels dwindle and TH transcription is
shut off.
We have gone on to show that NGF induces
other genes as well. One of these encodes periph-
erin, a neuron-specific intermediate filament
protein that is present in the axons of peripheral
neurons as a component of the neuronal cytoskel-
eton. Our studies of the developing rat nervous
system indicate that peripherin expression coin-
cides with the final steps of neuronal maturation
and acquisition of function.
The mechanisms that control peripherin ex-
pression appear to be quite distinct from those
that control c-Fos or TH. We detect no binding
site for the Fos-Jun complex in the peripherin
gene. Instead, a negative regulatory element ap-
pears to release an inhibitory factor, thus activat-
ing the gene.
Study of these mechanisms may give a clue to
how cells permanently exit from the cell cycle
and induce the expression of genes that they em-
ploy after losing the capacity to proliferate. One
event that may block exit from the cell cycle is
the loss of control of expression of c-myc, an-
other growth factor-induced gene. It encodes a
protein, c-Myc, that is distantly related to c-Fos
and has specialized, but poorly understood, func-
tion in inducing cell proliferation. It is expressed
at abnormally high levels in many tumors.
We have identified a DNA nucleotide sequence
to which c-Myc can bind and a protein partner of
c-Myc called Myn that can stimulate its DNA bind-
ing. Our studies of c-Myc and Myn and their com-
plexes with DNA indicate they are controlled at
many levels, including modification of the ma-
ture proteins by phosphorylation. Our studies
seek to reveal the role of c-Myc in normal cell
proliferation and in tumorigenesis. These studies
are supported by the National Institutes of
Health.
We find that c-Myc expression is elevated in the
naturally occurring childhood brain tumor me-
dulloblastoma. The tumor cells have been
blocked to differentiation and have proliferated
abnormally. We also find that the transforming
gene of adenovirus. El a, like c-myc, can block
the differentiation of our model neurons, mimick-
ing the state of the tumor.
We are introducing various sorts of mutants of
the c-Myc protein into cells to disrupt pathways
stimulated by the normal, wild-type c-Myc. These
experiments may reveal other protein factors that
cooperate with c-Myc in controlling cell prolifer-
ation and differentiation. We seek to understand
how cells coordinate their functional maturation
with a loss of ability to proliferate as they un-
dergo the final stages of differentiation.
470
Cell-Cell Interactions Determine Cell Fate in the
Drosophila Retina
S. Lawrence Zipursky, Ph.D. — Associate Investigator
Dr. Zipursky is also Associate Professor of Biological Chemistry at the University of California School of
Medicine, Los Angeles. He received his Ph.D. degree from Albert Einstein College of Medicine, where he
studied mechanisms ofDNA replication in bacteria in the laboratory of ferard Hurwitz. He moved to the
California Institute of Technology to pursue postdoctoral studies in Drosophila neurogenetics with
Seymour Benzer. He has been on the faculty of UCLA for seven years.
WE are interested in two questions in devel-
opmental biology. First, How are specific
cell fates established through cell-cell interac-
tions? And second, What are the mechanisms un-
derlying the specificity of neuronal connectivity?
To address these issues, we have focused our stud-
ies on the development of the Drosophila visual
system. I will briefly describe progress we have
made in understanding the cellular and molecu-
lar mechanisms underlying the specification of a
unique cell fate in the developing retina. This
work is supported in part by a grant from the Na-
tional Institutes of Health. We are using similar
genetic and molecular approaches toward under-
standing how different classes of photoreceptor
neurons identify their unique postsynaptic tar-
gets in the developing brain.
Development of the Drosophila Retina
The Drosophila retina has a near-crystalline
structure of some 800 identical units called om-
matidia. Each ommatidium contains a group of
cells organized in a stereotyped fashion, with
eight photoreceptor neurons, designated R1-R8,
forming its core, surrounded by accessory cells
that produce screening pigments and the lens.
Genetic studies in the 1970s established that
there were no strict cell lineage relationships be-
tween the cells in the fly's eye. Observations have
supported the view that the developmental mech-
anisms regulating the acquisition of cell identity
are dependent not on a cell's ancestry but rather
on its interactions with other developing cells.
The R cells are the first to develop within the
ommatidial unit. Their stereotyped and sequen-
tial pattern of diff'erentiation suggested to early
workers that cell fates were established as a con-
sequence of a cascade of inductive interactions.
Cell fates were proposed to be a result of unique
inductive cues provided by differentiating neigh-
boring cells. The first cell to differentiate is R8,
followed by the cells surrounding it. The last cell
to differentiate is R7. It contacts the differentiat-
ing R8, Rl, and R6 cells as well as a number of
other unpatterned and undifferentiated cells. It
was proposed that the unique R7 cell fate was
induced by specific signals requiring direct cell
contacts between the R7 precursor and the Rl,
R6, and R8 cells.
An Inductive Event Specifying
R7 Development
The first step toward a molecular description
of the developmental mechanisms regulating R7
development was the discovery of the sevenless
{sev) mutation. In flies carrying this mutation,
the R7 cell fails to assume its normal fate and
becomes a nonneuronal lens-secreting cone cell.
The sev gene was shown to encode a receptor
tyrosine kinase, which is expressed in the R7 pre-
cursor cell and many other cells of the develop-
ing eye. The molecular structure of the sev pro-
tein, its expression pattern, and its genetic
requirement in the R7 precursor cell led to the
proposal that it is a receptor for an inductive cue
specifying an R7 cell fate.
Several years ago we identified a mutation in
another gene, bride of sevenless (boss), which
resulted in a phenotype identical to sev. Genetic
studies indicated that boss is required in R8, not
for its own development but for that of the R7
cell. This raised the intriguing notion that the
boss protein is an inductive ligand to which the
sev receptor binds. Molecular analysis of the boss
protein revealed that it is an integral membrane
protein with a large extracellular domain, multi-
ple transmembrane segments, and a short cyto-
plasmic tail. Using antibodies, we showed that
boss is specifically expressed in the R8 neuron.
Three observations argue that the boss protein
is a ligand for the sev receptor. First, mixtures of
sev and boss-expressing tissue culture cells bind
specifically to one another. Second, membranes
containing the boss protein rapidly and specifi-
cally activate the sev tyrosine kinase activity. And
finally, boss was shown to be transferred from the
surface of the R8 cell to an organelle in R7 re-
ferred to as a multivesicular body. This internal-
ization is strictly dependent upon the presence of
the sev protein on the surface of the R7 cell.
Earlier workers had demonstrated that the sev
471
Cell-Cell Interactions Determine Cell Fate in the Drosophila Retina
protein is expressed on many cells within an om-
matidium in addition to the R7 precursor cell.
Nevertheless, only one cell assumes an R7 cell
fate. Four of these additional cells contact the
boss-expressing R8 cell, and the others are one or
more cell diameters away. We have shown that
multiple mechanisms limit the induction to only
one cell, through studies in which we examined
R7 development in various mutant backgrounds
and in response to ectopically expressed boss
protein. These studies revealed that the cells that
do not contact the R8 cells are competent to re-
spond to the inductive cue and do so if presented
with a surface-bound form of boss. Conversely,
the sev-expressing cells that normally contact R8
are restricted from responding to the cue as a re-
sult of a very early developmental event that
commits them to alternative fates.
We believe that similar molecular and develop-
mental strategies involving the carefully con-
trolled expression of surface-bound ligands and
competing alternative developmental pathways
will be important in establishing the remarkable
cellular diversity and organization of more-
complex nervous systems, such as the vertebrate
brain.
472
Molecular Genetics of Sensory Transduction
Charles S. Zuker, Ph.D. — Associate Investigator
Dr. Zuker is also Associate Professor of Biology and of Neurosciences at the University of California School
of Medicine, San Diego. He received his Ph.D. degree from the Massachusetts Institute of Technology for
studies with Harvey Lodish. He carried out postdoctoral research with Gerald Rubin in the Department of
Biochemistry at the University of California, Berkeley, before joining the Department of Biology at UCSD.
Dr. Zuker is currently a Pew Fellow in the Biomedical Sciences and a Fellow of the March of Dimes
Foundation.
AN understanding of signal transduction is es-
sential to elucidating the cellular and molec-
ular basis of information processing in biological
systems. The primary event in the processing of
visual stimuli is phototransduction, the conver-
sion of light energy into a change in the ionic
permeabilities of the photoreceptor cell mem-
brane. The aim of our research is to clarify mecha-
nisms used for signal transduction in the visual
system, using a combined molecular, genetic,
and physiological approach. The study of this pro-
cess in Drosophila applies powerful molecular
genetic techniques to identify novel transduc-
tion molecules and to examine their function in
vivo, in their normal cellular and organismal
environment.
Experimental Strategy
Over the past few years my colleagues and I
have been working on the isolation and character-
ization of genes important for photoreceptor cell
function. Most recently we have focused on genes
encoding proteins involved in the regulation of
the visual transduction cascade. We have identi-
fied a protein kinase C (PKC) that is expressed
exclusively in the Drosopbilavisual system. Anal-
ysis of the light response from mutants defective
in this PKC showed it to be required for the deac-
tivation and rapid desensitization of the visual
cascade. The availability of a PKC mutant in Dro
sophila provides the basis for genetic and bio-
chemical studies to identify biologically relevant
substrates and regulators of this enzyme.
Rhodopsin, like other G protein-coupled re-
ceptors, is phosphorylated by a specific kinase
upon activation (rhodopsin kinase). This reac-
tion is thought to be involved in the termination
of rhodopsin's active state. Phosphorylation, how-
ever, is not sufficient for receptor inactivation,
since the phosphorylated form can still activate
the G protein. Complete inactivation has been
shown {in vitro) to require the stoichiometric
interaction of rhodopsin with a protein called
arrestin, or S antigen. This protein has also been
implicated in a number of autoimmune retinal
disorders in mammals. We have isolated the
genes encoding two photoreceptor cell-specific
arrestin molecules in Drosophila and have gen-
erated mutants of these genes. Our results should
help us understand the molecular basis of G pro-
tein-coupled receptor regulation and of relevant
abnormalities in the human visual and nervous
systems.
Mechanotransduction
We have recently begun to study mechano-
transduction, the process by which specialized
sensory cells convert mechanical stimuli — for
instance, sound, touch, gravity, or movement —
into electrical (neuronal) signals. In contrast to
phototransduction, nothing is known about the
molecular basis of mechanosensitivity. Our aim is
to identify genes and proteins involved in mech-
anotransduction by isolating mutations that affect
mechanosensory behavior.
When Drosophila larvae are touched gently,
they contract and retreat. We have developed a
screen for mutant larvae defective in this behav-
ior and have isolated several mutant lines. Muta-
tions that specifically affect the working of the
sensory organs are of greatest interest. Mapping
the mutations — in prelude to isolating and clon-
ing the affected genes — is in progress. In addi-
tion, we have developed a molecular approach to
isolate genes expressed specifically in mechano-
sensory organs. We are using developmental mu-
tants that overproduce or lack mechanosensory
bristles to isolate such genes by subtractive
hybridization.
475
J
i
i
I
i
I
i
1
Investigators by Location
ALABAMA
Birmingham
University of Alabama at Birmingham
and associated hospitals
Cooper, Max D., M.D.
CALIFORNIA
Berkeley
University of California, Berkeley,
and associated hospitals
Goodman, Corey S., Ph.D.
Rubin, Gerald M., Ph.D.
Schekman, Randy W., Ph.D.
Tjian, Robert, Ph.D.
Los Angeles
University of California, Los Angeles, and
associated hospitals
Kaback, H. Ronald, M.D.
Simpson, Larry, Ph.D.
Smale, Stephen T., Ph.D.
Witte, Owen N., M.D.
Zipursky, S. Lawrence, Ph.D.
University of Southern California
and associated hospitals
Lai, Michael M.-C, M.D., Ph.D.
Palo Alto
Stanford University and the
Stanford University Hospital
Aldrich, Richard W., Ph.D.
Barsh, Gregory S., M.D., Ph.D.
Brown, Patrick O., M.D., Ph.D.
Crabtree, Gerald R., M.D.
Davis, Mark M., Ph.D.
Francke, Uta, M.D.
Goodnow, Christopher C, B.V.Sc, Ph.D.
Kobilka, Brian K., M.D.
Nusse, Roel, Ph.D.
Scheller, Richard H., Ph.D.
Schoolnik, Gary K., M.D.
Weissman, Irving L., M.D.
Pasadena
California Institute of Technology
and associated hospitals
Anderson, David J., Ph.D.
Bjorkman, Pamela J., Ph.D.
Sternberg, Paul W., Ph.D.
San Diego
The Salk Institute for Biological Studies
Evans, Ronald M., Ph.D.
Sejnowski, TerrenceJ., Ph.D.
Stevens, Charles F., M.D., Ph.D.
University of California, San Diego,
and the UCSD Medical Center
Bevilacqua, Michael P., M.D., Ph.D.
Emr, Scott D., Ph.D.
Rosenfeld, Michael G., M.D.
Tsien, Roger Y., Ph.D.
Zuker, Charles S., Ph.D.
San Francisco
University of California, San Francisco,
and associated hospitals
Agard, David A., Ph.D.
Ganem, Donald E., M.D.
Gitschier, Jane M., Ph.D.
Grosschedl, Rudolf, Ph.D.
Jan, Lily Y., Ph.D.
Jan, Yuh Nung, Ph.D.
Kan, Yuet Wai, M.D., D.Sc.
Littman, Dan R., M.D., Ph.D.
Payan, Donald G., M.D.
Peterlin, B. Matija, M.D.
Reichardt, Louis F., Ph.D.
Sedat, John W , Ph.D.
Weiss, Arthur, M.D., Ph.D.
Williams, Lewis T., M.D., Ph.D.
COLORADO
Boulder
University of Colorado at Boulder and the
University's Health Sciences Center
Cech, Thomas R., Ph.D.
Kirkegaard, Karia A., Ph.D.
Denver
National Jewish Center for Immunology
and Respiratory Medicine
Kappler, John W., Ph.D.
Marrack, Philippa, Ph.D.
University of Colorado Health Sciences Center
and associated hospitals
Mailer, James L., Ph.D.
CONNECTICUT
New Haven
Yale University and associated hospitals
Artavanis-Tsakonas, Spyridon, Ph.D.
Bottomly, H. Kim, Ph.D.
Briinger, Axel T., Ph.D.
Cresswell, Peter, Ph.D.
475
Investigators by Location
De Camilli, Pietro, M.D.
Flavell, Richard A., Ph.D.
Fox, Roben O., Ph.D.
Ghosh, Sankar, Ph.D.
Horwich, Arthur L., M.D.
Jahn, Reinhard, Ph.D.
Janeway, Charles A , Jr., M.D.
Lerner, Michael R., M.D., Ph.D.
Reeders, Stephen T., M.D.
Schatz, David G , Ph.D.
Sigler, Paul B., M.D., Ph.D.
Steitz, Joan A., Ph.D.
Steitz, Thomas A., Ph.D.
Tanabe, Tsutomu, Ph.D.
Wolin, Sandra L., M.D., Ph.D.
GEORGIA
Atlanta
Emory University School of Medicine
and associated hospitals
Warren, Stephen T., Ph.D.
ILLINOIS
Chicago
The University of Chicago and
The University of Chicago Hospitals
Bell, Graeme I., Ph.D.
Fuchs, Elaine, Ph.D.
Laimins, Laimonis A., Ph.D.
Lindquist, Susan L., Ph.D.
Singh, Harinder, Ph.D.
Steiner, Donald F., M.D.
Sukhatme, Vikas P., M.D., Ph.D.
Evanston
Northwestern University and associated hospitals
Lamb, Robert A., Ph.D.
Donelson, John E., Ph.D.
Welsh, Michael J., M.D.
MARYLAND
Baltimore
The Carnegie Institution of Washington
and The Johns Hopkins Hospital
McKnight, Steven Lanier, Ph.D.
Spradling, Allan C, Ph.D.
The Johns Hopkins University
and Hospital
Beachy, Philip A., Ph.D.
Corden, Jeffry L , Ph.D.
Craig, Nancy L., Ph.D.
Desiderio, Stephen V., M.D., Ph.D.
Huganir, Richard L., Ph.D.
Nathans, Daniel, M.D.
Nathans, Jeremy, M.D., Ph.D.
Reed, Randall R., Ph.D.
Valle, David L., M.D.
Yau, King-Wai, Ph.D.
Yellen, Gary, Ph.D.
MASSACHUSETTS
Boston
Brigham and Women's Hospital
Chin, William W., M.D.
Cunningham, James M., M.D.
Maas, Richard L., M.D., Ph.D.
The Children 's Hospital
Alt, Frederick W., Ph.D.
Kunkel, Louis M., Ph.D.
Nadal-Ginard, Bernardo, M.D.
Orkin, Stuart H., M.D.
Ph.D.
INDIANA
Bloomington
Indiana University and associated hospitals
Kaufman, Thomas C, Ph.D.
Indianapolis
Indiana University School of Medicine
and associated hospitals
Williams, David A., M.D.
IOWA
Iowa City
University of Iowa and associated hospitals
Harvard Medical School and
Brigham and Women 's Hospital
Church, George M., Ph.D.
Duyk, Geofifrey M., M.D., Ph.D.
Leder, Philip, M.D.
Perrimon, Norbert, Ph.D.
Seidman, Jonathan G., Ph.D.
Massachusetts General Hospital
Alexander-Bridges, Maria C, M.D.,
Corey, David P., Ph.D.
Habener, Joel F., M.D.
Ph.D.
Campbell, Kevin P., Ph.D.
476
Investigators by Location
Tufts University School of Medicine
and associated hospitals
Isberg, Ralph R., Ph.D.
Cambridge
Harvard College, Arts and Sciences,
and The Children 's Hospital
Harrison, Stephen C, Ph.D.
Wiley, Don C, Ph.D.
Massachusetts Institute of Technology
and associated hospitals
Horvitz, H. Robert, Ph.D.
Hynes, Richard O., Ph.D.
Kim, Peter S., Ph.D.
Lehmann, Ruth, Ph.D.
Pabo, Carl O., Ph.D.
Page, David C, M.D.
Steller, Hermann, Ph.D.
Tonegawa, Susumu, Ph.D.
Waltham
Brandeis University and associated hospitals
Miller, Christopher, Ph.D.
Rosbash, Michael, Ph.D.
Worcester
University of Massachusetts
and associated hospitals
Davis, Roger J., Ph.D.
MICfflGAN
Ann Arbor
University of Michigan and associated hospitals
Bonadio, Jeffrey F., M.D.
Collins, Francis S., M.D., Ph.D.
Feinberg, Andrev^^ P., M.D., M.P.H.
Ginsburg, David, M.D.
Kurnit, David M., M.D., Ph.D.
Leiden, Jeffrey M., M.D., Ph.D.
Lowe, John B., M.D.
Nabel, GaryJ., M.D., Ph.D.
Thompson, Craig B., M.D.
Wilson, James M., M.D., Ph.D.
MISSOURI
St. Louis
Washington University and associated hospitals
Atkinson, John P., M.D.
Chaplin, David D., M.D., Ph.D.
Holers, V. Michael, M.D.
Korsmeyer, Stanley J., M.D.
Loh, Dennis Y.-D., M.D.
Olson, Maynard V., Ph.D.
Pike, Linda J., Ph.D.
Sadler, J. Evan, M.D., Ph.D.
Thomas, Matthevi^ L., Ph.D.
NEW JERSEY
Princeton
Princeton University and associated medical centers
Shenk, Thomas E., Ph.D.
Tilghman, Shirley M., Ph.D.
NEW YORK
Bronx
Albert Einstein College of Medicine
of Yeshiva University and
associated hospitals
Bloom, Barry R., Ph.D.
Jacobs, William R., Jr., Ph.D.
Cold Spring Harbor
Cold Spring Harbor Laboratory
and associated hospitals
Beach, David H., Ph.D.
New York
Columbia University and associated hospitals
Axel, Richard, M.D.
Hendrickson, Wayne A., Ph.D.
Jessell, Thomas M., Ph.D.
Kandel, Eric R., M.D.
Siegelbaum, Steven A., Ph.D.
Struhl, Gary, Ph.D.
Cornell University Medical College
O'Donnell, Michael E., Ph.D.
Memorial Sloan-Kettering Cancer Center
Massague, Joan, Ph.D.
New York University (Medical Center and
Washington Square ) and associated hospitals
Movshon, J. Anthony, Ph.D.
Ziff, Edward B , Ph.D.
The Rockefeller University and
Rockefeller University Hospital
Blobel, Gunter, M.D., Ph.D.
Burley, Stephen K., Ph.D.
477
Investigators by Location
Desplan, Claude, Ph.D.
Friedman, Jefifrey M., M.D., Ph.D.
Geliebter, Jan, Ph.D.
Heintz, Nathaniel, Ph.D.
Kuriyan, John, Ph.D.
Nussenzweig, Michel C, M.D., Ph.D.
Sakmar, Thomas P., M.D.
Young, Michael W., Ph.D.
Stony Brook
State University of New York at Stony Brook and
University Hospital at Stony Brook
Adams, Paul R., Ph.D.
NORTH CAROLINA
Durham
Duke University, including Duke University
Medical Center
Bennett, G. Vann, M.D., Ph.D.
Blackshear, Perry J., M.D., D.Phil.
Caron, Marc G., Ph.D.
Cullen, Bryan R., Ph.D.
Davis, Laura I., Ph.D.
Lefkowitz, Robert J., M.D.
Nevins, Joseph R., Ph.D.
Parker, Keith L., M.D., Ph.D.
OKLAHOMA
Oklahoma City
Oklahoma Medical Research Foundation
and associated hospitals
Esmon, Charles T., Ph.D.
OREGON
Eugene
University of Oregon and associated hospitals
Matthews, Brian W., Ph.D.
PENNSYLVANIA
Philadelphia
University of Pennsylvania
and associated hospitals
Brugge, Joan S., Ph.D.
Dreyfuss, Gideon, Ph.D.
Kadesch, Thomas R., Ph.D.
Liebhaber, Stephen A., M.D.
Malim, Michael H., Ph.D.
Nussbaum, Robert L., M.D.
Taub, Rebecca A., M.D.
TENNESSEE
Memphis
St. Jude Children 's Research Hospital
Sherr, Charles J., M.D., Ph.D.
Nashville
Vanderbilt University, including Vanderbilt
University Hospital
Exton, John H., M.D., Ph.D.
TEXAS
Dallas
University of Texas Southwestern Medical Center
at Dallas and associated hospitals
Beutler, Bruce A., M.D.
Deisenhofer, Johann, Ph.D.
Fischer Lindahl, Kirsten, Ph.D.
Garbers, David L., Ph.D.
Gething, Mary-Jane H., Ph.D.
Sprang, Stephen R., Ph.D.
Siidhof, Thomas C, M.D.
Yanagisawa, Masashi, M.D., Ph.D.
Houston
Baylor College of Medicine and associated hospitals
Beaudet, Arthur L., M.D.
Bellen, Hugo J., D.V.M., Ph.D.
Belmont, John W., M.D., Ph.D.
Caskey, C. Thomas, M.D.
Cohen, Stephen M., Ph.D.
Ledley, Fred D., M.D.
Overbeek, Paul A , Ph.D.
Quiocho, Florante A., Ph.D.
Soriano, Philippe M., Ph.D., D.Sc.
Woo, Savio L. C, Ph.D.
Rice University and associated hospitals
Gomer, Richard H , Ph.D.
UTAH
Salt Lake City
University of Utah, including University of Utah
Medical Center
Capecchi, Mario R., Ph.D.
Gesteland, Raymond F., Ph.D.
Lalouel, Jean-Marc, M.D., D.Sc.
Sakonju, Shigeru, Ph.D.
Thummel, Carl S., Ph.D.
White, Raymond L., Ph.D.
478
Investigators by Location
WASfflNGTON
Seattle
Fred Hutchinson Cancer Research Center
Henikofr, Steven, Ph.D.
Weintraub, Harold M., M.D., Ph.D.
University of Washington
and associated hospitals
Bevan, Michael J., Ph.D.
Glomset, John A., M.D.
Hurley, James B., Ph.D.
Palmiter, Richard D , Ph.D.
Perlmutter, Roger M., M.D., Ph.D.
WISCONSIN
Madison
University of Wisconsin
and associated hospitals
Carroll, Sean B., Ph.D.
479
International Research Scholars
In recognition of the contributions of scientists outside of the United States to advances in biomedi-
cal science, the Institute initiated in 1991 its International Research Scholars Program. This is a
limited experimental effort that provides research support through five-year grants to promising
scientists working in areas of fundamental biomedical research related to the Institute's ongoing
medical research programs.
The investigators invited to compete for the initial grant awards were located in Canada and Mexico.
From those invited to compete, 24 were selected (14 in Canada, 10 in Mexico). Grants also were
awarded to the Mexican Academia de la Investigacion Cientifica and to the United States National
Academy of Sciences for joint activities over four years to promote the exchange of scientific informa-
tion and encourage cooperation between the scientific communities in each country, particularly in
the life sciences.
The next phase of the international program will focus on biomedical scientists in the United
Kingdom, Australia, and New Zealand. The emphasis remains on the support of outstanding scientists
whose research careers are still developing, rather than those in the later phases of a distinguished
career. Grant awards for this second group will be announced late in 1992.
Canada
Mexico
Bernstein, Alan, Ph.D.
Finlay, B. Brett, Ph.D.
Greenblatt, Jack, Ph.D.
Grinstein, Sergio, Ph.D
Gros, Philippe, Ph.D.
Arias, Carlos F., Ph.D.
Calva, Edmundo, Ph.D.
Cota, Gabriel, Ph.D.
Darszon, Alberto, Ph.D.
Guarneros Pena, Gabriel, Ph.D.
Herrera-Estrella, Luis R., Ph.D.
Lizardi, Paul M., Ph.D.
Orozco, Esther, Ph.D.
Possani, Lourival Domingos, Ph.D.
Romo, Ranulfo, M.D., Ph.D.
Joyner, Alexandra L., Ph.D.
McGhee, James D., Ph.D.
Mosmann, Tim R.., Ph.D.
Pawson, Tony, Ph.D.
Read, Randy J., Ph.D.
Rossant, Janet, Ph.D.
Roy, Jean-Pierre, M.D.
Snutch, Terry P., Ph.D
Tsui, Lap-Chee, Ph.D.
481
Molecular Biology and Epidemiology
for Control of Rotavirus Diarrhea
Carlos F. Arias, Ph.D. — International Research Scholar
Dr. Arias is Investigador Titular B, Department of Molecular Biology, at the Biotechnology Institute,
National Autonomous University of Mexico, Cuernavaca. He received his undergraduate degree in
biochemistry and his M.S. and Ph.D. degrees in biomedical research from the National University of
Mexico, Mexico City. He recently conducted sabbatical research with James Strauss at the California
Institute of Technology.
ACUTE, infectious diarrhea is the commonest
Lcause of morbidity and mortality among
young children living in developing countries,
accounting for as many as one billion illnesses
and between four and five million deaths an-
nually. Rotaviruses are the leading cause of se-
vere diarrheal disease in children under three
years of age, and it is estimated that an effective
vaccine would save about 800,000 children's
lives per year.
Because rotavirus plays such an important role
in severe infantile gastroenteritis, and because
even advanced levels of hygiene seem unable to
control the spread of rotavirus infections, there
has been considerable interest in developing ef-
fective vaccination strategies.
Our laboratory is interested in studying how
rotaviruses attach and enter their host cell and
how they replicate there to produce the viral
progeny. Also among our interests are the host
immune response to rotavirus infection and the
epidemiology of these viruses in Mexico.
As in other infectious agents, proteins located
at the surface of the rotavirus particles are in-
volved in the early interactions (attachment and
penetration) of the virions with the host cell.
These proteins contain antigenic determinants
that represent major immunological targets. The
structural and functional characterization of the
viral surface proteins, as well as the study of im-
mune response in the infected host, should con-
tribute to the development of rationally designed
therapeutic agents and new prevention strate-
gies. Moreover, the success of a vaccine may de-
pend on knowledge of the epidemiology of the
pathogen in the particular geographic area where
the vaccine is to be used.
The surface of the rotavirus contains two pro-
teins, VP4 and VP7. VP4 forms spikes that extend
from the viral surface. This protein has been asso-
ciated with a variety of viral functions, including
the determination of virulence in vivo and the
ability to agglutinate red blood cells (hemagglu-
tination). VP4 is also important in the penetra-
tion of the virion into the cell. On the other hand.
the protein responsible for the initial attachment
of rotavirus to its host cell remains controversial;
both VP4 and VP7 have been proposed.
The attachment of animal rotaviruses to cells in
culture is mediated by compounds containing
sialic acid, since treatment of cells with sialidases
greatly reduces the binding of virus particles to
the cell surface. In addition, the hemagglutinat-
ing activity of rotaviruses and their binding to
cultured epithelial cells can also be inhibited by
incubation of the virus with sialoglycoproteins
such as glycophorin.
We have isolated rotavirus mutants whose bind-
ing is no longer inhibited by treatment of cells
with sialidases or incubation with glycophorin.
The detailed analysis of these mutants should
give us information about the protein (s) in-
volved in the attachment of the virions to the cell.
Furthermore, we are constructing deletion mu-
tants and chimeras between VP4 genes of hemag-
glutinating and nonhemagglutinating rotavi-
ruses, to define further the protein domains
involved in the binding of the virus to epithelial
and red blood cells.
After the virus's initial attachment to the cell
surface, the next step in its infection cycle is to
enter the cell. The entry of the virus particle can
be augmented by treating it with trypsin, and
probably is trypsin dependent. Experiments in
progress have shown that this proteolytic treat-
ment induces three specific cleavages of VP4. We
are interested in identifying the cleavage (s) di-
rectly associated with enhancement of viral infec-
tivity and in learning about the mechanism
through which the cleaved VP4 protein mediates
the virion's penetration of the cell.
Both surface proteins have also been character-
ized immunologically. It has been found that anti-
bodies to either protein neutralize the virus in
vitro and are capable of passively protecting ani-
mals from rotavirus challenge. In addition, oral
infection with live rotavirus stimulates protec-
tive immunity, which can be mediated by VP4
and/or VP7. These observations make the two
proteins attractive candidates for recombinant
subunit vaccines.
483
Molecular Biology and Epidemiology for Control of Rotavirus Diarrhea
Rotavirus infections are limited to the gut, and
it is believed that a vaccine will have to be given
orally to stimulate a local (mucosal) immune re-
sponse in order to induce protection. An attrac-
tive approach to this, alternative to the use of live
attenuated virus, is the use of live oral vaccines
containing nonpathogenic strains of enteric bac-
teria carrying the genes that code for the protec-
tive rotavirus antigens. We are currently con-
structing expression vectors to direct the
synthesis of the surface proteins of rotavirus in
the enteric bacteria Salmonella and Lactobacil-
lus, and will test the potential of these recombi-
nant strains to induce a protective immune re-
sponse against rotavirus infection.
Finally, our laboratory is also interested in the
study of some aspects of the epidemiology of ro-
tavirus. Six different serotypes of human rotavir-
uses have been identified, four of which appear
to account for the majority of isolates. We have
found that the four major serotypes circulate in
Mexico. We are currently studying the molecular
determinants of virulence of rotavirus and inves-
tigating how the frequency of the different sero-
types changes over time and in relation to
virulence.
In situ localization o/c-kit and Steel ex-
pression in adult mouse cerebellum. The
protein products of the proto- oncogene
c-kit and its ligand, the protein product
of the Steel gene, interact to activate Kit
tyrosine kinase activity vital to the func-
tion of melanocytes, blood cells, and
germ cells. The abbreviations identify
the molecular layer (mol), granular
layer (gran), and Purkinje cells (Pur).
From Motro, B., van der Kooy, D.,
Rossant, f., Reith, A., and Bernstein, A.
1991. Development 115:1207-1221.
484
Molecular Genetics of Normal and Leukemic
Hematopoiesis
Alan Bernstein, Ph.D. — International Research Scholar
Dr. Bernstein is Head of the Division of Molecular and Developmental Biology and Associate Director of
the Samuel Lunenfeld Research Institute at Mount Sinai Hospital, Toronto. He received his Ph.D. degree
in medical biophysics from the University of Toronto and trained as a postdoctoral fellow at the Imperial
Cancer Research Fund Laboratories, London. Before moving to Mount Sinai he was Senior Scientist in
the Division of Biological Research at the Ontario Cancer Institute. He is the first incumbent of the Anne
Tanenbaum Chair in Molecular and Developmental Biology and was recently elected a Fellow of the Royal
Society of Canada.
