Full text of "Annual report : National Institute of Child Health and Human Development"

See other formats

.^^vW.#\

^<^^

^ \>.

^^ \ / ,^, % v^'

*«fSS*

r A

iVvw.-^v^*'/

^ ., .... ■%.

^^,

/ ^1

*^^

^^*

'^1

^ V"

.^^

.*<^

^^ >^' ^ti^t \ / .-A% ^ >^^ ^^:% \ >^^

^^" _„. %. #" ... %. ^^" „. %. ^^" ..

ffEASJ

./ ■'^-

■y^ v^

.^^

Qc,

^> <^" /i\ ^> ^" ^£% V </ /i^. V #^^ ^l^% \ y #^\ V ^-^ *

# ; : c^)

f V "«* Z' X '"-^ / X "^

'i^'o,.

'^s^

t«;;(K;

-Olj^^

^HV^

.MAft

Xc.

'■»/

o^ ^ vV 'Y

W^ y

.^^
&-

-0. \. ...^^^

■^c:^^^ </ '^^

%
%

^1^°^^

X. '^tJ / '-

>. ^'- fz\ \ <^ /f^% ^> <^'

-^ -"i-A

-^ v''

'lie

^^\

^ r f^ \

% r

r>^

\ .^^

^>^

^^/ / \

# \ ^^* .^.^ X ^

,6.L "Vis •^

-^^^

.-^^

*>.

# ^<:

\>. '^'^* #■

# ^<:

.£*."

■^.^

l»

J>^

^*:

.•^^

V \.

€

*?!,

..-^"^ ^:>^ ,^^ %> .v^

^ X w / X "^^ .-^^ X ^^"* / X '^^^* .*^" X ^-*' #^ X ''^' /

' ^.^^ X ^^"*;/ X '^^^* .*^" X ^-*'/ X '^fe^^/

^- -* -5^ X .^^* m\ -/fp^ \ /\-A.:\
tJ / X, *' ^,^ X. '^^ y^ X '^-*' .*^"

■• ^ -f i9':f i^^)f i^':f «
.. .^^- -^^j" x^-'/' v^~/ x;^- V \

_^" V .^" V ^-s?-" %. ^^" ^ %. ^^" V .^^ V

> ^

vS)"

*' /^ X'^^ ^^ X '^*' >*^" \'^^' ^^^ X "^^ /

\ ^*' ^^^ X '^^ / X ^-^' ***^ X *^^ #*^ X '^^' .*^" X '^fe^ .*^"
^ / \ *-»^ / \ ^^ / \ *--' / \ '<^ / \ "-" / \ '■^'

^^/s^^ ^*"^A ^^/^^ ^'%m"^ ^^m"^ ^' ^iV"^^ ■

\ \ ^* /^^ \ ^ tj\ S / -■S'. X / #» X ^ .& X / rj.

Ok- <*

1^1 ?^^ 1^1 ?^^ ! 1^1 I

' ^j\.. ^\^'.:\. J^ "- ^^^^ -^^ - ^^- -^ - ^^- -^ -

^^^ X ^^^^ .^^^^ X ^^^' .^^^^ X ^^*^ .^^^ X '^^^'.^^^ X ^*\.^^^ X ^^fe^.

.it/ ..V'A,^ %.

OfflH

44%^

51^^

National Institute of Child Health and Human Development

NICHD

ANNUAL REPORT

FY95

October 1, 1994 to September 30, 1995

PARTS

p^3

DEPARTMENT OF HEALTH A» HUHAM SERVICES - PUBLIC HEALTH SERVICE

NOrrCE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD 00707 11 LCMN

PERIOD COVERED

October 1, 1994 to September 30, 1995

TITLE OF PROJECT (80 characters or less. Title must fit on one tine between the borders.)

Pharmacological Studies of Synaptic Transmission In Vitro

PRINCIPAL INVESTIGATOR (List other professional personnel below the Principal Investigator.) (Name, title.

P.I.: M.L. Mayer Visiting Scientist

Others: K. Partin NRC fellow LCMN,

D. Bowie Visiting Fellcw LCMN,

M. Fleck IRTA Fellow LCMN,

R. Ba hring Visiting Fellow LCMN,

C. Winters Chemist LCMN,

LCMN, NICHD 12 months

NICE© 12 months

NICHD 12 mcxiths

NICHD 9 itonths

NICHD 8 months

NIOiD 8 months

COOPERATING UNITS (if any)

LAB/BRANCH

Laboratory of Cellular and Molecular Neurophysiology

SECTION

Section of Neurophysiology and Biophysics

INSTITUTE AND LOCATION

NICHD, NIH, Bethesda, Maryland 20892

TOTAL STAFF YEARS:
5.1

PROFESSIONAL:
4.4

OTHER:
0.7

CHECK APPROPRIATE BOX(ES)

D (a) Human D (b) Human tissues S (c) Neither
D (al) Minors
n (a7A Interviews

SUMMARY OF WORK (Use standard unreduced type. Do not exceed the space provided.)

This section investigates the functional properties of excitatory amino acid
receptors in the vertebrate CNS, utilizing electrophysiological and molecular

biological technigues. Concentration jump techniques are used to apply glutamate
receptor selective agonists and antagonists to cells and membrane patches under
voltage clamp. Preparations in use include recombinant receptors expressed in
transfected mammalian cells and Xenopus oocytes, and native receptors generated in
primary cultures of hippocampal neurons and glial cells. Analysis of the molecular
basis of allosteric regulation of AMPA receptors is being investigated by
generation of mutants in which conserved amino acids which differ between the
alternative flip/flop exons are tested for sensitivity to potentiation by
cyclothiazide and noncompetitive inhibition by 2 , 3-benzodiazepines . In experiments
on hippocampal neurons multiple rate constants for recovery from potentiation by
cyclothiazide suggest the existence of heteromeric receptors with different ratios
of subunits in their flip and flop splice versions. Inward rectification (
glutamate receptors unedited in the second membrane domain was shown to result from
ion channel block by micromolar concentrations of cytoplasmic polvamines.

ZOl HD 00707 11 LCMN

Project Description;

Objectives ; To characterize the functional properties of excitatory amino
acid receptors, and their role in synaptic transmission in the vertebrate
central nervous system. To investigate neurotransmitter and drug action at
glutamate receptors. To study the molecular properties of ion channels in
neurons and glia, and to characterize mechanisms regulating neuronal and
glial cell excitability.

Methods employed; Electrophysiological techniques are used to study
neurotransmitter action on a variety of preparations including native
receptors expressed in neurons and glial cells, and recombinant glutamate
receptors expressed in oocytes and transfected human embryonic kidney cells.
Ion channels activated by excitatory amino acids are studied using whole cell
patch clamp, the perforated patch technique, outside-out and nucleated
patches, and single channel recording as appropriate for individual
experiments. Drugs are applied by concentration jump using an array of
computer controlled valves and microperf usion tubes which form a rapid drug
application system; two methods are currently in use and involve either
stepper motor or piezoelectric systems. Simultaneous intracellular recording
from pairs of synaptically connected neurons, and analysis of autaptic
responses from neurons grown in microculture are used to study excitatory
synaptic transmission. Computer analysis of the kinetic properties of agonist
evoked membrane currents, single channel activity, synaptic release
probabilities and the simulation of the gating properties of glutamate
receptor channel plays an important role in the design and interpretation of
our experiments.

Major findings: Two major projects were undertaken this year. (1) Analysis
of the action of cyclothiazide and some 2, 3-benzodiazepines on mutant and
native AMPA receptors. (2) Characterization of the mechanisms underlying mRNA
editing sensitive inward rectification of AMPA and kainate receptors. Both
projects were considerably aided by the use of cotransf ection with a reporter
plasmid encoding a jellyfish green fluorescent protein, which allowed
transfected cells to be identified by fluorescence microscopy prior to
recording.

Allosteric modulation by cyclothiazide of mutant AMPA receptors. Recent
studies in our laboratory established that cyclothiazide, an allosteric
modulator of glutamate receptor desensitization with complete selectivity for
activity at AMPA versus kainate preferring receptors, preferentially
modulates the flip versus flop splice variants of AMPA receptors. Sequence
alignments of the flip/flop domains of each of the four AMPA receptor
subunits (which are generated via alternative splicing of two modules each
encoding 38 amino acids) reveal 7 amino acids consistently differing between
the two splice variants. These 7 amino acids are arranged in groups of 2 +
1+4 and are separated by sets of highly conserved amino acids which do not
differ appreciably between subunits or splice variants:

Consensus Flip
Consensus Flop

Thr-Pro-(8 AA)-Ser-(21 AA)-Lys-Asp-Ser-Gly-(2 AA)
Asn-Ala-(8 AA)-Asn-(21 AA) -Gly-Gly-Gly-Asp- ( 2 AA)

To determine the molecular basis for the difference in sensitivity to
cyclothiazide of the flip and flop splice variants we performed site-directed
mutagenesis, exchanging amino acids between flip and flop versions in each
of these three regions, and expressing the mutant receptors either in
Xenopus oocytes or transiently transfected 293 cells, using as an assay each
of the multiple changes in AMPA receptor activity which occur due to
allosteric modulation by cyclothiazide. Comparable sets of mutants were made
for the GluRA and GluRB subunits, with GluRA assayed alone and in combination

ZOl HD 00707 11 LCMN

with GluRB. These experiments revealed that the marked attenuation of
desensitization by cyclothiazide for flip splice variants requires the
isolated serine residue shown by *** in the above alignment, with no
difference in behavior when the C-terminal set of 4 conserved amino acids are
in their flip or flop forms. Conversely, in GluRAflop the presence of an
asparagine residue at this position is sufficient to produce lower
sensitivity to cyclothiazide with no effect observed on exchange of the C-
terminal set of 4 amino acids to their flip forms. In heteromeric AMPA
receptors assembled from wild type GluRA and GluRB the flip splice forms are
dominant for regulating sensitivity to block of desensitization by
cyclothiazide. Experiments for heteroroers generated from mutant GluRAflop and
GluRBflop subunits show that high sensitivity to cyclothiazide typical of
wild type AflipBflip is achieved when a serine residue is introduced into
either the GluRA or GluRB subunits again with no effect of the C-terminal set
of 4 amino acids. Together these experiments show convincingly that all of
the splice dependent effects of cyclothiazide are determined by exchange of
a single amino acid, Ser in flip and Asn in flop.

Agonist Efficacy for activation of mutant AMPA receptors. Prior work revealed
an 8-fold difference in the amplitude of equilibrium responses to maximally
effective concentrations of the agonists kainate and glutamate for AflipBflip
compared to AflopBflop, with intermediate behavior observed for AflipBflop
and AflopBflip. We performed experiments performed on mutant AMPA receptors
which indicate that compared to the site controlling sensitivity to
cyclothiazide, additional residues in the flip/flop modules of heteromeric
AMPA receptors determine the relative amplitude of equilibrium responses to
kainate versus glutamate. The amplitude ratio (kainate/glutamate) of
responses for these agonists was highest for wild type AflopBflop, decreased
approximately 3-fold on exchange of Asn to Ser at positions 750 and 754 in
both GluRA and GluRB, with a similar 3-fold decrease on exchange from GGGD
to KDSG of the C-terminal set of 4 amino acids in both subunits, but remained
below the value for AiBi. The combined exchange of both Asn to Ser and the
conserved set of the C-terminal 4 amino acids to generate Ao[ii]Bo[ii]
produced a 5.4-fold increase in the relative amplitude of equilibrium
responses to glutamate versus kainate. This was still significantly lower
than the 11-fold difference observed for wild type AflopBflop versus
AflipBflip and the additional exchange of residues at the amino terminal end
of the flip/flop modules which differ between splice variants in GluRA and
GluRB was required fully restore the behavior of wild type flip receptors.
Thus, although amino acids which regulate potentiation by cyclothiazide in
heteromers generated from GluRA and GluRB are also critical for attenuation
of desensitization, they are not sufficient to determine the relative
amplitude of responses to kainate versus glutamate,

A Gin residue blocks sensitivity to cyclothiazide.