MY laboratory is interested in the molecular
mechanisms that govern the orderly produc-
tion of hematopoietic cells in the adult mammal.
It is clear from earlier work involving the genera-
tion of unique clonal markers (either visible cyto-
genetic markers or the chromosomal integration
sites of retroviral vectors) that hematopoiesis in-
volves the proliferation and differentiation of
pluripotent stem cells. These cells have the devel-
opmental and proliferative capacity to produce
the millions of myeloid and lymphoid cells that
are continuously required in adult life. Thus the
hematopoietic system has served as an attractive
and important experimental model of develop-
mental processes. Defects in hematopoiesis can
also lead to disease, including anemias, leuke-
mias, and such genetic disorders as sickle cell
anemia and thalassemia, making the understand-
ing of this system important to both medicine and
biology.
We are taking two main approaches to eluci-
date the mechanisms that control the regulation
of hematopoietic stem cells. The first approach
involves the analysis of mouse mutations that
disrupt hematopoiesis, while the second involves
the study of the multistage erythroleukemias in-
duced by the various forms of Friend murine leu-
kemia virus, a murine retrovirus.
The white-spotting (IT) and Steel (^Sl) Loci
Mutations at either the dominant white-
spotting {W} or Steel (SI) loci can lead to coat
color defects, severe macrocytic anemia, and ste-
rility. This similarity in pleiotropic phenotype of
W^and SI mutants is striking, particularly as there
is no common developmental origin of cells that
give rise to melanocytes, blood cells, and germ
cells. Furthermore, the Wand SI loci are clearly
distinct genes, as they map on mouse chromo-
somes 5 and 10, respectively.
Initial insights into their mechanism of action
came from in vivo cell-mixing experiments, in-
volving either bone marrow transplantation or
aggregation chimeras with animals of different ge-
notypes. These early studies established that the
developmental defects in W mutant mice result
from a cell-autonomous, intrinsic stem cell de-
fect, whereas the cellular defect in 5/ mutant
mice is in the microenvironment in which these
cells develop in the embryo and function in the
adult animal.
Several years ago, we showed that H^was allelic
with the proto-oncogene c-kit, a member of the
family of genes that includes c-fms and pdgfr,
which encode transmembrane receptors with ty-
rosine kinase activity. Different W alleles are ei-
ther deletions or point mutations in the cytoplas-
mic tyrosine kinase domain that result in partial
or complete loss of tyrosine kinase activity and a
consequent reduction in the signaling capability
of the Kit receptor. Thus the W mutant mouse
provided the first example of a germline muta-
tion in a mammalian proto-oncogene. This work
is supported by a grant from the National Insti-
tutes of Health.
These findings predicted that SI might encode
the ligand for the Kit receptor and that this ligand
might be a potent growth or survival factor for
melanoblasts, germ cells, and hematopoietic
stem cells. The microenvironmental nature of the
defects in SI mutant mice also predicted that the
5/ protein product might be membrane bound.
All three predictions have turned out to be
correct. Thus the W-Sl gene pair has provided
strong molecular genetic evidence for the impor-
tant role that cell-cell interactions play in the de-
velopment and function of hematopoietic and
other stem cells. Their protein products are strik-
ingly similar in design and function to those of
the sev and boss genes, which control the devel-
opment of the R7 photoreceptor cell in the com-
pound eye of the fruit fly Drosophila.
Friend Leukemia Virus
In the second approach, we are identifying and
analyzing cellular genes important in the induc-
tion of the multistage erythroleukemias induced
by Friend leukemia virus. Several years ago, we
showed that the p53 gene was inactivated by de-
letion or retroviral insertion in approximately 30
485
Molecular Genetics of Normal and Leukemic Hematopoiesis
percent of independent leukemic clones isolated
from the spleens of mice infected with Friend
virus. These experiments suggested that />53 is a
tumor-suppressor gene, a conclusion supported
by recent experiments demonstrating allelic loss
and mutation in the p53 gene in a broad spectrum
and high proportion of human cancers.
Tlie importance of p53 in the evolution of
Friend leukemia also became evident from stud-
ies on transgenic mice that express high levels of
mutant forms of the p55 protein. In addition to
displaying an increased spontaneous predisposi-
tion to a variety of malignancies, these p53 trans-
genic mice are more susceptible to the late stages
of Friend leukemia. Thus inactivation of the p53
gene, either somatically after Friend virus infec-
tion or as the result of inheritance of dominant
negative alleles of p53, appears to be a central
event in the disease. These studies also suggested
that it is the accumulation of a specific set of mu-
tational events, rather than the order in which
they normally occur, that is critical for leukemia
induction.
In addition to inactivation of the p53 gene, ac-
tivation of one of two novel members of the ets
gene family of DNA-binding transcriptional acti-
vators occurs during the evolution of Friend leu-
kemia. Transcription of the Spi-\ gene is acti-
vated as a result of the integration of Friend
spleen focus-forming virus (SFFV) in 95 percent
of erythroleukemic clones induced in adult mice
by SFFV. In contrast, we showed last year that
transcription of another ets family member, Fli- 1 ,
is activated by insertion of the replication-
competent Friend MuLV in 75 percent of erythro-
leukemia clones induced after infection of new-
born mice. The members of the ets gene family,
which include c-ets- 1 and c-ets-2, elk- 1 and elf-1,
erg, E74, 5p/-l/PU.l, and Fli-\, contain a con-
served domain of 80-90 amino acids, the ETS do-
main, which is involved in specific DNA binding,
as well as a less-conserved transcriptional activa-
tion domain.
The strict specificity of the integration sites for
SFFV and Friend MuLV is intriguing, as Spi-\ and
Fli-\ are both members of the same gene family
and both SFFV and Friend MuLV induce erythro-
leukemias involving activation of the receptor for
erythropoietin and inactivation of the p53 gene.
We are currently pursuing the hypothesis that
Fli- 1 and Spi- 1 are functionally distinct genes, en-
coding proteins that transactivate a distinct set of
genes downstream in the leukemogenic pathway.
These studies involve the analysis of the DNA-
binding specificity of the Fli-\ and Spi- \ proteins
as well as the generation of mutant mice carrying
either gain- or loss-of-function mutations in these
genes. This work is supported by a grant from the
National Cancer Institute of Canada.
The ets gene family was first discovered by ana-
lyzing the genome of the avian erythroblastosis
virus E26, which contains a part of a myb fusion
protein, the \-ets oncogene. Thus at least three
members of the ets family, \-ets, Spi-1, and Fli-1,
are involved in the leukemic transformation of
erythroid cells. This specificity of ets genes for
the erythroid lineage suggests that members of
this gene family are involved in the regulation of
one or more genes critical to erythropoiesis. We
are interested in identifying these genes to
gain further insights into both the regulation
of normal hematopoiesis and the induction of
leukemia.
486
Molecular Biology of Two
Enteropathogenic Bacteria
Edmundo Calva, Ph.D. — International Research Scholar
Dr. Calva is Associate Professor and Chairman of Molecular Biology at the Biotechnology Institute,
National Autonomous University of Mexico (UNAM), Cuernavaca. He received his Ph.D. degree in
molecular biology from the University of Wisconsin-Madison under Richard Burgess and did postdoctoral
research at UNAM. He is currently President of the Mexican Biochemical Society.
THE study of infectious diseases provides op-
portunities to explore various basic biolog-
ical phenomena. During an infection, entero-
pathogenic bacteria participate in numerous
biological events that may lead to intestinal ill-
ness. Upon oral ingestion, the bacteria must un-
dergo a series of interactions with the host that
involve specific molecules, either on the surface
or in the interior of the bacterial and host cells.
For instance, bacteria adhere to epithelial gut
cells and produce enterotoxins that result in diar-
rhea. They can invade host tissue with the same
effect. Invasion sometimes results in a systemic
infection, in which the bacteria have means of
protection against the immune system, seriously
jeopardizing health.
Current research questions in the area of bacte-
rial pathogenesis address the molecular mecha-
nisms of the bacteria-host interaction. Our knowl-
edge of adherence, invasion, resistance to the
immune system, and enterotoxin production,
among other processes, is just being unveiled. We
know only scant details of some of these pro-
cesses in a few bacteria, and even then are not
completely aware of the genetic variation within
the species studied — variation that may result in
different strains with varying capabilities for
causing disease. Furthermore, some bacterial and
parasite antigens share structures with stress pro-
teins, like those expressed during heat shock. An
open area of study involves defining the global
genetic circuits that regulate virulence factors
and determining whether these "regulons" share
common features with those encountered in the
response to stresses other than infection.
Health biotechnology should benefit from the
definition of bacterial antigens that have a role in
infection or against which a specific immune re-
sponse is mounted, thus permitting the develop-
ment of better vaccines and diagnostic proce-
dures as well as furthering our understanding of
the structure and function of the immune system.
Benefits should also emerge from the isolation
and characterization of specific bacterial genes,
together with knowledge about their distribution
and polymorphism in bacterial populations. Such
information should permit the rapid and specific
detection of bacteria through nucleic acid ampli-
fication procedures. These can not only be useful
for monitoring infections in animals, plants, or
humans, or the contamination of foodstuff, but
should also be valuable in the definition of modes
of transmission and environmental reservoirs for
the bacteria. In this manner, molecular epidemi-
ology will very likely shed light on yet other bio-
logical phenomena.
Salmonella typhi
S. typhi is the causal agent of typhoid fever
(TF) in humans, a disease estimated to afflict an-
nually more than 1 2 million persons worldwide.
TF is the result of a systemic infection, in which S.
typhi can be isolated from blood cultures. As a
gram-negative bacterium, 5. typhi has an outer
membrane that surrounds an inner one and the
cell wall. Thus exposed to the cell's exterior envi-
ronment, outer membrane proteins (OMPs) have
been shown to be important immunogens for
protection against various bacterial infections in
laboratory models. We have demonstrated the
utility of OMPs for the rapid immunodiagnosis of
TF in patients from different parts of the world,
most of whom raise specific antibodies to OMP
preparations.
Our laboratory has reported the isolation and
characterization of an S. typhi gene coding for
OmpC, a major OMP. Amino acid sequence align-
ment with other OMPs that form pores (porins)
has allowed structure prediction in the porin su-
perfamily. Using site-directed mutagenesis, we
have constructed a gene that codes for a chimeric
protein containing a foreign epitope inserted in a
region predicted to be exposed on the cell sur-
face. This epitope is a segment of a capsid protein
from rotavirus, a causal agent of diarrhea, against
which neutralizing antibodies are made. The chi-
meric protein does indeed contain this epitope
on a segment exposed at the bacterial cell sur-
face, supporting the notion that the native region
is located toward the exterior and indicating that
OmpC could be used as a carrier of heterologous
epitopes. This information on protein topology
487
Molecular Biology of Two Enteropathogenic Bacteria
can be useful in the design of multivalent vac-
cines against several infectious agents.
We have used the S. typhi ompC gene to probe
the genetic variability of the salmonella genus.
In this manner we have been able to establish
some phylogenetic relationships among several
species.
We have also determined that expression of 5.
typhi OmpC is influenced differently by medium
osmolarity than its well-studied counterpart in
Escherichia coli. Nevertheless, expression of 5.
typhi ompC is also dependent on the E. coli
OmpR transcriptional activator. Apparently the
two bacteria differ in the mechanisms of gene ex-
pression in response to osmotic stress, although
common effectors appear to be shared. We are
presently studying this phenomenon in detail.
In addition, we have isolated other S. typhi
OMP genes, namely ompF and phoE, and are
characterizing these to gain a better understand-
ing of structure-function relations in porins and
their genes.
Campylobacter jejuni
C. jejuni is one of the major causal agents of
diarrhea throughout the world. Molecular biol-
ogy of this organism has developed slowly,
mostly because of the difficulty of maintaining
stably cloned C. jejuni DNA segments in E. coli.
Consequently, only a handful of C. jejuni genes
have been isolated. The mechanisms underlying
this DNA instability are not understood, and re-
search in this area might well yield valuable
results.
The isolation of C. jejuni, producing a cholera-
like enterotoxin, has been associated with clini-
cal symptoms of a watery secretory type of diar-
rhea. We have shown that the C. jejuni
chromosome contains sequences similar to the
enterotoxin genes of Vibrio cholerae (CT entero-
toxin) and E. coli (LT, heat-labile enterotoxin).
In cloning and characterizing these sequences,
we have explored different host cells and differ-
ent-size C. jejuni DNA fragments in order to find a
successful cloning procedure. In the process, we
have isolated two cryptic fragments, one highly
specific for C. jejuni and C. coli (another
diarrhea-producing Campylobacter) and the
other revealing genetic variability between the
two bacteria. We are testing the usefulness of
both DNA probes in field epidemiology and char-
acterizing them at the nucleotide level.
488
Functional Heterogeneity
in Prolactin-secreting Cells
Gabriel Cota, Ph.D. — International Research Scholar
Dr. Cota is Professor of Physiology, Biophysics, and Neurosciences at the Center for Research and
Advanced Studies, National Polytechnic Institute, Mexico City. He received his Ph.D. degree in physiology
and biophysics from the Center and subsequently carried out postdoctoral research with Clay M.
Armstrong at the University of Pennsylvania, as a Pogarty International Research Fellow.
OUR laboratory is interested in the cellular
mechanisms involved in the control of pro-
lactin secretion. Prolactin is a vertebrate hor-
mone that participates in the regulation of a di-
versity of physiological processes, including
lactation. This versatile chemical messenger is
produced in the pituitary gland by endocrine
cells called lactotropes or mammotropes. Basic
information about lactotrope function and its
control offers insight into the pathogenesis of hy-
perprolactinemia, a frequent hypothalamic pitu-
itary disorder in humans.
Until recently, lactotropes were commonly
thought to comprise a homogeneous cell popula-
tion in the normal pituitary gland. However, stud-
ies performed on cultured pituitary cells indicate
the existence of a considerable lactotrope hetero-
geneity. In these accessible model systems, sub-
sets of lactotropes that differ in basal secretory
activity or responsiveness to extracellular regula-
tory factors have been distinguished. Our work
over the last three years has focused on the origin
of such functional differences.
Lactotrope Subtypes
We have identified two subpopulations of lac-
totropes in pituitary cultures derived from adult
male rats. We used the reverse hemolytic plaque
assay, an immunological technique that permits
the microscopic visualization of single-cell se-
cretions. In this assay, cells releasing the appro-
priate hormone induce lysis of indicator erythro-
cytes. The size of the zone of hemolysis, or
plaque, around an individual secretor provides
an index of the cumulative amount of hormone
released by that cell. In keeping with previous
observations by Jimmy Neill (University of Ala-
bama) on the secretory behavior of female rat
lactotropes, we found that under basal con-
ditions some prolactin-secreting cells form small
plaques (SP lactotropes) and others induce large
plaques (LP lactotropes). Thus SP and LP lacto-
tropes secrete distinct amounts of prolactin per
unit of time in the basal state.
Calcium Channel Activity as a Determinant
of Lactotrope Heterogeneity
There is much evidence that calcium ion plays
a major role in pituitary cells as an intracellular
messenger in hormone secretion. In particular,
basal prolactin secretion is thought to be sus-
tained by calcium influx through plasma mem-
brane calcium channels, which transiently open
during spontaneous action potentials. This sug-
gests that differences in basal secretory rate
among lactotrope subtypes might arise, at least in
part, from a differential expression of calcium
channels. To test this possibility, we investigated
the calcium channel activity of lactotrope sub-
types, using electrophysiological techniques.
Our analysis indicates that both SP and LP
lactotropes express two classes of voltage-gated
calcium channels in the plasma membrane:
low-voltage-activated (LVA) and high-voltage-
activated (HVA) channels. The activity of LVA
channels does not significantly differ between
the two lactotrope subtypes. By contrast, the sur-
face density of HVA channels is markedly higher
in LP cells than in SP cells.
In addition, we tested the effect of nifedipine
on prolactin secretion. Nifedipine is a dihydro-
pyridine drug that selectively blocks HVA cal-
cium channels in many types of excitable cells,
including prolactin secretors. We found that ni-
fedipine inhibits prolactin secretion by preferen-
tially suppressing the LP lactotropes. In fact, a
large proportion (around 50 percent) of LP lacto-
tropes behave functionally as SP lactotropes in
the presence of nifedipine. Our results, taken
together, indicate that calcium entry through
dihydropyridine-sensitive HVA calcium channels
contributes to the high basal rate of prolactin se-
cretion in LP lactotropes.
Sodium Channels in Lactotropes
We have characterized an additional difference
in ion channel activity between the two lacto-
trope subtypes: membrane depolarization in-
duces larger whole-cell sodium currents in LP
lactotropes than in SP lactotropes. Such differ-
ences are not related to cell-to-cell variations in
the kinetic properties of the sodium currents and
persist after current amplitude is normalized by
cell capacitance, which eliminates membrane
489
Functional Heterogeneity in Prolactin-secreting Cells
area as a variable. Thus sodium channels, like
HVA calcium channels, are not uniformly ex-
pressed among SP and LP lactotropes.
Sodium channels should be functionally im-
portant for the secretory activity of lactotropes, as
they favor the triggering of action potentials and
thereby promote the opening of HVA calcium
channels. Indeed, we have found that blocking
the sodium channels of cultured cells with exter-
nal tetrodotoxin drastically decreases the total
amount of prolactin secreted. Furthermore, popu-
lation analysis of prolactin plaque sizes suggests
that tetrodotoxin, like nifedipine, preferentially
inhibits prolactin secretion from LP cells.
These findings raise the question of what fac-
tors regulate ion channel expression in pituitary
cells. They also renew our interest in the differ-
entiation of lactotrope subtypes in the develop-
ing pituitary gland. Our current research is fo-
cused on these topics.
490
Ionic Channels in Sea Urchin Sperm Physiology
Alberto Darszon, Ph.D. — International Research Scholar
Dr. Darszon is Professor of Biochemistry at the Biotechnology Institute, National Autonomous University
of Mexico, Cuernavaca, and Adjunct Professor of Biochemistry at the Center for Research and Advanced
Studies, National Polytechnic Institute, Mexico City. He received his undergraduate degree in chemistry
from the Universidad Iberoamericana in Mexico City, and his Ph.D. degree in biochemistry from the
Center for Research and Advanced Studies. His postdoctoral research was done with Mauricio Montal at
the University of California, San Diego. His honors include the National Science Award of the Mexican
Academy of Scientific Research and both Guggenheim and World Health Organization fellowships.
THE molecular mechanisms involved in cell
communication are at the forefront of re-
search in biology today, since they play a key role
in determining the behavior of organisms. For the
past decade the main goal in our laboratory has
been to understand the vital egg signals that allow
sperm to fuse with the egg and fertilize it.
The sea urchin has proved to be an excellent
model in which to study reproduction. From this
animal enormous quantities of sperm (10^°/
male) can be collected that respond to environ-
mental stimuli rapidly, synchronously, and in a
compulsory order. It has been shown that the
flow of ions through the plasma membrane of sea
urchin sperm participates crucially in the events
leading to fertilization. Indeed, these are excit-
able cells that quickly respond to components
from the outer layer of the egg, the jelly, with
changes in their plasma membrane permeability.
Ionic fluxes play a fundamental role in the acti-
vation of respiration and motility, in chemotaxis,
and in triggering the sperm acrosome reaction
(AR) . This latter reaction occurs within seconds
after sea urchin sperm reach the egg and interact
with its jelly coat. The AR involves important mor-
phological changes that allow sperm to fuse with
the egg. Participating in AR induction is an in-
crease in the uptake of Ca^"^ and Na^ and efflux of
K"^ and H^. We would like to learn how these
fluxes are related and elucidate the molecular
mechanisms that orchestrate them to trigger
the AR.
Previous results indicated indirectly that some
ion fluxes occur through ionic channels. These
are integral membrane proteins capable of form-
ing hydrophilic pores through the membrane that
allow the passive diffusion of ions at high rates
(10^"^ ions/s). We have used model membranes
(planar bilayers) formed from sperm compo-
nents, and patch-clamp techniques in whole
cells, to detect for the first time the activity of
single channels in the plasma membrane of sea
urchin sperm. These techniques, together with
studies of membrane potential, intracellular Ca^^
([Ca^+]j), and intracellular pH (pHj) in whole
sperm, have established the presence of K^, Ca^^,
and Cl~ channels in this cell and are allowing us
to explore their participation in the AR.
These findings have led us to propose a work-
ing hypothesis of how egg jelly-induced changes
in ionic permeability might trigger the AR in sea
urchin sperm. A receptor (to the inducing egg
factor) in the plasma membrane of the sperm cell
opens a Ca^^ channel, which deactivates in a few
seconds. The activation of this channel modu-
lates the opening of a second Ca^^ channel by
poorly understood mechanisms that include a
Ca^^-dependent change in pHj. At the same time
or immediately after the first Ca^"^ channel opens,
a channel is activated, hyperpolarizing the
cell and activating a voltage-dependent Na^/H^
exchange that increases pHj. This latter change in
pHj is linked to the opening of the second Ca^"^
channel and to a large depolarization.
In light of this working hypothesis, we thought
it should be possible to hyperpolarize sperm arti-
ficially and induce an increase in [Ca^"^]! after a
depolarization and AR. Valinomycin-induced hy-
perpolarization of sperm from the sea urchin Ly-
techinus pictus in K^-free sea water raised pHj,
caused a small increase in ^'Ca^"^ uptake, and trig-
gered some AR. When the cells were depolarized
with 30 mM KCl 40-60 seconds after the hyper-
polarization, the pHj decreased and there was a
significant increase in ''^Ca^^ uptake, [Ca^"^];, and
AR. Therefore the jelly-induced hyperpolariza-
tion may lead to the intracellular alkalinization
required to trigger the AR and may modulate, on
its own or via pHj, Ca^^ channels involved in this
process.
The sea urchin sperm offers distinct advantages
over more complex cell types as a basic model for
chemotaxis. As the spermatid matures, many in-
tracellular organelles and macromolecules not
involved in fertilization are eliminated. Sperm
are incapable of division and devoid of the ma-
chinery for genome expression. However, they
retain systems for sensing, swimming toward, and
fusing with the egg, as described above.
The sperm of the sea urchin Arbacia punctu-
lata is attracted at nanomolar concentrations to a
491
Ionic Channels in Sea Urchin Sperm Physiology
small peptide isolated from the homologous egg.
This peptide, called resact, is species specific and
induces in sperm a transient increase in [Ca^^];
and cGMP and an increase in pHj. The egg of
Strongylocentrotus purpuratus contains a pep-
tide, named speract, that also elevates cGMP and
[Ca^^ji and induces an alkalinization and a K"^-
dependent hyperpolarization.
Sea urchin sperm, however, are tiny cells (head
diameter ~2 ytvci). This has precluded a careful
characterization of their electrophysiological
properties that would shed light on the molecu-
lar mechanisms determining their fascinating
egg-induced behavioral changes. Recently, in col-
laboration with Donner Babcock and Martha
Bosma from the University of Washington in Seat-
tle, we found that it is possible to swell sea urchin
sperm in diluted sea water. The swollen cells are
spherical (~4 yum in diameter), immotile, and
metabolically active, and they can regulate their
[Ca'^^Ji, pHj, and membrane potential.
The swollen cells respond to pM concentra-
tions of speract with an increase in K"^-selective
permeability that lasts for many seconds. We
found that this permeability change, as well as
the changes in [Ca^^]; and pHj that are seen at
higher speract concentrations, also occurs in
nonswoUen sperm. An advantage of swollen
sperm is that they can be much more easily patch
clamped and single-channel-activity recorded.
We have observed that pM speract activates a
small K"^ channel. Thus swollen sperm provide
new avenues to study ionic channels and their
regulation by egg factors and second messengers.
This work is supported by grants from the Mex-
ican Council for Science and Technology, the
World Health Organization, and the Miguel Ale-
man Foundation.
Confocal immunofluorescent micrograph of
Salmonella typhimurium (yellow) interact-
ing with a cultured epithelial cell and
causing rearrangement of epithelial actin
filaments (blue) around the invading bac-
terium.
Research and photograph by B. Brett
Finlay.
492
Host-Pathogen Interactions in Microbial
Pathogenesis
B. Brett Finlay, Ph.D. — International Research Scholar
Dr. Finlay is Assistant Professor in the Biotechnology Laboratory and the Departments of Biochemistry
and Microbiology at the University of British Columbia, Vancouver. He is a member of the Canadian
Bacterial Diseases Center of Excellence. After receiving his B.Sc. and Ph.D. degrees in biochemistry from
the University of Alberta, Edmonton, he conducted postdoctoral work on microbial pathogenesis in the
laboratory of Stanley Falkow at Stanford University as a fellow of the Alberta Heritage Foundation for
Medical Research.
IN all cases of bacterial disease, the bacterium
or a bacterial product interacts with host cells
or surfaces in either of two ways. It may adhere to
the cell surface or may actually enter the cell and
grow within (intracellular pathogen) . Pathogens
residing in an intracellular environment are pro-
tected from the host's immune systems, antibi-
otics, and other therapeutic agents. They often
use this safe niche to multiply before disseminat-
ing to other sites and deeper tissue.
Our laboratory uses a multidisciplinary ap-
proach to define the interactions that occur be-
tween pathogenic bacteria and host cells. Essen-
tial to continuation of these interactions is the
exchange of signals between pathogen and cell.
We are studying the molecular nature of these
signal transduction events in an effort to deter-
mine the mechanisms involved. Knowledge of
these mechanisms should point the way to novel
therapeutic strategies.
Several bacterial pathogens are used in these
studies, since each organism has its individual
features. Additionally, comparison of the signals
generated by different pathogens can tell which
mechanisms are common and which are unique.
Salmonella typhimuriumi
A Model for Intracellular Parasitism
For several reasons, 5. typhimurium provides
an excellent model for the study of intracellular
parasitism. Salmonella species continue to cause
significant health problems in both developed
and less developed countries. These organisms
have the capacity to enter into, survive, and repli-
cate within host cells — features that contribute
to virulence. Since 5. typhimurium is z close velz-
tive of the nonpathogenic Escherichia colt, we
have been able to use established molecular ge-
netic techniques to study several aspects of Sal-
monella pathogenesis. Finally, 5. typhimurium
infection of the mouse closely mimics human ty-
phoid fever.
Since Salmonella species (and many other
pathogens) interact with intestinal epithelial
cells following oral ingestion, we have been able
to utilize monolayers of polarized epithelial cells
grown on permeable substrates to study the inter-
actions. These systems have several properties in
common with columnar epithelial cells of the in-
testine. Salmonella species have the capacity to
enter and penetrate through these polarized epi-
thelial monolayers. Moreover, transposon mu-
tants that are defective for such penetration have
been isolated from 5. typhimurium. Molecular
characterization of the genetic loci that are
disrupted by these transposons will provide clues
about the bacterial products required for cell
entry and penetration.
Once inside a vacuole within an epithelial cell,
5. typhimurium finds itself within a presumably
quite different environment. Very little is known
about the microenvironment of any intracellular
pathogen. To probe and define aspects of the va-
cuolar habitat, we have been using a bacterial re-
porter gene (tocZ) fused to several S. typhimur-
ium genes that are variously regulated. For
example, one can measure expression of reporter
genes that are affected by oxygen levels, carbon
source, pH, or iron or magnesium concentra-
tions. These studies are providing clues about the
nature of this intracellular niche. Additionally,
we have begun to search for other S. typhimur-
ium loci that are only induced when the bacte-
rium is inside host cells, in the hope of finding
other regulated genes.
After initial survival inside a host cell, 5. typhi-
murium begins to multiply within its intracellu-
lar vacuole. Virtually nothing is known about its
requirements there. We have identified three bac-
terial genes that are needed for S. typhimurium
to multiply within host cells but not without.
Characterization of these genes and other experi-
ments may provide information about bacterial
products necessary for intracellular growth.
Signal Transduction Between Host
and Pathogen: Involvement of the Host
Cytoskeleton and Tyrosine Kinases
Most intracellular pathogens require participa-
tion of the host cell for successful pathogen inter-
nalization. Functional actin filaments are often
involved in bacterial uptake, and 5. typhimur-
493
Host-Pathogen Interactions in Microbial Pathogenesis
ium is one pathogen that requires host actin fila-
ments for invasion. When S. typhimurium enters
cukured epitheUal cells, there is a large rear-
rangement in intracellular actin and the function-
ally related cytoskeletal proteins a-actinin, tropo-
myosin, and talin that surround the membrane
engulfing the invading organism. This rearrange-
ment, presumably needed for functional internal-
ization, is triggered by bound extracellular bacte-
ria and occurs in the region beneath adherent
organisms. Once the pathogen is inside a vacu-
ole, the epithelial cytoskeleton returns to its nor-
mal distribution.
We have been examining the signals trans-
duced through the host cell membrane that are
responsible for triggering this cytoskeletal rear-
rangement, and host cell signals that are neces-
sary for the uptake of other pathogenic bacteria
into epithelial cells. Host tyrosine kinases appear
to participate in the uptake of several pathogens,
including Yersinia species (causative agent of
gastrointestinal problems). Listeria monocyto-
genes (a gram-positive organism associated with
meningitis and serious infections of neonates),
and enteropathogenic Escherichia coli (caus-
ative agent of diarrhea). We are using various
kinase inhibitors, monoclonal antibodies that
recognize phosphotyrosine residues, and radiola-
beling of host phosphate pools to begin to iden-
tify the components of the signal transduction
pathways that these bacteria pirate for their
own use.
An in Vitro Blood-Brain Barrier to Study
Bacterial Meningitis
Bacterial meningitis (infection of the brain lin-
ing) is a common and serious disease in both
children and adults. In an effort to define more
clearly the molecular mechanisms mediating this
disease, we developed a model in vitro blood-
brain barrier (BBB) . This model uses primary iso-
lates of microcapillary endothelial cells from bo-
vine brain. Reseeded onto permeable substrates,
these cells form impermeable monolayers with
tight junctions and have several features charac-
teristic of the BBB.
When Haemophilus influenzae (the most
common cause of pediatric meningitis) is added
to these monolayers, they are completely
disrupted by a cytotoxic mechanism — similar to
the effects observed in vivo. We have found that
this cytotoxicity is mediated by bacterial lipo-
polysaccharide (LPS) in concert with a soluble
serum factor. Current research has a twofold aim:
first, to block this cytotoxicity with monoclonal
antibodies against the serum factor in a primate
meningitis model and second, to define the signal
transduction pathways that are triggered in the
endothelial cell by the LPS. In addition, the role
of cytokines in this cytotoxic event is being
studied.
Collectively this work provides several insights
as to how pathogenic bacteria manifest disease.
Studies with 5. typhimurium have afforded new
perceptions into the molecular biology of intra-
cellular parasitism. Definition of the signal trans-
duction pathways of invasive pathogenic bacteria
is providing evidence that many pathogens ex-
ploit existing pathways, often utilizing them to
pirate the host cytoskeleton for their own advan-
tage. Work with the BBB has offered new opportu-
nities for therapeutic intervention in treating
meningitis.
494
Mechanisms of Transcriptional Regulation
Jack Greenblatt, Ph.D. — International Research Scholar
Dr. Greenblatt is Professor in the Banting and Best Department of Medical Research and the Department of
Molecular and Medical Genetics at the University of Toronto. He earned his undergraduate degree in
physics from McGill University, Montreal, and his Ph.D. degree in biophysics from Harvard University,
where he studied bacterial gene regulation with Walter Gilbert. He pursued postdoctoral studies with
Alfred Tissieres at the University of Geneva. He has received the Ayerst Award of the Canadian
Biochemical Society.
THE ultimate focus of transcriptional regula-
tory mechanisms is RNA polymerase, a com-
plex multisubunit enzyme whose activity is
guided by interactions with a myriad of regula-
tory proteins and regulatory sequences in DNA or
RNA. Much of our research is devoted to the basic
enzymology of initiation and termination of tran-
scription and to identifying and characterizing
some of the key protein-protein interactions in-
volved. In addition, we are examining how
model regulatory proteins interact with the basic
transcriptional apparatus. This work is also sup-
ported by grants from the Medical Research
Council of Canada and the National Cancer Insti-
tute of Canada.
Initiation of Transcription by Human RNA
Polymerase II
A set of general transcription factors (TFILA, B,
D, E, F, H) is necessary for RNA polymerase II to
initiate the transcription of protein-coding genes.
TFIID is the general factor that recognizes TATA
sequences present in the promoters of many
genes. After TFIID binds to the DNA, TFIIA and
TFIIB recognize and bind to the TFIID-DNA com-
plex. Subsequently the other general factors and
RNA polymerase II assemble into a multiprotein
complex at the promoter. We use protein affinity
chromatography as a technique to identify direct
protein-protein interactions involved in the as-
sembly of this complex.
TFIID has been highly conserved during evolu-
tion. TFIID molecules from most or all eukar-
yotes, including fungi, insects, and plants, can
function in transcription reactions containing
RNA polymerase II and other general factors of
human origin. By using yeast TFIID as a ligand for
affinity chromatography, we have identified
three human polypeptides that interact with
TFIID and constitute human TFIIA. Curiously,
TFIID columns do not retain TFIIB, nor do TFIIB
columns retain TFIID, suggesting that a confor-
mational change in TFIID induced by DNA bind-
ing facilitates its interaction with TFIIB.
By using RNA polymerase II as a ligand for affinity
chromatography, we identified human RAP30 and
RAP74, the small and large subunits of TFIIF. Hu-
man cDNAs encoding RAP30 and RAP74 have both
been cloned, the latter in collaboration with
Zachary Burton (Michigan State University) . Using
recombinant RAP30, we found that RAP30 is the
subunit of TFIIF that binds RNA polymerase II.
RAP30 prevents RNA polymerase II from associat-
ing with and transcribing nonpromoter sequences
in DNA, a property also of bacterial a factors. In-
deed, we found that RAP30 can bind to Escherichia
coli RNA polymerase and be displaced by the major
bacterial a factor, known as a""^.
RAP30 also has a central role in promoter rec-
ognition by RNA polymerase II. In collaboration
with Danny Reinberg (University of Medicine and
Dentistry of New Jersey), we found that RAP30
can recruit RNA polymerase II to a preinitiation
complex containing TFIIA, TFIIB, and TFIID. In
fact, recognition of a promoter containing a TATA
sequence by RNA polymerase II can be achieved
with recombinant TFIIB, RAP30, and TBP, the
TATA sequence-binding subunit of TFIID, all
produced in E. coli. These three general factors,
therefore, constitute a minimal set of proteins
necessary and sufficient for promoter binding by
RNA polymerase II. However, this preinitiation
complex containing RNA polymerase II will not
initiate transcription unless supplied with
RAP74, TFIIF, TFIIH, and other factors. The roles,
subunit compositions, and protein-protein inter-
actions of some of these factors remain to be
identified.
Regulation of Initiation
by RNA Polymerase II
Many transcriptional activator proteins have
two domains. One binds to regulatory sequences
in DNA, and the other, known as an activation
domain, provides an activating signal to the basic
transcriptional apparatus. How these activation
domains function is a fascinating question in reg-
ulatory biology. It has been the major focus of
collaborative studies with my colleague, C.James
Ingles (University of Toronto) .
Many activation domains are highly acidic. A
particularly potent one is found in the Herpes
495
Mechanisms of Transcriptional Regulation
simplex virus protein VP 16. By using the VP 16
activation domain as a ligand for affinity chroma-
tography, v^e found that VP 16 interacts directly
with the TBP subunit of TFIID . Mutations in VP 1 6
that reduce gene activation by VP 1 6 also reduce
its binding to TBP, and preliminary work has
identified a mutation in TBP that reduces its abil-
ity to interact with and respond to VPl6. Interac-
tion of VP 16 with TBP alters the association of
TBP with the promoter. Therefore, VP 16 may in-
fluence the ability of TBP to recruit other general
factors and RNA polymerase II to the promoter.
A potent acidic activation domain is also found
in p53, the product of a human anti-oncogene (a
tumor-suppressing gene) that is mutated in about
half of all human cancers. Like VP16, the p53
activation domain also binds TBP. By using affin-
ity chromatography to search for other proteins
that interact with p53, we recently discovered a
new protein that interacts with the p5 3 and VP 1 6
activation domains. At least some oncogenic mu-
tations in p53 prevent binding of this protein to
the p53 activation domain. The precise role of
this novel protein in transcription is still under
investigation.
Transcriptional Antitermination in E. coli
The N protein of bacteriophage X prevents ter-
mination by RNA polymerase during transcrip-
tion of X operons expressed immediately after in-
fection of E. coli cells. We have reconstituted
antitermination by N in vitro in a system contain-
ing seven purified proteins: E. coli RNA polymer-
ase, the E. coli transcription termination factor
Rho, N, and four E. co/? cofactors for antitermina-
tion (NusA, NusB, SIO, and NusG). By using pro-
tein affinity chromatography and other methods,
we have identified many of the protein-protein
interactions in this system. Three factors, NusA,
NusG, and SIO, all bind to RNA polymerase. NusA
is important for termination of transcription at
some terminators. Since N binds to NusA, NusA is
also an adapter that couples the antitermination
factor N to RNA polymerase. In addition, NusB
binds to SIO and NusG to Rho factor. In fact,
NusG aids termination by Rho in vitro and is es-
sential for termination by Rho in vivo.
The stable association of N with RNA polymer-
ase requires an N utilization site (nw? site) in the
transcribed DNA and all four bacterial cofactors.
The nut site RNA in the growing RNA transcript is
recognized and bound by the proteins. N recog-
nizes the boxB component of the nut site RNA,
while NusB and SIO recognize a better version of
the box A component of nut site RNA, which is
found in the antiterminator sequences of bacte-
rial ribosomal RNA operons. Since mutations in
RNA polymerase can prevent the formation of a
stable multiprotein complex and can prevent
binding of protein to the nut site RNA during
transcription, we infer that a stable ribonucleo-
protein complex containing nut site RNA and
five antitermination proteins assembles on the
surface of RNA polymerase. This extensively
modified RNA polymerase can then transcribe
through kilobases of X DNA containing many tran-
scriptional terminators.
The human immunodeficiency viruses (HIV-1
and HIV-2) produce antitermination factors,
known as Tat, that recognize regulatory se-
quences in viral RNA, known as TAR. The human
host factors involved in antitermination by Tat are
unknown. Antitermination of HIV transcription
by Tat may well resemble antitermination of X
transcription by N in many respects.
496
Ionic Homeostasis in White Blood Cells
Sergio Grinstein, Ph.D. — International Research Scholar
Dr. Grinstein is Head of the Division of Cell Biology at the Research Institute of the Hospital for Sick
Children, Toronto, and Professor of Biochemistry at the University of Toronto. He received his Ph.D. degree
at the National Polytechnic Institute, Mexico City, where he studied with David Erlij. His postdoctoral
training was in two stages: initially at the Hospital for Sick Children under the supervision of Aser
Rothstein, and later at the Federal Institute of Technology in Zurich with Giorgio Semenza. He has received
the Ayerst Award of the Canadian Biochemical Society.
LEUKOCYTES constitute the body's first line of
defense against invading microorganisms.
These white blood cells first detect and engulf
bacteria and other microbes, then secrete lytic
enzymes and synthesize reduced oxygen metabo-
lites to kill them.
These microbicidal processes call for pro-
nounced changes in the generation and intracel-
lular distribution of acid equivalents. First, the
phagocytic vacuole wherein the microorganisms
are trapped becomes markedly acidic. This evi-
dently promotes the activity of the lytic enzymes
released into the phagosome and may also facili-
tate its fusion with the vesicles containing bacteri-
cidal agents. Then too, the leukocytes' rate of
metabolic acid production increases greatly dur-
ing infection, threatening the stability of the cyto-
solic compartment, which must remain slightly
alkaline to preserve optimal cell function.
The purpose of our research is to understand
the mechanisms underlying phagosomal acidifi-
cation, the pathways responsible for excess meta-
bolic acid during leukocyte activation, and partic-
ularly the processes involved in the maintenance
of the cytoplasmic pH under both resting and ac-
tivated conditions.
Antiports and Channels in the Regulation
of Cytosolic pH
The pronounced metabolic burst that leuko-
cytes undergo when confronted by microorgan-
isms or their products can be largely attributed to
activation of an otherwise quiescent enzyme, the
NADPH oxidase. The one-electron reduction of
oxygen catalyzed by this enzyme is accompanied
by oxidation of NADPH to NADP^ and release of
protons. Regeneration of NADPH through the
hexose monophosphate shunt is a source of fur-
ther proton production. If uncompensated, the
proton production by these pathways would pro-
duce a massive intracellular acidification, incom-
patible with normal leukocyte function and possi-
bly even their viability.
Three primary pathways appear to be involved
in proton (equivalent) extrusion in activated leu-
kocytes. The first and perhaps most important is
an electroneutral exchanger (antiport) that trans-
ports protons out of the cells in exchange for ex-
tracellular sodium. A major isoform of this anti-
port has been identified in fibroblasts as a
1 10-kDa membrane glycoprotein. We have found
that the antiport is active not only after stimula-
tion but also in resting cells. Its activity, however,
is greatly enhanced following the addition of bac-
terial chemoattractants or of molecules that
mimic events in the intracellular signaling cas-
cade triggered by microorganisms.
Our current and future research eff'orts in this
area deal with the molecular characterization of
the antiport in leukocytes and of the mechanisms
underlying its activation during infection and in-
flammation. We are particularly interested in the
subcellular localization of the antiports before
and after stimulation, in the mechanisms
whereby chemoattractants signal activation, and
in the segregation and/or inactivation of anti-
ports in compartments where sodium/proton ex-
change activity is counterindicated (such as the
phagosome) .
We have recently detected a second pathway
that appears to be important in the extrusion of
protons from activated leukocytes, namely a pro-
ton conductance, possibly a channel. This con-
ductive path is essentially undetectable in quies-
cent cells but becomes clearly apparent when the
cells are stimulated. Preliminary data indicate
that this putative channel is present in neutro-
phils, macrophages, and the human leukemic
cell line HL60. The conductive proton pathway
could serve two important functions in the stimu-
lated leukocyte: it could contribute to the extru-
sion of net acid equivalents from the cell, and it
could also serve as a source for counterions to
neutralize the voltage generated by the NADPH
oxidase, proposed to be electrogenic.
Little is known at present about the conductive
pathway. We are interested in defining its molec-
ular identity, physiological significance, intra-
cellular distribution, developmental pattern, and
the molecular basis of its activation. In this re-
gard, it is noteworthy that activation of the con-
ductance closely mirrors the behavior of the
497
Ionic Homeostasis in White Blood Cells
NADPH oxidase. We are in the process of explor-
ing the relationship between these events, with
particular interest in whether the putative chan-
nels are a component of the oxidase complex or
whether assembly of the latter is required for ac-
tivation of the conductance.
For this purpose, we have initiated studies us-
ing-cells from patients with chronic granuloma-
tous disease. These cells are defective in specific
components of the NADPH oxidase. In future ex-
periments we will attempt to detect the conduc-
tance electrophysiologically and to reconstitute
its activity in cell-free systems. If a linkage be-
tween the two processes is established, purified
or recombinant components of the oxidase will
be used.
Proton Pumps in Cytoplasmic
and Organellar pH Regulation
Recent experiments have also suggested that a
third mechanism of proton extrusion is opera-
tional in stimulated leukocytes. Pharmacological
evidence indicates that this latter pathway may be
a proton-pumping ATPase of the vacuolar type. As
described for the other systems, the activity of the
pumps becomes clearly apparent after stimula-
tion. At present, neither the subcellular location
of the pumps is known nor the mechanism of ac-
tivation understood. We will try to determine
whether activation results from post-translational
modification of pumps present in the relevant
membrane or whether translocation between in-
active and active compartments occurs. Immuno-
chemical and molecular biological means will be
used to localize the pumps and identify the
type(s) of isozyme involved.
Proton pumps are also seemingly responsible
for phagosomal acidification. We are currently in-
terested in the source of the pumps that underlie
this process and their mode of activation. We are
also planning to study the determinants of the
internal pH of the phagosome and other endo-
membrane compartments. Our current evidence
suggests that differential counterion permeabil-
ity, which was claimed to be the main source of
pH heterogeneity, is not an important factor dic-
tating intraorganellar pH. Differential pH sensitiv-
ity of the pumps, due to varying subunit compo-
sition, or a regulated proton leak permeability are
our preferred hypotheses.
These studies are expected to contribute to our
understanding of immune cell function and intra-
cellular pH regulation in these and other cells.
498
Genetic Basis of Multidrug Resistance
Philippe Gros, Ph.D. — International Research Scholar
Dr. Gros is Associate Professor of Biochemistry at McGill University, Montreal, and a member of the McGill
Cancer Center and the McGill Center for the Study of Host Resistance. He received his Ph.D. degree from
McGill University and pursued postdoctoral training in molecular biology and cancer research at
Massachusetts General Hospital with Joel Habener and at the Massachusetts Institute of Technology with
David Housman.
TUMOR cells in vivo and in vitro can develop
simultaneous resistance to a wide range of
structurally and functionally unrelated cytotoxic
drugs. Such multidrug resistance (MDR) severely
impedes the chemotherapeutic treatment of
many types of tumors. Structural or functional
characteristics common to drugs of the MDR
spectrum are few. In general these drugs are
small, hydrophobic natural products that often
contain a basic nitrogen atom and penetrate
the cell by passive diffusion across the mem-
brane. MDR is associated with a decreased in-
tracellular drug accumulation and concomitant
increased drug efflux from resistant cells, both
ATP-dependent .
MDR is caused by the overexpression of a high-
molecular-weight membrane phosphoglycopro-
tein called P-glycoprotein (P-gp) . P-gp has been
found capable of binding photoactivatable ana-
logues of ATP and cytotoxic drugs, suggesting
that it functions as an ATP-driven efflux pump
that reduces the intracellular accumulation of
drugs in resistant cancer cells. Recent studies
have shown that increased P-gp expression in neu-
roblastomas and soft-tissue sarcomas causes lack
of response to chemotherapy and is associated
with very poor prognosis and outcome of these
diseases.
P-gp is encoded by a small family of closely
related genes, termed mdr or pgp, that share con-
siderable sequence homology and common an-
cestral origins. This gene family has three
members in rodents (mdrl, mdr 2, and mdr 3)
and two in humans {MDRl and MDR2) .We have
isolated and characterized full-length cDNA
clones corresponding to the three mouse genes
and have deduced the amino acid sequences of
the three predicted polypeptides. P-gps share
considerable sequence homology (80-85 per-
cent identity) and common structural features,
including 1 2 predicted transmembrane domains
and two nucleotide binding sites.
Each P-gp is formed by two homologous halves
that show sequence conservation with a large
group of bacterial transport proteins participat-
ing in the import and export of specific substrates
in Escherichia coli. This evolutionary conserva-
tion is in keeping with P-gp's proposed drug ef-
flux function.
The normal physiological function of P-gps has
yet to be elucidated. Each P-gp isoform is ex-
pressed in a tissue-specific fashion, generally on
the apical surface of secretory epithelial cells
such as those of the bile canalicular, the brush
border of the intestine, and the proximal tubule
of the kidney. It has also been found in endothe-
lial cells of the blood-brain barrier and in early
pluripotent stem cells of the hematopoietic sys-
tem. From these findings it appears that P-gp ei-
ther plays a normal detoxifying role against envi-
ronmental xenobiotics or transports normal
physiological substrates yet to be identified.
Recently it was shown that the mrfr gene family
is itself part of a larger family of sequence-related
genes shown to play key physiological functions
in normal cells and tissues. These include the
STE6 gene of the yeast Saccharomyces cerevi-
siae, responsible for the transmembrane trans-
port of the "a" mating pheromone; the pfmdrl
gene of the malarial parasite Plasmodium falci-
parum, associated with chloroquine efflux from
resistant isolates of this parasite; and in humans,
the CFTR chloride channel gene, mutations of
which cause cystic fibrosis, and the RING family
genes, which code for peptide pumps implicated
in antigen presentation by T lymphocytes. There-
fore it appears that the mdr supergene family
codes for membrane-associated transport pro-
teins that may transport different types of sub-
strates by the same mechanism.
We have carried out functional analyses of indi-
vidual members of the mouse mdr gene family.
For this, we have transfected and overexpressed
cDNAs that correspond to each member of the
family. We observed that mdrl and mdr3, but
not mdr2, could directly confer MDR to other-
wise drug-sensitive cells. In addition, the profile
of drug resistance conferred by mdrl and mdr3
appeared distinct.
One of the key unanswered questions about P-
gp and MDR is how a single transport protein can
apparently recognize and transport a large group
499
Genetic Basis of Multidrug Resistance
of structurally and functionally unrelated com-
pounds. The identification of P-gp segments and
residues implicated in drug recognition and trans-
port is a necessary prerequisite to the rational de-
sign of new cytotoxic compounds capable of
blocking or bypassing the action of P-gp in drug-
resistant tumor cells. We have exploited the high
degree of sequence similarity and striking func-
tional differences detected among members of
the mouse mdr family to identify, in chimeric
and mutant proteins, segments and residues im-
portant for drug recognition. To identify the do-
mains of mdrl that are essential for MDR and that
may be functionally distinct in the biologically
inactive mdr2, we have constructed 1 1 chimeric
molecules in which discrete domains of mdr2
have been introduced into the homologous re-
gion of mdrl. We have analyzed these chimeras
for their capacity to confer MDR.
The two predicted ATP-binding sites of mdr2
were found to be functional, as either could com-
plement the biological activity of mdrl. How-
ever, the replacement of either the amino- or the
carboxyl-terminus transmembrane (TM) domain
region of mdrl by the homologous segment of
mdr2 resulted in inactive chimeras. Replace-
ment of as few as two TM domains of mdrl from
either the amino- or carboxyl-terminal halves by
the corresponding segment of mdr2 was suffi-
cient to destroy mdrVs activity. These observa-
tions suggest that the functional differences de-
tected between mdrl and mrfr2 reside within the
TM domains of the two proteins.
P-gps encoded by mouse mdrl and mdr3
confer distinct drug resistance profiles. While
both clones confer comparable levels of resis-
tance to vinblastine (VBL), mdr 3 confers prefer-
ential resistance to actinomycin D (ACT), and
mdrl to colchicine (COL).
To identify protein domains implicated in the
preferential drug resistance encoded by either pa-
rental mdr clone, homologous protein domains
were exchanged in a series of 16 hybrid cDNA
clones, and the drug resistance profiles encoded
by the corresponding chimeric proteins were an-
alyzed. While all chimeric clones conferred simi-
lar levels of VBL resistance, the levels of ACT and
COL resistance conferred by the various clones
were heterogeneous, being either similar to the
parental mdrl or mdr3 clones or, in many cases,
intermediate between the two.
Only those chimeric proteins carrying seg-
ments that overlapped both the amino and car-
boxyl sets of TM domains of the respective parent
conveyed the parent's full preferential drug re-
sistance profile. These results suggest that
the resistance profiles encoded by mdrl or
mdr 3, possibly representing sites of drug- protein
interactions, involve several determinants asso-
ciated with TM domains from both homologous
halves of P-gp. Recently we have tentatively iden-
tified one of these sites. We have observed that a
simple serine-to-phenylalanine substitution at
position 941 (mdrl) or 939 (m<^r3), within pre-
dicted TMl 1 , had a dramatic effect on the overall
activity of the two pumps.
The modulating effect of this mutation on
mdrl and mdr 3 varied for the drugs tested. It was
very strong for COL and adriamycin (ADR) but
only moderate for VBL. For mdrl, the serine-to-
phenylalanine replacement produced a unique
mutant protein that retained the capacity to
confer VBL resistance but lost the ability to confer
ADR or COL resistance. These results suggest that
ADR-COL and VBL may have distinct binding sites
on P-gp and that the serine residue within TMl 1
plays a key role in P-gp's recognition and trans-
port of the former drugs. We are currently prob-
ing drug-P-gp interactions at this residue, using
additional mutants together with COL and ADR
analogues modified at key positions on their re-
spective backbone.
Other investigators have found the same resi-
due mutated in the pfmdrl gene from isolates of
the human malarial parasite Plasmodium falci-
parum that are resistant to chloroquine (CLQ).
Such resistance is caused by an increased ATP-
dependent CLQ efflux and is associated with mu-
tant alleles of the mdr homologue pfmdrl, map-
ping near TMl or within TMl 1 . Taken together,
these and our studies indicate that the TMl 1 do-
main of mdr and mdr-like genes is critical for
drug recognition and transport. Besides their
high degree of hydrophobicity, no significant ho-
mology is detected between TMll domains of
mdrl-3 and pfmdrl proteins. Both, however,
have the potential of forming amphipathic
helices.
The mdr Ser^'^"^^' residues and the mutated
residues in pfmdrl fall within the hydrophilic
side of this helix, and the mutant residues map
near what appears to be the boundary of the hy-
drophilic side. These amphipathic helices may
be important for the recognition of hydrophobic
compounds, such as MDR drugs, that readily par-
tition within the cell's lipid bilayer.
500
Control of Bacterial Protein Synthesis
During Viral Infection
Gabriel Guameros Pena, Ph.D. — International Research Scholar
Dr. Guameros is Professor of Genetics and Molecular Biology at the Center for Research and Advanced
Studies, National Polytechnic Institute, Mexico City. He received his undergraduate and M.Sc. degrees in
microbiology, chemistry, and biochemistry in Mexico City, and his Ph.D. degree in molecular biology
from the University of California, Berkeley, where he studied with Harrison Echols. He joined the staff of
the Center after doing postdoctoral work in molecular biology at the University of Geneva, Switzerland,
in Harvey Eisen's laboratory. He has been awarded fellowships from the Guggenheim Memorial
Foundation, the Commission of the European Communities, and the Sistema Nacional de
Investigadores, Mexico.
INFECTING viruses divert the functions of the
host cell for their own development. To
achieve this, the viral genome directs the synthe-
sis of regulatory molecules, which reorient cell
functions to conform to the specific viral develop-
ment program. There is abundant evidence of
transcriptional control during phage X infection
of Escherichia coli, where the phage genome di-
rects the synthesis of regulatory proteins that
alter the cells' transcriptional pattern. Little is
known, however, about the translational control
in X-infected cells. Our laboratory has pursued a
case of translational control involving X RNA se-
quences, named bar, and peptidyl-tRNA hydro-
lase (Pth), a bacterial enzyme essential for pro-
tein synthesis.
The Target in the Cell
Bacterial mutants partially defective in the ac-
tivity of Pth are unable to support the growth of X
phage. We have shown that the corresponding
mutations are located in the pth gene region. This
result was further confirmed by sequence analy-
sis; the mutations are base substitutions within a
translational open reading frame that corre-
sponds to the pth gene. Pth, the gene product,
was isolated and characterized chemically and
enzymatically.
Pth hydrolyzes peptidyl-tRNAs to yield free
tRNAs and peptides. It has been proposed that the
enzyme is a scavenger of peptidyl-tRNAs that have
dropped off the ribosomes during editing of mis-
incorporated amino acids in polypeptide chains.
The Pth function is essential for the cell, as in-
ferred from the fact that a heat-sensitive mutant of
Pth accumulates peptidyl-tRNAs and stops pro-
tein synthesis upon shift to the nonpermissive
temperature. Moreover, the enzyme is ubiquitous
among organisms from bacteria to mammals.
The possible role of Pth in ribosome-bound hy-
drolysis of peptidyl-tRNA at the step of polypep-
tide chain termination has not been supported by
the work of others. Preliminary results, however,
indicate that Pth may be involved in polypeptide
chain termination (see below), and we aim to
investigate this possible participation.
Nature of the X Regulator
We have isolated phage mutants that overcome
the bacterial Pth defect. These mutations defined
several genetic sites named bar. DNA sequence
analysis of two of these, barl and barll, revealed
that the mutations affect nearly identical l6-bp
segments having dyad symmetry.
The inhibition of phage development by mu-
tants defective in Pth requires transcription
through wild-type X bar regions. Transcripts
themselves, not polypeptides, seem to be the ac-
tive molecules. Aplasmid system in which short X
bar sequences were cloned in front of an active
promoter somehow mimics the X inhibition
effect.
Transcription of wild-type bar in plasmids is
lethal to Pth-defective (but not wild-type) cells.
This effect is specific, because constructs carry-
ing mutant bar sequences are not lethal. Tran-
scription through a vector harboring a synthetic
nucleotide sequence as short as 20 bp mimicking
barl caused pth mutant lethality; therefore we
think that the 1 6-bp bar sequence is the core of
the inhibitory transcripts. Protein synthesis is
shut off soon after bar transcript induction, but
RNA synthesis continues for several hours. Thus
the lethal effect is probably caused by a rapid
inhibition of protein synthesis.
How does bar RNA inhibit protein synthesis?
Preliminary data on nonsense codon-specific
suppression, obtained in collaboration with
Emanuel Murgola (M. D. Anderson Cancer Cen-
ter, Houston), have led us to propose a unifying
model that implicates foar RNAand Pth in peptide
chain termination. The model postulates, first,
that bar RNA can interfere with the termination
of UGA-mediated translation by antiparallel base-
pairing with ribosomal 1 6S RNA, and second, that
mutant Pth causes a defect in polypeptide termi-
nation facilitating bar RNA- 1 6S RNA interaction.
501
Control of Bacterial Protein Synthesis During Viral Infection
We are attempting to test this model through a
direct assay of bar RNA imeraction with cell com-
ponents and the effect of Pth on polypeptide
chain termination. We will also investigate possi-
ble genes and gene products that interact with
Pth through the isolation of second-site muta-
tions that suppress pth defect.
Biochemistry of Pth and Regulation
of pth Expression
To purify Pth protein, we have exploited the
fact that cells harboring pth plasmids overpro-
duce the wild-type enzyme. Part of our current
efforts are devoted to overproducing and purify-
ing mutant enzymes to compare their biochemi-
cal properties. Mutant Pths may have different
patterns of specificity for various amino acyl-
tRNAs. Also our laboratory is working on the con-
trol of pth expression by analyzing the transcrip-
tional and translational properties of the gene.
What Does bar Do for X?
We have considered two possible roles for bar
regulation in X biology, taking into account the
two functions proposed for Pth. First, bar RNA
may control the relative levels of specific tRNAs
to fit the profile of codon usage in X. This func-
tion, perfectly tolerated in normal cells, may lead
to inhibition of phage development or to lethality
in cells defective for Pth. Second, bar RNA may
act on termination and/or initiation of polypep-
tide chains directed by phage transcripts. Among
X genes, UGA is the most frequent termination
codon (not UAA, as in the host), and in the X
genome it is not uncommon for a gene's termina-
tion codon to overlap partially the initiation co-
don of the next gene, producing the sequence
AUGA. The fact that the bar core RNA contains
the sequence AUGA, and the alleged interaction
of this sequence with ribosomal 1 6S RNA, suggest
a role for these overlapping genes in polypeptide
chain-termination (-initiation) events.
502
Molecular Genetics of Photosynthesis
and Carbon Assimilation in Plants
Luis R. Herrera-Estrella, Ph.D. — International Research Scholar
Dr. Herrera-Estrella is Professor and Head of the Department of Plant Genetic Engineering at the Center
for Research and Advanced Studies, National Polytechnic Institute, Irapuato. He received his
undergraduate degree as biochemical engineer from the National Polytechnic Institute in Mexico and his
Ph.D. degree from the State University of Ghent, Belgium, where his thesis advisors were Marc Van
Montagu and Jeff Schell. The subject of his thesis was the expression of foreign genes in plant cells. As a
postdoctoral researcher in Ghent, he studied the regulation of light- inducible plant genes and the
movement of proteins into the chloroplast. His honors include the Minoru and Ethel Tsutsui Distinguished
Graduate Research Award in Science from the New York Academy of Sciences and the Javed Husain Award
for Young Scientists from UNESCO.
PHOTOSYNTHESIS and carbon assimilation
are the most important biochemical and mo-
lecular events in the life cycle of higher plants
and, indeed, are key to the provision of nutrients
for the entire food chain. Solar energy is first col-
lected in the chloroplasts of photosynthetic tis-
sues, mainly of leaves, by light-harvesting anten-
nas composed of chlorophyll and protein
molecules. The collected energy is then used to
convert atmospheric carbon dioxide (CO2) into
triose phosphate molecules. These three-carbon
molecules proceed through a series of reactions,
called the Calvin-Benson cycle, that culminates
in the production of sugars from v^^hich all the
organic molecules required for plant life are
synthesized.
More specifically, triose phosphate molecules
are converted in the cytoplasm of photosynthe-
tic, or source, cells into sucrose, which is translo-
cated through the phloem to feed the nonphoto-
synthetic, or consumer, tissues (i.e., roots,
flowers, seeds, tubers). Assimilated carbon is
stored temporarily or permanently in the form of
starch in both source and consumer tissues. The
starch in seeds or tubers provides most of the car-
bon and energy for the germination and develop-
ment of new plants.
The light-dependent production of ATP and
NADPH, the reductive assimilation of CO2, and
sucrose and starch synthesis are interlinked and
interdependent. These processes must be coordi-
nated in vivo at both the biochemical and genetic
level (i.e., at the level of gene expression). The
balance between the efficiency of CO2 fixation,
sucrose translocation and uptake, and assimila-
tion of sucrose in consumer tissues plays a funda-
mental role in determining the productivity of
any given plant species. This balance is affected
by both genetic determinants of the individual
and its interaction with the environment.
Our laboratory is interested in studying the mo-
lecular events that control the biochemical pro-
cesses involved in carbon assimilation in plant
cells. One aspect that we are investigating is how
light regulates genes involved in photosynthesis.
Ribulose 1 ,5-bisphosphate carboxylase oxygen-
ase (RuBisCO) is a multimeric enzyme composed
of eight identical small subunits (SS) and eight
identical large subunits (LS) . In the so-called C3
plants, RuBisCO carries out the initial CO2 fixa-
tion step, utilizing the five-carbon sugar ribulose
bisphosphate to produce two three-carbon deriva-
tives (triose phosphate molecules). The gene
family encoding the SS is located in the nuclear
genome; the gene encoding the LS is located in
the plastid genome. How genes located in differ-
ent cellular compartments are coordinately regu-
lated (those for both SS and LS are regulated by
light and should produce the corresponding sub-
units in equimolar amounts) , and how the active
RuBisCO enzyme is assembled from its subunits,
are some of the questions we wish to answer.
In collaboration with June Simpson, we
showed previously that the 5'-flanking sequences
of genes encoding the small subunit of the RuBis-
CO {ss genes) and the chlorophyll «/fo-binding
proteins (cabSO gene) are responsible for the
light-inducible transcription of these genes. We
have also shown that enhancer and silencer ele-
ments are involved in the tissue-specific and
light-inducible expression of the cabSO gene.
The cis- and trans-acting elements involved in the
regulation of the cabSO gene are being analyzed.
From these studies a 7-base pair repeated ele-
ment within the 5'-flanking region of the cabSO
gene has been identified as the site of interaction
with a putative regulatory DNA-binding protein.
Many attempts to assemble RuBisCO in vitro
from its isolated subunits have failed, suggesting
that the plant cell in vivo provides other req-
uisite components. It has been suggested that
RuBisCO assembly requires the participation of
proteins that act as molecular chaperones to pro-
mote the correct interaction between the sub-
units, ensuring the assembly of the catalytically
active enzyme. Two 60-kDa proteins termed
503
Molecular Genetics of Photosynthesis and Carbon Assimilation in Plants
chaperonins (Cpn60) a and ^ have been sug-
gested as participating in RuBisCO assembly. Since
the availability of functional RuBisCO is important
for carbon assimilation, we have isolated two
cDNAs and a genomic clone for Cpn60 /?.
To study whether Cpn60 ^ is indeed involved
in RuBisCO assembly, and whether it is specific
for'RuBisCO or plays a more general role in pro-
tein assembly, we are currently generating trans-
genic plants containing a chimeric gene to pro-
duce an antisense RNA for Cpn60 /3. The RNA
should interact with the Cpn60 /3 mRNA and
arrest the production of Cpn60 /3 polypeptides,
thus producing transgenic plants that will allow
us to assess the role of Cpn60 (8 in RuBisCO assem-
bly. We are also constructing chimeric genes com-
posed of the promoter of a Cpn60 ^ gene and the
coding sequence of the /^-glucuronidase bacterial
reporter gene, to study the tissue-specific and en-
vironmental regulation of this gene in transgenic
plants.