Considerable homology between kainate and AMPA receptors in the sequences
which align with the alternatively spliced region in AMPA receptors raised
the possibility that the lack of sensitivity of kainate receptors to
cyclothiazide might reflect small sequence differences between kainate and
AMPA receptors over this region. At the position corresponding to the Ser/Asn
site in AMPA receptors all 5 kainate receptor subunits express a glutamine
residue. Substitution of a Gin at position 750 in both AflipS750Q and
AflopN750Q essentially abolished sensitivity to cyclothiazide without
altering agonist potency, indicating that although the introduction of Gin
at position 750 does not confer the high sensitivity to kainate typical of
wild type kainate receptors and which is seen in AMPA receptor chimeras
containing kainate receptor sequences thought to encode the agonist binding
site. To control for non-specific effects resulting from mutations at
position 750 in AMPA receptors, we introduced an alanine at position 750; for
this mutant potentiation by cyclothiazide was intermediate between that for
GluRAi and GluRAo (data not shown), indicating that substitution with an
amino acid other than Ser or Asn is net sufficient to prevent modulation by

ZOl HD 00707 11 LCMN

cyclothiazide. To determine if the absence of kainate receptor sensitivity
to cyclothiazide might arise from the presence of the conserved Gin residue
at the position which in kainate receptor subunits aligns with the Ser/Asn
site in AMPA receptors we constructed the mutant GluR6Q755S. In contrast to
results obtained for the AMPA receptor mutant AiS750Q, the desensitization
kinetics of control responses to 1 mM kainate for GluR6Q755S were not
identical to those for wild type GluR6, with GluR6Q755S showing a reduced
rate of onset of desensitization, and with less desensitization at
equilibrium for GluR6Q755S than for wild type GluR6. However, in contrast to
the cyclothiazide insensitive responses obtained for wild type GluR6, the
rate of onset of desensitization for GluR6Q755S was slowed approximately
five-fold in the presence of 100 pM cyclothiazide. Other functional
properties characteristic of GluR6, including a high affinity for kainate,
lack of sensitivity to AMPA, and pronounced block of desensitization by
concanavalin A were maintained in GluR6Q755S.

Noncompetitive inhibition by 2, 3-benzodiazepines . AMPA receptors show
negative allosteric regulation by 2 , 3-benzodiazepines such as GYKI 53655 and
GYKI 52466 which act as non-competitive antagonists. It has been hypothesized
that cyclothiazide and GYKI 53655 act at a common site. Our identification
of a serine residue that is critical for directing the interaction of
cyclothiazide with AMPA receptors has allowed us to test whether the mutation
of this residue to glutamine, which abolishes potentiation by cyclothiazide,
can in addition block antagonism by 2 , 3-benzodiazepines, as would be
predicted for action at a common site. We found that the S to Q mutation
does not alter antagonism by 2 , 3-benzodiazepines, suggesting that the
molecular determinants directing the interaction between GYKI 53655 and AMPA
receptors are not identical to those controlling sensitivity to
cyclothiazide. Additional support for this was obtained from analysis of the
responses of AMPA receptor flip/flop splice variants, which despite
differences in equilibrium desensitization and sensitivity to cyclothiazide,
do not show differential sensitivity to 2 , 3-benzodiazepines . Of interest,
experiments with AMPA receptors generated from forebrain poly A mRNA revealed
greater sensitivity to 2 , 3-benzodiazepines than receptors generated by
expression of recombinant AMPA receptors, possibly indicating the existence
of an accessory protein or novel receptor subunit which modulates the
activity of AMPA receptors.

Multiple populations of heteroaeric AMPA receptors revealed by kinetic
analysis of sensitivity to cyclothiazide. The rate of recovery from
potentiation by cyclothiazide for two populations of cultured hippocampal
neurons, PI pyramidal-shaped neurons and P8 multipolar intemeurons dissected
from stratum radiatum area CAl shows considerable variability suggesting the
occurrence of multiple populations of AMPA receptors with different
sensitivity to cyclothiazide. Recovery kinetics ranged from predominantly
fast, flop-like fast dissociation, to predominantly slow, flip-like
dissociation, but included combinations with both fast and slow components
as well as intermediate dissociation rates which are not seen with homomeric
receptors. To determine the likely molecular basis for this effect we
examined the rate of recovery from potentiation by cyclothiazide for
recombinant AMPA receptors formed by coexpression of GluRA and GluRB.
Recovery kinetics were found to vary with both subunit and splice-variant
composition. Recovery time constants were 0.2 sec for AoBo, 3.3 sec for AiBo
and > 20 s for AoBi or AiBi. Experiments with GluRA coexpressed at flip/flop
ratios from 1:1 to 1:4 revealed that the intermediate time constant is not
a result of heteromerization between GluRA and GluRB, but reflects the
flip/flop ratio in these heteromers, with recovery becoming faster in
heteromers with high levels of flop subunits. The differential expression of
recovery rate constants in hippocampal neurons likely reflects a similar
mechanism.

Rectification mediated by cytoplasmic polyamines. When AMPA and kainate

ZOl HD 00707 11 LCMN

receptor subunits were first cloned it was discovered that they contain an
RNA editing site in an exon encoding part of the ion channel pore. The
resulting codon change, which causes substitution of an Arg for Gin residue
in the GluRB, GluRB and GluR6 subunits, has profound functional consequences
for ion flow. Receptors with a GLN residue at this site exhibit pronounced
inward rectification, which is absent in subunits with an ARG residue
introduced by RNA editing. Although functional studies on native AMPA
receptors typically reveal outward rectification strongly inward rectifying
AMPA receptor responses with high Ca^* permeability have been observed in
subpopulations of hippocampal interneurons and in Bergmann glial cells, as
well as in native kainate receptors in embryonic hippocampal neurons and in
glial progenitor cells. The mechanisms underlying inward rectification of
glutamate receptor responses have only now been established, and suggest that
the pore region of glutamate receptors has structural and functional
characteristics similar to those for inward rectifying potassium channels,
a gene family previously thought to have no homology to glutamate receptors.
The strong inward rectification typical of whole-cell responses for the
unedited forms of the kainate receptor GluR6 and the AMPA receptor GluRA is
lost following the excision of outside-out membrane patches, suggesting that
a cytoplasmic diffusible factor accounts for inward rectification of
glutamate receptor responses. Consistent with this hypothesis, and similar
to recent studies on inward rectifying potassium channels, the inclusion of
the polyamine spermine in the patch-pipette solution restored strong inward
rectification. In outside-out patches the IV relationship for glutamate
receptor responses in the absence of spermine was close to linear at
potentials between -100 to +50 mV, with weak outward rectification at
membrane potentials between +50 and +100 mV. However, with spermine added to
the internal solution, there was rectification with pronounced voltage
dependent block over the membrane potential range -50 to +50 mV. However,
different from results obtained for inward rectifying potassium channels
there was essentially no block by 1 mM internal Mg^* for glutamate receptors,
indicating that the pore regions are similar, not not functionally identical.
Further evidence for this was obtained from analysis of the effects of
extreme depolarization. This revealed pronounced relief of block over the
membrane potential range +50 to +100 mV for both AMPA and kainate receptors,
strikingly different from the maintained responses observed for potassium
channels. This most likely reflects differences in the pore sizes for
potassium channels, which are narrow and have high selectivity, while
glutamate receptors, which are known to show weaker selectivity, would be
expected to have large pores, and perhaps allow polyamines to permeate at
high membrane electric field strengths.

Estimation of intracellular polyamine concentrations. By using a Boltzman
model for voltage dependent block by internal polyamines in outside patches
we were able to determine values for the affinity (Kd values calculated for
OmV membrane potential) of spermine (S.SpM) , spermidine (25.4/jM) and
putrescine (1.2 mM) for kainate receptors generated by GluR5(Q). For AMPA
receptors assembled from GluRAflip spermine affinity (1.5 pM) was slightly
higher. To analyze the effect of membrane potential on receptor affinity for
spermine, the Kd was determined by dose response analysis at membrane
potentials of -80, -40 and +40 mV. At any given membrane potential,
concentration-dependent block by spemine was well fit by a single binding
site isotherm, with Kd values decreasing progressively with depolarisation,
from 467 /jM at -80 mV, to 1.15 pM at +40 mV. Comparison with data calculated
for potassium channels at +50 mV reveals the affinity of spermine for GluRA
flip to be only 3-fold lower than for the IRKl and BIRIO inward rectifying
potassium channels.

Knowledge of the affinity and voltage dependence of binding of spermine and
spermidine were then used to estimate the cytoplasmic concentration of
polyamines which would account for the rectification of whole-cell responses.
Assuming both spermine and spermidine to be present in the cytoplasm at a

ZOl HD 00707 11 LCMN

ratio of 1:3 (as determined by Watanabe and colleagues in biochemical
experiments) we obtained an estimate for the free spermine and spermidine
concentrations in intact cells of 51 and 153 ^/M respectively, in excellent
agreement with the results of biochemical analysis which revealed that in
mammalian cells the free concentrations of polyamines closely match these
values, but are typically < 10% of total because RNA and DNA generate binding
sites with high capacity and low affinity for spermine and spermidine. The
latter observation most likely explains why the washout of rectification
during whole cell recording is exceedingly slow. Our results help to explain
a paradoxical observation concerning the high Ca2+ permeability but outward
rectifying properties of AMPA receptor responses in patches removed from
cortical interneurons. Based on the properties of recombinant AMPA receptors,
which link Ca2+ permeability to inward rectification, it would have been
predicted that native AMPA receptor responses with high Ca2+ permeability
would show strong inward rectification. That this was not seen in outside out
patches removed from cortical interneurons can be explained by loss of the
cytoplasmic gating mechanism generated by polyamine mediated ion channel
block. In contrast, the extent of inward rectification of glutamate receptor
responses observed during whole-cell recording for hippocampal neurons
expressing native kainate receptors, and for Bergmann glial cells which lack
the GluRB subunit, agrees well with that expected for intracellular polyamine
concentrations in the range 50 - 200 pM.

Significance to Biomedical Research and the Program of the Institute. The

role of glutamate receptors in mediating excitatory synaptic transmission in
all parts of the brain is now widely appreciated, but far from completely
understood. Experiments with molecular cloning have identified multiple
families of receptor subunits, many of which exist in several splice
variants. The functional role of this diversity is still largely unknown but
provides rich potential for regulation during development, and for
therapeutic manipulation. In addition to a basic signalling function
regulating information flow across synapses, excitatory amino acids are
known to markedly influence the development of neuronal circuitry; to
participate in the genesis of abnormal electrical activity associated with
disease states; and to contribute to the pathophysiology of stroke and
cerebral anoxia. The profound changes in central nervous system development
which occur following the experimental manipulation of glutamate receptor
function in embryos and in juvenile animals points to an important role in
regulating growth of the human nervous system during childhood. Recent
experiments with selective NMDA and AMPA receptor antagonists suggest that
drugs targeted at both subtypes of the glutamate receptor family have
anticonvulsant activity, and show neuroprotection in animal models of stroke.
In addition, 7VMPA receptor antagonists produced a profound amelioration of
motor dysfunction in animals models of Parkinson' s disease. Preliminary gene
mapping experiments to locate glutamate receptors in human chromosomes
suggest that abnormalities in their expression could underlie CNS diseases
of previously unknown origin; a clearer understanding of this issue is likely
to emerge as maps are constructed for more subtypes.

A basic understanding of the mechanism of action of excitatory synaptic
transmitters, and factors controlling their release, will be needed to
understand the cellular basis of the complex behaviors regulated by the
family of glutamate receptor ion channels. Research in the laboratory is
directed towards this goal.

Proposed course;

(1) Currently we are making and analysing additional mutations of the
residues in around the critical serine/asparagine site in the flip flop
module of GluRA to probe this region for structural requirements involved in
block of desensitization by cyclothiazide. Our major task during in the
immediate future will be to assay these mutants by rapid perfusion using

ZOl HD 00707 11 LCMN

transfected fibroblasts for desensitization properties in the presence and
absence of cyclothiazide. Information gained from the effects of the amino
acid side chain at position 750 on the kinetics of recovery from potentiation
by cyclothiazide would be expected to give insight into the molecular nature
and function of this site.