Again, the main assimilates from photosyn-
thesis are triose phosphate molecules that are
converted into sucrose to be translocated to con-
sumer tissues. Several physiological and biochem-
ical studies indicate that sucrose 6-phosphate syn-
thase (SPS) is the limiting enzyme for this
conversion. To study the role of SPS in carbon
assimilation, and the interrelation of the genes
encoding this enzyme with photosynthetic genes,
we are currently isolating SPS cDNA clones, using
monospecific SPS antibodies and polymerase
chain reaction technology in collaboration with
Horacio Pontis from Mar del Plata-Argentina.
(The work on SPS is supported in part by a grant
from the Rockefeller Foundation.)
Opposite: Light microscopy of transverse stem
sections of transgenic tobacco plants harbor-
ing a chimeric gene in which the P-glucuroni-
dase-coding sequence is under control of the
chaperonin 60 0 promoter. The blue staining
indicates the cell- type- specific expression di-
rected by the promoter at the basal (A) and
apical (B) stem regions of transgenic plants.
Research and photograph by Eduardo Zaba-
leta in the laboratory of Luis Herrera-Estrella.
504
505
Expression of the mouse En- 2 gene in a band of cells across
the midbrain- hindbrain junction. The 10-day transgenic
embryo contains a lacZ reporter construct expressed from
En-2 DNA regulatory sequences. The cells expressing lacZ
( dark ) are visualized by histochemical staining of the
lacZ gene product.
From Sedivy, f., and foyner, A. 1992. Gene Targeting.
New York: W.H. Freeman, p 161.
506
Gene Pattern Expression in Early Embryogenesis
Alexandra L.Joyner, Ph.D. — International Research Scholar
Dr. Joyner is a Senior Scientist at the Samuel Lunenfeld Research Institute of Mount Sinai Hospital,
Toronto, and Associate Professor of Molecular and Medical Genetics at the University of Toronto. She
received her B.Sc. degree in zoology and her Ph.D. degree in medical biophysics from the University of
Toronto. She did postdoctoral work in mammalian development with Gail Martin at the University of
California, San Francisco.
THE establishment of the basic body plan re-
quires an intricate coordination of cell-cell
interactions that appear to be controlled largely
by the genetic program handed down from gener-
ation to generation in our DNA. Many of the genes
that run the program of pattern formation have
been identified in Drosophila and shown by mu-
tant analysis to regulate the development of em-
bryonic regions rather than the differentiation of
cell types. In keeping with this, many of these
genes are expressed early in embryogenesis in
spatially defined patterns.
The primary focus of research in my laboratory
has been to identify and study mouse homologues
of Drosophila pattern-formation genes. This
work has been based on the premise that a con-
servation of gene structure through evolution re-
flects a corresponding conservation of gene func-
tion. Work over the last few years in many
laboratories, including mine, has shown this to
be the case.
A second research project has involved devel-
oping and applying a new type of random screen,
called the gene trap, for genes expressed in a spa-
tially defined manner during early mouse
embryogenesis.
Mouse Homologues of the Drosophila
Gene engrailed
The fruit fly body is divided into a number of
repeated units referred to as segments, and the
engrailed (en) genes are known to be required
for proper development of the posterior half of
each segment. The en gene has characteristics of
a "switch" that can direct cells down a posterior,
as opposed to anterior, developmental pathway.
Molecular characterization of the gene's product
has shown it to be a transcription factor that binds
DNA through a motif called a homeodomain. We
have been studying En 1 and En-2, the mouse
homologues of the Drosophila en gene, to deter-
mine whether they also act as transcription fac-
tors controlling pattern formation.
One striking difference between these homolo-
gous mouse and fruit fly genes is that the fly en
gene is expressed during embryogenesis in 14
stripes (one for each segment) , whereas the ver-
tebrate En genes are first expressed in a single
band across the developing mid- and hindbrain
junction. En l and En-2 continue to have a spa-
tially defined expression pattern in the brain un-
til the cells begin to differentiate into particular
cell types. En expression then switches and be-
comes cell-type specific.
Many other mouse homologues of Drosophila
pattern-formation genes have also been analyzed.
A recurring theme is expression in spatially de-
fined patterns early in development, particularly
in the nervous system. This suggests that at least
part of the mechanism for laying down the basic
body plan, and especially for specifying different
regions of the nervous system along the anterior-
posterior axis, involves the coordinate expres-
sion of different sets of these developmental
switch genes.
One of our objectives has been to make trans-
genic mice that lack the En genes. To date we
have made mice that are deleted for the En-2
homeodomain-coding DNA sequences. These
mice do not show a major disruption of the whole
mid- and hindbrain region but show a distinct
disruption in the pattern of folds in the cerebel-
lum. We are now studying the developmental pro-
gression of the defect to determine the cellular
basis of the mutant phenotype. That the cerebel-
lum, uniquely among adult brain structures, ex-
presses En-2 in the absence of En-1 indicates a
functional redundancy between En-1 and En-2.
Thus both En genes must be deleted before their
function in development of the mid- and hind-
brain can be studied.
A second aspect of our En research is to iden-
tify other genes in the same genetic pathways as
En-1 and En-2. One of our approaches has been
to seek transcription factors that regulate En ex-
pression. As a first step, we have analyzed the
DNA sequences around the En-1 and En-2 genes
and have identified fragments that will direct ex-
pression of the /acZ reporter gene to the mid- and
hindbrain junction. These fragments can now be
further subdivided and used to identify the tran-
scription factors that bind these sequences and
regulate En expression.
507
Gene Pattern Expression in Early Embryogenesis
In addition, we are using these DNA transcrip-
tion regulators to express other pattern-formation
genes aberrantly in the mid- and hindbrain re-
gion. In this way we can test, for example,
whether other homeodomain-containing genes,
expressed in more-posterior regions than En, can
switch the developmental program of mid- and
hindbrain cells to more-posterior regions.
We are also interested in identifying genes that
are regulated by En. One approach we are taking
is to analyze mouse homologues of Drosophila
genes that are regulated by en. To this end, we
have cloned homologues of the Drosophila cubi-
tus interruptus (Ci) gene. The mammalian genes
are called Gli, since one member of the family
was found to be overexpressed in gliomas. We
can now analyze the expression of the Gli and
other potential £n-regulated genes in mice lack-
ing En- 2.
Screening for Genes Expressed in Spatially
Defined Patterns
With a view to identifying other types of genes
involved in pattern formation, we devised a
screen that takes advantage of mammalian gene
structure and mouse embryonic stem (ES) cells.
The screen involves randomly integrating a vec-
tor we refer to as a gene trap into the ES genome.
The vector has a splice acceptor site upstream of
the reporter gene lacZ. We reasoned that if the
vector integrated in the correct orientation into
the intron, then splicing of the gene would be
directed to the reporter gene. We have now con-
firmed this by cloning and characterizing a num-
ber of transcripts for /flcZ-endogenous gene
fusions.
Cells expressing lacZ can be revealed with a
histochemical stain. ES cells expressing the re-
porter are then reintroduced into a host embryo
for lacZ expression pattern analysis. We have
now analyzed the expression pattern of the
cloned endogenous genes from two lacZ inser-
tions and have shown that the lacZ pattern faith-
fully mimics that of the endogenous gene.
Finally, lacZ insertions that show a spatially
defined expression pattern can be transmitted
into transgenic mice and the insertions analyzed
for mutant phenotypes. Two out of three such
mice showed mutant defects. Thus we have dem-
onstrated the feasibility of the gene trap approach
and are now carrying out a larger screen for can-
didate pattern-formation genes.
All of this work is also funded by grants from
the Medical Research Council and the National
Cancer Institute of Canada, the National Insti-
tutes of Health, and Bristol-Myers Squibb.
508
Diagnostic Use of RNA Replication
in Infectious Diseases
Paul M. Lizardi, Ph.D. — International Research Scholar
Dr. Lizardi is Professor of Biochemistry at the Biotechnology Institute, National Autonomous University
of Mexico, Cuernavaca. He received his Ph.D. degree from the Rockefeller University and conducted
postdoctoral research in embryology with Donald Brown at the Carnegie Institution of Washington,
Baltimore. After serving as Associate Professor at Rockefeller as an Andrew Mellon Foundation fellow.
Dr. Lizardi spent a sabbatical year at Massachusetts General Hospital, Boston, and held a visiting
professorship in genetics at Harvard Medical School.
INFECTIOUS diseases are frequently managed
by health professionals without definitive
identification of the pathogen. Classical ap-
proaches to diagnosis, such as direct microscopic
observation, cultivation, or infection of suscepti-
ble hosts, are often too slow or cumbersome for
routine medical care. Modern laboratory tests
based on the use of antibodies provide effective
tools for diagnosis of a number of infectious dis-
eases, but most antibody tests are designed to de-
tect the presence of a host immune response
rather than the pathogen itself. Thus there is a
need for techniques permitting the rapid and reli-
able detection of infectious agents so that epide-
miological monitoring and patient management
can be more effective.
A direct way to detect a pathogen is to identify
its genetic material, which invariably contains
unique sequence patterns. However, the genetic
material is usually so minute in a biological sam-
ple that its detection presents an extraordinary
technical challenge. Molecular biologists have
been up to the task, and techniques developed
recently permit the generation of millions of cop-
ies of DNA or RNA segments in the test tube by a
process of exponential amplification. The best-
known amplification method is the polymerase
chain reaction (PCR) , in which DNA strands are
sequentially separated by heating and then cop-
ied with DNA polymerase. The PCR limit of de-
tection is about 50 molecules of target, which is
over 100,000 times as sensitive as a typical en-
zyme-linked immunoassay.
Amplified RNA Binary Probes
My laboratory is developing alternative ampli-
fication methods for the detection of RNA or DNA
in biological samples. The work is being carried
out in close collaboration with Fred Kramer at the
Public Health Research Institute in New York
City and Jack Szostak at Massachusetts General
Hospital in Boston. Our methods exploit several
interesting properties of RNA molecules:
• Relatively short molecules of RNA (15-30
nucleotides) have long been known to form
stable helical structures when bound to perfectly
complementary single strands of DNA or RNA.
The thermodynamic stability of these helical
complexes distinguishes them from similar com-
plexes containing mismatched bases. Hence one
can design an RNA probe that will bind very
strongly, and uniquely, to a segment of the DNA
or RNA of an infectious agent in a biological
sample.
• RNA probe molecules can be joined to other
RNA molecules in a target-dependent manner.
That is, joining will only take place if the mole-
cules are aligned on a complementary target
strand that serves as a guide for the joining event.
We catalyze the joining by an enzyme called a
ribozyme ligase, isolated in Jack Szostak's labora-
tory. The ligase is itself an RNA, derived from a
naturally occurring intron called group I, origi-
nally discovered in Tetrahymena thermophyla
by Thomas Cech (HHMI, University of Colorado
at Boulder) .
• A specific class of molecules known as repli-
catable RNAs can be produced exponentially
under natural conditions, so that millions of cop-
ies are generated in minutes. An enzyme called
RNA replicase catalyzes reactions in which RNA
single strands, parent and daughter, are forced
apart during synthesis, in contrast to DNA-depen-
dent reactions, in which the two strands remain
annealed. The best characterized of these en-
zymes is Q-beta replicase. We have shown that
replicatable RNAs, in the presence of this en-
zyme, can harbor RNA probe inserts without loss
of replicative efficiency.
We have devised schemes in which probe bind-
ing, RNA joining, and exponential replication of
the joined RNA are used in an assay to detect the
presence of a specific target sequence from an
infectious agent. Probes that are not joined are
not replicated. The signal in these assays is repli-
cated RNA, generated by joined probes and
readily quantitated by fluorescence staining. The
intensity of the signal is proportional to the num-
ber of targets present in the original sample.
However, a number of technical problems re-
main to be solved before optimal sensitivity and
specificity can be achieved in these assays. For
example, the ligation step involving the ribo-
509
Diagnostic Use of RNA Replication in Infectious Diseases
zyme ligase is still relatively inefficient. (About
25 percent of the ligation-competent binary
probe molecules are joined in a 1-hour incuba-
tion.) An actual assay would require a ligation
efficiency closer to 85 percent. Fortunately, re-
cent developments in RNA biochemistry suggest
a possible solution to this problem.
Directed Evolution of Ribozymes
We propose to develop more-efficient ribo-
zyme ligases for our assays by directed evolution
in vitro. Rachel Green, in Jack Szostak's labora-
tory, has selected novel ribozymes with altered
catalytic efficiency by Darwinian selection in a
test tube. A Darwinian selection experiment be-
gins with the synthesis of a large number of DNA
molecules, each containing the sequence of a ri-
bozyme. Known as a mutant pool, the population
of molecules contains point mutations generated
during chemical synthesis. Each molecule con-
tains just a few mutations, located at random po-
sitions in the sequence, and thus the population
contains over a trillion (10'^) variants. The en-
zyme T7 RNA polymerase is then used to generate
RNA copies of the DNA, creating over a trillion
dififerent mutant ribozymes. These are briefly in-
cubated under RNA ligation conditions, where
they are given the opportunity to catalyze a chem-
ical reaction that joins them to a special piece
of RNA.
Those ribozymes that succeed in carrying out
the ligation reaction are chosen as survivors in a
subsequent step, while those that fail to partici-
pate in catalysis are lost by dilution. Survivors are
allowed to increase in number by reverse tran-
scription (which converts RNA into DNA), fol-
lowed by PGR amplification of the DNA. This se-
ries of steps effectively rewards all competent,
efficient ribozyme mutants by reconverting them
into DNA and copying them many times.
At this point the DNA is transcribed again into
RNA, and a second cycle of Darwinian selection
begins, exactly as above. After four cycles of se-
lection, a relatively small number of mutant ribo-
zyme sequences will be present in the DNA popu-
lation, instead of the original millions. These
molecules are the fittest: they were able to cata-
lyze ligation reactions in all four rounds of selec-
tion. The molecules are then sequenced in order
to compare them to the original parental
ribozyme.
Using the methods outlined above, Rachel
Green isolated a number of interesting mutant
ribozymes that display high efficiency in a spe-
cific type of ligation reaction. The experiments
used to generate these ribozymes mimic natural
variation and selection. While it takes nature
hundreds, thousands, or even millions of years to
evolve better enzyme catalysts, RNA enzymes can
be improved in the laboratory by a Darwinian se-
lection process in a matter of months. The term in
vitro genetics has been coined to describe these
man-made selection schemes. We are working on
improvements that should enable us to generate
ribozyme ligases better suited to our assays, in
order to improve the process of joining RNA bi-
nary probes.
Multiplexed Diagnostic Assays
Epidemiological monitoring of infectious dis-
eases is a complex undertaking in a developing
country like Mexico. Diseases must be monitored
both in large, modern urban settings, like Mexico
City, and in sparsely populated, rural areas like
the Ghiapas countryside. The list of diseases caus-
ing significant mortality and morbidity includes
AIDS, malaria, amebiasis, tuberculosis, hepatitis,
typhoid fever, diverse intestinal infections, and
more recently cholera. There is a need for devel-
oping accurate, low-cost technologies to detect
pathogens in the field, and also a need for ex-
tremely efficient, high-tech tools for the epidemi-
ological laboratory and the blood supply centers.
A valuable epidemiological tool would be a diag-
nostic assay capable of detecting any of a number
of pathogens in a single clinical sample. Such a
tool is known as a multiplex assay.
Under the auspices of the Rockefeller Founda-
tion and the World Health Organization, our insti-
tute has established a collaborative research pro-
gram with Stanford University School of Medicine
to develop and implement state-of-the-art tech-
niques for epidemiological assessment in Mex-
ico. The Stanford group is headed by Gary
Schoolnik from the Division of Geographic Medi-
cine and HHMI. Using a multiplex PGR assay,
they have conducted a series of experiments dem-
onstrating the simultaneous detection of any of
several bacterial pathogens in diarrheal stool.
Assays based on RNA amplification could be
multiplexed by using a combination of binary
probes. Each binary probe pair would contain
RNA sequences designed for binding to an indi-
vidual pathogen, and as many as 10 probe pairs
could be mixed together in a single assay. Eventu-
ally multiplexed assays could be automated, and
cost-effectiveness would be very favorable be-
cause several candidate infectious agents would
be assayed at once.
510
Lineage-Specific Gene Expression
in Caenorhabditis elegans
James D. McGhee, Ph.D. — International Research Scholar
Dr. McGhee is Professor in the Department of Medical Biochemistry at the University of Calgary, Alberta.
He received his B.Sc. degree in physiology and biochemistry from the University of Toronto and his Ph.D.
in molecular biology from the University of Oregon. His postdoctoral research was done in the laboratory
of Gary Felsenfeld at the National Institutes of Health. He is a Medical Scientist of the Alberta Heritage
Foundation for Medical Research.
ONE of the most important problems in devel-
opmental biology is to understand the mech-
anism of lineage-specific gene expression. That
is, how does a developing embryo manage to ex-
press a particular gene in one cell or cell lineage
and not in the many other cells of its body? Fur-
thermore, how does it tell time, in order to ex-
press this gene only at the correct point in
development?
We are approaching these problems in the sim-
ple nematode, or roundworm, Caenorhabditis
elegans. The reasons for choosing such a simple
animal are that it comprises only about a thou-
sand cells and, more importantly, that the divi-
sion pattern and cell lineage of every one of these
is known. We are studying the expression of
genes in the C. elegans intestine, since this is the
simplest lineage available.
When the developing embryo has only eight
cells, one of these (called the E cell) is the pro-
genitor of the animal's entire intestine. As a
marker for biochemical differentiation of the gut,
we have characterized a simple hydrolytic en-
zyme, a nonspecific carboxylesterase. As shown
in the figure, the gene coding for this esterase
(called ges-1, standing for gut esterase) is only
expressed in the developing intestine. Presum-
ably the natural function of the enzyme has some-
thing to do with the worm's digestion, but mu-
tants that do not express esterase appear to be
perfectly viable.
The ges-1 gene is transcribed from the genome
of the embryo at a point in development when
the embryo has only 100-200 cells and the devel-
oping gut lineage consists of only four cells. Mi-
cromanipulation experiments have shown that
ges-1 is expressed normally in embryos in which
non-gut cells have been removed or destroyed. In
other words, we have no evidence for any interac-
tion between gut and non-gut cells. An intriguing
feature of ges-1 expression is that it is completely
dependent on DNA synthesis during the cell cycle
in which the embryo has a total of eight cells, i.e.,
the cell cycle just after the gut has been clonally
established.
The problem of understanding intestinal-
specific expression of the ges-1 gene is twofold.
One must define regions in the DNA to which
transcription factors bind, and then identify and
characterize the transcription factors that bind to
these regions. To address the first part of the
problem, we have injected the cloned DNA from
the ges-1 gene back into a ges-i-nonexpressing
mutant and have been able to reconstitute accu-
rate gut-specific expression. We have used this
transformation assay to identify sequences that
appear to be necessary for correct gut expression.
We have also found regions, however, whose
deletion causes the ges-1 gene to be expressed not
in the gut but in specific sets of cells in the phar-
ynx, in the body wall musculature, and in the
hypodermis. These cells belong to either the sis-
ter or cousin lineage of the gut.
The simplest model to explain these ectopic
staining patterns invokes lineage-specific repres-
sors that would normally keep the ges-1 gene si-
lent. When the repressor-binding site is deleted,
the normally silent gene can be expressed. We are
testing this model by attempting to identify short
discrete sequences that cause lineage-specific re-
pression of a gene that is otherwise ubiquitously
expressed. We are also attempting to provoke ec-
topic expression of the ges-1 gene by injecting
large amounts of putative repressor-binding sites
as a competitor.
There is a curious and unexpected feature of
these ectopic ges-1 staining patterns. With con-
structs that produce staining in pharynx, mus-
cles, and hypodermis, the three staining patterns
are by and large exclusive: individual embryos
show only one of the patterns, and combinations
of patterns are rare. Whatever the molecular ex-
planation turns out to be, the phenomenon sug-
gests that the embryo is making a decision very
early in development and stably propagating this
decision thereafter. Indeed, this decision must ac-
tually be made at or before the two-cell stage of
the embryo, since this is the last point when the
different lineages are common.
We are also working on the second part of the
problem of lineage-specific gene expression,
namely to identify the protein factors that bind to
511
Lineage-Specific Gene Expression in Caenorhabditis elegans
the ges-1 gene and control its expression only in
the gut and only at the correct embryonic stage.
Here one of the few limitations of C. elegans as an
experimental organism becomes apparent: it is
essentially impossible to obtain large amounts of
synchronized embryos on which to do biochemi-
cal experiments. We are taking two routes past
this'obstacle.
The first route is to turn part of our efforts to
Ascaris suum, the well-known parasite of pig in-
testines. Ascaris adults are millions of times
larger than adult C. elegans, yet their embryos are
remarkably similar. Furthermore, Ascaris fe-
males produce hundreds of thousands of ferti-
lized eggs per day, which develop more-or-less
synchronously. We hope to use this system to
watch the arrival and departure of transcription
factors during early nematode development.
The second route is based on our finding that
C. elegans embryos can be blocked in mid-
development by exposure to the chemotherapeu-
tic agent fluorodeoxyuridine. This has allowed us
to produce nuclear extracts from these blocked
embryos and to detect a number of protein factors
(possibly as many as a dozen) that bind to the
5'-flanking sequence of the ges-1 gene. We have
concentrated on two such factors.
The first factor binds to a DNA sequence that
we have tentatively identified as a gut "activa-
tor." The protein is not detectable in oocyte cyto-
plasm and thus must be produced (transcribed or
translated) sometime during early development.
The second factor binds to a sequence that is com-
pletely conserved between C. elegans and the re-
lated nematode Caenorhabditis briggsae. Un-
like the first factor, this second factor is present
in the oocyte cytoplasm and must migrate at some
point in early development into the embryonic
nuclei. The second factor would thus be a candi-
date for a classic cytoplasmic "determinant" pro-
duced by the mother worm and then segregated
into the gut lineage during early cell cycles.
If we can indeed describe the constellation of
transcription factors responsible for activating
the ges-1 gene in the developing gut, will we re-
ally be much further ahead in understanding its
control? In some respects, all this work will only
move the problem one step further back in devel-
opment, to the question of how the transcription
factors themselves got where they did. However,
one of the main advantages of ges-1 as an experi-
mental system is that we only have 3-5 cell cy-
cles through which to regress, until we reach the
oocyte. Thus it should be possible to describe the
complete molecular logic that controls this sim-
ple gene.
This work was supported by the Medical Re-
search Council of Canada and by the Alberta Heri-
tage Foundation for Medical Research.
A newly hatched larva of the nematode Caeno-
rhabditis elegans. The dark red precipitate re-
flects the esterase activity of the ges- 1 gene, ex-
pressed in all 20 cells of the gut.
From Edgar, L.G., and McGhee, J.D. 1986.
DevBiol 114:109-118.
512
Cytokine Regulation of Effector Functions
in Immune Responses
Tim R. Mosmann, Ph.D. — International Research Scholar
Dr. Mosmann is Professor and Chair of the Department of Immunology at the University of Alberta,
Edmonton. He received his Ph.D. degree in microbiology at the University of British Columbia, Vancouver,
and obtained postdoctoral training in Toronto and Glasgow as a fellow of the Medical Research Council.
After four years as Assistant Professor in Immunology at Alberta, he spent eight years as a senior scientist
at DNAX Research Institute in Palo Alto. He then returned to the University of Alberta.
AFTER infection the host's immune system
must respond by specifically recognizing
the invading organism and inducing the effector
mechanisms that will most effectively destroy the
pathogen. Determination of the correct antigen-
specific recognition structures is mediated by the
process of clonal selection, whereby B or T cells
are selectively activated if they already express
receptors specific for the infecting agent. Choice
of the appropriate effector mechanism is deter-
mined to a large degree by T cells and the cyto-
kines they secrete.
The immune system can attack infectious
agents by a number of means. Antibody is highly
effective at neutralizing toxins and free virus and
at coating bacteria to enhance their recognition
and destruction by phagocytic cells. There are
also various cell-mediated cytotoxic mecha-
nisms, such as killer T cells, macrophages, and
granulocytes, that are most useful against intra-
cellular infectious agents. Cytokines secreted by
activated T cells regulate the accumulation and
activation of these cells at sites of infection, and
some cytokines also have direct cytotoxic
functions.
Recruitment and activation of macrophages
and granulocytes occur strongly during a delayed-
type hypersensitivity (DTH) reaction, which pre-
sents an effective response against intracellular
infections. In contrast, antibody responses are
normally more effective against extracellular
pathogens. It has long been known that antibody
and DTH responses are often "either-or" re-
sponses of the immune system, although the
mechanism of this reciprocal regulation has only
been partially resolved.
Several years ago we and others described two
types of T cell that secrete very different patterns
of cytokines. THl cells produce interIeukin-2
(IL-2), interferon-7 (IFN-7), and lymphotoxin
(LT), whereas only TH2 cells produce IL-4, IL-5,
IL-6, IL-10, and P6OO. These cytokines have mul-
tiple and profound effects on various aspects of
the immune response, and so THl and TH2 cells
have markedly different functions.
TH2 cells induce antibody production by B
cells. In particular, a strong TH2 response is asso-
ciated with high production of IgE, the antibody
that causes allergy. Although THl cells can also
induce antibody production under some circum-
stances, these cells are much more effective at
inducing DTH. Thus the choice of THl or TH2
responses influences the balance between DTH
and antibody responses.
Although the THl and TH2 patterns account
for a major part of the immune response during
certain infections, these probably represent ex-
tremes, and there are other T cells with different
cytokine secretion phenotypes. These other pat-
terns may be involved in the many possible types
of immune response that occur against different
infectious agents.
Since the type of immune response induced
against a particular pathogen is often characteris-
tic for that agent, the choice of effector function
must be precisely regulated. It appears that cross-
regulation by TH 1 and TH2 cytokines plays a ma-
jor role in determining the effector functions.
Some of this cross-regulation is now understood:
IFN-7 produced by THl cells can inhibit prolifer-
ation of TH2 cells, and IL-10 produced by TH2
cells inhibits the activation of THl cells.
Effector T cells such as TH 1 and TH2 are proba-
bly derived from precursor cells that secrete only
IL-2. After initial stimulation with antigen, these
cells presumably differentiate into THl, TH2,
and other cells. This process is influenced by cy-
tokines— e.g., IL-4 encourages the production of
more cells that secrete IL-4. However, the full
regulation of this process is largely unknown, and
this is currently an active area of investigation in
our laboratory and others. Since the differentia-
tion process is quite rapid (e.g., a few days), we
are developing single-cell cytokine detection
methods so that we can follow differentiation of
single clones of cells.
Two methods are currently possible: the mRNA
for different cytokines can be measured in single
cells after amplification by polymerase chain re-
action, and cytokine protein secreted by single
cells can be analyzed by a colorimetric spot assay
using anticytokine monoclonal antibodies. Both
513
Cytokine Regulation of Effector Functions in Immune Responses
methods are feasible for detecting single cyto-
kines produced by one cell, and we are currently
adapting them to simultaneous measurement of
multiple cytokines. Since there is considerable
evidence of several different cytokine secretion
phenotypes, 5-10 cytokines must be analyzed
from each cell to obtain a true picture of T cell
diversity. Once developed, this methodology
will be applied to the differentiation of T cells
during normal immune responses, allowing the
signals that influence differentiation to be
determined.
The characteristic cytokines of THl and TH2
are major determinants of the function of these
cells. For example, IL-4, IL-5, and IL-10, pro-
duced by TH2 cells, contribute to the production
of IgE, mast cells, and eosinophils — all asso-
ciated with the development of allergy. IFN-7 is a
major contributor to DTH reactions, inhibits
many of the functions of IL-4, and inhibits TH2
cells directly. We have been interested for some
time in additional cytokines of the two T cell
types.
Recently we have been working on P6OO, origi-
nally discovered as a clone in a cDNA library of
activated TH2 cells. P600 is particularly interest-
ing, since it is expressed only after activation and
is not produced by THl cells. The sequence of
the open reading frame in the P600 cDNA clone
encodes a small protein with a leader sequence
characteristic of secreted or membrane proteins.
Since all of these properties suggest that P600
may be an additional TH2-specific cytokine, we
decided to test for possible functions of the P6OO
protein.
We transfected the P6OO cDNA clone into mon-
key cells and observed secretion of a new protein
of the expected size. After screening in a wide
variety of biological assays, we found that P600
induces the production of large numbers of ad-
herent cells from bone marrow. The functions of
these cells are currently under investigation to
determine how the effects of P6OO fit into the
overall pattern of TH2 functions during immune
responses.
The different T cell cytokine secretion pheno-
types were originally discovered in the mouse,
and initially there was some doubt that similar
phenotypes exist in humans. However, recent
data on human T cell clones and immune re-
sponses have made it clear that the THl and TH2
patterns are also important in himian diseases. Ap-
parently several parasitic diseases strongly in-
duce TH 1 - or TH 2 -biased responses, and only one
of these responses is usually able to clear the in-
fectious agent. Recent data indicate that these
two cytokine patterns are also important in lep-
rosy. Thus a knowledge of the regulation of these
patterns will have considerable potential for the
design of better immunomodulatory treatments
and vaccines.
514
Cellular and Molecular Basis of Variability
in Entamoeba histolytica
Esther Orozco, Ph.D. — International Research Scholar
Dr. Orozco is Professor of Genetics and Molecular Biology in the Department of Experimental Pathology at
the Center for Research and Advanced Studies, National Polytechnic Institute, Mexico City. She received
her bachelor's degree in chemistry and biology from the University of Chihuahua, Mexico, and her Ph.D.
degree from the National Polytechnic Institute. She has been a visiting professor in several institutes
around the world, including Harvard School of Public Health in Boston and the Weizmann Institute of
Science in Israel. Dr. Orozco counts among her honors a Guggenheim Foundation Award and the
Dr. J. Rosenkranz 1991 Award given by Syntex.
MORE than 500 million people throughout
the world are infected by Entamoeba his-
tolytica, the protozoan parasite responsible for
human amebiasis. Ninety percent of those in-
fected do not present clinical symptoms, while
the rest develop colitis, dysentery, or hepatic
abscesses.
In 1925 Brumpt proposed the possible exis-
tence of two E. histolytica species, one patho-
genic (P) and the other nonpathogenic (NP), and
he and others have continued to advance this pos-
sibility. There are data both supporting and op-
posing this concept. For instance, migration pat-
terns of several enzymes in a high number of
amebic isolates has led to the establishment of
more than 20 E. histolytica zymodemes. Interest-
ingly, the zymodemes of P and NP trophozoites
differ.
According to the opinion of several groups, zy-
modemes and virulence are stable phenotypes.
However, experimental evidence against the sta-
bility of zymodemes and virulence has also been
documented. Resolution of this controversy is of
fundamental importance clinically. If E. histoly-
tica comprises two species, cyst passers spread-
ing the NP form are not a public health menace,
and asymptomatic carriers need not be treated.
However, if the harmless E. histolytica can turn
into a virulent one, a different decision may be
indicated. In any case, the molecular basis of con-
version, if it occurs, would be a highly interesting
phenomenon to study.
We have focused this controversy by 1 ) investi-
gating cloned E. histolytica populations after
their isolation from asymptomatic carriers through-
out the axenization process, 2) comparing E. his-
tolytica clones with diverse origins, and 3)
detecting and cloning DNA sequences present
only in E. histolytica clones that express
pathogenicity.
Study of NP E. histolytica Throughout
Axenization
Cloned trophozoites (MAV-1), with an initial
NP zymodeme I and isolated from an asymptom-
atic carrier, were cultured under polyxenic,
monoxenic, and axenic conditions. Zymodeme I
of polyxenic MAV-1 trophozoites, cultured in
Robinson's medium, presented an NP zymodeme
XII when trophozoites were cultured in Jones'
medium. Monoxenic trophozoites grown in the
presence of Fusobacterium symbiosum re-
tained NP zymodeme I (with slow-running hexo-
kinases [HKs]), an a-phosphoglucomutase (PGM)
band, and lack of pathogenicity. Surprisingly,
when MAV-1 trophozoites were grown under
axenic conditions, the original NP zymodeme
switched to a P zymodeme, with fast-running HKs
and a /3 PGM. Virulence, defined as the tropho-
zoites' ability to infect experimental animals and
damage target cells, was also expressed.