(2) Our analysis of the rectification properties of AMPA and kainate
receptors raises the obvious question as to residues involved in binding
polyamines. This will be addressed by constructing mutants in the second
membrane domain, initially targeting conserved aspartate and glutamate
residues surrounding the RNA editing site,

(3) The most likely explanation for our observation that the inward
rectification of glutamate receptors shows relief with strong depolarization
is permeation of the blocker at strong membrane electric field strengths.
Currently little is known concerning the permeability properties of AMPA and
kainate receptors. This will be addressed using organic cations of known
sizes in an attempt to construct a realistic model which can describe the
action of polyamines.

(4) As time and resources permit we still intend to analyse the molecular
pharmacology of kainate receptors. Structure activity analysis of the
responses of native kainate receptors on DRG neurons has revealed a
hydrophobic pocket in the agonist binding site which is likely to contribute
to the substantial difference in affinity for the action of 5-substituted
willardiine derivatives on DRG versus hippocampal neurons. Because the
subunit composition of kainate receptors in DRG neurons has not been
established it is not known whether this represents a general feature of
kainate preferring receptors from the family GluR-5 through GluR-7, whether
coassembly with subunits from the KA-1 KA-2 family is required or alters the
relative selectivity of action of individual willardiines. To understand
more about agonist action at kainate receptor subtypes dose response curves
will be obtained for the key compounds 5-fluoro, 5-iodo, and 5-
methylwillardiine, using oocytes treated with concanavalin A to block
desensitization, and expressing GluR-5Q, GluR-5/KA-l, GluR-5/KA-2, GluR-6Q,
GluR-6R, GluR-6/KA-l, GluR-6/KA-2. In addition the same agonists will be
tested in oocytes expressing the AMPA preferring subunit GluR-A flip, using
cyclothiazide to block desensitization.

Publications

Benveniste, M and Mayer, ML. Trapping of glutamate and glycine during open
channel block of rat hippocampal neuron NMDA receptors by 9-aminoacridine.
Journal of Physiology 1995;483:367-384.

Bowie, D and Mayer, ML. Inward Rectification of Both AMPA and Kainate Subtype
Glutamate Receptors by Intracellular Polyamine Mediated Ion Channel Block.
Neuron 1995 (In Press).

Dani, JA and Mayer ML. Structure and function of glutamate and nicotinic
acetylcholine receptors. Current Opinion in Neurobiology 1995;5:310-317.

Mayer, ML, Partin, KM, Patneau, DK, Wong, LA, Vyklicky, L Jr., Benveniste, MJ
and Bowie, D. Desensitization at AMPA, Kainate and NMDA Receptors. In:
Excitatory Amino Acids and Synaptic Function. Ed. H. Wheal and A. Thomson.
Academic Press (In press).

Mayer, ML, Benveniste, M and Patneau, DK. Molecular pharmacology of NMDA
receptors. In: The NMDA Receptor. 2nd Edition. Ed. G.L. Ccllingridge and J.C.
Watkins. Oxford University Press, 1994; 132-146.

ZOl HD 00707 11 LCMN

Partin, KM, Bowie, D and Mayer, ML. Structural determinants of allosteric
regulation in alternatively spliced AMPA receptors. Neuron 1995;14:833-843.

DEPARTMENT OF HEALTH AND HUKAN SERVICES - PUBLIC HEALTH SERVICE

NOTICE OF IMIRflMURAL RESEARCH PROJBCr

PROJECT NUMBER

ZOl HD 01205-03LCMN

PERIOD COVERED

October 1, 1994 to September 30, 1995

TITLE OF PROJECT (80 characters or Less. Title must fit on one line between the borders.)
Cellular and Synaptic Physiology of Hippocampal Interneurons

PRINCIPAL INVESTIGATOR (List other professional personnel below the Principal Investigator.) (Name, title,

C. J. McBain, Ph.D. Senior Staff Fellow;NICHD-IjCMN 12 months

Others:

Lei Zhang, Ph.D. Visiting Fellow, NICHD-LCMN 7inonths
Gianitiaria Mat^caferri, Ph.D. Visiting Fellcw, NICHD-LCMN 9 months
Jing Du, Ph.D. Visiting Fellow, NICHD-LCMN 12 months
Adaora ChikwencJu, Pre-IRTA Fellow, NICHD-LCMN 10 months

COOPERATING UNITS (if any)

LAB/BRANCH

Laboratory of Cellular and Molecular Neurophysiology

SECTION

Unit on Cellular and Synaptic Physiology

INSTITUTE AND LOCATION

NICHD, NIH Bethesda MD 20892

TOTAL STAFF YEARS:
4.1

PROFESSIONAL:
3.3

OTHER:
0.8

CHEC< APPROPRIATE BOX(ES)

D (a) Human

D (al) Minors
D ra2^ Interviews

D (b) Human tissues S (c) Neither

SUMMARY OF WORK (Use Standard unreduced type. Do not exceed the space provided.)

The aim of this project is to characterize the properties of voltage gated channels

and synaptic transmission of hippocampal inhibitory neurons in the developing brain
and how these properties impact hippocampal function under both physiological and
pathological conditions. Although much is known about their neurochemistry, their
role in the local circuits and the basic electrophysiological properties of
inhibitory interneurons, little is known about the specific ionic or ligand gated
channels expressed on this highly divergent population of cells. A major part of
our effort is to understand the ionic mechanisms which regulate the activity of
these cells and how these mechanisms impact hippocampal function using patch clamp,
immunohistochemical and molecular technigues. Our work over the past year has
focussed on particular populations of inhibitory neurons of the CAl stratum
oriens/alveus and pyramidal cell layers. Particularly we have characterized the
complement of potassium channels present on these cells using both a combined
electrophysiological and immunohistochemical approach. We have determined the roles
of voltage-gated currents in st. oriens-alveus interneurons in determining the
action potential and firing patterns of these cells. In addition we have
investigated the developmental expression and functional role of the K channel
2subunit Kv3 . 1 in parvalbumin containing interneurons of st. pyramidale.The role of
various types of interneuron during the plastic phenomena of long term depression

and potentiation using perforated patch and paired reording technigues has also
been studied .

ZOl HD 01205 03 LCMN

Project Description

Objectives;

The aim of this project is to characterize the properties of voltage gated
channels and synaptic transmission of hippocampal inhibitory neurons in the
developing brain and how these properties impact hippocampal function under both
physiological and pathological conditions. Although much is known about their
neurochemistry, their role in the local circuits and the basic
electrophysiological properties of inhibitory interneurons, little is known about
the specific ionic or ligand gated channels expressed on this highly divergent
population of cells. A major part of our effort is to understand the ionic
mechanisms which regulate the activity of these cells and how these mechanisms
impact hippocampal function.

Methods Employed;

Whole cell patch clamp studies under current and voltage clamp configurations
have been performed from visually identified inhibitory neurons of the st. oriens
/alveus, St. lacunosum-moleculare and stratum pyramidale, maintained in
hippocampal slices from neonatal animals. In addition outside-out macroscopic
patches have been excised from the soma of interneurons held in the slice
preparation in order to provide an accurate estimation of the activation and
inactivation profiles of novel potassium channels expressed on these cells.
Simultaneous whole cell patch clamp from interneurons and pyramidal neurons
together with field recordings have been made in studies observing the role of
interneurons in synaptic plasticity. Perforated patch techniques are now employed
in studies investigating long-term potentiation in three different subtypes of
interneuron. Combined electrophysiology, morphology and immunohistochemistry
techniques are used on single inhibitory cells in the slice preparation to
correlate interneuron physiology with the expression of particular potassium
channel subunits. Patch clamp recordings from primary cultures of st. lacunosum-
moleculare interneurons are being used to characterize the complement of
potassium channels expressed on these cells.

Major Findings:

1. Passive propagation of LTD to st. oriens-alveus inhibitory neurons modulates
the temporoammonic input to the hippocampal CAl region.

In order to study the role of st . oriens cells during a long-term depression
(LTD) -inducing protocol Gianmaria Maccaferri has made whole cell patch clamp
recordings from horizontally oriented st. oriens interneurons while
simultaneously monitoring the field potential (fPS) of CAl pyramidal neurons in
the hippocampal slice. Morphological characterization was determined by post hoc
biocytin staining techniques. Application of a low-frequency stimulation protocol
to St. radiatum afferents has been shown to be a reliable method to elicit
homosynaptic LTD in CAl pyramidal cells. Under our experimental conditions, this
protocol induced LTD of both the fPS and St. oriens/alveus EPSPs. The
simultaneous depression of the fPS and st. oriens/alveus EPSP could be explained
either by the induction of an identical phenomenon at the feed-forward synapse
between the St. radiatum afferents and the St. oriens interneuron or
alternatively, St. oriens/alveus EPSPs could be generated by the feed-back
recurrent collaterals of CAl pyramidal cells. In the former hypothesis st . oriens
EPSPs would undergo a real, direct LTD process, while in the latter the effect
would be the result of the feed back circuitry synapses onto these interneurons
and not a direct induction of LTD.

If the observed LTD of st. oriens/alveus was a direct cellular phenomenon similar
to that described in CAl pyramidal cells, its induction should be Ca dependent.
Accordingly, strong buffering of [Ca 'i levels in the St. oriens/alveus should

ZOl HD 01205 03 LCMN

disrupt the observed synaptic plasticity. Under conditions to buffer
intracellular calcium levels, both LTD of the fPS and the St. oriens/alveus EPSPs
were still observed. Moreover, application of a repeated high frequency tetanic
stimulus after 30 min. of stable depression could restore or even potentiate the
original synaptic response. This experiment ruled out the possibility that the
LTD elicited in the St. oriens cells was a direct cellular phenomenon.
Alternatively it may simply result from a reduction of the feedback excitation
of St. oriens cells by pyramidal neurons due to the direct depression of the CAl
pyramidal neurons as suggested above.

Analysis of the temporal relationship between the peak of the fPS and the onset
of the interneuron EPSP revealed that the onset of the St. oriens EPSP always
occurred after the peak of the fPS. Since the field population spike reflects the
discharge of a large population of pyramidal cells, we conclude that under our
experimental conditions horizontal st. oriens interneurons receive their
excitatory drive from CAl pyramidal neurons and are involved in pure feedback
mechanisms of inhibition. This was an extremely important observation since until
now interneurons of the St. oriens-alveus subfield have been considered to
additionally receive feedforward afferents from the CA3 pyramidal subfield.

To determine whether the monosynaptic input to the st . oriens interneurons could
undergo cellular LTD we directly stimulated the axon collaterals of the CAl
pyramidal neurons which drive the excitation of the St. oriens/alveus
interneurons. Under these conditions pure antidromic fPS were recorded from St.
pyramidale however the low-frequency protocol consistently failed to elicit any
form of synaptic plasticity suggesting that st. oriens cells do not demonstrate
a direct form of LTD. We next made perforated patch recordings to prevent the
possible washout of some essential intracellular component necessary for the
induction of LTD. Using the antibiotics nystatin, amphotericin and gramicidin to
"perforate" the area of membrane under our electrodes we were still unable to
identify any direct form of cellular LTD in these neurons. This is the first
demonstration of a network effect that results in a long-term modification of the
excitatory synaptic responses of CAl interneurons.

This form of plasticity which we refer to as "passively propagated plasticity"
could play an important role in the regulation of information processing in the
hippocampal circuit. Morphological analysis of the cells used in the present
study show that recordings were obtained from cells with a horizontal dendritic
arborization restricted to the St. oriens-alveus layer and whose axons terminate
on the distal dendrites of pyramidal cells in st. lacunosum-moleculare. This
region of the pyramidal neuron dendritic tree receives direct excitatory input
from the temporoammonic pathway arising in the entorhinal cortex (EC) . We next
demonstrated that the induction of LTD in CAl pyramidal neurons by Schaffer
collaterals, not only depresses synaptic activity onto pyramidal neurons, but
depress excitation of the horizontal interneurons of St. oriens which results
in a reduction of their inhibitory output onto the distal portion of the
pyramidal neuron dendrites. The net result of the LTD formation acts as a
"switch" in the source of synaptic input onto CAl pyramidal neurons from the CA3
region to the EC. This allows afferent activity to bypass the classic trisynaptic
circuit and to directly activate the distal dendrites of the CAl pyramidal cells
by a mechanism involving the disinhibition of the temporoammonic pathway.

2. Do direct forms of synaptic plasticity occur on any of the subpopulations of
CAl interneuron?

During the course of performing the above experiments two reports were published
stating that in contrast to our observations, cellular plasticity could indeed
be observed in two distinct populations of CAl interneurons. Neither of these
studies sufficiently controlled for the possibility of a passive propagation of
plasticity induced in the CAl pyramidal neurons to these types of interneurons.
To directly address this possibility and to determine whether any other
subpopulation of interneuron with the CAl subfield did indeed posses cellular

1..

ZOl HD 01205 03 LCMN

plasticity, we designed experiments utilizing perforated patch clamp techniques.
Using this technique Giatunaria Maccaferri was able to discriminate between direct
cellular forms of plasticity and those induced by the passive propagation of
plasticity from the principal pyramidal neurons. Using a pairing protocol (to
elicit direct forms of plasticity) followed by a tetanic stimulation paradigm (to
elicit a large population plasticity) we identified a cellular form of long-term
potentiation in only one of the three morphologically identified cell types so
far tested. In an interneuron type located in stratum radiatum, the pairing
protocol induced robust long lasting cellular LTP. In contrast, a distinct
population of st. radiatum interneurons lacked both cellular and 'passively
propagated" forms of LTP. In all interneurons of the stratum oriens we were
unable to find a cellular form of LTP but similar to the data reported above
these cells only posses a 'passively propagated" form of LTP. Experiments are in
progress to determine the underlying mechanisms involved in the cellular form of
LTP observed in the stratum radiatum interneurons,

3. Potassium conductances underlying action potential repolarization and
afterhyperpolarization in rat CAl hippocampal interneurones

Lei Zhang and Chris McBain have determined the role of a variety of potassium
conductances in action potential repolarization and afterhyperpolarization.
Whole-cell current clamp recordings were made from visually identified
interneurons in hippocampal slices (300fjm) of neonatal rats (11-20 days old).
Biocytin was introduced during the whole cell recording to permit visualization
of the interneurons of interest. 4-Aminopyridine prolonged the action potential
repolarization. The effects of 4-AP persisted in Ca -free conditions. Action
potentials evoked from hyperpolarized potentials possessed an increased rate of
repolarization. These data suggest an involvement of the rapidly activating
transient current, I, in spike repolarization. Action potential duration was
increased in the presence of Ca -free, Cd -containing solution, iberiotoxin or
ImM TEA. The fast component of the AHP was attenuated by these agents suggesting
that the Ca^*-activated K* conductance I. underlies both the spike repolarization
and fast AHP.

In Ca -free conditions, TEA (>lmM) dose-dependently prolonged the action
potential duration by blocking a late conductance in action potential
repolarization, suggesting a role for the sustained current, 1^.
The slow AHP was attenuated by Ca^*-free medium, apamin or the Ca^* chelator EGTA,
suggesting a role for the Ca^*-activated K conductance, I^^^^p. We conclude that
action potential repolarization and AHPs of St. oriens-alveus interneurones
result from the activation of pharmacologically distinct, temporally overlapping
potassium conductances. These findings were then correlated with the voltage
clamp data obtained in a parallel study from the same neuron class.

4. Voltage-gated potassium currents in stratum oriens-alveus inhibitory neurones
of the rat CAl hippocampus.

Lei Zhang and Chris McBain have recorded voltage-activated K* currents from
visually identified inhibitory interneurones of the CAl St. oriens-alveus region
in neonatal rat hippocampal slices using outside-out patch and whole-cell voltage
clamp techniques. Voltage-activated outward K* currents comprised both a
transient and a sustained component when elicited from a holding potential of -90
mV. Tail current analysis of both transient and sustained current reversal
potentials showed that both outward currents were predominantly carried by
potassium ions. The transient current, I,, was activated with a time to peak
within 5 ms, inactivated with a time constant "15 ms at 0 mV and possessed half
activation at -14 mV. Half-inactivation of the transient current occurred at -71
mV. At -90 mV, the transient current recovered from inactivation with a time
constant of 142 ms. The properties of the transient current are virtually
identical to the transient current observed in the principal pyramidal neurons

ZOl HD 01205 03 LCMN

of the hippocampus. Activation of currents from a holding potential of -50mV
permitted isolation of a sustained current, I|^. In Ca^*-free conditions the
sustained current showed rapid activation, reaching about 80% of its maximum
within 1.5 ms, and showed little inactivation during 1 second depolarizing steps.
The majority of sustained outward currents showed no voltage-dependent
inactivation. In "20% of cells, a slow time-dependent inactivation of the
sustained current was observed, suggesting the presence of a second type of
sustained current in these cells. A Ca'^-dependent K* current comprised a
significant portion of the total sustained current, this current was activated
at voltages positive to -30 mV and showed no time dependent inactivation over a
1 second depolarizing step. This current component was removed in Ca -free
conditions or by iberiotoxin. Low concentrations of 4-AP (50 pM) attenuated both
the transient and sustained current components recorded in a Ca^*-free solution.
Higher concentrations of 4-AP (< 10 mM) were without further effect on the
sustained current but completely blocked the transient current with an ICcg of
1.8mM. TEA blocked the sustained current with an ICcq of 7 . 9mM without
significantly reducing the transient current. Both current components were
resistant to dendrotoxin (500nM). The properties of the sustained current
components are distinct from the sustained currents recorded in pyramidal neurons
suggesting that these currents in part determine the fast-spiking properties of
this interneuron population

5. Properties of the hvperpolarization-activated current (Ih) in CAl hippocampal
St.-oriens-alveus interneurons.

Despite the fact that interneurons constitute only about the 15 % of the total
hippocampal cell population, their role in the regulation of excitability is
believed to be crucial. The intrinsic electrophysiological properties of these
cells play an important role in the generation of their spontaneous activity.
Among the many different kind of voltage-operated conductances which are present
in these interneurons, the hyperpolarization-activated current (Ih) is
particularly interesting in this respect. In fact, Ih has been shown to be
involved in the regulation of the firing pattern of thalamic neurones, is present
in many different neuronal preparations, and was firstly described as the current
responsible for spontaneous cardiac pacemaking. In order to study the basic
properties of this current in St. oriens-alveus inhibitory neurons, Gianmaria
Maccaferri has made whole-cell voltage-clamp recordings from visually identified
interneurons in the hippocampal "in vitro" slice preparation. Ih was activated
as an inward current by hyperpolarizing voltage pulses in the range -60 to -130
mV from a holding potential of -40 mV. The activation kinetics was very slow (t
200 ms at -130 mV) and no inactivation properties were detected. The activation
curve showed a threshold at ~ -60 mV, was saturated at ~ -130 mV and had a mid-
activation point at "-90 mV. Consistent with its mixed ionic nature (Na* and K*),
its reversal potential was "-30 mV, as obtained by extrapolation of the fully
activated Ih I/V relationship. Finally, complete blockade of the Ih conductance
could be achieved after cell exposure to external Cs^* (0.25-5 mM) or the
specific Ih antagonist Zeneca ZD7288 (100 um) . To determine whether Ih played a
role in the regulation of any action potential firing properties in these cells,
recordings were made under current clamp conditions in the presence of these
selective Ih antagonists. Following blockade of excitatory synaptic transmission
with DNQX (20 /jM) , application of either Cs or ZD7288 caused a dramatic decrease
in the rate of spontaneous firing, together with a large membrane
hyperpolarization. In conclusion, these experiments reveal the presence of the
hyperpolarization-activated current in st. oriens-alveus interneurons and
demonstrate its role in determining the spontaneous firing pattern. Finally,
experiments are in progress to determine any possible modulation of this
conductance by putative neurotransmitters, which could add a new tool in the
regulation of the physiology of these interneurons.

6. Developmental Expression and Functional Characterization of the Potassium
Channel Subunit Kv3 . lb in Parvalbumin-Containinq Interneurons of the Rat

ZOl HD 01205 03 LCMN
Hippocampus.

Previous in situ hybridization studies of the distribution of the Shaw (Kv3) K
channels have shown that two members of this subfamily, KvS.l and Kv3.2, are
expressed in low levels in the principal cells of the hippocampus i.e. the
pyramidal cells. The expression of these subunits is diffuse and only a small
number of densely labeled cells, scattered throughout the hippocampus were
observed, tentatively identified as interneurons. In the CAl region, cells
labeled with Kv3 . 2 probes were more abundant in the st. oriens and occasionally
along the border of st. pyramidale and were rarely seen in the st. radiatum or
lacunosum. In contrast cells hybridizing with Kv3.1 are seen both along the
stratum radiatum-pyramidale border and in stratum oriens/alveus. The significance
of the selective expression of these potassium channel subunits into the
inhibitory neuron population is at present unclear.

In a study led by Jing Du, the expression of the voltage-gated potassium channel
subunit Kv3.1b in the developing hippocampus was determined by immunoblot and
immunohistochemical techniques. Kv3.1b protein was first detected at postnatal
(P) day 8 using a specific polyclonal antibody obtained from Dr. Bernardo Rudy
(NYU) . The Kv3 . Ib-immunopositive cell number per tissue section reached a maximum
at P14 and was maintained through P40. In contrast, the KvS.lb protein content
of isolated membrane vesicles in immunoblots progressively increased through P40
suggesting an increase in Kv3.1b content per cell throughout this time period.
Kv3.1b protein was selectively expressed in the somata, proximal dendrites and
axons of cells lying within or close to the pyramidal cell layer consistent with
their being GABAergic inhibitory interneurons. Kv3.1b was present in "80% of
parvalbumin-positive interneurons. The developmental onset of Kv3 . lb and
parvalbumin immunoreactivity was identical. In contrast, Kv3.1b was largely
absent from the subset of somatostatin-positive inhibitory interneurons, which
are known to be a distinct class of inhibitory interneurons from the parvalbumin
containing cells. Electrophysiological recordings were made from St. pyramidale
interneurons whose morphology and Kv3 . lb-positive immunoreactivity were confirmed
post hoc. Outward currents in these cells had voltage-dependent and biophysical
properties resembling those of channels formed by Kv3.1b. The current blocked by
low concentrations of 4AP showed marked inactivation, suggesting that KvS.lb may
coassemble with other members of the Kv3 subfamily. In current clamp recordings,
concentrations of 4AP which blocked the current through Kv3.1b channels allowed
us to tentatively assign a role to Kv3.1b containing channels in action potential
repolarization. The developmental onset of parvalbumin-immunoreactivity in the
cells has long been associated with the acquisition of their " fast-spiking"
characteristics. In the present study however, we have demonstrated that the
onset of expression of Kv3.1b is coincident with acquisition of parvalbumin
immunoreactivity. This suggests that the fast-spiking activity of these
interneurons is not determined by the onset of parvalbumin-immunoreactivity per
se as previously thought, but may instead be determined by the onset of Kv3 . lb
expression, a channel which may have a direct involvement in the determination
of the action potential characteristics in these cells. The observation that the
Kv3.1 subunit is preferentially expressed in many cells known to be GABAergic and
possess high firing rates is probably not coincidental. The activation of Kv3.1
channels may act to keep the action potential short. In addition, the calcium-
binding protein, parvalbumin, could serve to buffer the intracellular calcium
concentration. Both factors could act to prevent the accumulation of Ca^* and the
activation of Ca^*-activated outward currents. Furthermore, the rapid
deactivation of Kv3.1 channels, will help decrease the membrane conductance soon
after the peak of the spike. Taken together these electrical properties of these
neurons will ensure a high frequency of action potential firing. More likely, the
observation that Kv3 . lb-containing channels help to shape the action potential
of " fast-spiking' interneurons will also extend to action potentials of other
cells strongly expressing Kv3.1b or other Kv3 related subunits. These data
demonstrate that Kv3 . lb is developmentally regulated in a specific subpopulation
of hippocampal interneurons and that channels containing this subunit may be a
major determinant in imparting " fast-spiking" characteristics to these and other

ZOl HD 01205 03 LCMN

cells throughout the CNS containing the Kv3.1b subunit.