The analysis of E. histolytica zymodemes indi-
cates that P as well as NP trophozoites have a set
of genes for a given enzyme. Excluding the fast-
running HKs and the /3 PGM that have not been
reported for NP isolates, all other isoenzymes are
expressed in P and NP trophozoites, including
slow-running HKs. A certain type of bacteria in
the medium could be participating in the regula-
tion of isoenzyme expression. We will continue
our studies on the factors involved in the expres-
sion of isoenzymes in P and NP trophozoites, in
order to look for conclusive evidence on zymo-
deme and virulence switching — evidence that
should dissipate skepticism about conversion.
Differences in Genetically Related Clones
Functional and biochemical dissimilarities
mark clones of diverse origin. P trophozoites
show a high rate of phagocytosis, destroy cell cul-
ture monolayers, and produce hepatic abscesses
in experimental animals. On the contrary, NP tro-
phozoites generally do not show these virulence-
related properties. Fortuitously, they express low
virulence.
Monoclonal and polyclonal antibodies against
certain antigens can distinguish between P and
NP trophozoites. The presence of the major ame-
bic cysteine protease involved in the cytopathic
effect of trophozoites is also specific to P tropho-
515
Cellular and Molecular Basis of Variability in Entamoeba histolytica
zoites. However, the so-called P trophozoites
form a heterogeneous group of E. histolytica iso-
lates, which show virulence in a broad spectrum.
This indicates that virulence-involved genes are
differentially regulated.
On the other hand, amebic strains are consti-
tuted by trophozoites displaying different pheno-
types. Clones from a given strain express viru-
lence in the same broad spectrum as amebic
strains. Some clones are avirulent (are they NP?);
others are highly virulent, and specific monoclo-
nal and polyclonal antibodies are able to discrimi-
nate between them.
The simplest interpretation for this is that
strains are a mixture of genetically unrelated tro-
phozoites living together in the human intestine.
However, clones derived from cloned popula-
tions differ phenotypically and genotypically.
The differences include virulence, antigens, and
genetic divergence. Clones isolated from clone A
(strain HM1:IMSS) show the P zymodeme but
differ in virulence and in the expression of the
1 12-kDa adhesin, involved in both phagocytosis
and virulence of E. histolytica.
Detection and Cloning of DNA Sequences
Present Only in Pathogenic Trophozoites
Several E. histolytica genes show high similar-
ity among P strains but differ from their homolo-
gous genes in NP strains. Molecular studies have
given the strongest evidence to suppon the exis-
tence of two E. histolytica species. To date, there
is no explanation for the genomic divergence
shown by P and NP trophozoites. However, reti-
cence to accept that there are two species based
on their polymorphism is generated by the re-
markable genomic plasticity of this parasite, even
among clones derived from cloned populations.
Genomic divergence is showed by strains and
clones cultured under different conditions. Ge-
netically related clones obtained from a cloned
population show polymorphism in several genes
of the same clone cultured in different laborato-
ries (clone A, strain HM1:IMSS). They also differ
in their molecular karyotype and in the chromo-
somal location of some genes, indicating genome
rearrangements. Finally, we have identified,
cloned, and sequenced a 1.6-kilobase DNA frag-
ment from clone A (virulent) that is altered
in nonvirulent (NP?) clones derived also from
clone A.
We are currently studying genetic mechanisms
underlying this intriguing phenomenon. One hy-
pothesis, highly speculative, is that there are
genes in E. histolytica that, under pressure, are
selectively rearranged, modified, or amplified
before being transcribed. If certain genes were
amplified in P trophozoites and not in NP, single-
copy or slightly modified genes could not be de-
tected in comparative experiments. If this were
true, it would explain, at least in part, the poly-
morphism found in P and NP strains.
The presence of "cassettes" carrying different
genes that could be amplified and expressed se-
lectively is another possibility. We are trying to
develop a genetic transformation system in E. his-
tolytica to introduce foreign genes as landmarks
to follow up the genome modifications that will
explain the erratic behavior of this parasite,
or at least the differences between P and NP
trophozoites.
This work is supported by a grant from the Na-
tional Council for Science and Technology,
Mexico.
516
Phosphorylation and Protein-Protein Interactions
in Signal Transduction
Tony Pawson, Ph.D. — International Research Scholar
Dr. Pawson is Senior Scientist at the Samuel Lunenfeld Research Institute of Mount Sinai Hospital,
Toronto, where he holds the Apotex Chair of Molecular Oncology; Professor of Molecular and Medical
Genetics at the University of Toronto, and a Terry Fox Cancer Research Scientist of the National Cancer
Institute of Canada. He received his Ph.D. degree from the University of London while working at the
Imperial Cancer Research Fund. Prior to moving to Toronto, Dr. Pawson did postdoctoral work at the
University of California, Berkeley, and was a faculty member of the University of British Columbia.
MANY of the polypeptide hormones that con-
trol development in the embryo and cell
growth, differentiation, and metabolism in the
adult bind to the extracellular region of receptors
that span the plasma membrane. These receptors,
in their cytoplasmic region, contain a protein ki-
nase domain that phosphorylates tyrosine resi-
dues. Normally these catalytic receptors are only
active when engaged by the appropriate growth
factor. However, structural alterations, resulting
from cancer-causing mutations in the relevant
genes, can render them active even in the absence
of a growth factor. Such aberrantly active recep-
tors can trigger unrestricted cancerous cell
Activated growth factor receptors can elicit
many changes in stimulated cells, including
changes in gene expression, cellular architec-
ture, cell-cell communications, and cellular me-
tabolism. An important question involves the
identities of the intracellular targets of receptor
tyrosine kinases — targets that control signal
transduction pathways within the cell.
SH2 -containing Proteins Bind Activated
Growth Factor Receptors
The first thing growth factor receptors do upon
activation, apparently, is phosphorylate them-
selves on tyrosine. Receptor autophosphoryla-
tion serves as a switch to elicit the tight binding
of intracellular signaling proteins to the receptor.
These receptor-binding proteins include Ras
GTPase-activating protein (GAP), phospholipase
C-7, phosphatidylinositol (PI) 3'-kinase, and Src-
family cytoplasmic tyrosine kinases. They have
the attributes expected of signaling proteins, in
the sense that they regulate specific intracellular
signal transduction pathways and in some cases
are known to be phosphorylated, and activated,
by receptor tyrosine kinases. How are these func-
tionally diverse signaling proteins all able to rec-
ognize autophosphorylated receptors? They each
contain one or two copies of a protein domain,
the Src homology 2 (SH2) domain, which recog-
nizes specific autophosphorylated sites. Thus ty-
rosine phosphorylation of the receptor itself
creates a tight binding site for SH2 domains and
promotes a strong interaction between the acti-
vated receptor and its target SH2-containing
proteins.
There is evidently some specificity to these
SH2-receptor interactions. Different SH2 do-
mains bind tenaciously to distinct phosphory-
lated sites on a receptor, suggesting that the
amino acid sequence surrounding the phospho-
tyrosine residue is important in determining tar-
get specificity.
A Network of SH2 -mediated Interaction
An increasing number of SH2-containing pro-
teins that can bind to activated growth factor re-
ceptors have been identified. These are all likely
to regulate some aspect of intracellular signaling,
though in several cases the relevant biochemical
pathways have not been defined.
The association of SH2 domains with auto-
phosphorylated growTih factor receptors is one in-
dication of their general ability to bind
tyrosine-phosphorylated sites. As an example,
considerable data suggest that the SH2 domain of
the c-Src cytoplasmic tyrosine kinase interacts
with an inhibitory tyrosine phosphorylated site in
its own carboxyl-terminal tail, thereby repressing
c-Src tyrosine kinase activity. However, once
c-Src is enzymatically (and oncogenically) acti-
vated by removal of the carboxyl-terminal phos-
phorylation site, the SH2 domain has a positive
transforming effect, probably by virtue of its abil-
ity to bind substrates of the kinase domain and
participate in the formation of signaling com-
plexes. Thus the c-Src SH2 domain has multiple
functions. GAP is another example of a signaling
protein whose SH2 domains can interact with
several different phosphotyrosine-containing
proteins.
We are pursuing the structure and function of
SH2 domains, using a variety of biochemical, cel-
lular, and genetic approaches.
517
Phosphorylation and Protein-Protein Interactions in Signal Transduction
Receptor Tyrosine Kinases
in Embryonic Development
We have recently cloned cDNAs for two novel
receptor tyrosine kinases, elk and nuk, which are
primarily expressed in the mammalian nervous
system. During formation of the nervous system
in the early mouse embryo, nuk assumes a very
specific and interesting pattern of expression. We
are using genetic manipulation in the mouse to
determine whether nuk is involved in control-
ling neural differentiation.
Dual-Specificity Protein Kinases
Protein kinases are typically divided into those
that can phosphorylate serine/threonine residues
and those that phosphorylate tyrosine. We have
recently identified two mammalian protein ki-
nases that can phosphorylate both serine/threo-
nine and tyrosine when artificially expressed in
bacteria. We are investigating the significance of
this observation for mammalian cells.
One of these kinases, Clk, is related to
members of the cdc2 protein kinase family. The
second. Nek, is similar to the product of the
nimA gene, required for initiation of mitosis in
the fungus Aspergillus nidulans. The identifica-
tion of nek as a mouse relative of nimA suggests
that this gene may have been conserved in evolu-
tion, as have other members of the cell cycle ma-
chinery. These observations prompt the specula-
tion that control of the cell cycle involves protein
kinases with dual specificity.
Additional funding for this work was provided
by the National Cancer Institute of Canada and
the Medical Research Council of Canada.
518
Chemical and Functional Characterization
of Scorpion Toxins
Lourival Domingos Possani, Ph.D. — International Research Scholar
Dr. Possani is Professor and Chairman of the Department of Biochemistry at the Biotechnology Institute,
National Autonomous University of Mexico, Cuernavaca. He received his B.S. degree in natural history
from the Universidade Federal do Rio Grande do Sul, Porto Alegre, Brazil, and his Ph.D. degree in
molecular biophysics from the University of Paris, France. He completed the academic requirements for
his Ph.D. degree in biochemistry at the Rockefeller University, working with Edward Reich on the isolation
of the acetylcholine receptor.
SCORPIONISM is a serious public health prob-
lem in certain areas of the world. In Mexico
alone, more than 200,000 people are stung by
scorpions each year. Thanks to serotherapy with
horse antisera, the number of deaths is kept to
approximately 700-800 annually. Not all scor-
pions are dangerous to humans. Among the 136
species known to live in Mexico, only 6 present a
real risk. They belong to the family Buthidae,
genus Centruroides. The species dangerous to
humans are Centruroides elegans, C. infamatus
infamatus, C. limpidus limpidus, C. limpidus
tecomanus, C. noxius, C. sculpturatus (which
occurs also in Arizona), and C. suffusus suffusus.
Half of the casualties are due to one species,
represented by two subspecies: C. limpidus lim-
pidus from the states of Guerrero and Morelos,
and C. limpidus tecomanus from the state of Co-
lima. All dangerous species abide in the Pacific
coast region. No dangerous scorpions are found
in the Gulf of Mexico region, the Yucatan Penin-
sula, or Mexico City.
Envenomation by the sting of these arachnids is
due to low-molecular-weight peptides called
toxins. More than 1 00 such toxins have been puri-
fied and characterized. There are three well-
known classes of scorpion toxins: short-chain
peptides (38-39 amino acid residues long),
which are K"^ channel blockers; medium-chain
peptides (61-66 amino acid residues), specific
for Na^ channels; and long-chain peptides (about
70 amino acid residues), toxic to insects and
crustaceans. Recently a fourth class seems to have
emerged with the discovery of low-molecular-
weight proteins (about 1 20 amino acid residues)
that affect the Ca^^ release channel from sarco-
plasmic reticulum.
The study of scorpion toxins is scientifically,
medically, and biotechnologically interesting.
They represent several families of structurally re-
lated polypeptides with exquisite preference for
ion channel molecules, which makes them a
model for future new drugs to control cell excit-
ability or for new types of specific insecticides.
And they represent excellent tools for studying
cellular responses, permitting discrimination
among receptor molecules in the membranes and
investigation of various types of ion channels.
The contribution of our research group, which
comprises a multidisciplinary interaction with
investigators from several countries, began with
the isolation and chemical characterization of
toxin gamma, a potent Na^ channel blocker, from
the Brazilian scorpion Tityus serrulatus. Our
work was extended at the beginning of the 1 980s
by the discovery of the first peptide capable of
blocking the delayed rectifier channel from
the squid axon. It is a short-chain peptide (39
amino acid residues) and was subsequently
shown to affect other types of channels. This
peptide was named noxiustoxin, after the scor-
pion C. noxius.
Our studies continued with the purification
and characterization of more than 25 different
peptides from scorpion venoms, specific block-
ers of Na^ channels from a variety of different
tissues. At least eight additional small-chain pep-
tides similar to noxiustoxin were also isolated
and characterized. To study the function of these
toxins, we used neurotransmitter-release and
-binding studies with brain synaptosomes and
mainly electrophysiological studies with a vari-
ety of excitable cells. Our more recent progress
in this field is focused on the three lines of re-
search mentioned below.
Chemical Synthesis of Peptides
and Monoclonal Antibodies
We have recently synthesized, by the solid-
phase method of Merrifield, more than 1 00 dif-
ferent peptides corresponding to segments of the
amino acid sequence of the scorpion toxins.
These peptides and the preparation of a dozen
distinct monoclonal antibodies allowed us to
probe for specific structural regions of the toxins.
For example, the nonapeptide at the amino-ter-
minal region of noxiustoxin was shown to recog-
nize and affect channels in a manner similar to
the native peptide. These results are promising in
the context of possible development of new
drugs aimed at controlling cellular excitability
through channels.
519
Chemical and Functional Characterization of Scorpion Toxins
Using monoclonal antibodies, we found four
main distinct antigenic determinants in the Na"'"
channel target toxins of eight different species of
scorpion from the American continent. One of
the antibodies was shown to produce an impor-
tant protection to experimental animals enveno-
mated with highly purified toxin. This result is
also important for developing possible synthetic
peptides for vaccination.
Cloning of Scorpion Toxin Genes
The most recent advance of our laboratory, in
collaboration with Francisco Bolivar and Baltazar
Becerril, was the successful isolation of several
clones that code for peptides corresponding to
the amino acid sequence of scorpion toxins. A
cDNA library was prepared from the telsons of the
scorpions C. noxius and C. limpidus limpidus.
In the future we hope to use directed mutagene-
sis to obtain specific modified toxins in order to
study the structure-function relationship of these
ligands with specific ion channels.
Search for New Toxins
New, unknown peptides toxic to insects and
crustaceans were purified and are being se-
quenced. These are from several species of scor-
pion, including some not dangerous to humans.
Also, in collaboration with Hector Valdivia and
Roberto Coronado from Madison University, we
are conducting a search for specific toxins that
target the Ca^"'" release channel from the sarco-
plasmic reticulum.
This research is supported in part by grants
from the Consejo Nacional de Ciencia y Techno-
logia, Mexico, and Direccion General del Per-
sonal Academico, from the National Autonomous
University of Mexico.
Venom extraction by electrical stimulation of a scorpion, species Hadrurus concolorous.
Photograph by Manuel Varela and Lourival Possani.
520
Protein Crystallography in the Study
of Infectious Diseases
Randy J. Read, Ph.D. — International Research Scholar
Dr. Read is Assistant Professor of Medical Microbiology and Infectious Diseases and of Biochemistry at the
University of Alberta, Edmonton. He obtained his undergraduate and doctoral training in the Department
of Biochemistry at the University of Alberta, then pursued postdoctoral training in protein crystallography
with Wim Hoi at the University of Groningen in the Netherlands, before returning to Edmonton.
INFECTIOUS organisms — viruses, bacteria, and
parasites — must overcome many obstacles to
cause disease. They must gain entry to their fa-
vored niche in the host, obtain nutrients, evade
attacks of the immune system, and spread to new
hosts. These are complex problems with com-
plex and varied solutions. But the pathogenic
mechanisms used by microorganisms to infect
and cause disease are gradually coming to light,
in some cases at the level of the key molecules.
The interest in pathogenesis is more than aca-
demic, since the understanding of these pro-
cesses can be exploited to prevent or treat disease.
Our interest is in advancing the understanding
of infectious disease at the molecular level, using
the technique of x-ray crystallography to study
the three-dimensional structure of important
proteins. There are two major aspects to our
work. We are studying the crystal structures of a
number of proteins that are involved in pathogen-
esis, some of which are described below, and we
are developing methods to exploit this kind of
structural information in the design of new drugs.
Bacterial Toxins
Many pathogenic bacteria produce toxins that
cause cell and tissue damage and can be responsi-
ble for the most severe effects of the illness. Bacte-
rial toxins often belong to the A-B class, having a
two-part structure in which the B (binding) sub-
unit binds to the surface of a target cell and the A
(active) subunit enters the cell, carrying out the
toxic action. We are involved in studying the
crystal structures of pertussis toxin (in collabora-
tion with Glen Armstrong at the University of Al-
berta, and with Connaught Laboratories in To-
ronto) and verotoxin (in collaboration with
James Brunton, University of Toronto).
Pertussis toxin (PT) is produced by Bordetella
pertussis, the bacterium that causes whooping
cough. There is a major interest in the role of this
toxin in improved vaccines. Currently, killed
whole-cell vaccines are used for whooping
cough. They are effective, but have an undesir-
able level of toxicity. The side effects should be
reduced or eliminated in a defined vaccine pro-
duced from genetically engineered proteins. It has
been shown that PT is a necessary component of
effective whooping cough vaccines, but it must be
rendered nontoxic for safe use. A three-dimensional
structure of PT would help show how to remove its
toxicity while preserving the surface features of the
molecule recognized by the immune system. In ad-
dition, we would gain a better understanding of how
the B component of the toxin recognizes the surface
of target cells, how the A subunit enters the cell, and
how it carries out its toxic action. Exploiting the
intense radiation produced by synchrotrons, we
have collected x-ray data from crystals of this toxin
and from several potential heavy-atom derivatives,
but the determination of the phases needed for the
visualization of the structure is still in progress.
The Shiga toxin family is a group of closely re-
lated toxins produced by Shigella dysenteriae
type 1 (Shiga toxin) and by certain strains of Esch-
erichia coli (verotoxins, or VTs). Shiga toxin is
associated with bacterial dysentery, a serious
problem in many developing countries. The
strains of E. coli that produce VTs cause a disease
often referred to as "hamburger disease," be-
cause it can be acquired from contaminated ham-
burger. VTs can provoke the hemolytic uremic
syndrome and are thereby the major cause of
acute kidney failure in children.
We have crystallized and solved the structure
of the B subunit of VT-1 , a member of the Shiga
toxin family. The B subunit forms a pentamer that
recognizes target cell surfaces by binding to the
carbohydrate component of a cell-surface glyco-
lipid, globotriaosylceramide (Gbj). Since this in-
teraction largely determines which cells the
toxin will attack, we are interested in under-
standing its molecular details. Comparison of the
amino acid sequences of all the toxins in this fam-
ily has enabled us to predict that a surface cleft
between B-subunit monomers will prove respon-
sible for binding. Crystallographic binding stud-
ies should allow us to test this prediction.
The most surprising result is an unexpected
structural similarity to the B subunit of members
of the cholera toxin family. This is unexpected
because the associated A subunits of the two
toxin families are completely unrelated, the B
521
Protein Crystallography in the Study of Infectious Diseases
subunits are very different in size, and the degree
of amino acid sequence similarity is, if anything,
less than what one would expect for two random
sequences. Nonetheless, the structural similarity
implies some distant evolutionary relationship
between these families.
Pseudomonas aeruginosa Pilin
Pseudomonas aeruginosa is an opportunistic
pathogen that infects burn victims and immuno-
compromised patients. It is also one of the major
pathogens infecting the lungs of cystic fibrosis
patients. Filaments on its surface, termed pili, at-
tach to epithelial cell surfaces, promoting colon-
ization. Pili are formed from a helical array of
identical pilin subunits. We have crystallized pi-
lin from this organism (in collaboration with Wil-
liam Paranchych, University of Alberta) and are
attempting to improve the quality of the crystals
to allow us to determine its structure. The three-
dimensional structure would help us understand
the details of cell-surface binding, which could
be used to devise strategies to interfere with
colonization.
Computer-aided Drug Design
Most of the drugs in use today were discovered
by trial and error, and many of these have un-
known mechanisms of action. Those with known
mechanisms usually act by binding specifically to
a drug receptor, often a protein molecule. Since
crystallography allows us to examine the struc-
tures of receptors and the details of their interac-
tions with drugs, it should help in improving ex-
isting drugs or even in inventing new ones.
Progress has been made through crystallography
in the former objective, but designing new drugs
from scratch is still extremely difficult. We are
trying to address this problem with the aid of
computers.
To design a new drug, one must first choose an
appropriate receptor or target. In the design of an
antibacterial drug, the key factor is selective tox-
icity-, the drug should poison the pathogen but
not the patient. One might choose, as a drug tar-
get, an essential enzyme in a biochemical path-
way unique to the bacterium, because that would
minimize the chances of side reactions with host
enzymes. In our work on drug design, we are as-
suming that a good potential drug target has been
chosen and its crystal structure determined. The
problem, then, is how to exploit the structural
information. There are probably billions of com-
pounds that might be used as drugs; determining
which of these might bind to, and interfere with,
the target protein is far from trivial.
We have chosen a "divide and conquer" ap-
proach to drug design to reduce the magnitude of
this problem. The vast set of possible compounds
is made up of various combinations of a much
smaller set of molecular fragments. We propose
to design drugs by using a computer to dock
members of a library of fragments to the region of
the desired binding site, then to combine docked
fragments to form chemically sensible mole-
cules. We have tested the feasibility of fragment
docking, and the results are sufficiently promis-
ing that we will go on to test the feasibility of the
next step. Fragment docking requires the calcula-
tion of binding energies, which as yet can only be
approximate and requires a great deal of com-
puter time. But we believe that this method can
provide a useful tool for drug design in the not-
too-distant future, in view of the acceleration in
the speed of computers and the improvements
that are continually being made in the under-
standing of molecular interactions.
Our work is also supported by the Medical Re-
search Council of Canada and the Alberta Heri-
tage Foundation for Medical Research.
A schematic illustration of the crystal structure
of the B (binding) subunit of verotoxin-1,
viewed from the top of the pentamer. Arrows
indicate strands of j3-sheet, and cylinders indi-
cate ahelices.
From Stein, P.E., Boodhoo, A., Tyrrell, GJ.,
Brunton, f.L., and Read, R.J. 1992. Nature
355:748-750. Copyright © 1992 Macmillan
Magazines Limited.
522
Representations and Transformations of Tactile
Signals in Somatic and Frontal Motor Cortices
Ranulfo Romo, M.D., Ph.D. — International Research Scholar
Dr. Romo is Professor of Neurosctence at the Institute of Cellular Physiology, National Autonomous
University of Mexico, Mexico City. He received his M.D. degree from the National University of Mexico and
a Ph.D. degree in neuroscience from the University of Paris, France. His postdoctoral work was done with
Jacques Glowinski at the College of France in Paris, Wolfram Schultz at the University of Pribourg,
Switzerland, and Vernon B. Mountcastle at the Johns Hopkins University in Baltimore. Dr. Romo has
received the Demuth Prize of the Swiss Medical Research Foundation and a Guggenheim Memorial
Foundation fellowship.
PRIMATES have access to events occurring in
the external world through specialized sen-
sory systems. The events are first transduced by
the sensory receptors and encoded and transmit-
ted to the central nervous system by the primary
afferent fibers. These messages are orderly distrib-
uted and processed in brain centers, where the
external events are represented and, under some
conditions, lead to sensation, perception, mem-
ory, and purposeful motor acts. These phenom-
ena can only be studied in highly evolved brains.
Our laboratory is investigating the representation
of sensory signals in the brain and the mecha-
nisms by which the motor centers process them
in order to guide behavior.
The somatic sensory system of subhuman pri-
mates appears to be an appropriate model for ap-
proaching the question of how tactile signals are
represented in the cerebral cortex, since the
hands and relevant brain structures are much like
those of the human. Similar sensory performance
in somatesthetic tasks has been observed in both.
Moreover, the exploratory hand movements of
both primates have similar characteristics, re-
flecting the fact that their somatic and motor sys-
tems are similarly linked. These parallels set the
base for studying the dynamic neural operations
of the sensory-motor interface — in other words,
the way sensory representations guide motor
behavior.
We have selected an experimental paradigm in
which monkeys discriminate among sensory stim-
uli delivered to the skin of their hands as we re-
cord the associated cortical activity. This allows
us 1) to define the relevant stimuli among which
monkeys can discriminate, 2) to follow the trans-
formation of the initial cortical display in the dis-
tributed cortical system separating it from those
cortical areas that drive the difi'erential motor re-
sponses required for successful execution of the
task, and 3) to study the details of intracolumnar
operations in the sensory-association areas of the
parietal lobe. I will refer to the first point.
Experiments are in progress to define how the
direction and speed of a probe moving across gla-
brous skin of behaving monkeys is represented in
the activity of somatosensory cortex neurons
(areas 3b and 1). For this, we have designed and
constructed a Cartesian robot that allows me-
chanical stimuli to be presented to the skin of an
awake primate's hand at specified traverse dis-
tance, speeds, and directions. We have quantita-
tively studied many neurons in areas 3b and 1 of
two alert monkeys performing a behavioral task
unrelated to the tactile stimuli. The receptive
fields were scanned with a probe moving at dif-
ferent speeds in eight directions at preselected
levels of force exerted by the probe in the skin.
The first objective in the analysis of the re-
sponding neurons was to reconstruct the initial
representations of the peripheral events in areas
3b and 1 and to identify the possible transforma-
tions occurring in these stages of the somatic pro-
cessing system. The preliminary analysis indi-
cated that it was possible to quantify the
representation of the physical aspects of the stim-
uli in the discharges evoked: velocity and posi-
tion (kinematics) and force (dynamics) . The ac-
tivity of a large percentage of the recorded cells
in areas 3b and 1 varied with the speed of the
stimulus and displayed directional preference.
This finding suggests that certain sets of neurons
of the primary somatic areas respond and make a
neural replica of the mechanical stimuli and that
there are already some transformations (prefer-
ence for direction) at this level. Moreover, the
effects of these two variables (speed and direc-
tion) were modulated by the force exerted by the
stimulus on the skin.
What is interesting in our data is the fact that
directionality can already be detected at the very
beginning of the cortical somatic processing sys-
tem. This finding may explain the presence of
some nonisomorphic images of complex tactile
signals in cortical areas 3b and 1 .
We are presently studying the representation of
these signals in areas 2, 5, and 7b, applying the
same experimental protocol. We are also imple-
525
Representations and Transformations of Tactile Signals in Somatic
and Frontal Motor Cortices
meriting a method for recording the simultaneous 3b and 1 as animals detect and categorize be-
activity of up to eight neurons in order to study tween different directions and velocities of the
the input-output operations of columns of areas stimuli moving across the skin.
524
Anterior-Posterior Patterning
in the Early Mammalian Embryo
Janet Rossant, Ph.D. — International Research Scholar
Dr. Rossant is Professor of Molecular and Medical Genetics at the University of Toronto and Senior
Scientist at the Samuel Lunenfeld Research Institute of Mount Sinai Hospital, Toronto. She received her
undergraduate training in zoology at Oxford University and her Ph.D. degree from Cambridge University.
Her postdoctoral training was in Richard Gardner's laboratory at Oxford. Before joining Mount Sinai
Hospital, she was Associate Professor of Biological Sciences at Brock University, St. Catherines.
SHORTLY after implantation in the mother's
uterus, the mammaHan embryo is transformed
from an undifferentiated group of cells with no
axis of symmetry into a trilayered structure with
anterior-posterior (A-P) and dorsal -ventral (D-V)
polarity and the beginnings of segmentation. One
of the main challenges in mammalian develop-
ment is to understand the cellular processes that
underlie these events and how they are geneti-
cally controlled.
Studies in vertebrates other than the mouse
have implicated inductive interactions between
tissue layers as important in both determining
new tissue types and establishing regional do-
mains along the A-P axis. We have devised an ex-
plant-recombination system that allows us to
address the importance of such inductive interac-
tions in the mouse. Using the mouse homeobox-
containing engrailed-like (En) genes as markers
for a specific anterior domain of the nervous sys-
tem, we have shown that expression of these
genes in ectoderm depends on interaction with
underlying mesoderm.
Isolated anterior ectoderm from premesoderm
stages of development will not initiate En gene
expression in vitro but will express En genes
after aggregation with later anterior mesoderm.
Anterior mesoderm will also induce En expres-
sion in posterior ectoderm, which never nor-
mally expresses En proteins. Posterior mesoderm
is incapable of En induction when combined
with either early ectoderm or later posterior
ectoderm.
These experiments and others in progress indi-
cate that there is regionalization in the capacity
of mesoderm to induce specific neural struc-
tures. Furthermore, preliminary evidence sug-
gests that the basic patterning of the forebrain,
midbrain, and anterior hindbrain structures is
laid down by mesoderm induction at the early
neural plate stage and that the later interac-
tions that lead to the localized expression of
other anterior genes may be confined to the
neuroectoderm .
We are continuing these kinds of studies on
both mesoderm and neural induction in the
mouse, making use of the increasing number of
early marker genes for the processes of A-P pat-
terning. We expect these studies to define in
more detail the temporal and spatial parameters
of inductive interactions. The data will help in
the search for the underlying molecular basis of
patterning in the embryo.
One factor thought to play a role in A-P pattern-
ing is retinoic acid (RA). There is considerable
circumstantial evidence to implicate RA in help-
ing to establish the boundaries of specific A-P do-
mains in the developing embryo. We have pro-
vided more such evidence by showing that
transgenic mice, carrying an RA-responsive ele-
ment upstream of a neutral promoter-/<2cZ con-
struct, express the bacterial lacZ gene in a spe-
cific posterior domain of the embryo.
The boundary of this expression domain coin-
cides, at the early neural plate stage, with the
anterior boundary of the Hox gene, Hox-2.9. As
development proceeds, the boundary of expres-
sion of the transgene recedes in concert with the
establishment of more-posterior //ox gene bound-
aries in the hindbrain. Hox genes are thought to
be involved in A-P patterning, especially in defin-
ing the identity of the segmental hindbrain rhom-
bomeres. The coincidence of Hox gene bound-
aries and the receding RA-responsive transgene
boundary suggests a possible involvement of RA
in establishing Hox gene expression domains.
We have shown that anterior members of the
Hox- 2 gene cluster respond within four hours to
exogenous RA in vivo. This response involves an
anterior shift in their expression boundaries, and
the altered expression domain persists in later
embryos and can be correlated with the specific
teratogenic effects of RA in the hindbrain. Further
experiments are planned to confirm that endoge-
nous RA is important for Hox gene patterning.
Our knowledge of the genetic control of pat-
terning in the gastrulating mouse embryo is still
very limited. If one could screen rapidly through
large numbers of genes for their pattern of ex-
pression at gastrulation, one could hope to iden-
tify overlapping expression patterns that indicate
fundamental developmental domains. In coUabo-
525
Anterior-Posterior Patterning in the Early Mammalian Embryo
ration with Alexandra Joyner (HHMI, Interna-
tional Research Scholar), we are using gene-trap
vectors inserted into embryonic stem (ES) cells as
one strategy for identifying expression domains.
In these vectors the bacterial lacZ gene lacks a
promoter but has a splice acceptor sequence up-
stream, such that /«cZ can be activated as a fusion
transcript when the vector inserts in the intron of
an active host gene. Many /acZ-expressing ES
clones with different integrations are being as-
sessed in chimeras for their pattern of expression
at gastrulation. We have screened more than 200
such lines to date and find about 20 percent with
spatially restricted expression patterns. The
screen will continue, and the most interesting
lines will be cloned from the fusion transcript
and taken through the germline to analyze the
mutant phenotype induced by the vector
integration.
Other more rapid screens for expression pat-
terns using whole-mount in situ hybridization
with cDNA clones are in the pilot stages. The com-
bination of large-scale searches for new patterns
of gene expression with embryonic manipulation
experiments using such genes as markers should
give new insights into the basis of patterning in
the mouse embryo. This broad approach does
not, of course, eliminate the need to study further
the role of specific genes already identified as
potential regulators of development, and we are
also undertaking the targeted mutagenesis of a
number of such genes.