7. Two distinct * delaved-rectif iers' determine the voltage-dependent potassium
current phenotvpes in st. lacunosum-moleculare interneurons in primary culture

Adaora Chikwendu has established primary cultures of interneurons derived from
the St. lacunosum-moleculare subfield of the CAl hippocampus. Whole cell,
voltage-clamp recordings were made from cells 4-8 days in vitro. In cells
possessing a bipolar morphology, "delayed rectifier" outward K* currents were
activated in all cell types at test potentials positive to -40mV (V„ = -60mV) .
One of two current phenotypes was usually observed in any given cell which were
termed " sustained" and ' slowly-inactivating" . The time to peak of the sustained
current was identical at all test potentials. In contrast the time to peak of the
slowly inactivating current was markedly voltage dependent, decreasing at more
positive test potentials. The voltage-dependence of activation of either current
phenotype, could be fit by the Boltzmann equation both of which possessed a half
activation close to +4mV (n = 52 and 38 for each current type). These outward
currents showed minimal ("20%) voltage-dependent inactivation (Vhalf = -55 and -
67mV) . In cells possessing predominantly "slowly-inactivating" outward currents,
4-AP dose-dependently (10^M-30mM) blocked a fraction of the total outward current
with an IC^q of 312 ± 101/iM. At a maximal concentration of 30mM, 4-AP selectively
blocked 41% of the total current. The block by 4AP was not voltage-dependent at
any concentration tested, and blocked an identical fraction of current at all
test potentials. The current was however use dependent and the total block was
only observed following numerous steps to a given test potential. Isolation of
the 4-AP sensitive component yielded a ' slowly-inactivating" current with a
positive voltage dependence of activation (Vhalf = +10.9 ± 1.7mV). The current
remaining in 30mM 4-AP was "sustained" and also possessed a positive Vhalf of 7.2
± 1.2mV. In cells where a "sustained" outward current dominated in control
conditions, 4-AP (30 mM) again removed a "slowly-inactivating" component which
represented only 23% (n = 5) of the total outward current. TEA dose-dependently
(10/jM - 30mM) blocked the total outward currents with an ICcg of 142 ±47/jM. In
30mM TEA, 85% of the total outward current was blocked regardless of whether the
current phenotype was predominantly "sustained" or 'slowly inactivating".
However, low concentrations of TEA (10 - lOOmM) selectively removed a "slowly-
inactivating" current component (Vhalf = 10.1 ± 1.3mV, n=5). The block by TEA,
like 4AP was not voltage-dependent. Inclusion of TEA (1 and lOmM) in the internal
solution caused an apparent increase in the fraction of current blocked by
external 4AP, suggesting that these two antagonists act at different sites on the
ion channel. In contrast, external TEA blocked a similar fraction of current
whether TEA was included in the internal solution or not, suggesting that
external TEA may displace internal TEA from its site of action with the ion
channel. Inclusion of 4AP in the internal solution caused an apparent reduction
in the fraction of current blocked by external TEA suggesting that these agents
are blocking a similar current component. Internal 4AP however was without effect
on the current fraction blocked by external 4AP suggesting a mechanism similar
to that when TEA was present both on the inside and outside of the channel. I n
conclusion, bipolar st. L-M interneurons have at least two "delayed rectifier"
currents with similar activation profiles which can be differentiated based on
their sensitivity to 4-AP and TEA. The differing proportions of either current
component usually determines the overall current phenotype in any given cell.

Proposed Course

The future directions of this research program are two-fold. First we propose to
continue our studies involving the synaptic plasticity properties of interneurons
within the hippocampal slice formation, secondly to continue our studies on the
characterization of voltage-dependent potassium currents present on hippocampal
interneurons

1. An understanding of the role of each interneuron type in various forms
of synaptic plasticity is an important step in elucidating the precise

ZOl HD 01205 03 LCMN

roles of these cells in the network properties of the hippocampal
formation. To this end we are presently studying the roles of several
distinct interneuron types of the CAl subfield during the formation of
long-term potentiation. A detailed pharmacological investigation will be
made on those cells to determine whether or not this type of LTP is
calcium and NMDA receptor dependent. These studies will then be extended
to characterize the spiny and aspiny neurons of the dentate hilar region.
Since little more than the basic cellular properties of these important
cell types are known, an understanding of the types of long-term plastic
changes associated with these cells will be important in determining the
permissive role these cells play in determining information flow into the
tri-synaptic circuit.

2. We propose to use paired whole cell patch clamp recordings from
functionally connected interneurons of st. oriens and alveus and pyramidal
neurons. Since we are the first lab to demonstrate that horizontal
interneurons of the st. oriens are driven purely by the recurrent
collaterals of CAl pyramidal neurons we propose to study this selective
innervation more carefully using recordings between connected pairs under
both physiological and high-K+ seizure inducing conditions. The use of
pairs of whole cell voltage and/or current clamp recording from these
cells will permit the unequivocal identification of their role in the
synaptic circuit, the "strength* of the synaptic transmission between
these two cell types and the identity of the glutamate receptors involved.

3. Since one of our long term goals is to elucidate the specific role of
K* channels in interneuron physiology, we will continue the
electrophysiological characterization of specific K* channel conductances
present on a variety of interneuron subtypes concomitant with
immunological detection of the appropriate subunit on the cells from which
electrophysiological recordings have been made. In a continuation of our
Kv3.1 subunit study on basket cells of the CAl pyramidal cell type, we
will characterize the role of the subunit Kvl.4 which would appear also to
be expressed on these cells. We anticipate having a functional antibody to
this channel in the near future and will perform experiments similar to
those reported above for Kv3.1. The identification of Kvl.4 in these cells
is an important one since its gating mechanism is directly modulated by
changes in the extracellular potassium concentration. This property is
peculiar to channels formed by this subunit and will likely have important
consequences during electrographic activity. To determine any potential
role for this channel in regulating interneuron activity during
electrographic seizure activity we will directly monitor the properties of
this channel in the high-K* model of electrographic seizure already used
in this laboratory.

4. In a continuation of the study involving potassium currents in primary
cultures of st. lacunosum-moleculare interneurons, we propose to interrupt
putative potassium channel function in these cells with the delivery of
antisense oligonucleotides. In addition, a parallel study will investigate
antisense oligonucleotide knock-out of K channels subunits in cultures of
cerebellar granule cells (cells rich in the potassium channel subunits
Kv3.1 and Kvl.4). Initially antisense oligonucleotides targeted to both
Kv3.1 and Kv3 . 2 will be synthesized and used to block expression of these
channels in both these primary cultures since both subunits are thought to
be strongly expressed in these cell types. Immunohistochemical and
molecular approaches will determine the level of expression of down-
regulated K+ channels on these cells. Subsequent electrophysiological
experiments will determine the physiological consequences of preventing
the protein synthesis of these K channels. As the expression patterns of
K* channels subunits on hippocampal interneurons is characterized, more
relevant K* channel subunits as targets for antisense knockout will be
identified and tested.

J.O

ZOl HD 01205 03 LCMN

Significance to Biomedical Research and the Program of the Institute

Inhibitory interneurons represent '10% of the total neuron population of the
hippocampus proper and several distinct subpopulations of cell type have been
characterized in this lab and others. The high levels of tonic inhibition present
within the hippocampal formation under physiological conditions highlights the
importance of an understanding of interneuron physiology. The fast spiking
characteristics of the st. oriens interneurons present the principal neurons of
the hippocampus; the pyramidal neuron, a 10 fold higher level of inhibition
synaptic input than excitatory input. The critical role inhibitory neurons play
in the regulation of hippocampal excitability is underscored by the large number
of existing seizure models which compromise the GABAergic inhibitory mechanism.
Thus, an understanding of the basic physiological and pathological properties of
these neurons may elucidate mechanisms which could be useful for the management
or the prevention of seizure activity in the hippocampus. To this end we have
focused on the compliment of voltage gated potassium channels on these cells and
their roles in various forms of synaptic plasticity. We anticipate that the
characteristic physiological properties of these cells are due largely in part
to the compliment of voltage gated channels present on these neurons. It is a
long term goal of this lab to exploit the novel potassium channels present on
these cells responsible for the regulation of their excitability.

ZOl HD 01205 03 LCMN

Publications

Journal Articles

Desai, M, McBain, CJ, Kauer, J & Conn, J. Metabotropic glutamate receptor-
induced disinhibition is mediated by reduced transmission at excitatory synapses
onto interneurons and inhibitory synapses onto pyramidal cells. Neuroscience
Letters 1994;181:78-82.

Maccaferri G & McBain CJ. Passive propagation of LTD to st . oriens-alveus
inhibitory neurons modulates the temporoammonic input to the hippocampal CAl
region. Neuron 1995;15:137-145.

McBain, CJ. Hippocampal interneuron activity in the elevated potassium model of
epilepsy. Journal of Neurophysiology 1994;72:2853-2863. Erratum appears in vol.
73.

Zhang, L & McBain, CJ. Potassium conductances underlying action potential
repolarization and af terhyperpolarization in stratum oriens-alveus inhibitory
neurones of the rat CAl hippocampus. Journal of Physiology 1995, in press.

Zhang, L & McBain, CJ. Voltage-gated potassium currents in stratum oriens-alveus
inhibitory neurones of the rat CAl hippocampus. Journal of Physiology 1995, in
press.

Il5

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBLIC HEALTH SERVICE

NOmCE OF INIRAMURAL RESE?VRCH FROOBCT

PROJECT NUMBER

ZOl HD 1206 02 LCMN

PERIOD COVERED

October 1, 1994 to September 30, 1995

TITLE OF PROJECT (80 characters or less. Title must fit on one line between the borders.)
RECEPTOR MEDIATED CALCIUM SIGNALLING IN GLIA AND NEURONS

PRINCIPAL INVESTIGATOR (List other professional personnel below the Principal Investigator.) (Name, title.

P.I.: Dr. J. T. Russell

Others: Ms. L. A. Holtzclaw
Dr. C. A. Shepperd
Dr. N. Darvish
Dr. P. B. Simpson

Head, Section on LCMN, NICHD 12 months
neuronal secretory systems

Biologist
^JRC Fellow
Visiting Fellow
Visiting Fellow

LCMN, NICHD 12 months

LCMN, NICHD 10 months

LCMN, NICHD 8 months

COOPERATING UNITS (if any)

Dr. V. Gallo
Dr. D. C. Klein

Head, Unit on Neurobiology
Head Section on
Neuroendocrinology

LCMN, NICHD
LDN, NICHD

LAB/BRANCH

Laboratory of Cellular and Molecular Neurophysiology

SECTION

Section on Neuronal secretory systems

INSTITUTE AND LOCATION

NICHD, NIH. Bethesda, Maryland 20892

TOTAL STAFF YEARS:
4.0

PROFESSIONAL:
3.5

OTHER:
0.5

CHECK APPROPRIATE BOX(ES)

D (a) Human D (b) Human tissues S (c) Neither

D (al) Minors
D Ca2) Interviews

SUMMARY OF WORK (Use standard unreduced type. Do not exceed the space provided.)

This Section aims to understand the principles and mechanisms that govern
intracellular calcium signalling in glial cells and neurons. Although neurons are
the central players in cellular communication and information processing in the
brain, recent discoveries of glial cell types, excitability and signalling
properties have shown that glial cells may play an important role in long distance
signalling in the brain. Glial cells do not show electrically excitable properties
similar to neurons. They do, however, possess a form calcium based excitability in
response to activation of neurotransmitter receptors. Most receptor mediated
regulation of physiological processes utilize cellular calcium signals as triggers
In glia, like in other cells, such calcium signals have recently been recognized to
have complex temporal and spatial characteristics. The research focus of the
Section includes studies on the signal transduction mechanisms of various receptor
systems in glial cells as well as the cell biology of calcium signalling in both
glial cells and neurons. One objective is to understand cellular processes that
control signal transduction mechanisms for Ca^* signalling and cellular processes
that regulate temporal and spatial signalling within cells and between cells at the
molecular level. A second objective is to understand the processes that regulate
calcium based excitability in glial cells resulting from neuronal activity
Thirdly, we would like to understand the changes in the signalling mechanisms that
occur in reactive astrocytes during brain injury. Finally, this Section aims to
describe in detail the ionic basis for cellular excitability in pineal cells.