Mouse embryos at day 8.5, showing expression domains of
RAREhsplacZ (top), Hox-2.9 (middle), andKxox-20 (bottom).
Research and photograph by Ron Conlon in the laboratory of
fanet Rossant.
526
Response of the Cerebral Cortex to Spatial
Information
Jean-Pierre Roy, M.D. — International Research Scholar
Dr. Roy is Assistant Professor of Neurology and Neurosurgery at the Montreal Neurological Institute,
McGill University, and Neurologist at Montreal Neurological Hospital, Montreal, Quebec. After receiving
his M.D. degree from Laval University in Quebec City, he spent two years in Mircea Steriade's laboratory
at Laval investigating the origin of rhythmic oscillations in the thalamic relay cells. He then went on to a
neurology residency at the Montreal Neurological Hospital. This was followed by fellowship studies at
NIH with Robert Wurtz on neuronal sensitivity to visual stimuli in the cerebral cortex.
THE main goal of my work is to understand the
transformation of simple information, such as
the speed and direction of motion registered by
individual neurons, into information about the
motion of the subject and the spatial properties of
his environment. My approach is to examine the
response of single cells in one area of a rhesus
monkey's parietal cortex — the medial superior
temporal area (MST) — to different visually pre-
sented stimuli. The neurons in the MST are inter-
esting in that about half of them respond to mo-
tion in one direction, for example left, when the
motion is in front of where the subject is looking,
and respond to motion in the opposite direction,
right in this case, when the motion is behind
where the subject is looking. This corresponds,
we think, to the apparent motion of the environ-
ment experienced during self-motion.
When a subject moves in the environment
while looking at an object to the side, the objects
in the foreground will move in the direction op-
posite to that of self-motion while those in the
background will move in the same direction as
the self-motion. The correlation between direc-
tion of motion and disparity in these neurons sug-
gests that they detect these environmental mo-
tions. But more than that, it suggests that these
cells are involved not so much in indicating the
movement of the environment as such, but rather
the movement of the subject himself. Indeed,
these cells will respond if foreground moves left
or if background moves right, or both. Each of
these three conditions will be observed when the
subject moves to the right. Hence we propose
that those neurons signal motion of the subject
himself.
Within one direction and one disparity re-
sponse, we have evidence that speed is another
property the cells record. Indeed, there appears
to be a preference for lower speeds when the dis-
parity of the motion is of a small absolute value,
i.e., when it corresponds to motion close to the
point of fixation, and a preference for higher
speeds when the disparity is of a large absolute
value, i.e., when it corresponds to motion far
from the point of fixation. This last property is
very interesting in that it suggests that those cells
could be signaling not only the direction of the
self-motion but possibly also the structure of the
environment.
This requires elaboration. I am proposing that
the preliminary evidence of a differential re-
sponse of cells to different speeds could repre-
sent a role for them in transforming the speed
information into depth (or more precisely, rela-
tive depth) information.
During the condition described above, objects
in the environment that are very near or very far
will move fast, while those that are close to the
plane where the subject fixates will move slowly.
Speed, then, contains information about the
depth of objects in the environment. In order to
test this hypothesis, individual planes moving in-
dependently, as were used before, are inade-
quate. What is needed are multiple planes mov-
ing simultaneously at different speeds. To
explore whether MST neurons do respond to
these stimuli representing speed and disparity
gradients, we have developed stimuli with ap-
propriate characteristics.
With those stimuli, we should be able to exam-
ine the response of neurons to different simple
properties of the stimulus. Our prediction is that
the cells should respond optimally to a stimulus
that has the correct direction of motion of the
foreground versus the background, but also a
speed gradient and possibly a disparity gradient.
This would suggest that those cells are indeed
involved in transforming the information they re-
ceive about speed into information about depth.
It would not be the ultimate answer, which could
only come from lesion studies we are planning
for the future.
This work is also supported by a grant from the
Medical Research Council of Canada.
527
Molecular Studies on Neuronal Calcium Channels
Terry p. Snutch, Ph.D. — International Research Scholar
Dr. Snutch is Assistant Professor of Zoology and Neuroscience at the Biotechnology Laboratory, University
of British Columbia, Vancouver. He obtained a B.Sc. degree in biochemistry from Simon Fraser University,
Vancouver, and remained there to complete his Ph.D. degree for studies on the molecular genetics of the
heat- shock response o/ Caenorhabditis elegans with David Baillie. After further research training in the
laboratory of Norman Davidson at the California Institute of Technology, Dr. Snutch joined the newly
formed Biotechnology Laboratory. He recently received a fellowship in neuroscience from the Alfred P.
Sloan Research Foundation and the Killam Research Prize from the University of British Columbia.
THE entry of calcium ions (Ca^^) into cells
mediates a wide variety of cellular and physi-
ological responses, including muscle contraction
and hormone secretion. In the nervous system, an
increase in intracellular Cd}^ concentration di-
rectly affects the electrical properties of neurons.
Ca^"^ entry has also been shown to have a role in
regulating gene expression, modulating Ca^^-
dependent enzymes, and mediating nerve grov^h
and regeneration. Furthermore, an increase in
C?}* concentration at the presynaptic nerve ter-
minal triggers the release of neurotransmitter, in-
hibiting or exciting postsynaptic neurons.
The rapid entry of Ca^"*^ into neurons is me-
diated by membrane proteins called Ca^"*^ chan-
nels. These diverse molecules respond to voltage
changes across the cell membrane by opening
Ca'^"'^-selective transmembrane pores. Besides the
wide variety of normal physiological effects that
Ca^"^ channels mediate, they are implicated in a
number of disorders, such as angina, hyperten-
sion, migraine, and certain arrhythmias. The clin-
ical treatment of these disorders is aided by Ca^"^
channel-blocking drugs. Our present studies uti-
lize molecular cloning techniques to address ba-
sic questions concerning the structure, function,
and expression of Ca^"^ channels in the mamma-
lian nervous system.
Molecular Diversity of Neuronal
Ca^* Channels
Four types of Ca^"^ channel, called T, L, N, and
P, have been identified, with differing electro-
physiological and pharmacological properties.
Among their subunits is the aj-subunit, which
both responds to voltage changes and forms the
ion-conducting pore. Although the differences
among various Ca^^ channels may be due to a
number of factors, we hypothesize that distinct
a 1 -subunits account for most Ca^^ channel
diversity.
Utilizing molecular cloning techniques, we
have isolated cDNAs encoding four distinct
classes of Ca^^ channel from rat brain: classes
A, B, C, and D. Their amino acid sequences show
that they are large proteins (2,100-2,300 amino
acids) and that they share a number of conserved
features with cloned sodium and potassium
channels.
The class C channel found in brain is nearly
identical to a Ca^^ channel previously found in
heart and lung, while classes A, B, and D repre-
sent novel forms. Hybridization of the clones to
rat genomic DNA indicates that each of the classes
of brain Ca^^ channel is encoded by a distinct
gene.
One interesting result of these studies is that
within each class of aj-subunit cloned, several
varieties have been identified, and the actual
number of distinct a, -subunits is much larger
than previously thought. Using molecular ge-
netic techniques, we have demonstrated that
these varieties, at least in the class C and D in-
stances, are generated by alternative splicing. For
class C Ca^"^ channels, two distinct isoforms are
expressed at different levels in various regions of
the rat brain and thus may make unique contribu-
tions to distinct populations of neurons.
Differential Localization of Ca^^ Channel
Subtypes
In an attempt to provide information concern-
ing the physiological roles of the neuronal Ca^"*"
channel subtypes, we are determining their cel-
lular and subcellular localizations. Studies local-
izing Ca^"^ channel gene expression by examina-
tion of RNA levels show that the various classes of
Ca^^ channel are differentially expressed in the
brain. For example, in the cerebellum the class A
channel is mostly localized to Purkinje cells,
while class C channels are more widespread. In
addition to their different spatial distributions in
the brain, studies using antibodies generated
against the cloned Ca^* channel aj -subunits show
that the various Ca^^ channel proteins have dis-
tinct subcellular localizations on individual neu-
rons. For example, some Ca^* channel subtypes
appear to be localized to dendritic regions, while
others are located on both cell bodies and den-
drites. We believe that the distinct cellular and
529
Molecular Studies on Neuronal Calcium Channels
subcellular distributions of the Ca channel sub-
types reflect their unique contributions to neuro-
nal physiology. A long-term goal of this research
is to define how the individual subtypes uniquely
contribute to neuronal functioning.
Functional Expression of Neuronal
Ca^^-Channels
We are taking two approaches to determining
the functional properties of the cloned Ca^"^
channel aj-subunits. First, full-length clones for
the four main classes of rat brain aj-subunit are
being introduced into mammalian cell lines that
normally do not express any Ca^"^ channels. The
electrical and pharmacological properties of the
Cd}^ channels will then be determined with the
patch-clamp technique. Second, we are using the
DNA sequences derived from the cloned Ca^^
channels to generate small synthetic pieces of
DNA (oligonucleotides) that will hybridize to
mRNAs encoding these molecules.
Microinjection of oligonucleotides into cells
that already express defined Ca^^ channels re-
sults in the oligonucleotide hybridizing to Ca^"*"
channel mRNA and inhibiting expression. By in-
hibiting the expression of a single type of Ca^"*"
channel in cells that normally express several
types, we hope both to identify the specific type
of Ca^"^ channel blocked by the oligonucleotide
and also to obtain some insight into the physiolog-
ical roles that individual Ca^"^ channel types con-
tribute to neurons. Using this technique we have
found that the rat brain class C aj-subunit en-
codes a dihydropyridine-sensitive (L-type) Ca^"^
channel. Similar studies with the other classes of
brain Ca^"^ channel are under way.
Detection of single- copy sequences in fluores-
cence micrographs of metaphase chromo-
somes, interphase nuclei, and free chromatin,
a ) Metaphase chromosomes from a human dip-
loid lymphocyte culture, hybridized with four
cosmid probes (cM58-3-6, CF14, cf21, and
cWlO-20) together, spanning 341 kb. b) Hy-
bridization of the same probes with interphase
nucleus. Arrows indicate two sets of hybridiza-
tion signals, c-f) Results of hybridization with
different combinations of cosmids in the 7q31
region, c) cNH24 and cf21; d) cM58-3-6 and
cf21; e) CM58-3.6, CF14 and cf21 (note: two
sets of hybridization signals); f) cM58-3-6,
CF14, cf21, and cW 10-20.
Research and photograph by Henry Heng in
the laboratory of Lap-Chee Tsui.
530
Cystic Fibrosis, Gene Expression in the Mammalian
Lens, and Mapping of Chromosome 7
Lap-Chee Tsui, Ph.D. — International Research Scholar
Dr. Tsui is Senior Scientist and Sellers Chair of Cystic Fibrosis Research in the Department of Genetics at
the Research Institute of the Hospital for Sick Children, Toronto, and Professor of Molecular and Medical
Genetics at the University of Toronto. He was born in Shanghai, raised and educated in Hong Kong, and
there awarded degrees from the Chinese University. His Ph.D. degree is from the University of Pittsburgh,
where his thesis was on the structure and assembly of bacteriophage X (with Roger Hendrix). After
training briefly in the Biology Division of Oak Ridge National Laboratory, he joined the laboratories of
Manuel Buchwald and Jack Riordan at the Hospital for Sick Children to work on cystic fibrosis. Dr. Tsui's
honors include the titles of Scientist of the Medical Research Council of Canada, Fellow of the Royal Society
of Canada, and Fellow of the Royal Society of London.
THE research interests of my laboratory consist
of three general topics in the molecular biol-
ogy of mammalian gene regulation and function.
Molecular Genetics of Cystic Fibrosis
Through classical genetic linkage analysis and
various molecular cloning strategies, the gene re-
sponsible for cystic fibrosis (CF) , the commonest
severe autosomal recessive disorder among Cau-
casians, has been localized and identified. We
have shown that the major CF mutation, account-
ing for approximately 70 percent of all mutant
alleles, is a deletion of the phenylalanine residue
at amino acid position 508 of the predicted poly-
peptide, which is named cystic fibrosis trans-
membrane conductance regulator (CFTR) .
To investigate the basic defect in CF, we are
continuing our search for the other CFTR muta-
tions. Since the CFTR gene contains 27 exons and
spans 230 kb of DNA, detecting a microscopic
mutation is not straightforward. Furthermore, the
lack of a convenient functional assay for CFTR
makes it difficult to distinguish a truly disease-
causing mutation from a benign amino acid sub-
stitution. To coordinate the detection eff'ort, our
laboratory has taken a central role in the forma-
tion of an international consortium of 90 groups
of researchers from 26 countries. Through active
exchange of gene sequences and mutation data
before submission for publication, the consor-
tium has already identified more than 150 appar-
ent CF mutations and 30 sequence variations. The
collective data from two years of operation also
show that most of the remaining 30 percent of
mutant alleles are individually rare and highly
heterogeneous among different populations.
Despite our intensive DNA sequencing effort,
covering each of the exons plus their flanking
regions, we have not been able to identify the
mutations for 6 of the 94 CF chromosomes exam-
ined. These alleles probably harbor different mu-
tations located in regions that affect transcription
or RNA processing. Nevertheless, we have devel-
oped a highly informative marker for use in ge-
netic counseling of CF families. The observed he-
terozygosity for this marker, a dinucleotide
repeat polymorphism in one of the introns of the
CFTR gene, is over 95 percent. In a case of prena-
tal diagnosis, we applied the marker successfully
after all other available tests failed.
The common CF symptoms include chronic
obstructive lung disease, pancreatic enzyme in-
sufficiency, and elevated sweat electrolytes.
Other organs and tissues, such as the hepatobili-
ary tree, intestines, and vas deferens, may also be
involved. The varied degree of severity of the
symptoms among CF patients suggests that the
phenotypes are at least partly conferred by the
genotypes at CFTR. Furthermore, information
about genotypes and phenotypes should also pro-
vide important clues to the function of CFTR and
the basic defect in the various affected organs and
tissues. Based on this assumption and the use of
the extensive clinical data collected at the CF
clinic in our hospital, we have demonstrated a
good correlation between pancreatic involve-
ment and genotypes at the CFTR locus.
To facilitate direct biochemical and physiologi-
cal analysis, we have constructed a full-length
cDNA for CFTR in several expression vectors.
Site-directed mutagenesis has been used to re-
move sequences that are toxic to the host bacte-
rium as well as to generate mutant constructs for
functional evaluation. Using DNA transfection
into heterologous cell types, we showed that this
cDNA could confer a cAMP-regulated chloride
channel activity de novo, suggesting that CFTR is
a chloride channel itself. Preliminary data from
the analysis of constructs reproducing some of
the naturally occurring mutations appear to be in
good agreement with those predicted from the
severity of pancreatic involvement.
In order to understand the regulation of the
CFTR gene, we have performed a series of dele-
tion and transfection studies to determine the se-
quence elements responsible for basal promoter
531
Cystic Fibrosis, Gene Expression in the Mammalian Lens, and Mapping
of Chromosome 7
activity and tissue specificity. Our current data
suggest that the basal promoter element is within
250 bp of the major transcription initiation site.
There also seems to be a negative regulatory ele-
ment immediately upstream of this sequence,
and proper expression of CFTR in vivo may re-
quire additional cis-regulatory element(s) yet to
be identified. A thorough understanding of the
regulation of CFTR transcription may provide ad-
ditional means for treatment of CF, particularly in
some mild cases where an increase of CFTR syn-
thesis may compensate for the partial defect.
Experiments are also in progress to exploit the
yeast STE6 gene as a genetic system to gain some
insight into the structure and function of the first
ATP-binding domain (NBFl) in CFTR. A system-
atic survey of second site mutations (which may
rescue the primary mutation) should provide im-
portant information about the structure of NBFl
and the possibility of its application in drug
design.
Lastly, in order to generate an animal model for
the study of CF, we have been trying to inactivate
the mouse Cftr gene via homologous recombina-
tion in embryonic stem cells. Our attempt to in-
terrupt exon 1 0 has so far been unsuccessful, and
our current targets are exons 1 and 13. The avail-
ability of a mutant mouse strain should greatly
facilitate studies to clarify the pathophysiology of
CF and improve means of treatment.
Regulation of Gene Expression
in Mammalian Lens Development
Transparency of the vertebrate eye lens is at
least partly conferred by the short-range ordering
of water-soluble crystallin molecules in the lens
fiber cells. In order to understand the role these
proteins play in maintaining lens transparency,
one of our interests has been to identify the muta-
tion responsible for a dominant lens defect in a
mouse strain called Elo {Eye lens obsolescence) .
We have performed genetic linkage analysis to
show that the Elo mutation is closely linked to the
7-crystallin gene cluster and have excluded the
first five of the six genes in the cluster as being
the location of the mutation. Through subse-
quent sequence analysis, we have identified a
frameshift mutation in the gene, the last
member of the six-membered cluster. Although
the significance of the latter observation is pres-
ently unknown, it strongly argues that 7-crystal-
lin plays a major rather than a generally assumed
passive role in lens development.
Physical Characterization of Human
Chromosome 7
In order to generate a more complete set of
reagents for the study of genes on human chro-
mosome 7, a chromosome-specific yeast artificial
chromosome (YAC) library has been constructed.
Using a human-hamster somatic cell hybrid with
a single human chromosome 7, we have isolated
more than 1,000 YAC clones containing human
DNA inserts averaging 475 kb. The clones are be-
ing mapped to specific chromosome regions by
hybridization with previously localized DNA seg-
ments and with a somatic cell hybrid mapping
panel. Over 100 clones have thus far been identi-
fied. In addition to the generation of a long-range
physical map of chromosome 7, we are develop-
ing efficient techniques for detecting gene se-
quences based on YAC cloning.
We have also developed a novel in situ hybrid-
ization procedure to facilitate our physical map-
ping effort. Using chromatin fibers released from
interphase nuclei with specific reagents (mostly
known as topoisomerase II inhibitors) and an al-
kaline buffer, we have been able to perform fluo-
rescent in situ hybridization to order DNA
segments less than 20 kb apart. Immediate appli-
cation of this technique (named free chromatin
mapping) should allow us to estimate physical
distance between any given DNA segments and to
study complex sequence arrangement, such as
that in the centromeric regions.
532
Index
abd-A gene, Drosophila development, 549
ABL oncogene, lymphocyte growth regulation, 452-454
Acetylcholine, neuronal function, Iv
Acetylcholine receptor (AChR)
ion channel function, 463-464
neuronal function, Iv, lix
achaete scute gene complex (AS-C), 12
Acquired immune deficiency syndrome (AIDS)
CD4 and CDS molecules, 182
gene expression, cellular transcription, 285-286
immune response and, lii-liii
mycobacterial disease, 200-202
retroviral replication, 53-54
T cell development biology, 265-266
Acrosome reaction, sperm physiology, 491-492
a-Actinin, neuromuscular disease, 233-234
Action potentials
definition of, Iv
generation in olfaction and, 332
research on, 1-2
Activin, transforming growth factor and, 275-276
Adducin
egg structure-function studies, 386
spectrin skeleton assembly, 34
Adenomatous polyposis coli (APC), 441-442
Adenosine, neuromodulation, caffeine consumption,
401-402
Adenosine deaminase
deficiency
blood cell formation, 446
gene therapy, 69- 70
structure/function studies, 329-330
Adenovirus
gene expression control model, 369-370
transcriptional regulation, oncogenesis, 293-294
Adrenal cortex, steroid hormone gene expression, 317-318
Adrenergic receptors
molecular biology, 25/-252
structure and function, lix, 229-230
tVj-Adrenergic receptor, structure and function, 229-230
/3-Adrenergic receptor kinase (/3ARK), 252
(32-Adrenergic receptor, cellular biology, 230
Adult respiratory distress syndrome (ARDS), in inflammation
and metastasis, 40
Affective disorders, neurological dysfunction and, Ix
Agammaglobulinemia, immune system development and, 84
agouti genes, mouse development and, 19-20
Agrin
molecular neuroimmunology, 319-320
synaptic transmission, 355-356
AIDS. 5ee Acquired immune deficiency syndrome
Albinism, tyrosinase and, 307-308
Aldose reductase, structure/function studies, 329-330
Aldosterone, cardiovascular disease and, 244
Allelic exclusion
B cell development, 299-300
immune response and, ///
Allosteric enzymes, structural studies, 388
al protein, DNA interactions, 309-311
Alternative splicing, protein diversity and, 288
Amblyopia, neurophysiology of visual systems, 284
Amino acids
protein structure, xxx
transport, retroviral infections, 97-98
Aminopeptidase A (APA), 84
Animal cells
example of, xxx-xxxi
gene regulation, 293-294
Ankyrin, structure, 34
Ankyrins, plasma membranes, 33-34
Antennapedia locus, morphogenesis, 223-224
Anterior-posterior patterning, embryogenesis, 525-526
Antibodies
complement system, 15-16
gene sequences and, xlviii, li
lymphocyte difl'erentiation, 110
molecular mimicry by, 110
Antibody-antigen interactions, structure/function studies,
329-330
Antigens
cytotoxic T cells, 37-38
determinant, immune response and, li
gene sequences and, xlix, li
immune response and, xlvi, li-liii
processing mechanisms, 93-94
Antigen-specific receptors
generation of, 351-352
genetic mechanisms, 9-10
Antiports, leukocyte homeostasis, 497-498
Aplysia. cell biological studies of memory, 219-220
Apolipoprotein E, chromosome structural studies, 4
Apoptosis
inflammatory cytokines, 73- 74
visual systems, neurogenesis and, 398
Arachidonic acid, phospholipids, 165-166
Archaebacteria, protein folding, 189-190
Atherosclerosis, platelet-derived growth factor (PDGF),
447-448
Atrial natriuretic peptides (ANPs), 151-152
aWTw 7 attachment, transposition mechanism, 91-92
Autoantibody probes. See also Small nuclear
ribonucleoproteins (snRNPs)
gene expression, 391-392
Autoimmune disease
biology of T cell development, 265-266
CD4 T cell activation, 209-210
complement system, 15-16
immune response, 11, 136-138
mechanisms, 171-172
T cell function in, 273-274
tissue-type plasminogen activator (t-PA), 157-158
Autosomal dominant polycystic kidney disease (ADPKD),
333-334
Avian leukemia virus (ALV), 95-96
Axenization, Entamoeba histolytica, 515-516
Axon guidance
neural development, 207-208
structure and function, liv
visual systems, neuronal cell recognition, 396-398
B cells
complement receptors, 185-186
development, 9-10
gene regulation, 379-380
immune response, xlvi
immune system development and, 83-84
immunoglobulin heavy-chain gene control. 215-216
immunological self-tolerance and autoimmunity, / 71- 1 72
molecular regulation, 299-300, 415-416
signal transduction pathways, 159-160
B7 molecule, T cell activation, 209-210
Bacterial toxins, protein crystallography, 521-522
fcrt/- genetic sites, phage X, 501-502
Bare lymphocyte .syndrome (BLS), 325-326
Base pairing, DNA structure and, xxxix
bed gene, transcriptional activation, 403
BCG (bacille Calmette-Guerin) vaccine
leprosy and tuberculosis immunity and pathogenesis,
4 7-48
mycobacterial disease, 200-202
BCR/ABL protein, lymphocyte growth regulation, 452-454
Becker muscular dystrophy
gene identification and correction, 69
533
Index
Becker muscular dystrophy {continued )
molecular genetics, 233-234
neuronal function, Ix
Beckwith-Wiedemann syndrome (BWS), 131
Betaglycan, transforming growth factor and, 275-276
i3-turn, genetic analysis of, 140
6«co/rf gene, transcription control, 112-114
Binary probes, RNA replication, 509-510
Biological clock, biochemical actions of, 465-466
Bioph^ics, eukaryotic gene regulation, 59-60
Bithorax complex
Drosophila melanogaster, 349-350
limb development genetics in Drosophila, 80
Blazing a Genetic Trail, xix
Blood-brain activators, host-pathogen interactions, 493-494
Blood cells
formation mechanisms, 445-446
molecular genetics of, 305-306
Blood clotting
cell adhesion and, 195-196
genetic regulation, 344-346
molecular genetics, 161-162
prevention mechanisms, 123-124
Blood groups, glycosyltransferase molecular genetics and,
267-268
Blood vessels, paracrine control, 459-460
Bone marrow
blood cell formation, 445
transplants, coagulation genetics, 162
Borrelia burgdorferi, immune tolerance, 136-138
Breast cancer, gene mapping techniques and, 82
bride of sevenless (boss) gene, retinal cell-cell interactions,
471-472
C4b-binding protein (C4bBP), clotting prevention, 123- 124
Cadherin, neuron development, 335-336
Caenorhabditis elegans
developmental genetics, 187-188
lineage-specific gene expression, 511-512
molecular genetics, 399-400
neuronal function, lix
research with, xliv
Caffeine, neuromodulation, 401-402
Calcitonin gene-related peptide (CGRP)
neuroendocrine system, 339-340
synaptic transmission, 192
Calcium
hormone mediation, 129-130
nerve cell electrical activity, 1-2
signal transduction, visual systems, 193-194
Calcium channels
dystrophin-glycoprotein complex, 61-62
molecular engineering, 426-428
molecular studies, 529-530
neurotransmitter storage and release mechanisms, 203
phototransduction mechanism, retinal rods and cones,
461-462
prolactin secretion, 489-490
scorpion toxin characterization, 519-520
structure-function studies, 409
Calmodulin, protein structure and folding, 139-140
Calmodulin kinase II (CaMKII) gene, mammalian memory,
424
cAMP (cyclic AMP)
chemical communication, sex pheromones, 257-258
cystic fibrosis transmembrane conductance regulator
(CFTR), 439-440
DNA-binding proteins, gene expression, 175-176
molecular engineering, 426-428
olfactory system, ion channel regulation, 373-374
second messenger system, lix
cAMP recognition element (CRE)
cell biological studies of memory, 219-220
T cell receptor regulation and, 255-256
cAMP-responsive enhancer-binding proteins (CREBs), 176
Campylobacter jejuni, molecular biology, 487-488
Cancer
cell adhesion and, 195-196
genetic modification, 247-248
growth factors, 289-290
lymphocyte development, molecular genetics, 231-232
mammalian development and, 314, 315-316
multidrug resistance, 499-500
Carbohydrate compounds, inflammation and, 40
Carbohydrate ligands, selectins and, 40
Carbon assimilation, photosynthesis genetics, 503-505
Carboxyl-terminal domain (CTD), RNA polymerase II
structure and function, 85-86
Carboxypeptidase Y (CPY), lysosomal hydrolase sorting, 121
Cardiovascular disease, genetic studies, 243-244
CCAAT displacement protein (CDP), molecular genetics of
blood cells and, 306
CcN motif, cell fate control, 14
CD2 molecule, T cell recognition, 101-102
CD4 molecule
activation, 209-210
cellular immune response, 181-182
development and infection, 263-264
human immunodeficiency virus (HIV), 178
immune response and. Hi
in T cells, 51-52
CDS molecule
cellular immune response, 181-182
development and infection, 263-264
immune response and. Hi
CD 18 molecule, genetic disease and, 26
CD28 pathway, lymphocyte development, 415-416
CD45 antigen, protein-tyrosine phosphatases, 413-414
cdc2 gene, cell cycle control and, 271-272
cdk2 protein, cell cycle control and, 272
cDNA clones
calcium channel molecular studies, 529-530
cerebellar gene expression and cell cycle, 180
Down syndrome and, 237-238
Cell adhesion
cell structure and, xxxviii
glycosyltransferase molecular genetics and, 267-268
molecular basis for, 195-196
motor neuron differentiation, 211-212
neuron development, 335-336
Cell adhesion molecules (CAM), 26
Cell asymmetry, molecular genetics, 399-400
Cell biology and regulation, research programs in,
xxix-xxxviii
Cell-cell interaction
Drosophila retina, 471-472
early embryogenesis, oncogenes, 297-298
neural development, 207-208
spectrin skeleton a.ssembly, 34
transforming growth factor and, 275-276
Cell cycle
cell division and, xxxv-xxxvi
colony-stimulating factor 1 receptor (CSF-IR), 371-372
DNA replication and, 21-22
gene expression and, 179-180, 271-272
research trends in, xxx
Cell death. See also Apoptosis
Caenorhabditis elegans development, 187- 188
visual systems, neurogenesis and, 396-398
Cell density, differentiation and, 166-168
Cell fate
differentiation and, 166-168
Drosophila retina, 471-472
554
Index
Cell growth
cell structure and, xxxv-xxxvi
liver regeneration, 411-412
Cell injury, inflammatory cytokines, 73-74
Cell lineage, Caenorhabditis elegans development,
187-188
Cell migration, Caenorhabditis elegans development, 188
Cell motility, defined, xxxv
Cell regulation
chemistry, 375-576
phospholipids, 165-166
Cell signaling, Caenorhabditis elegans development,
187-188
Cell-mediated immunity
CD4-bearing T lymphocytes, 51-52
research on, xlvi
Cellular metabolism, peptide activation, 175-176
Central nervous system (CNS), growth cone guidance and,
169-170
Centromere, position efi'ects, 184
Cerebellum development, gene expression and cell cycle,
179-180
Cerebral cortex
sensory representations, visual system, 367-368
spatial information, 527
Cervical cancer, papillomavirus molecular biology, 241-242
c-Fos protein, transcription control, 470
CGG repeat, fragile X syndrome, 431-432
cGMP (cyclic GMP)
phototransduction mechanism, retinal rods and cones,
461-462
second messengers and cell regulation, 151-152
Chagas disease, kinetoplast genome, 378
Chaperone genes, heat-shock proteins (HSPs), 261-262
Charcot-Marie-Tooth disease, gene mapping, 141
Chemical communication, sex pheromones, 257-258
Chemotaxis, sperm physiology, 491-492
Chloramphenicol acetyltransferase (CAT), 35-36
Chloride channels
cystic fibrosis transmembrane conductance regulator
(CFTR), 439-440
purification and reconstitution, 281-282
Chloroplasts, protein translocons and, 45-46
Cholecystokinin, molecular biology of, 143-144
Cholera toxins, protein crystallography, 521-522
Cholesterol, metabolism, 4
Choroideremia, molecular genetics, 295-296
Chromaffin cells, cell fate control, 11-12
Chromatin
cell cycle control, 21-22
three-dimensional structure, eukaryotic chromosomes,
362
Chromosome 7, mapping in cystic fibrosis (CF), 531-532
Chromosome puffs, metamorphosis, 417-418
Chromosomes
end structure, 71-72
structure and function, position efl'ects, 183-184
three-dimensional studies, 3-4
eukaryotic chromosomes, 360-362
unstable, position effects, 183-184
Chylomicronemia syndrome, genetic studies of
cardiovascular disease, 243-244
Cilia, ion channel activation, 87-88
Circular dichroism, extracellular matrix, 49
c-jun gene, epidermal growth factor (EGF), 104
Clathrin, Yersinia pseudotuberculosis, 197-198
C-less permease, energy-transducing membrane proteins,
213-214
Clonal deletion, immune response and, //
Clonal selection hypothesis, immune response and, //
Cloning
DNA, yeast artificial chromosomes (YACs), 303-304
genetically related. Entamoeba histolytica, 515-516
scorpion toxin receptor genes, 520
c-myc gene
epidermal growth factor (EGF), 104
transcription control, 470
Collagen, hereditary kidney disease, 333-334
Colon cancer, APC gene and, 441-442
Colony-stimulating factor (CSF), myeloid cell growth
control, 371-372
Colony-stimulating factor (CSF) 1 receptor (CSF-IR),
371-372
Colorblindness, X-linked disorders, 163-164
Common variable immunodeficiency (CVID), 84
Complement component C3, gene regulation, 185-186
Complement receptor 2 (CR2), 185-186
Complement system
immune response and, lii-liii
research on, 15-16
Complementation-determining region (CDR), 101
Computational modeling
sensory representations, 367-368
visual systems neurophysiology, 283-284
Computer-aided drug design, protein crystallography,
521-522
Constant (invariant) domain, protein chains, xlvi
a-Constant Spring (aCS), a-globin gene expression, 260
Contractile protein genes, transcriptional regulation,
287-288
Co-receptors. See also CD4 and CDS molecules
immune response and. Hi
Coronavirus, replication and pathogenesis, 239-240
couch potato {cpo) gene, peripheral nervous system
development, 29-30
Cro repressor protein, macromolecular interaction, 278
c-src proto-oncogene, 383-384
Cyclic nucleotide-gated channel (CNG), second-messenger
ion channel regulation, 373-374
Cyclins
cell cycle control, 21-22
G2 cyclins, 271-272
Cyclosporin A, T cell activation and differentiation, 90
Cysteinyl residues, energy-transducing membrane proteins,
213-214
Cystic fibrosis (CF)
gene expression, chromosome 7 mapping, 531-532
linkage analysis and, 81
molecular studies of, 25-26
somatic gene therapy, 449-451
Cystic fibrosis transmembrane conductance regulator (CFTR)
function and regulation, 439-440
mapping in cystic fibrosis (CF), 531-532
multidrug resistance, 499-500
somatic gene therapy, 449-451
Cytochrome b/c, complex, photosynthesis and respiration, 107
Cytochrome P-450 enzymes, macromolecular structure, 108
Cytokines
CD4-bearing T lymphocytes, 51-52
effector functions in immune responses, 513-514
inflammation and, 73-74
Cytoplasm, structure, xxxv
Cytoskeleton
cell adhesion and, 196
host-pathogen interactions, 493-494
structure, xxxv
Cytotoxic T cells
immune response and, xlvi
recognition of, 37-38
D-type cyclins, 371-372
DCoH dimerization cofactor, genetic regulatory
mechanisms, 8V-90
Decay-accelerating factor, complement system, 16
535
Index
"Decoding code," mRNA, 155-156
Deformed gene, Drosophila morphogenesis, 24
Delayed-early (DE) genes
growth factors, 289-290
viral gene regulation, 279-280
Deletion mutagenesis, cell fate control, 13-14
Delta genes, cell fate control, 13-14
Dendrites, structure and function, liv
Density-sensing factor (DSF), differentiation and, 166-168
Desensitization, epidermal growth factor (EGF), 328
Developmental genetics, mutagenesis projects, 383-384
Diabetes mellitus
aldose reductase functions, 330
immune tolerance, 136-138
IRE-A DNA-binding protein, 8
molecular genetics, 27-28, 143-144
stiff-man syndrome and, 106
tissue-type plasminogen activator (t-PA), 157-158
type II, polygenic inheritance, 144
Diacylglycerol (DAG)
calcium-mediated hormones, 130
sex pheromones, 257-258
Diarrhea, guanylyl cyclase receptors, 152
Dictyostelium discoideum, differentiation and, 166-168
Differentiation
developmental control of gene expression and, / 73
genetic regulatory mechanisms, 89-90
keratin expression, skin development, 146-148
kidney cells, transcription factors, 407-408
in lymphocytes, 109-110
motor neurons, 211-212
T cell recognition and, 101-102
visual system development, 341-342
DiGeorge syndrome, gene targeting and, 63-64
Dihydropyridine receptor
calcium channels, 409
voltage-gated calcium channels, 61-62
Disease genes. See Genetic disease
Distal-less gene, limb development genetics, 79-80
DNA
breakage, bacterial transposition reaction, 91-92
fingerprinting, gene mapping, xliv
genome sequencing, 77
probes, adenomatous polyposis coli (APC), 441-442
repair, antibody genes, 9-10
replication mechanisms, 301-302
strand transfer, bacterial transposition reaction, 91-92
structure of, xxxix-xl
synthesis, cell cycle control and, 271-272
DNA photolyase, DNA repair and, 107
DNA polymerase III
(3-subunit structural studies, 235-236
replication mechanisms, 301-302
DNA-binding proteins. See also Protein-DNA interaction
Drosophila melanogaster, developmental genetics,
349-350
immunoglobulin heavy-chain gene control, 215-216
structural studies, 309-311, 393-395
transcription regulation, 421-422
dorsal gene, signal transduction in B cells, 159-160
Down syndrome, molecular analysis, 237-238
Drosophila melanogaster
axis formation and germline determination, 253-254
behavior and neuromuscular development, 465-466
cell adhesion and, 196
cell fate control
embryonic development, 13-14
regulatory molecules, 12
developmental genetics, 349-350
egg structure-function studies, 385-386
embryogenesis
oncogenes, 297-298
transcription control, 112-114
eukaryotic chromosomes, three-dimensional structure,
360-362
growth cone guidance and neuronal development,
169-170
heat-shock proteins (HSPs), 261-262
ion channels, molecular mechanisms, 5-6
limb development genetics, 79-80
morphogen gradients and pattern development, 403-404
morphogenesis, genetic control, 23-24, 223-224
neural development, 207-208
neuronal function, lix
voltage-sensitive channels, 205-206
pattern-formation genes, 65-67, 507-508
peripheral nervous system development, 29-30
position effects, 183-184
research with, xliv
retinal cell-cell interactions, 471-472
sensory transduction, 473
signal transduction
intercellular communication, 323
visual systems, 193-194
visual systems
embryogenesis, 341-342
pattern formation and neuronal cell recognition,
396-398
Drugs, receptors, molecular biology, 251-252
Duchenne muscular dystrophy (DMD)
dystrophin, 62
gene identification and correction, 69-70
molecular genetics, 233-234
neuronal function, Ix
Dysentery. See also Diarrhea
Entamoeba histolytica variability, 515-516
protein cr^'stallography, 521-522
Dystroglycan, Duchenne muscular dystrophy (DMD), 62
Dystrophin
Duchenne muscular dystrophy (DMD), 62
identification and correction, 69-70
neuronal function, Ix
research on, xliii
Dystrophin-associated glycoproteins (DAGs), 62
calcium channels, 61-62
Dystrophin-related protein (DRP), neuromuscular disease,
233-234
ElA protein, gene expression, 369-370
E2 transcription factor, cellular regulation, 375-376
E74 gene, metamorphosis, 417-418
Early B cell factor (EBF), gene expression and, 174
EBERs, autoantibody probes, 392
Ecdysone, metamorphosis, 417-418
Egg cells, structure-function studies, 385-386
Egg-laying hormone (ELH), neuropeptide processing and
packaging, 355-356
Egr gene family, kidney cell growth and differentiation,
407-408
Electron microscopy (EM), chromosome structure, 3-4
Electrostatic interactions, ligand molecular recognition, 330
Elliptocytosis, hemoglobin synthesis, 218
Embryogenesis
anterior-posterior patterning, 525-526
oncogenesis in, 297-298
Embryonic induction mechanisms, vertebrates, 269-270
Embryonic stem (ES) cells
anterior-posterior patterning, 525-526
developmental genetics and, 20, 383-384
Drosophila embryogenesis, 508-509
mammalian memory, 423-424
molecular studies of cystic fibrosis, 25-26
Emery-Dreifuss muscular dystrophy, X chromosome,
431-432
Endocrine cells, synaptic vesicle traffic in, 105-106
536
Index
Endoplasmic reticulum (ER)
cytotoxic T cells, 37-38
protein translocons and, 45-46
Endothelial cells
blood clotting regulation, 344-346
blood vessel function, 459-460
in inflammation and metastasis, 39-40
Endothelial leukocyte adhesion molecule- 1 (ELAM-1),
39-40
Endothelins, blood vessel function, 459-460
engrailed {en) genes
DNA-protein interactions, 309-311
Drosophila morphogenesis, 23-24
embryogenesis, 507-508
anterior- posterior patterning, 525-526
Enhancer genes
peripheral nervous system development, 29-30
viral gene regulation, 279-280
Enhancer of split code, cell fate control, 13-14
Entamoeba histolytica, variability in, 515-516
Enteropathogenic bacteria
molecular biology, 487-488
pathogenicity studies, 357-359
Enzyme specificity, chromosome structure, 3-4
Epidermal growth factor (EGF)
action mechanisms in, 327-328
cell fate control, 13-14
signal transduction of receptor, 103-104
Epidermolysis bullosa simplex (EBS), 148
Epigenetic changes, tumor-suppressor genes, 132
Epinephrine, receptors for, 251-252
Epithelial cells
cystic fibrosis transmembrane conductance regulator
(CFTR), 439-440
papillomavirus molecular biology, 241-242
Epitope, immune response, xlvi
Epstein-Barr virus (EBV)
autoantibody probes, 392
complement receptors, 185-186
familial hypertrophic cardiomyopathy (FHC), 363-364
T cell function, 273- 274
erbB oncogene, epidermal growth factor (EGF), 104
Escherichia coli
chromosome structural studies, 4
enteropathic (EPEC), 357-359
genome sequencing, 77
protein-DNA interaction, 394-395
transposition mechanism, 91-92
E-selectin, glycosyl transferase molecular genetics and, 268
ets genes, hematopoiesis genetics, 486
Eukaryotic chromosomes, three-dimensional structure,
360-362
Excitation-contraction (E-C) coupling, 409
Exocytosis, synaptic vesicle mechanisms, 106
Exons, structure, xxx
Extracellular matrix
cell structure and, xxxvi-xxxvii
molecular biology, 49
neuron development, 335-336
Factor VIII
coagulation genetics, 161-162
X-linked disorders, 163-164
Familial hypercholesterolemia (FH)
genetic studies of cardiovascular disease, 243-244
somatic gene therapy, 449-451
Familial hypertrophic cardiomyopathy (FHC), 363-364
Fasciclin, growth cone guidance and, 170
FBPase, intracellular protein transport, 353-354
Fc receptors, cell surface recognition, 41-42
Fibrinogen, tyrosine phosphorylation in platelets, 55-56
Fibroblast growth factor (EGF)
blood vessel formation, 448
cell fate control, 11-12
embryonic mouse development, gene targeting, 63-64
macromolecular interaction, 278
structural studies, 387-388
tyrosine phosphorylation, 56
Fibronectins, cell adhesion and, 195-196
Finding the Critical Shapes, xix
FK-506 immunosuppressants, 90
Flavin-adenine dinucleotide (FAD), DNA repair and, 107
Floor plate-specific genes, 211-212
Fluorescence, molecular engineering, 426-428
FMR l gene, fragile X syndrome, 431-432
Follicle-stimulating hormone (FSH), gene mapping, 142
Foreskin epithelial cells, papillomavirus molecular biology,
241-242
Fas genes, kidney cell growth and differentiation, 407-408
Fragile X syndrome, 431-432
gene identification and correction, 69-70
Free-energy perturbation, structural biology, 58
Free R value, NMR accuracy, 57-58
Friend leukemia virus, hematopoiesis, 485-486
From Egg to Adult, xix
Fructose transport, molecular genetics of diabetes and, 28
Fucosyltransferase, molecular genetics, 267-268
Fused gene, mammalian development, 419-420
G protein
adrenergic receptors, structure and function, 229-230
basic fibroblast growth factor (bFGF)
structural studies, 387-388
calcium-mediated hormones, 129-130
cellular regulation, 375-376
molecular genetics, nematode development, 399-400
nerve cell electrical activity, 1-2
neuronal function, lix
olfaction and, 331-332
signal transduction
transmembrane transduction, 347
visual systems, 193-194
visual pigments, 291-292
G, cyclins, colony-stimulating factor 1 receptor (CSF-IR),
371-372
GABA neurotransmitter, synapse-like microvesicles (SLMVs),
106
GA-binding protein (GABP), viral gene regulation, 279-280
GAD, stiff-man syndrome and, 106
GAL4 protein, macromolecular assembly, 177-178
gap genes, Drosophila embryogenesis, transcription control,
112-114
Garrod, Archibald, xxxix
GCN4 molecule
structural predictions, 58
yeast transcriptional regulator, macromolecular assembly,
177-178
Gene cloning, research trends in, xliii
Gene expression
adenovirus as control model, 369-370
animal cells, 293-294
autoantibody probes, 391-392
B cell development, 379-380
calcium channel molecular studies, 529-530
cell cycle and, 179-180
cell differentiation and activation, 12, 255-256
cell lineage in Caenorhabditis elegans, 511-512
chromosome 7 mapping, 531-532
developmental control, 173-174
embryogenesis, 507-508
eukaryotic, 59-60
hemoglobin .synthesis, 217-218
hormonal regulation, 75-76
keratin, 146-148
liver regeneration, 411-412
lymphocyte development, 381-382
537
Index
Gene expression {continued^
mammalian development, 419-420
mechanisms of, in animal cells, 421-422
neural development, 208
pattern regulation in Drosophila and, 65-67
protein-RNA/DNA interaction, 393-395
research on, xliii
retroviral
cellular transcription, 285-286
rggulation of, in humans, 95-96
steroid hormones, biosynthesis, 317-318
viruses and, 279-280
visual pigments, 291-292
Gene isolation, Down syndrome, 237-238
Gene mapping
genetic basis of hearing loss and, 1 19-120
genetic disease and, 141-142
protein structure and folding, 139-140
retroviral replication and, 54
Gene regulation. See Gene expression
Gene segments, receptor structure and, xlvi-xlvii, li
Gene structure, xxx, xxxii
antibody genes, 9-10
Gene targeting
antibody genes, 9-10
embryonic mouse development, 63-64
Gene therapy. See also Somatic gene therapy
albinism and, 307-308
human disease and, 249-250
inherited disease, 69-70
T cell receptor functions, 435-436
Gene transfer, blood cell formation, 446
Gene trapping, developmental genetics and, 20
Genetic disease
cancer as, 247-248
gene identification and correction, 69-70
gene mapping techniques, 81-82
metabolic disorders, gene therapy, 457-458
molecular studies, 25-26
research trends in, xlii-xliii
skin, keratin gene expression, 146-148
X-linked disorders, 163-164
Genetic engineering, structural biology and, xli
Genetics, research programs in, xxxix-xlv
Genomes
DNA structure and, xxxix, Ixi
fragments, Down syndrome, 237-238
mapping of, xliv
rearrangement, antibody gene structure, 9-10
sequencing projects, 77
structure, xxx
ges- 1 gene, cell lineage in Caenorhabditis elegans,
511-512
Glanzmann's thrombocytopenia, 55-56
GLI protein, DNA interactions, 310
Glial cells, function, Iv
Globin genes, structural determinants, 259-260
a-Globin gene expression, structural determinants, 259-260
Glomerular basement membrane, hereditary kidney disease,
333-334
Glucagon, gene expression, 175
Glucocorticoid-remediable aldosteronism (GRA), 244
Glucocorticoids, steroid hormone gene expression, 317-318
Glucose transport, molecular genetics of diabetes and, 28
Glyceraldehyde-3-phosphate dehydrogenase (GAPDH),
transcriptional regulation, 7-8
Glycogen phosphorylase, structural studies, 388
Glycolysis/gluconeogenesis, IRE-A DNA-binding protein, 8
Glycoproteins
T cell recognition, 443
viral structure and replication, intracellular transport,
245-246
Glycosyltransferases, molecular genetics, 267-268
Golgi apparatus, structure, xxv
Gonadotropin-releasing hormone (GnRH), gene mapping,
142
Goodpasture syndrome, molecular basis for, 333-334
GP Ilb-IIIa receptor, tyrosine phosphorylation in platelets,
55-56
Graft-versus-host disease (GVHD), coagulation genetics, 162
Grants for Science Education, xix
gRNA, RNA editing, 377-378
groEL protein, protein folding, 189-190
Growth cone guidance, neuronal recognition and, 169-170
Growth factors. See also specific growth factors
cell proliferation and, 447-448
genomic response, 289-290
insulin mechanisms and, 43-44
signal transduction, 517-518
transcription control, 469-470
Growth-regulation proteins, blood cell formation, 446
GTP, transmembrane signal transduction, 347
GTP-binding protein
neurotransmitter storage and release mechanisms, 203
olfaction and, 331-332
Guanylyl cyclase, second messengers and cell regulation,
151-152
Gyrate atrophy (GA), urea-tricarboxylic acid pathways,
429-430
HI transcription factor (H1TF2), 179-180
H-2 molecules, major histocompatibility complex and,
153-154
1119 gene, mammalian development, 419-420
Haemophilus influenzae, host-pathogen interactions,
493-494
hairy gene, pattern regulation in Drosophila and, 66
Hearing loss, genetic basis for, 1 19-120
Heart function, neuronal excitability, 206
Heat-shock proteins (HSPs)
protein 60 (hsp60), 189-190
stress tolerance and, 261-262
T cell epitopes, 358
heavy-chain (a-chain) segments, li
hedgehog gene, Drosophila morphogenesis, 23-24
Helix-helix association and stability, 57-58
Helix-loop-helix
cell fate control, 14
DNA interactions, 311
MyoD gene activation, 433-434
Helper T cells. See also T cells
immune response and, xlvi
Hemagglutinin, influenza virus, 157-158
Hematopoiesis
gene expression, 381-382
normal and leukemic genetics, 485-486
Hematopoietic stem cells
blood cell formation, 445
genetic manipulation, 31-32
lymphocyte life cycle, 437-438
molecular genetics of, 305-306
Hemoglobin
genetic disease and, xlii-xliii
molecular genetics of blood cells and, 305-306
synthesis, genetic control, 217-218
Hemophilia
coagulation genetics, 161-162
X-linked disorders, 163-164
Heparin, basic fibroblast growth factor (bFGF) and, 387-388
Hepatitis B virus (HBV)
replication and pathogenesis, 149-150
RNA replication and pathogenesis, 240
surface antigens, 315-316
Hepatitis delta antigen (HDAg), RNA replication and
pathogenesis, 240
538
Index
Hepatitis delta virus (HDV), RNA replication and
pathogenesis, 240
Hepatocellular carcinoma, 149-150
Hepatocellular transplantation, gene therapy and, 249-250
Hepatocyte nuclear factor 3, eukaryotic gene regulation,
59-60
Heterochromatin, position effects, 183-184
Heterogeneous nuclear ribonucleoproteins (hnRNPs),
117-118
High-mobility group (HMG) box, 7-8
Hippocampus, neuronal excitability, 206
Histidine, visual pigments, 291-292
Histocompatibility antigens (H antigens)
genetics, structure and function, 1 33- 1 34
molecular structure, 153-154
T cell receptors, 41
HLA-B27 glycoprotein, 443
HNF-1|8 transcription factor, 89-90
Holoenzyme, DNA replication and, 301-302
Homeobox genes
Drosophila melanogaster, 349-350
embryonic induction mechanisms, 269-270
motor neuron differentiation, 211-212
Homeodomain proteins
Drosophila morphogenesis, 23-24
DNA interactions, crystal structures, 309-31 1
DNA-binding specificity, 112-114
genetic regulatory mechanisms, 89-90
structural biology and, Ixiii
Homeotic genes
Drosophila morphogenesis, 23-24
genetic control. 223-224
limb development genetics in Drosophila, 79-80
Homologous chromosomes, three-dimensional structure,
360-362
Homologous recombination, research trends, xliv
Hormone response elements (HREs), 126
Hormones
calcium-mediated, molecular mechanisms, 129-130
gene expression regulation, 75-76
molecular biology, 251-252
Host-pathogen interactions, microbial pathogenesis,
493-494
"Housekeeping" genes, complement system, 16
Hox genes, embryogenesis
anterior- posterior patterning, 525-526
induction mechanisms, 269-270
mouse development, gene targeting, 63-64
hox- 1 .5 muvmt, 63-64
hox- 1 .6 mutant, 63-64
Human foamy virus (HFV), 95-96
Human growth hormone (hGH), 256
Human immunodeficiency virus (HIV)
AIDS and. Hi
CD4 molecule and, 178
gene expression, cellular transcription, 285-286
protein-DNA interaction, 394
retroviral replication, 53-54
HIV- 1 and HIV-2, 95-96
surface glycoproteins, 263-264
T cell epitopes, 358
trans-activation, 325-326
Human leukocyte antigens (HLA), processing mechanisms,
93-94
Human myeloid leukemias, lymphocyte growth regulation,
452-454
Human T cell leukemia virus (HTLV-I and HTLV-II)
retroviral gene expression, 95-96
surface glycoproteins, development and infection,
263-264
Humoral immune response
CD4-bearing T lymphocytes, 5/- 52
research on, xlvi
hunchback {hb) gene, Drosophila embryogenesis
segmentation, 113-114
transcription control, 113-114, 403
Huntington disease, gene mapping, 81-82
Hyperlipidemia, genetic studies of cardiovascular disease,
243-244
Hypertension
blood vessel function, endothelins and, 459-460
genetic studies, 243-244
steroid hormone gene expression, 317-318
Hypoxanthine guanine phosphoribosyltransferase (HPRT),
69
IgA deficiency (IgA-D), 84
Imaginal wing discs, pattern regulation in Drosophila and,
66-67
Immediate-early genes (lEGs)
growth factors, 289-290
kidney cell growth and differentiation, 407-408
viral gene regulation, 279-280
Immune response
CD4 and CDS molecules, 181-182
cytokine regulation of effector functions, 513-514
genetic mechanisms in antibodies, 9-10, 171-172
genetic research and, 136-138
leprosy and tuberculosis immunity and pathogenesis,
47-48
research on, xlvi-xlvii
Immune system
cell development, 83-84
cell surface recognition, 41
complement system, 15-16
research programs on, xlvi
Immunodeficiency
gene regulation and, 325-326
immune system development and, 84
Immunofluorescence techniques
gene mapping, 81
post-transcriptional regulation, 118
Immunoglobulin
antigen receptor molecule studies, 351-352
B cell development, gene expression, 379-380
developmental control of gene expression and, 173-174
heavy-chain gene control, 215-216
K light-chain gene, 159-160
lymphocyte development, 415-416
activation and, 109-110
neoplasia and, 231-232
Immunology, research programs in, xlvi-liv
Immunoprecipitation, synaptic transmission, 356
Immunosuppression, lymphocyte life cycle, 437-438
Imprinting, cancer as genetic disease and, 248
In situ hybridization, eukaryotic chromosomes, 360-362
In vivo gene expression, cellular and retroviral genes, 286
Inactivation process, ion channels, 5
Inborn errors of metabolism, gyrate atrophy, 429-430
Inducible cell adhesion molecule- 1 10 (INCAM-1 10), 39-40
Infectious disease
protein crystallography, 521-522
protein synthesis, 501-502
RNA replication, 509-510
Inflammation, vascular endothelium and, 39-40
Influenza virus
glycoprotein mechanisms, 443
structure and replication, 245-246
Inositol 1,4,5-trisphosphatc (IP3), 257-258
Insertional mutation
albinism and, 307-308
developmental genetics and, 20
Insulin. See also Insulin-responsive element A (IRE-A)
mechanisms of action, 43-44
molecular genetics of diabetes and. 27-28
production mechanisms for, 389-390
539
Index
Insulin {continued)
receptor studies, 390
Insulin-like growth factors, evolution of, 390
Insulin-responsive DNA-binding protein (IRP-A), 7-8
Insulin-responsive element A (IRE-A), 7-8
Insulinotropic peptides, gene expression, / 76
int-1 gene, embryonic mouse development, 63-64
int-2 gene, embryonic mouse development, 63-64
INT-1 oncogene, limb development genetics, 79-80
Integrase, structure, 53-54
Integration, retroviral replication, 53-54
Integrin
cell adhesion and, 195-196
neuron development, 335-336
Yersinia pseudotuberculosis, 197-198
Intercellular communication, signal transduction, J23
Interferon-7
cytokine regulation of effector functions, 513-514
immunodeficiency and, 325-326
InterIeukin-1 (IL-1 )
B cell development, 379-380
immune response and, xlvi-xlvii
inflammatory cytokines, 73-74
Interleukin-6 (IL-6), 73-74
Interleukin-7 (IL-7), 84
lnterleukin-10 (IL-10), 513-514
Interleukins
as host defense in cancer, 248
lymphocyte activation and, 109-110
Intermediate filaments, cytoskeleton structure, xxxv
Intermediate voltage electron microscopy (IVEM)
tomography, 3-4
International Research Scholars Program, xix, 481
Intracellular adhesion molecule-1 (ICAM-I), 26
Intracellular microorganisms
cytotoxic T cells, 38
molecular genetics, 197-198
Intracellular parasitism, microbial pathogenesis, 493-494
Intracellular signals, molecular engineering, 428
Intracellular transpon
proteins, 353-354
viral structure and replication, 245-246
Introns, structure, .v.r.v
Invariant chains, antigen processing, 94
Invasin, Yersinia pseudotuberculosis, 197-198
Ion channels
function, Iv, Iviii
leukocyte homeostasis, 497-498
mechanical activation, 87-88
molecular mechanisms, 5-6, 463-464
nerve cell electrical activity, 1-2
proteins, functional mechanisms, 281-282
scorpion toxin receptor genes, 519-520
second messenger control, 373-374
sperm physiology, 491-492
synaptic transmission, 191-192
viral structure and replication, 245-246
Islet amyloid polypeptide (LAPP) (amylin)
biosynthesis, 390
molecular genetics of diabetes and, 27-28
Islets of Langerhans, insulin production, 389-390
Junk DNA, xxx
K locus, signal transduction in B cells, 159-160
Keratin, gene expression, 146-148
Keratinocyte-stimulating factor (KRF-1), 242
Kidney
cell growth and differentiation, 407-408
hereditary disease in, 333-334
Kinase, See Protein kinases
Kinetics, T cell recognition, 101-102
Kinetoplast DNA, rr)'/)««osowe genomes, 377-378
"knocked-out" genes, 265-266
kreisler genes, 19-20
l( 1 Jcorkscrew gene, signal transduction, 323
l( 1 )pole hole gene, signal transduction, 323
labial gene, morphogenesis, 224
Lactate dehydrogenase (LDH), inflammatory cytokines,
73-74
Lactotropes, prolactin secretion, 489-490
lacZ gene, anterior-posterior patterning, 526
Lamina neurons, visual systems, 398
Lamins, eukaryotic chromosomes, 362
Large dense-core vesicles (LDCVs), 105-106
Laron syndrome, gene mapping, 142
Lateral geniculate nucleus (LGN), 283-284
a-Latrotoxin, neuronal secretory pathways, 405-406
Lectins, carbohydrate ligands and, 40
Legionella pneumophila, molecular genetics, 197-198
Leishmaniasis
immune evasion and, 115-116
kinetoplast genome targeting, 3 78
Leprosy
genetic control of, 202
immunity and pathogenesis, 47-48
Lesch-Nyhan syndrome, 69
Leucine zipper proteins
growth factors, 289-290
MyoD gene activation, 433-434
protein folding, 225-226
structural biology, 58
Leukemia
immune system and, liv
lymphocyte development, 231-232
Leukemia inhibitory factor (LIF), 31-32
Leukocyte adhesion deficiency (LAD), genetic disease and, 26
Leukocytes
genetic disease and, 26
ionic homeostasis, 497-498
Light-chain segments, gene sequences and, li
Linkage analysis, gene mapping and, 81-82
Lipoprotein lipase (LPL), 243-244
Listeria monocytogenes, cytotoxic T cells, 38
Liver, molecular biology, 411-412
Liver regeneration factor (LRF-1), 412
Long terminal repeat (LTR)
developmental genetics and, 20
human immunodeficiency virus (HIV), 325-326
Long-term potentiation, mammalian memory, 423-424, 425
Low-density lipoprotein (LDL), 449-451
Lowe's syndrome, molecular genetics, 295-296
LRP transmembrane phosphatase, 413-414
Luciferase reporter mycobacteriophages (LRMs), 200-202
Luteinizing hormone (LH), gene mapping, 142
Lyme disease, immune tolerance, 136-138
Lymphocyte function antigen 1 (LFA-1), 102
Lymphocyte receptors, immune response, xlvi
Lymphocytes
activation mechanisms, 109-110
development
antibody genes, 9-10
gene expression, 381-382
life cycle, 437-438
molecular regulation, 415-416
normal and abnormal growth regulation, 452-454
signal transduction, 321-322
Lymphoid enhancer-binding factor 1 (LEF-1), 174
Lymphoid-specific regulation, 173-174
Lymphoma
immune system and, liv
molecular genetics, 231-232
Lysosomal hydrolase, sorting, 121-122
Lysosome structure, xxxi>
a-Lytic protease, chromosome structure, 3-4
540
Index
M channel, nerve cell electrical activity, 1-2
M2 protein, viral structure and replication, 245-246
M3 molecule, histocompatibility antigen research, 133-134
Macromolecules
recognition, protein folding, 225-226
structural basis of interaction, 277-278
structural studies, Ixi, 177-178
three-dimensional structures, 107-108
Macrophages, immune system, xlvi
Major histocompatibility complex (MHC)
antigen processing, 93-94
B cell development, 299-300
CD4 and CDS molecules, 182
class I molecules
antigen processing, 93-94
histocompatibility antigen research, 133-134
class II molecules
antigen processing, 93-94
immune tolerance, 136-138
macromolecular assembly, / 78
cytotoxic T cells, 37-38
immune response and, li-lii
mammalian memory, 423-424
molecular genetics of, 153-154
surface glycoproteins, 263-264
T cells
activation, 209-210
development biology, 265-266
receptor, 435-436
recognition, 101-102
Mammalian development
disease and, 313-314, 315-316
gene regulation, 419-420
genetics, 19-20
X and Y chromosomes, 312-314
MAP kinases, tyrosine phosphorylation, 56
MARCKS (myristoylated alanine-rich C-kinase substrates),
43-44
Marfan syndrome, gene mapping, 142
Master regulatory genes, 207-208
Maternal genes, transcription control, 112-114
Maternally transmitted antigen (Mta), 133-134
Maturation-promoting factor (MPF), 271-272
MCAT protein, retroviral infections, 97-98
mdr gene family, multidrug resistance, 499-500
Mechanical adjustment hypothesis, 87-88
Mechanotransduction, sensory cells, 473
Medial superior temporal area (MST), 527
Medical Research Organization (MRO), xix
Membrane proteins
cofactor protein, complement system, 16
energy transduction, 213-214
Membrane-bound growth factors, 276
Memory
cell biological studies, 219-220
T cells and, 423-424
in vitro generation, 102
Mendelian genetics, genetic structure and, xxxix
Mental retardation, X-linked disease, 295-296
Messenger RNA (mRNA)
decoding code, 155-156
gene expression and, 117-118
a-globin gene expression, 259-260
splicing studies, 337-338
transcription and, xxx, xxxiii
Metamorphosis, molecular regulation, 417-418
Metastasis
tissue-type plasminogen activator (t-PA), 157-158
vascular endothelium and, 39-40
Methyl-malonyl CoA mutase (MCM), gene therapy with,
249-250
Microfilaments, cytoskeleton structure, xxxv
/Sj-Microglobulin, polymorphism, 134
Microtubules
cytoskeleton structure, xxxv
linkage analysis and, 81
Missense mutations, familial hypertrophic cardiomyopathy
(FHC), 363-364
Mitochondria
histocompatibility antigen research, 134
matrix, protein folding, 189-190
structure, xxxv
Try panosome genome, 378
Mitosis, cell cycle control, 21-22
Molecular biology, research trends in, xxx
Molecular engineering, cell structure and neurobiology,
426-428
Molecular fractionation, synaptic transmission, 355-356