ZOl HD 1206 02 LCMN

Project Description;

Objectives: The aim of this Section is to describe in detail glial cell
signalling in response to neuronal activity in the normal brain and during brain
injury. During neuronal activity, it is believed that astrocytes monitor
synaptic activity and respond to released transmitters with intracellular calcium
signals by the activation of metabotropic receptor systems. These signals are
propagated as waves within single astrocytes and are transmitted through a
complex network of glial cells connected by gap junctions over long distances,
albeit at rates considerably slower than (approximately 1000 times) neuronal
signals. This has been termed calcium based excitability. This Section has
characterized such signals in tissue cultured glial cells in some detail and has
described a mechanism of saltatory propagation of calcium waves within cells.
A number of other laboratories also have studied such glial cell signals. All
these experiments, however, have been carried out on cells in tissue culture, as
purified glial cell cultures or as mixed cultures with neurons. The singular
question that remains unanswered is, if such signalling occurs in the intact
brain during normal brain function. One set of experiments were carried out in
a more intact preparation of hippocampal slices in organotypic culture. Studies
on acutely isolated brain slices are not yet available. This Section proposes
to examine if glial cells In situ participate in signalling as well as to study
the cellular mechanisms and processes that support such long distance signalling
by glial cells. Secondly, we propose to examine if changes in glial phenotype
following neuronal injury (reactive gliosis) result in changes in the signalling
modalities in these cells. During the year 1995- ' 96, we aim to study the
following aspects of signal transduction and glial cell calcium signalling: (i)
Investigate the receptor repertoire and mechanisms of calcium signalling in the
different types of glial cells isolated from rat brain. (ii) Investigate the
calcium release process by the endoplasmic reticulum membrane system in glia.
(iii) Investigate the structure of endoplasmic reticulum in astrocytes with
reference to its Ca^* signalling function. (iv) Investigate calcium based
excitability in neuronal-glial networks in situ, (v) Investigate the differences
in calcium signalling in reactive astrocytes compared with normal cells. (iv)
Characterize the ionic conductances that form the basis for cellular excitability
in rat pineal cells with particular focus the cyclic-GMP gated Ca^* channel
activated by VIP receptors.

Methods Employed;

Primary culture of astrocytes isolated from neonatal rats, high resolution
fluorescence microscopy, digital image processing, immunocytochemistry,
electrophysiological recordings.

Major Findings:

A. Calcium signalling in astrocytes.

1. Agonist-induced calcium waves

This Section has studied the qualitative and quantitative aspects of cytoplasmic
calcium wave generation and propagation within single astroglial cells isolated
from new born rat cerebral cortices. Stimulation of astrocytes with glutamate
or norepinephrine elicited intracellular calcium signals as waves that always
originated at an initiation locus and propagated through the cell. Our analysis
has revealed that agonist induced cytoplasmic Ca^* waves initiated at a single
locus in a cell and are propagated in a non-linear fashion with regenerative
amplification at specific cellular loci. High resolution analysis of the

ZOl HD 1206 02 LCMN

cellular loci in a number of experiments showed that the diameter of these
regenerative loci ranged from 7 /j m to 15 p m ( 12 ± 4 u m, mean ± S.D., n=28).
Repeated application of an agonist always initiated Ca'* waves at the same locus
with an identical spatio-temporal oscillatory pattern. In any given cell the
wave initiation locus and amplification sites remained invariant, We also found
that the active loci have independent oscillatory properties and can oscillate
at very different frequencies and possess different latencies. Nevertheless,
these loci are influenced by diffusional waves reaching them from adjacent
cellular regions and thus act as partially coupled oscillators. The local waves
generated at the various cellular loci, when out of phase with each other, and
are propagated in opposite directions, collide and annihilate.

We extended a previously developed model of calcium oscillations in excitable and
non-excitable cells to include spatial diffusion of calcium in a cell with
discrete active loci of wave amplification. This model was then used to analyze
experimental data and to gain insight into the mechanism of wave collisions and
annihilations. Qualitative and quantitative comparisons were made of the
computational results with calcium waves measured in astrocytes. The measured
and calculated calcium dynamics are strikingly similar in many respects.
Furthermore, the model predicted that the intrinsic latencies of each locus are
gives the appearance of wave behavior. Nearly simultaneous excitation of the
entire cell at high agonist concentrations argues against the idea of a "wave"
through an excitable medium, but rather suggests differences in the kinetic
behaviour of the different oscillators. No matter which terminology is used, the
model predicted behavior consistent with the experimental data.

2. Endoplasmic reticulum structure in astrocytes

The realization that the endoplasmic reticulum might have specialized regions
prompted us to study the distribution of InsPj receptors (IPCR) on the ER of
astrocytes. In addition, we wanted to examine the extent of endoplasmic
reticulum within astrocytes using ER specific dyes. IPCR was localized by
immunocytochemistry and the ER membrane was f luorescently labelled using Di-O-C.-

3. Carol Sheppard has completed the initial phase of this work and she is
currently preparing a manuscript for publication.

We obtained a number of different antibodies against the different types of IPCR
from different laboratories in order to study receptor distribution in astrocytes
using immunocytochemical techniques. We asked the following questions: (1) What
is the IPCR distribution in the ER of tissue cultured astrocytes? (2) Does this
distribution support the kinetic pattern of wave propagation measured in the same
cell? (3) What is the extent of IPCR containing endoplasmic reticulum calcium
stores in astrocytes? (4) Are the different types of IPCR expressed in the same
cells, and if so do they show different InsP, sensitivities in different regions
of the cell? In addition, antibodies against specific cell identification
markers such as glial fibrillary acidic protein (GFAP) and ganglioside-3 (GD-3)
were used to positively identify cells as either type-1 or type-2 astrocytes.

Endoplasmic reticulum in type-1 astrocytes was decorated by antibodies against
all three types of InsPj receptors ie. type 1 IPCR, IPCRl; type-2 IPCR, IPCR2;
and type-3 IPCR, IPCR3. The overall distribution pattern of these receptor
subtypes, however, was different in any given cell (see later). When the
endoplasmic reticulum was stained with Di-O-C^-3, the ER membrane system was
found to extend over the entire cell, including small cellular processes, and the
IPCR antibodies in general appeared to decorate this membrane system (see
later). This finding would suggest that, like in other cells, in astrocytes the
ER is a continuous cistern and participates in calcium signalling. In general

ZOl HD 1206 02 LCMN

the distribution pattern of the receptors was punctate with occasional clustering
and the size of the clusters was not uniform. In some regions of the cell,
relatively large clusters of receptors were found compared with other regions.

All three receptor subtypes were distributed throughout the endoplasmic reticulum
in type 1 astroglia. There, however, were differences in the intensity of
labelling by the antibodies in different regions of the cells. While IPCRl was
distributed intensely around the nucleus and relatively sparsely in the
periphery, IPCR2 was distributed relativelyevenly throughout the cell. IPCR3,
however was distributed throughout the cell with intense labelling around the
nucleus which was different from the distribution pattern observed with IPCRl.
IPCRl labelling appeared to extend into the nuclear matrix or at least the
nuclear inner membrane. High resolution analysis of the distribution pattern
showed that the different receptor subtypes decorate the same membrane system.
It is not clear, however, whether the receptors are made up of heteromeric
complexes of different IPCR subtypes.

In an attempt to compare the distribution of IPCR with the wave propagation
kinetics, in twelve different type 1 astrocytes, norepinephrine induced calcium
waves were analysed in detail to characterize wave kinetics. At the end of the
experiment, these cells were fixed and developed for dual labelled
immunocytochemistry. One of the labels used was the cell identification marker,
GFAP and the other antibodies against IPCR2. This study showed that regions of
high calcium release during wave propagation was associated with dense labelling
with anti-IPCR2. This was clearly shown in cellular processes where regions
corresponding to regeneration loci were found to have high density of receptors.
Nevertheless, ample staining was observed in regions where the calcium release
kinetics were several fold smaller. This observation may suggest that the ER
specializations in regions of regenerative loci may be not only an increased
density of IPCR but also differences in the intraluminal contents. b. It is
also likely that there is considerable receptor reserve on the ER and only some
are functional at any one time. This study on the structural characterization
of endoplasmic reticulum is currently in preparation for publication.

3. Functional aspects of calcium release channels on the endoplasmic reticulum.

(i) Ryanodine receptors in glia

In addition to studies on distribution of IPCR on astrocytes, Carol Sheppard and
Peter Simpson in the Section have shown that the endoplasmic reticulum of both
type 1 and type 2 astrocytes also possess ryanodine receptors. Their function
in receptor mediated calcium signalling in these cells, however, has been
unclear. We have shown by immunocytochemistry that type 2 astrocytes express
ryanodine receptors at a high level in the perinuclear region and at a lower
level throughout the cytoplasm. Peter Simpson has shown that perfusion with
caffeine activates a peak and plateau [Ca^*]- elevation in approximately 24% of
these cells, the peak being a Ca wave which propagates along processes into the
cell body in a non-linear, saltatory manner. This is the first report of
ryanodine receptor activation in astrocytes. The caffeine-activated [Ca'^Jj peak
is unaffected by removal of extracellular Ca^*, while the plateau is abolished,
consistent with this phase of the response being due to store depletion-activated
Ca^* entry. When caffeine is perfused, followed by perfusion of a phospholipase-
0 coupled agonist, and the waves activated by each are examined, it is apparent
that while initiation sites may differ, the propagation sites for each are at the
same loci, and the rates of rise of the responses follow the same pattern through
the cell. It has previously been suggested that propagation sites for waves
activated by phospholipase-C coupled receptors might correspond to regions of

ZOl HD 1206 02 LCNN

IPCR specialization, in type 1 astrocytes. The present finding indicates that a
more general ER specialization might be the cause of Ca response
characteristics, at least in type 2 astrocytes, the nature of which remains under
investigation. Staining with the ER marker Di-O-C^-3 indicates that the ER in
type 2 astrocytes is not evenly distributed along processes, perhaps consistent
with this hypothesized ER specialization. This study is continuing, the aim
being to understand the participation of ryanodine receptors in calcium
signalling by glial cells.

(ii) Effects of ER calcium pump inhibition

Our studies on protein distribution have shown that type 2 sarcoplasmic-
endoplasmic reticulum Ca^* ATPase (SERCA) pumps are expressed in type 1 and 2
astrocytes. The pattern of distribution is somewhat correlated with ryanodine
receptor expression. Peter Simpson showed that thapsigargin evokes a transient
[Ca ]■ elevation in type 1 and type 2 astrocytes in nominally Ca -free medium
by inhibition of the SERCA ATPase, whereas a sustained elevation is apparent in
normal Ca , consistent with an ability of store depletion in both cell types to
activate Ca entry. The subtracted Ca entry component of the response displays
a very similar temporal pattern in the two cell types. Unexpectedly, the time
required for thapsigargin to mobilize Ca^* differed markedly along the length of
a type 1 astrocyte occurring much less rapidly than for agonist-evoked
responses, indicative of a, perhaps novel kind of wave phenomenon, different from
those described above (t to 50% of maximum response varies from 50 to 75s after
onset of agonist perfusion) .

4. Calcium responses in other glial cell types

(i) InsP3-mediated Ca^* waves in type 2 astrocytes

By immunocytochemistry with specific antibodies, we have shown that type 2
astrocytes, like type 1 astrocytes, express at least three subtypes of IPCR,
with IPCRl apparently being expressed in the nucleus, IPCR2 in the cytoplasm and
IPCR3 in both regions. We have previously demonstrated the existence of Ca
waves with invariant spatial and temporal characteristics in type 1 astrocytes.
It has now been demonstrated that type 2 astrocytes also respond to agonists of
phosphoinositide-coupled receptors, such as carbachol or bradykinin, by
activation of waves of Ca^* which begin 2-lOs after onset of perfusion, and
propagate in a saltatory manner in both directions from small initiation sites,
typically near the end of processes, usually permeating into the cell body. Waves
are seen either in the absence or presence of extracellular Ca^*, and are
inhibited by prior store depletion. The average wave velocity is increased at
high agonist concentration compared to concentrations around EC50. Cells,
particularly those possessing several processes, typically respond with waves
initiated in numerous different loci at different times. The rate of wave
propagation is non-linear, such that as each wavefront reaches certain specific
sites the rise-time is reduced and the response magnitude is increased several-
fold. Waves appear to propagate passively between these sites. Both the
initiation sites and amplification sites are found at identical loci for several
oscillations evoked by a single agonist perfusion, for responses to successive
application of a given agonist, or for the responses to different Pl-coupled
receptors in a given cell.