Molecular recognition
eukaryotic gene regulation, 59-60
structural biology and, Ixiii
Moloney murine leukemia virus (MoMuLV)
developmental genetics and, 20
hematopoietic stem cell regulation, 31-32
retroviral infections, 97-98
Monoclonal antibodies, scorpion toxin receptor genes,
519-520
Monocytes, HIV gene expression, 285-286
Morphogen gradients, Drosophila body patterns, 403-404
Morphogenesis, genetic control, 223-224
Motor neurons, differentiation, 211-212
Mouse development
antibody genes, 9-10
gene targeting and, 63-64
genetics, 19-20
Multiplex sequencing, genomes, 77
Muscle development and function
molecular analysis, 287-288
MyoD gene family, 433-434
Muscular dysgenesis (rndg) gene, 409
Muscular dystrophies. See also specific muscular dystrophies
gene mapping, 141
molecular genetics, 233-234
neuronal function, Ix
Mutations
detection, gene mapping, 141-142
gene cloning and, xliii
ion channels, molecular mechanisms, 5
Myasthenia gravis, neuronal function, Iv
Mycobacteria
genetic control of disease, 200-202
leprosy and tuberculosis immunity and pathogenesis,
4 7-48
Myeloid cells, growth control, 371-372
Myoblasts, as recombinant protein delivery system, 256
Myocyte-specific enhancer-binding factor 2 (MEF2),
287-288
MyoD gene family
muscle cell lineage, 433-434
transcriptional regulation, 287-288
Myosin heavy chain (MHC)
familial hypertrophic cardiomyopathy (FHC), 363-364
transcriptional regulation, 287-288
nana gene, Drosophila development and, 253-254
Nectins, cell adhesion and, 195-196
Nematode development
genetic control, 187-188
molecular genetics, 399-400
Nerve cell structure, liv
Nerve grov^^h factor (NGF)
cell fate control, 11-12
transcription control, 469-470
tyrosine phosphorylation, 56
Nerve impulses, structure and function, Iv
541
Index
Nervous system
biological clock system, 465-466
cell fate in, 13-14
Neural crest cells, cell fate control, 11-12
Neural networks, sensory representations, visual system,
367-368
Neuroendocrine system, molecular mechanisms, 339-340
Neurogenesis
cell pattern control, 211-212
pattern regulation in Drosophila and proneural genes,
66-67
visual systems, 341-342, 396-398
Neuroimmunology, molecular studies, 319-320
Neuromodulation, caffeine consumption and, 401-402
neuromusculin (nrm) gene, 29-30
Neuron doctrine, liv
Neuronal development
calcium channel molecular studies, 529-530
cell fate control, 11-12
extracellular matrix, 335-336
Neuronal excitability, voltage-sensitive potassium channels,
205-206
Neuronal recognition
growth cone guidance and, 169-170
visual systems, 396-398
Neuronal survival, neuron development, 335-336
Neurons
differentiation, olfaction and, 332
molecular engineering, 426-428
secretory pathways, 405-406
structure and function, liv
synaptic vesicle traffic in, 105-106
Neuropeptides, synaptic transmission, 355-356
Neurophysiology, visual systems, 283-284
Neuroscience, research programs in, liv-lx
Neurotransmitters
function, Iv
neuronal secretory pathways, 405-406
receptors, synaptic transmission, 192
storage and release mechanisms, 203
synaptic vesicle proteins, 105-106
transcription control, 469-470
Neurotrophic factors, cell fate control, 11-12
NF-kB protein, signal transduction in B cells, 159-160
Nitric oxide, retroviral infections, 97-98
Non-self lymphocytes, immune response, xlvi
nos gene, translation repression of bb, 403-404
Notch gene
biological clock system, 465-466
cell fate control, 13-14
viral gene regulation, 280
Nuclear envelope, eukaryotic chromosomes, 360-362
Nuclear factor
of activated T cells (NFAT), 90
gene expression regulation, 75-76
Nuclear lamina, protein translocons and, 45-46
Nuclear magnetic resonance (NMR)
accuracy of crystal and solution, 57
chromosome structure, 3-4
extracellular matrix structure, 49
protein structure and folding, 139-140
structural biology and, Ixiii
Nuclear pore complexes (NPC)
protein translocons and, 45-46
structure and function, 99-100
Nuclear structures, xxx
post-transcriptional regulation, 117-118
Nucleoporins, nuclear pore complex (NPC), 99-100
NUPl gene, nuclear pore complex (NPC), 99-100
Obesity
IRE-A DNA-binding protein, 8
molecular biology and, 143-144
Oct-2 protein, B cell development, 379-380
Olfaction
molecular approaches to, 331-332
molecular biology, 17-18
second-messenger ion channel regulation, 373-374
Oligonucleotide, calcium channel molecular studies,
529-530
Oligosaccharide
epitopes, protein structure and function, 329-330
glycosyltransferase molecular genetics and, 268
Onchocerciasis, immune evasion and, 116
Oncogenes. See also specific genes, e.g., src gene
adenovirus transcriptional regulation, 293-294
cell structure and, xxxviii
colony-stimulating factor 1 receptor (CSF-IR), 371-372
early embryogenesis and, 297-298
gene regulation mechanisms and, 421-422
lymphocyte development, molecular genetics, 231-232
mammalian development studies and, 315-316
Optical/electron microscopy, eukaryotic chromosomes,
360-362
Organelles, structure, xxxv
Organogenesis, transcription control, 114
Ornithine transcarbamylase (OTC) deficiency, 70
Ornithine-5-aminotransferase (OAT), 429-430
Outer membrane proteins (OMP), 487-488
Outer surface protein (OspA & B), 138
Oxidative stress, redox proteins and, 236
Ox}'R protein, oxidative stress mechanism, 236
Oxytricba nova. DNA molecular structure, 72
p53 gene
hematopoiesis, 486
transcription regulation mechanisms, 496
paired ^ene, Drosopbila embryogenesis, 114
Pair-rule segmentation genes, 65-67
Pancreatic /3-cell, diabetes and, 27-28
Papillomaviruses, human (HPV), molecular biology of,
241-242
Paramyxovirus, structure and replication, 245-246
Parasites, immune evasion by, 115-116
patcbed gene, Drosopbila morphogenesis, 23-24
Pathway recognition, growth cone guidance and, 169-170
Pattern formation
Drosopbila body patterns, morphogen gradients, 403-404
limb development genetics in Drosopbila, 79-80
fax genes, embryonic induction mechanisms, 269-270
PC 12 cells, tyrosine phosphorylation, 56
Peptides
neurotoxins, ion channels, 281-282
structure, protein folding, 225-226
Peptidyl-tRNA hydrolase (Pth), protein synthesis, 501-502
per gene
biological clock system, 465-466
mRNA processing and behavior, 337-338
Perception, neurophysiology, 283-284
Peripheral nervous system
Drosopbila development, 29-30
growth cone guidance and, 169-170
olfaction and, 17-18
Peyer's patches, gene regulation of, 437-438
P-glycoprotein (P-gp), multidrug resistance, 499-500
pH regulation, leukocyte homeostasis, 498
Phage X, protein synthesis, 501-502
Phagocytes, Legionella pnuemopbila growth in, 198
Phase determination, structural biology and, Ixi
Phenotype, neuronal function, lix
Phenylalanine hydroxylase (PAH), somatic gene therapy,
457-458
Phenylketonuria (PKU), somatic gene therapy, 457-458
Phosphatidic acid (PA), calcium-mediated hormones, 130
Phosphatidylcholine (PC), calcium-mediated hormones, 130
542
Index
Phosphatidylinositol (PI)
metabolism, 327-3-28
phospholipids and cell regulation, 165-166
Phosphatidylinositol 3-kinase (PI3-kinase), 122
Phosphatidylinositol 4,5-bisphosphate (PIP2), 129-130
Phosphatidylinositol monophosphate phosphatase, 327-328
Phospholipase A^, cellular regulation, 375-376
Phospholipase C, calcium-mediated hormones, 129-130
Phospholipids, cell regulation, 165-166
Phosphorylation. See also Tyrosine phosphorylation
epidermal growth factor (EGF), 103-104
RNA polymerase II structure and function, 85-86
signal transduction, 517-518
Phosphotyrosine protein phosphatase, synaptic transmission,
192
Photosynthesis, molecular genetics, 503-505
Phototransduction, visual systems, 193-194
piebald gene, mammalian development, 419-420
Pilin, protein crystallography, 521-522
Pit-1 gene, neuroendocrine system, 339-340
Plasma cells, immune response and, xlvi
Plasma membranes
ankyrins in, 33-34
protein translocons and, 45-46
Plasminogen activator inhibitor- 1 (PAI-1), 161-162
Platelet-derived growth factor (PDGF)
Drosophila melanogaster signal transduction, 323
genomic response, 289-290
Platelets, tyrosine phosphorylation, 55-56
Pole plasm, Drosophila development and, 253-254
Poliovirus, genetics, 227-228
Polyadenylation, molecular mechanisms, 293-294
Polymerase chain reaction (PGR)
gene mapping, xliv
genetic basis of hearing loss and, 119-120
hematopoietic stem cell regulation, 32
major histocompatibility complex, 153-154
viral genetics, 227-228
Polypeptide growth factors, action mechanisms in, 327-328
Polypeptide hormones, gene regulation, 175-176
Ponder-Richards rotamers, chromosome structure, 4
Population dynamics, PKU gene distribution, 457
Positional cloning
gene mapping and, 81-82
obesity and diabetes, 143-144
Potassium channels
cystic fibrosis transmembrane conductance regulator
(CFTR), 440
functional mechanisms, 281-282
insulin production, 390
molecular mechanisms, 5-6
nerve cell electrical activity, 1-2
neuronal excitability, 205-206
scorpion toxin receptor genes, 519-520
POU domain, neuroendocrine system, 339-340
Prader-Willi syndrome, 141
Pre-B cell, gene expression and, 175
Pre-messenger RNA, splicing mechanisms, 337-338
Prenatal diagnosis, genetic disease, 457-458
Prepattern genes, neural development, 207-208
proboscipedia gene, morphogenesis, 223-224
Prohormone-converting enzymes, 389-390
Prokaryotic repressors, protein-DNA interactions, 309-3 1 1
Prolactin secretion, functional heterogeneity, 489-490
Promoters
peripheral nervous system development, 29-30
trapping, developmental genetics, 383-384
Proncural genes, pattern regulation in Drosophila and,
66-67
Protein kinase A (PKA), gene expression and cell cycle,
179-180
Protein kinase C, sensory transduction, 473
Protein kinases
carboxyl-terminal domain (GTD), 85-86
cell cycle control, 21-22
cell division and, xxxvi
epidermal growth factor (EGF) desensitization, 328
insulin mechanisms and, 43-44
protein sorting and, 1 22
signal transduction, dual-specificity, 518
synaptic transmission, 192
tyrosine phosphorylation in platelets, 55-56
Protein-DNA interaction. See also DNA-binding proteins
cAMP-dependent, gene expression, 175-176
Drosophila embryogenesis, 114
macromolecules, 278
RNA molecular structure, 72
Protein-protein interactions, signal transduction, 517-518
Proteins
design, protein folding, 225-226
folding
chromosome structure, 4
genetic research and, 139-140
histocompatibility antigen research, 133-134
in vivo research, 189-190
macromolecular interaction, 276-278
intracellular transport, 353-354
ion channels, functional mechanisms, 281-282
macromolecular interaction, 276-278
phosphorylation
cell cycle control and, 27 1-272
synaptic transmission, 191-192
regulatory proteins, 387-388
sorting, molecular genetics, 121-122
structure/function studies, xxx, xxxiv, Ixi, 329-330
synthesis, viral infection, 501-502
as transcriptional factor, xxxv
translocons, nuclear organelles and, 45-46
Protein-tyrosine kinase (PTK)
lymphocyte signaling, molecular basis for, 321-322
T cell receptor functions, 435-436
Protein-tyrosine phosphatases, lymphocyte activation,
413-414
Proton pumps, leukocyte homeostasis, 498
Proto-oncogenes, embryonic mouse development, 63-64
P-selectin gene, genetic disease and, 26
Pseudoknots, mRNA decoding, 1 56
Pseudomonas aeruginosa, protein crystallography,
521-522
pumilio gene, 253-254
Q-beta replicase, RNA replication, 509-510
R values, NMR accuracy, 57
R7 cells, retinal cell-cell interactions, 471-472
RAG genes
antibody genes, 9-10
lymphocyte differentiation, 110
RAG-1 protein, 351-352
RAG-2 protein, 351-352
ras protein
tyrosine phosphorylation, 56
visual system development, 342
reck protein, protein-DNA interaction, 394-395
Receptor-ligand interaction
chromosome structural studies, 4
macromolecules, 278
Receptors. See also specific receptors, e.g., T cell receptor
adrenergic receptors, biosynthesis, 229-230
antigen-specific, 351-352
cell structure and, xxxvi
gene segments and, xlvi, li
macromolecular assembly, 178
olfaction and, 17-18
periplasmic, active transport and chemotaxis, 330
Index
Receptors {continued)
retroviral infections, 97-98
transforming growth factor (TGF), 103-104
Recombination
antigen receptor molecule studies, 351-352
DNA technology, xxxix, xliii
Down syndrome and, 237-238
genetic disease research, 25-26
protein-DNA interaction, 394
RNA viruses, 239-240
transposition mechanism, 91-92
Recoverin
signal transduction, visual systems, 193-194
urea/tricarboxylic acid cycles, 430
Redox proteins, transcriptional response, 236
Regenerating liver inhibitory factor {RL/IF-1), 411-412
Regulation, molecular mechanisms, 339-340
Regulatory molecules
cell fate control, 12
structural studies, 387-388
Replication
cell cycle control, 21-22
in DNA, 301-302
protein-DNA interaction, 393-395
viral pathogenesis and, 149-150
Resolvase, protein-DNA interaction, 394-395
Restriction fragment length polymorphism (RFLP)
cerebellum development, 180
gene mapping, xliv
neurological disorders, Ix
obesity and diabetes, 143-144
tuberculosis infections and, 200-202
Wilms' tumor genes, 131
Retinal degeneration, molecular genetics, 295-296
Retinal rods and cones, phototransduction mechanism,
461-462
Retinitis pigmentosa, gene expression in visual pigments,
292
Retinoblastoma, adenovirus transcriptional regulation,
293-294
Retinoic acid (RA), embryogenesis, 525-526
Retinoic acid receptors (RARs), response pathway, 125-127
Retinoid X receptors (RXRs), response pathway, 125-127
Retrovirus
correction of methyl-malonyl CoA mutase (MCM),
249-250
developmental genetics and, 20
gene expression, 95-96
cellular transcription, 285-286
hematopoietic stem cell regulation, 31-32
infection mechanisms, 97-98
replication and, 53-54
Rev protein, retroviral gene expression, 95-96
Rev response elements (RRE), 96
"Reverse genetics." See also Positional cloning
genetic basis of hearing loss and, 119-120
Rheumatoid arthritis (RA)
immune tolerance, 136-138
T cell function in, 273-274
Rhodopsin, transmembrane signal transduction, 347
Rhombomeres, developmental genetics and, 19-20
Ribosomal jumping, mRNA decoding, 155-156
Ribosomal pausing, tRNA molecules, 455
Ribosome frameshifting, translational regulation, 455
Ribozymes, RNA replication, 509-510
Ring complexes, protein folding, 190
RMA-S mutant cells, histocompatibility antigen research, 134
RNA
catalysis, 71-72
chromosome end structure, 71-72
editing, Trypanosome genomes, 377-378
recombination, viral genetics, 227-228
replication in infectious disease, 509-510
synthesis, RNA viruses, 239-240
viruses
genetics, 227-228
replication and pathogenesis, 239-240
RNA polymerase, structural biology and, Ixi
RNA polymerase II
post-transcriptional regulation, 118
structure and function, 85-86
transcription regulation mechanisms, 495-496
RNA-binding domain (RBD), post-transcriptional regulation,
117-118
RNA-protein interaction
post-transcriptional regulation, 117-118
structural biology and, Ixi, 393-395
Rotavirus diarrhea, control epidemiology, 483-484
Rough endoplasmic reticulum (RER), xxxv
RPS4X, 314
RPS4Y, 314
RuBisCO
NMR accuracy, 57
photosynthesis genetics, 503-505
Saccharomyces cerevisiae
DNA cloning, 303-304
heat-shock proteins (HSPs), 261-262
lysosomal hydrolase sorting, 121
nuclear pore complex (NPC), 99-100
research with, xliv
RNA polymerase II structure and function, 85-86
secretory process, 353-354
Salmonella typhi. See Typhoid fever
Sarcomere, muscle development and function, 287-288
Scleroderma, autoantibody probes, 392
Second messenger systems
cell regulation, 151-152
cell structure and, xxxvi
ion channel regulation, 373-374
neuronal function, lix
caffeine consumption, 401-402
synaptic transmission, 192
Secretory process
intracellular transport, 353-354
neuronal pathways, 405-406
vesicle transport, 354
Segment polarity genes
cell-cell interactions, 298
Drosophila morphogenesis, 23-24
Segmentation
developmental genetics and, 19-20
Drosophila embryogenesis, 113-114
Drosophila morphogenesis, 23-24
Selectins
cell adhesion and, 196
in inflammation and metastasis, 39-40
Self antigens, immune response and, li
Self lymphocytes, immune response, xlvi
Self-reactive T cells, 136-138
Self-tolerance
biology of T cell development, 265-266
mechanisms, 171-172
Selfish (junk) DNA, xxx
Sensory perception, olfaction and, 17-18
Sensory-motor interface, representation-transformation,
523-524
Sequencing
gene cloning and, xliii
recent technology trends, 77
Serotonin, molecular engineering, 426-428
seven-in-absentia (sina) gene, 342
sevenless gene, visual system development, 341-342
Severe combined immunodeficiency (SCID)
antibody gene structure, 9-10
544
Index
blood cell formation, 446
lymphocyte growth regulation, 452-454
Sex combs reduced gene, 223-224
Sex determination, X and Y chromosomes and, 312-314
Sex pheromones, chemical communication, 257-258
Sexual dimorphism, IRE-A DNA-binding protein, 8
SH2 proteins, signal transduction, 517-518
Shaker gene
ion channels, molecular mechanisms, 5-6
neuronal excitability, 205-206
Shiga toxins, protein crystallography, 521-522
Shock, tumor necrosis factor (TNF) and, 35-36
Sialyl-Lewis X determinant, carbohydrate ligands and, 40
Sickle cell anemia, prenatal diagnosis, 217-218
Signal recognition factor (SRF), protein translocons and,
45-46
Signal recognition particle (SRP)
antigen processing, 93-94
protein translocons and, 45-46
translational regulation, 455
Signal transduction
Bcell pathways, 159-160
CD4 and CDS molecules, 181-182
colony-stimulating factor (CSF) 1 receptor (CSF-IR), 370
Drosophila melanogaster, 323
epidermal growth factor (EGF) receptor, 103-104
IRE-A DNA-binding protein, 8
lymphocytes, molecular basis for, 321-322
microbial pathogenesis, 493-494
olfaction and, 331-332
ion channel regulation, 373-374
phosphorylation and protein-protein interactions,
517-518
transmembrane, G protein-coupled receptors, 347
visual system, 193-194
development, 341-342
Signaling systems, neural development, 208
Simian virus 40 (SV40)
macromolecular assembly, 178
T antigen, immune tolerance, 138
Simple sequence repeats (SSRs)
diabetes and, 144
genetic basis of hearing loss and, 1 19- 120
Site-directed mutagenesis, 214
situs inversus, gene therapy and, 308
Skin, keratin expression, 146-148
Small eye (Sey) gene, 270
Small nuclear ribonucleoproteins (snRNPs)
gene expression, 391-392
post-transcriptional regulation, 117-118
RNA processing and behavior, 337-338
Smell. See Olfaction
Sodium channel
lactotropes, 489-490
molecular mechanisms, 5-6
nerve cell electrical activity, 1-2
scorpion toxin receptor genes, 519-520
Somatic gene therapy
blood cell formation, 446
genetic disease and, 449-451
metabolic disorders, 457-458
Somatic-motor cortices, representation-transformation,
523-524
Somatostatin
gene expression, 175
molecular genetics of diabetes and, 27-28
Spectrin, membrane skeleton proteins, 33-34
S-peptide mutants, structural biology, 58
Sperm physiology, ion channels, 491-492
Spermatozoa, second messengers and cell regulation,
151-152
Spherocytosis, hemoglobin synthesis, 218
Spinal muscular atrophy (SMA), 233-234
Spleen focus-forming virus (SFFV), 486
Split genes, RNA editing and, 378
src gene family, 235-236
protein-tyrosine phosphatases, 413-414
SRY genes
insulin-responsive element A (IRE-A), 8
X and Y chromosomes, 312-314
Steel gene
blood cell formation, 445-446
lymphocyte growth regulation, 454
normal and leukemic hematopoiesis, 485-486
rotavirus diarrhea, 484
Stem-loop structures, mRNA decoding, 156
Steroid hormones
gene expression in, 317-318
receptor genetics, 125-127
Stiff-man syndrome, diabetes and, 106
Stress tolerance, heat-shock proteins (HSPs), 261-262
Structural biology, research programs in, Ixi-lxiii
Structure-function studies
calcium channels, 409
egg cells, 385-386
energy-transducing membrane proteins, 213-214
neuronal excitability, 205-206
T cell receptor, 435-436
visual pigments, 291-292
Substance P, molecular neuroimmunology, 319-320
Substrate specificity, chromosome structure, 3-4
Sucrose 6-phosphate synthase (SPS) , 504
Superantigen
T cell function in health and disease and, 273-274
T cell repertoire, 221-222
Superfragment cloning, tumor-suppressor genes, 131
Surface membranes, cell structure and, xxxvi
SV40. See Simian virus 40
"Switch kinases," tyrosine phosphorylation, 56
Sympathetic neurons, cell fate control, 11-12
Synapse-like microvesicles (SLMVs), 105-106
Synapses, function, Iv-lvii
Synaptic plasticity, 219-220
Synaptic potential, Iv
Synaptic transmission
development and function, 355-356
molecular mechanisms, 191-192
Synaptic vesicles (SVs)
neuromodulation, caffeine consumption, 401-402
neuronal secretory pathways, 405-406
traffic in neurons and endocrine cells, 105-106
Synchrotron x-ray sources, Ixi
Synthetase-tRNA complex, protein-RNA interaction,
393-395
Synthetic peptides, scorpion toxin receptor genes, 519-520
Systemic lupus erythematosus (SLE), 391-392
T cell receptor (TCR)
antigen receptors, 101-102
lymphocyte activation, 413-414
molecule studies, 351-352
immune system development and, 138
/3 diversity, 83-84
mammalian memory, 423-424
structure-function studies, 435-436
superantigens and, 221-222
suppression, leprosy and tuberculosis immunity and
pathogenesis, 47-48
T cell receptor a enhancer
gene expression and, 175, 255-256
IRE-A DNA-binding protein, 8
lymphocyte development and neoplasia, 231-232
molecular structure, 41
V/3 portions, 221-222
T cell-specific DNA-binding protein, 175
545
Index
T cells. See also Cytotoxic T cells; Helper T cells
activation and differentiation, 89-90, 209-210
cell surface molecule regulation, 436
afi, mammalian memory, 423-424
biology of development, 265-266
CD4-bearing, 51-52
development, 9-10
epitopes. Salmonella typhi attenuated strains as, 358
76, mammalian memory, 423-424
gene.sequences and, li
HIV gene expression, 285-286
immune system development and, xlvi, 83-84
memory and. 423-424
molecular regulation, 415-416
recognition and differentiation, 101-102
role of, in health and sickness, 273-214
subsets, cytokine regulation of effector functions,
513-514
surface glycoproteins, development and infection,
263-264
V/3 superantigen, function in health and disease, 273-274
Target recognition
growth cone guidance and, 170
neural development, 207-208
Tat protein, retroviral gene expression, 95-96
TATA box, lymphocyte development, 382
TATA-binding protein (TBP), transcription regulation, 422
TdT (terminal deoxynucleotidyltransferase) gene, 381-382
Telomeres
hereditary kidney disease, 333-334
RNA molecules, 7/- 72
Testis-determining factor, IRE-A DNA-binding protein, 7-8
Tetrahymena thermophila enzyme, 71
TflllA, macromolecular assembly, 177-178
Thalassemia
o:-globin gene expression, 259-260
prenatal diagnosis, hemoglobin synthesis, 217-218
(3-Thalassemia (Cooley's anemia), 305-306
Three-dimensional structure, eukaryotic chromosomes,
360-362
"3-4-5" rule, hormone receptor genetics, 126
Thrombin, blood clotting regulation, 344-346
Thrombolysis, tissue-type plasminogen activator (t-PA),
157-158
Thrombomodulin, blood clotting regulation, 344-346
Thrombosis, prevention of clotting, 123-124
Thymidine kinase (TK), viral gene regulation, 279-280
Thyroid hormone, receptor genetics, 125-127
Thyroid hormone receptors (TRs), 75-76
Thyroid hormone-responsive elements (TREs), 75-76
Thyroid-stimulating hormone (TSH)
gene expression regulation, 75-76
post-transcriptional action, 76
Time-resolved imaging, structural biology and, Ixiii
Tissue factor, blood clotting regulation, 344-346
Tissue specificity, epidermal growth factor (EGF), 104
Tissue-type plasminogen activator (t-PA)
coagulation genetics, 161-162
human disease and, 157
Tn7 transposon, 91-92
toll gene, signal transduction in B cells, 159-160
Topoisomerase II, eukaryotic chromosomes, 361
torso gene, embryogenesis transcription control, 112-114
Trans-activating proteins, gene regulation mechanisms and,
421-422
Transacylase, phospholipids and cell regulation, 165-166
Transcription
cell structure and, xxx
cellular
gene expression, 285-286
cellular regulation, 375-376
developmentally regulated genes, 422
Drosophila embryogenesis, 112-114
genetic disease and, xlii-xliv
insulin mechanisms and, 43-44
metamorphosis, 417-418
molecular mechanisms, 339-340
muscle development and function, 287-288
MyoD gene activation, 433-434
promoter-specific regulation, 421-422
regulation mechanisms, 495-496
RNA polymerase 11 structure and function, 85-86
transmembrane signals, 469-470
Transcription factor IID (TFIID), 59-60
Transcription factors. See also Proteins
gene expression, adenovirus as control model, 369-370
genomic response and, 289-290
kidney cell growth and differentiation, 407-408
molecular genetics of blood cells and, 305-306
polypeptide hormones, / 75- / 76
viral gene regulation, 279-280
Transcriptional regulatory complexes, 177-178
Transducin
signal transduction, visual systems, 193-194
transmembrane signal transduction, 347
Transfer RNA (tRNAs)
ribosomal pausing, 455
structural biology and, Ixi
Transforming genes, papillomavirus molecular biology, 242
Transforming growth factor-a (TGF-a:)
cell regulation, 275-276
signal transduction of receptor, 103-104
Transforming growth factor-/? (TGF-/3)
cell regulation, 275-276
limb development genetics in Drosophila, 79-80
Transgenic mice
albinism and, 307-308
B cell development, 299-300
developmental genetics, 383-384
extracellular matrix structure, 49
genetic basis of cancer and, 247-248
a-globin gene expression, 260
immunological self-tolerance and autoimmunity, 1 71- 1 72
mammalian development studies and, 313-314, 315-316
phagocytic cells, molecular genetics of blood cells and,
306
research with, xliv
shock pathogenesis, tumor necrosis factor (TNF) and,
35-36
Transgenic plants, photosynthesis genetics, 503-505
Transient gene expression, gene therapy and, 250
Trans-inactivation, position effects, 183-184
Translation
cell structure and, xxx
insulin mechanisms and, 43-44
mRNA protein synthesis, 455
Translocation
intracellular protein transpon, 353-354
lymphocyte development, molecular genetics, 231-232
Transmembrane signaling, cellular regulation, 375-376
Transmitters, cell structure and. xxxvi-xxxvii
Transplantation
antigens, immunodeficiency and, 325-326
coagulation genetics, 162
complement system, 16
Transport functions, egg structure-function studies, 386
Transporters, cell structure and, xxxvi
Transposition, mechanism of, 91-92
Trembler gene, Charcot-Marie-Tooth disease and, 141
Tricarboxylic acid cycle, metabolic pathways, 429-430
Trophozoites, Entamoeba histolytica detection and
cloning, 516
cv-Tropomyosin (TM) gene, protein diversity and, 288
Trypanosoma brucei rhodesiense, 140
Trypanosomes, mitochondrial genome, 377-378
Trypanosomiasis, immune evasion and, 115-116
546
Index
Tuberculosis
epidemiological analysis, 200-202
immunity and pathogenesis, 47-48
Tumor necrosis factor (TNF)
inflammatory cytokines, 73- 74
regulatory proteins, 387-388
shock pathogenesis, 35-36
Tumor necrosis factor-a (TNF-a), structural studies,
387-388
Tumor necrosis factor-/3 (TNF-/J), structural studies,
387-388
Tumors, cholecystokinin regulation and, 144
Tumor-suppressor genes
current research on, 131-132
p53 protein, tissue-type plasminogen activator (t-PA),
157-158
Turner syndrome, X and Y chromosomes, 314
Type 1 collagen, extracellular matrix, 49
Typhoid fever
microbial pathogenesis, 493-494
molecular biology, 487-488
vaccines with attenuated typhoid strains, 358
Tyrosinase, albinism and, 307-308
Tyrosine hydroxylase (TH), transcription control, 469-470
Tyrosine kinases
cell cycle control, 21-22
host-pathogen interactions, 493-494
signal transduction, 518
Tyrosine phosphorylation, cell regulation, 55-56
Ultrabithorax gene , Drosophila morphogenesis, 24
Upstream stimulatory factor (USF), 59-60
Urea cycle, metabolic pathways, 429-430
Vaccines
immune response and, liv
leprosy and tuberculosis immunity and pathogenesis,
47-48
rotavirus diarrhea, 483-484
Salmonella typhi attenuated strains as, 358
T cell repertoire, superantigens, 221-222
Variability, Entamoeba histolytica, 515-516
Variable domains, protein chains, xlvi
Variable surface glycoprotein (VSG), 115-116
Vascular disease, prevention of clotting, 123-124
Vascular smooth muscle cells, blood vessel function,
459-460
Vasospasm, blood vessel function, 459-460
VDJ recombination, antigen receptor molecule studies,
9-10, 351-352
Vertebrates
embryonic induction mechanisms, 269-270
nervous system, cell fate control, 11-12
Vesicle transport, secretory pathways, 354
vestigial gene . pattern regulation and, 66-67
Viral packaging, RNA viral genetics, 227-228
Viral protein 16 (VPl6), 279-280
Viral replication and pathogenesis, molecular basis of,
149-150
Viral snRNPs, autoantibody probes, 392
Viruses. See also specific viruses, e.g.. Papillomaviruses
genetics, RNA, 227-228
macromolecular assembly, 178
mammalian gene regulation and, 279-280
Visual systems
chromosome 7 mapping, 531-532
Drosophila
development in, 341-342
pattern formation and neuronal cell recognition,
396-398
neural foundations, 283-284
phototransduction of retinal rods and cones, 461-462
sensory representations, 367-368
signal transduction, 193-194, 473
spatial information, 527
visual pigment molecular biology, 291-292
Vitamin D receptor (VDR), hormone receptor genetics, 126
Voltage-clamp technique, nerve cell electrical activity, 1-2
Voltage-dependent sodium channels, 33-34
Voltage-gated calcium channels, 61-62
Voltage-gated ion channels, 5-6
von Recklinghausen neurofibromatosis (NFl), 81
von Willebrand disease
blood clotting regulation, 344-346
molecular genetics, 161-162
von Willebrand factor
blood clotting regulation, 344-346
molecular genetics, 161-162
VPl protein, macromolecular assembly, 178
VP4 protein, rotavirus diarrhea, 483-484
VP7 protein, rotavirus diarrhea, 483-484
Vpsl5 protein (Vpsl5p), 122
Vps34 protein (Vps34p), 122
White blood cells, molecular genetics of, 305-306
white-spotting (W) gene, normal and leukemic
hematopoiesis, 485-486
Wilms' tumor (WT), 131-132
wingless gene
Drosophila morphogenesis, 23-24
limb development genetics in Drosophila, 79-80
oncogenesis, 297-298
Wnt-\ gene family, early embryogenesis, 297-298
Wound healing, cell adhesion and, 195-196
X chromosome
disorders, gene identification and correction, 69- 70
mammalian development, 312-314
molecular genetics, 295-296
structure, xxx, xxxiii
X-linked disorders, 163-164, 431-432
X-linked disorders, 163-164, 431-432
Xq28 chromosome region, X-linked disorders, 163-164
X-ray crystallography
cell-surface recognition, 41
structural biology and, Ixi, 57-58
Y chromosomes, mammalian development, 312-314
Yeast artificial chromosomes (YACs)
chromosome 7 mapping, 531-532
cloning techniques, 82
DNA, large-scale analysis, 303-304
gene expression and cell cycle, cerebellum development,
180
mitochondria, protein translocons and, 45-46
molecular genetics, neuromuscular disease, 233-234
obesity and diabetes, 144
research with, xliv
RNA viral genetics, 227-228
tumor-suppressor genes, 131
urea/tricarboxylic acid cycles, 429-430
Yersinia pseudotuberculosis. 197-198
YY-1 transcription factor, gene expression, 369-370
Zellweger syndrome, urea/tricarboxylic acid cycles, 430
ZFX genes, sex determination, 312-314
ZFY genes, sex determination, 312-314
zif 268 gene, DNA interactions, 309-31 1
Zinc finger proteins
DNA interactions, crystal structures, 309-31 1
kidney cell growth and difl'erentiation, 408
structural biology and, Ixiii
Zygotic genes, Drosophila embryogenesis, 112-114
547
I
I