(ii) Ca'* responses in other glial cell types

Initial studies other cell types indicate that several of the signalling
characteristics discussed above may be general features of glial cell behavior.

;^o

ZOl HD 1206 02 LCMN

Ca^* waves have been measured in oligodendrocytes in response to norepinephrine
and bradykinin, being initiated at several different sites in the extensive
process network of these cells and propagating in a similar manner to that
described above for InsP,-mediated waves in type 2 astrocytes. Successive
addition of these two Pl-coupled agonists each activate very similar wave
response patterns in a given oligodendrocyte cell, again suggestive of ER
specializations causing wave initiation and amplification. We have also detected
caffeine responses in oligodendrocytes, and such responses have also been found
in 0-2A glial progenitor cells. Very few type 1 astrocytes cultured from neonatal
rats respond to caffeine. A significant percentage of type 1 astrocytes cultured
from adult rats, however, do respond to this agent with a characteristic peak
[Ca ]■ elevation. Overall these results suggest that ryanodine receptor-
mediated Ca^* waves may be widespread among glia, and potentially play an
important role in signalling mechanisms in these cells.

In addition, we have investigated intracellular calcium signals triggered by
glutamate receptor activation in primary cortical oligodendrocyte lineage cells
and in the oligodendrocyte cell line CG-4. Glutamate, kainate and AMPA (30-300
/J M) increased [Ca ]j in both types of cells at the stage of oligodendrocyte
progenitors (0-2A; GD3+) or pro-oligodendroblasts (04+). The peak amplitude of
Ca^* responses to glutamate receptor agonists was significantly larger in
cortical cells. In CG-4 and in cortical cells, the majority (more than 90%) of
bipolar GD3+- or multipolar 04+ cells responded to kainate. In all the cells
analyzed, kainate was more efficacious than AMPA and glutamate. The percentage
of bipolar or multipolar cells responding to glutamate was significantly lower
in the CG-4 cell line than in primary cultures. Cellular responses typical of
metabotropic glutamate receptor activation were observed in 20% of the cortical
0-2A progenitors, but in none of the CG-4 cells. The AMPA-selective antagonist
GYKI 52466 blocked kainate-induced Ca2+ responses in cortical 0-2A cells. The
selective AMPA receptor modulator cyclothiazide (30 /j M) greatly potentiated the
effects of AMPA (30-100 ^j M) on [Ca^*]- in cortical and CG-4 cells. Our findings
indicate that Ca responses in cells of the oligodendrocyte lineage are
primarily shaped by functional AMPA receptors. This is a collaborative study
carried out together with Dr. V. Gallo and a manuscript describing these results
is in Press in the Journal of Neuroscience Research.

B. Pineal cell calcium signals

Nissim Darvish, a visiting fellow from Israel, has established
electrophysiological methodologies in the Section. This has been a much
anticipated addition to the laboratory, since it allows for exploring ionic
channel mechanisms involved in calcium signalling. Although this approach was
initially undertaken to characterize the cyclic GMP gated cation channel in
pineal cells, the methodology provides a new dimension in our analysis of glial
neuronal interactions as well. This system is currently fully functional and
only some software upgrades remain. In addition. Dr. Darvish has refined
techniques to obtain pineal cells from rats and maintain them in culture for long
periods of time. The following major findings have emerged from studies on these
tissue cultured pineal cells:

1. PACAP receptor mediated calcium signals in pineal cells

PACAP (Pituitary Adenylyl Cyclase Activating Peptide) is a 38 amino acid peptide
which shares considerable sequence homology with VIP and is found in nerve fibres
innervating rodent pineal. Since our previous studies showed that the signal
transduction of VIP receptor may utilize the cyclic GMP-gated cation channels,
we wanted to examine if PACAP also stimulated this receptor complex. Preliminary

»~. "i

ZOl HO 1206 02 LCMN

experiments showed that PACAP at a concentration of 0.1 ^ M caused an increase
in intracellular calcium signals in isolated rat pineal cells. The pattern of
these signals, however, showed a peak followed by plateau type of response. This
pattern is distinct from that seen when VIP receptors are stimulated, which
produced a transient peak response. In addition, unlike the VIP stimulated
calcium signals, the PACAP response persisted in the absence of extracellular
Ca^*. Furthermore, depletion of cellular calcium stores using thapsigargin (0.5
/J M) , an inhibitor of ER calcium pumps, abolished the PACAP response. These
observations suggested that the PACAP response is dependent upon release of
calcium from intracellular stores. The signal transduction mechanisms that
underlie this PACAP-induced calcium release from pineal cell ER are under
investigation.

The patch clamp studies on pineal cells utilized various different
configurations, namely, on-cell, whole-cell, perforated patch and isolated
patches. By the use of these approaches the following ionic conductances have
been identified in pineal cells. These are in various stages of completion and
is outlined hereunder.

2. cyclic GMP-gated currents

At least three different genes encoding VIP receptors have been identified so far
and all of them bear the G-protein associated receptor motif. The classical
signal transduction mechanism for VIP receptors is activation of adenylyl cyclase
and cAMP production. Experiments conducted in our laboratory have shown that one
of the VIP receptors may be coupled to cyclic GMP and a cation channel. We
concluded that gating of a cyclic-GMP dependent cation channel appears to mediate
the VIP induced transient (Ca^*]- signal. The following evidence from pineal
studies support this conclusion: (1) VIP elevates cGMP more than 10-fold, (2) the
cGMP analogues, 8-Br-c-GMP and DB-cGMP elevate [Ca^*]j; (3) SNAP, which liberates
NO and increases guanylyl cyclase activity in pinealocytes also increases [Ca ]j
in thee cells; (4) VIP, cGMP and SNAP all depend upon the presence of mM
concentrations of [Ca^*]^ for their induction of [Ca^*]j signals; (5) The effects
of VIP and cGMP are blocked by i-cis-diltiazem, which is known to specifically
block the cGMP gated cation channel in the retinal rods; (6) RT-PCR analysis
suggests that the rod-type cyclic nucleotide gated channel is expressed in pineal
cells. Accordingly, it seems reasonable to conclude that a VIP-cGMP- [Ca^*] j
mechanism exists in the pineal gland, and that the rod type cyclic nucleotide
gated cation channels mediate this effect. In an attempt to characterize this
cation channel perforated patch experiments have been carried out in isolated
pineal cells.

In the whole cell recording mode, under voltage clamp conditions, 8-bromo-cyclic
GMP ellicited an inward current in approximately 50% of pineal cells examined (6
out of 14). These measurements were carried out under conditions where the
voltage dependent sodium currents were blocked with tetrodotoxin (20 /j M) and the
the recording medium also contained tetraethyl ammonium (10 mM) CsCl (10 mM) and
4-aminopyridine (5 mM) and ibrotoxin (100 nM) . This combination of ion channel
blockers were added to completely abolish all potassium conductances. Under
these conditions, in pineal cells a type of potassium current still persisted.
Addition of 8-bromo-cyclicGMP (1 mM) shifted the reversal potential of the whole
cell current by 5 mV towards zero. This result will suggest activation of a non-
selective cation conductance by cyclic GMP. subtracting the currents measured
in the absence of cyclic GMP from that measured in its presence revealed that the
residual current had a linear I/V relationship. The reversal potential of this
current was close to zero (6 ± 3 mV) , and this result also is consistent with
a non-selective cationic conductance.

ZOl HD 1206 02 LCMN

Under current clamp conditions, addition of 8-bromo-cyclic GMP (1 mM) resulted
in depolarization of the membrane potential by 10 mV in three separate trials.
In one experiment, this depolarization was transient, however, and the membrane
potential returned to resting values even in the presence of cyclic GMP. In two
other experiments, the depolarization was sustained in the presence of 8-bromo-
cyclic GMP. Currently these limited experimental observations are being
extended.

3. Outwardly directed K* conductances

We have identified, at least, two different outwardly directed K* conductances
in rat pineal cells. One is the delayed rectifier K* channel and the other a
distinct voltage dependent large K* conductance. The latter was found in all the
30 pineal cells so far examined in the perforated whole cell configuration. We
have investigated this channel at the whole cell level using amphoterecine-B
perforated patches and single channel recordings in isolated patches. The whole
cell recordings revealed an I/V relationship where the reversal potential was
near -60 mV which is close to the calculated reversal potential for K* currents
under our experimental conditions. Furthermore, the result showed significant
outward rectification. The magnitude of the observed current varied a great deal
from cell to cell suggesting differences in channel density. This channel was
not blocked by pharmacological agents that are used to block other types of K*
channels ie. TEA (10 mM) , CsCl (10 mM), 4-AP (5 mM) , Ibrotoxin ( l/J M) . This
channel is completely blocked by cobalt (1 mM C0CI2) and not by barium.

At the single channel level, the I/V relationship of this conductance could be
fitted to a polynomial and showed an ' N' shaped curve with remarkable outward
rectification. The single channel conductance was measured to be 180 pS in the
linear range. The channel gating was voltage dependent and the open probability
(?) change e-fold for a 50 mv membrane potential change. Plotting the P^
against the command voltage revealed that the channel does not gate at or near
the resting potential of the cell. These specific properties are reminiscent of
calcium activated K* conductances observed in other cell types. This conductance
is being further characterized.

4. Inwardly directed K* conductance

In pineal cells, while the large outward K* conductance was found in all the
cells examined, in a subpopulation (20%, 6 out of 30) an inward K* current was
observed. The I/V relationship showed significant inward rectification and the
reversal potential as measured was close to calculated for a K* selective
conductance. The reversal potential shifted upon changing [K*] constant with
K* selectivity. Changing [K*]^^ also changed the slope of the I/V relationship
in the inward direction whereby increasing [K*] increased the slope. Channel
activity can be blocked by Barium (1 mM) added to the recording medium. These
observations are reminiscent of the inward rectifier K* channels described in
various excitable cells.

Significance to Biomedical Research and the Program of the Institute;

The experimental system being established is directed towards the understanding
of the cellular basis for glial cell signalling mechanisms, and neuronal and
glial cell interactions. Glial cell derived trophic factors are essential in
neural development and maintenance of the cytoarchitecture. Understanding of the
nature of the glial-neuronal interactions is essential to understanding the
development of cellular organization in the central nervous system. The methods
developed is expected to provide fundamental insights into the functional and

id'^J

ZOl HD 1206 02 LCMN

Structural organization of calcium signalling processes in cells. The cyclic
nucleotide gated Ca^* channels activated by VIP receptors is a novel mechanism
of signal transduction.

Proposed Course;

A. Calcium signalling in astrocytes

This year we aim to set up paradigms to study calcium based excitability in glial
cells in response to neuronal activity and to compare cellular calcium signals
in normal glial cells with those in reactive glia. For the former, we will use
both acutely isolated glial cells and confocal microscopy of glia in isolated
brain slices. In studies on reactive glia, we will utilize paradigms that are
currently in development in the laboratory to investigate kinetic parameters and
specific ion channel types involved in calcium based excitability. Thirdly, we
plan to expand our immunocytochemical studies on endoplasmic reticulum structure
with respect to distribution of ion channels and ion pumps involved in calcium
signalling. These approaches are described in detail hereafter.

B . Agonist-induced calcium waves in astrocytes

1. Ion channels involved in glial cell calcium signalling

Some of the major findings to emerge from our previous work on glial cell calcium
waves remain experimentally untested. A major question remaining to be addressed
is the nature of the initiation and amplification sites involved in Ca waves
in astrocytes and oligodendrocytes. By combining immunocytochemistry and Ca^*
wave analysis in individual cells, we hope to identify a correlation between
protein or ER expression patterns and Ca wave characteristics. A variety of ER
proteins which could underlie wave phenomena will be analysed in this manner.
While agonist-evoked activation of IPCR waves is well characterized and caffeine
activated waves have been identified, it is unclear what if any role ryanodine
receptors play in agonist responses in glial cells.

(i) Calcium waves in cells of the 0-2A lineage induced by caffeine. (Peter B.
Simpson)

We have found that although type 1 astrocytes possess ample ryanodine receptors
on the endoplasmic reticulum membranes, these channels are not sensitive to
caffeine, nor do they appear to participate in agonist-induced calcium signals.
In glial cells of the 0-2A lineage, although the CICR channels are sensitive to
caffeine, it is not clear if they participate in agonist evoked calcium
signalling, eg. during activation of ionotropic receptors. The objective of
these experiments is to describe in detail the participation of the CICR channels
in calcium signalling in the different types of glial cells. The response to
caffeine in type 1 astrocytes, type 2 astrocytes, oligodendrocytes and 0-2A
progenitor cells will be catalogued in detail. In each cell type, concentration
response curves will be constructed for caffeine. These experiments will be
followed by examination of the source of the calcium signals induced by caffeine.
Thapsigargin will be used to deplete the iER store to test if all of the caffeine
response is dependent upon release from intracellular stores. Similarly, the
effect of depriving extracellular Ca^* on the caffeine response will be tested.
Thirdly, the effect of the specific inhibitor of the CICR channel, ryanodine,
will be tested to delineate if these channels are sensitive to this
pharmacological agent.

The interaction between ionotropic receptor activation and caffeine induced

ZOl HD 1206 02 LCMN

calcium release will be investigated to assess the contribution of CICR channel
mediated calcium release to ionotropic responses. Kainate will be used as the
model agonist since all the cells of the 0-2A lineage have been shown to have
functional kainate receptors. The effect of kainate pretreatment of cells on
caffeine responses will be investigated. Similarly, the effect of ryanodine on
the kainate-induced calcium signals will be investigated. Finally,
characteristics of caffeine induced calcium waves will be studied in detail in
type 2 astrocytes and oligodendrocytes. Kinetics of wave propagation will be
analyzed, using paradigms previously developed in the laboratory. From these
experiments we hope to obtain wave amplitude and local slopes of calcium release
in individual cells. Waves produced by consecutive challenges with caffeine will
be analyzed to test for fidelity in the responses. This study will provide, for
the first time, a comprehensive analysis of CICR mediated calcium signals in
glial cells and would set the stage for studies on acutely isolated glial cells
from normal rat brain and from animals in which reactive gliosis has been
induced.

(ii) InsPj-dependent and caffeine-activated calcium waves in type 2 astrocytes
(Peter B. Simpson)

Since type 2 astrocyte CICR channels respond to caffeine, it allows for
comparison of calcium waves initiated by two different mechanisms ie. CICR vs
IPCR. This comparison will differentiate between specializations in ion channel
complement versus ER membrane or ER luminal specialization in the calcium wave
regenerative loci and wave initiation sites. Experiments will be carried out to
analyze in detail calcium waves induced by methacholine and bradykinin. These
results will be compared with caffeine-stimulated waves in the same cells.
Specific cellular loci of wave initiation and regenerative release sites will be
compared. Initially, dose response relationship will be constructed for the two
agonists. This set of experiments will be followed by characterization of the
dependence on the ER store for the response by the use of thapsigargin and
removal of extracellular Ca . Wave propagation kinetics will then be
characterized using previously established experimental paradigms. In this way,
wave initiation and regenerative propagation will be analyzed. These kinetics
will be compared in the same cell for waves initiated with the two agonists and
consecutive stimulations with the same agonist. Finally wave kinetics will be
compared between bradykinin and methacholine induced waves and caffeine induced
waves. If the two waves show identical wave initiation and regenerative release
loci, the result would support the hypothesis that the kinetic differences may
be due to ER membrane specialization or luminal contents in terms of protein
buffers and perhaps not due to differences in the distribution of the IPCR
channels and CICR channels.

(iii) Signal transduction in purinergic receptors on glial cells (David B.
Langley, Nissim Darvish)

Among the different types of receptor systems found on glial cells, purinergic
receptors are the most abundant and are expressed by all the different types of
glial cells so far examined. It is believed by some that activation of
purinergic receptors by ATP released from damaged and dying neurons acts as the
trigger to change glial cell phenotype to a reactive one. Activation of these
receptors show a robust [Ca^*]. response. Two principal mechanisms for ATP-
induced increases in [Ca^*]- have been described: influx of Ca^* through ligand-
gated ion channels activated by ATP and release of Ca from intracellular
stores, a metabotropic response. The metabotropic receptors are coupled to
phospholipase-C via G . Since ATP has been recognized to act as an excitatory
neurotransmitter in both peripheral and central nervous systems, we would like

ZOl HD 1206 02 LCMN

to characterize the glial cell receptors to ATP.

As mentioned earlier, we have shown that primary cultured type-1 and type-2
astrocytes respond to ATP, and 2-methylthio-ATP with a large increase in [Ca^*]..
Furthermore, an immortalized type-1 astrocyte cell line (DTNCl) also expresses
this receptor in abundance and we have established this cell line in culture in
our laboratory. In preliminary experiments we have shown that activation of
these receptors in DTNCl cells induces a robust [Ca^*]- response. We propose to
use both the cellular systems to characterize the ATP receptor mediated response
and take advantage of the homogeneity afforded by the cell line.

To determine the type of ATP receptor coupled to [Ca^*]j increase in glial cells
dose response curves for ATP and the non hydrolyzable ATP analogue, 2-methylthio-
ATP will be constructed in the same cells using [Ca^*]j measurements. These will
be compared with dose response curves constructed in the absence of extracellular
Ca^* ([Ca^*] = < 5;^ M) . The results of this analysis will differentiate between
the metabotropic response from ionotropic response. If the (Ca ]• signal is due
to a metabotropic response, a pharmacological characterization of the receptor
will be carried out by comparison of dose response curves for a,b-methylene-ATP
with 2-methylthio-ATP followed by construction of dose response curves in the
presence of selective antagonists for the P,- and P^^ subtypes of purinergic
receptors.

On the other hand, if the response is due to a ionotropic mechanism, we propose
to characterize this signal transduction mechanism using a combined patch clamp
and [Ca ]j measurement approach. Whole cell voltage clamp experiments will be
carried out to determine if the [Ca^*)^ signal is due to gating of voltage-
dependent Ca^* channels secondary to depolarization. Simultaneous electrical and
[Ca ]■ measurements will be used to study agonist mediated currents and
accompanying Ca^* changes. Cells will be stimulated with 2-methylthio-ATP in the
presence and absence of [Ca^*]^^ which wil determine if other ions are flowing
through the ligand-gated channel. Ion replacement experiments will follow these
studies to establish the relative permeabilities of the different monovalent and
divalent cations. A screen for possible modulators of this receptor channel will
be made to discover regulatory mechanisms. Attempts will also be made to study
channels in isolation using detached patches from the glial cell line.

(iv) Calcium signalling in reactive glia (Peter B. Simpson, David B. Langley)

Astrocytes are known to change their protein expression patterns and behavior in
response to brain insult, becoming what are known as reactive astrocytes. In
order to investigate whether changes in Ca^* signalling may underlie some of the
changes which occur in such conditions, preliminary experiments have been
initiated, in collaboration with Dr. Joan Schwartz and Dr. Vivian Wu (NINDS), to
compare Ca^* responses in astrocytes from normal rat brain with those from
animals subjected to a defined chemical lesion. In these experiments, the
receptor repertoire and calcium signals will be compared using the paradigms
described in detail in sections (i) and (ii) above.

Similarly, we plan to examine acutely isolated glial cells from areas where
reactive gliosis has been induced as soon as possible after isolation.
Particularly, we propose to study purinergic receptor mediated responses in these
cells. In collaboration with Dr. Paul Britton at Walter Reed Army Medical
Research Center, we propose to obtain lesioned rats and isolate glial cells from
lesioned areas. Cells will be isolated at various times after lesioning and
their responses to both ionotropic and metabotropic agonists will be examined.
Calcium waves and local kinetics will be measured and compared with cells

ZOl HD 1206 02 LCMN

isolated from non-lesioned regions.

2. Endoplasmic reticulum structure in astrocytes (James T. Russell, Peter B.
Simpson)

We plan to extend our current studies on the ER membranes in astrocytes in
culture. Dr. Carol Sheppard has described the distribution of all three types
of IPCR and one CICR subtype on endoplasmic reticulum membranes of type 1 and
type 2 astrocytes. These studies have raised a number of interesting questions.
Chief among them are : 1. What is the extend of the receptor clustering in
cellular regions? 2. Does the clustering of IPCR and CICR occur in the same
cellular regions? 3. Does the IPCRl distribution seen in the nucleus extend into
the inner nuclear membrane? 4. Can we obtain a three dimensional map of receptor
distribution on the ER? 5. Are the SERCA pumps colocalized with the ryanodine
receptor channel?

We propose to analyze all of the above questions in the coming year. Experiments
will be designed to immunocytochemically decorate the different subtypes of IPCR
and CICR. Both single labelling and dual labelling technicpjes will be utilized
as appropriate. Fluorescently labelled microbeads (0.1 /j m diameter) will be
used as internal standards to allow for calibration axial resolution. The
samples will be visualized using a high resolution cooled CCD camera and
digitized. Images will be acquired at multiple focal planes every 0.25 p m using
a piezo-electric device mounted beneath the objective lens. An extensive photon
reassignment algorithm will be used to deblurr these wide angle images to obtain
confocal images. The confocal images will be utilized to render in 3-D the
distribution of the channel types. Subregions in the cell will be analyzed for
specific localization, three dimensional reconstruction and receptor clustering.
Furthermore, using Z dimensional calibration, the extend of channel distribution
will be analyzed in cellular organelles including the nuclear membrane. This
analysis is expected to yield a high resolution (0.15 /i m in X and Y; 0.25 /j m
in Z) view of the calcium release channels in the ER of astrocytes. This
analysis will form the basis for comparisons between different glial cell
subtypes and reactive glia. These studies will be extended to acutely isolated
glial cells.

3. Glial cell signalling in situ

Until now all experiments showing calcium based signals in glial cell have been
carried out in tissue cultured cells in purified cultures or in mixed cultures
with neurons. Only the set of experiments were carried out in a more intact
preparation of hippocampal slices in organotypic culture currently supports the
notion that glial cells respond to neuronal activity. Glial cells in culture
continue dividing unlike cells in situ and express antigens such as vimentin and
nestin which are the hallmarks of 'reactive astrocytes'. In addition, the
effects of unidentified growth factors and trophic substances found in fetal
bovine serum on glial cells, their receptor expression and signal transduction
are uncontrollable under experimental conditions. Because of these problems we
would like to study this question of glial cell signalling using two different
approaches. In the first stage, experiments will be carried out to ask if
acutely isolated glial cells from brian have the expected complement of receptor
systems that respond to neurotransmitters. In the second approach, we intend
to use the brain slice preparation and confocal microscopy to study glial cell
signalling in response to neuronal activity.

(i). Acutely isolated glial cells (Lynne Holtzclaw, James T. Russell)

ZOl HD 1206 02 LCMN

In order to study acutely isolated glial cells, we will use the tissue print
technique. Brains from new born rats (1 to 10 days old) will be isolated and
tissue prints prepared in 25 mm glass cover slips. In this technique chunks of
tissue are made and treated with pronase and DNAase for a brief period of time.
The tissue chunks are placed in test tubes at 37 C and gently agitated. The
dissociated cells will be plated on substrate coated cover slips and examined
under the microscope following loading with calcium indicator dyes. It is
expected that the cells will be examined within 45 to 60 minutes of isolation
from the animal. This method has now been set up in the laboratory and our yield
of cells which are glial cells on the cover slips is approximately 20 - 25%. The
rest of the cells are neurons as shown by dual immunocytochemical staining using
antibodies against GFAP and neurofilament. We propose to analyse each cover slip
by GFAP immunocytochemistry following calcium measurement to identify cells with
certainty. Simultaneously improvements in the isolation technique will be
attempted to improve the yield of glial cells over neurons.

Cells on cover slips will be perfused while measuring [Ca ]• using fluorescence
microscopy. Dose response curves will be constructed using neurotransmitters
appropriate for the brain region and other candidate agonists. These experiments
will establish the receptor repertoire of glial cells from the different regions
of the central nervous system. Depending upon the responses that we record, we
propose to test if the signal transduction mechanisms for these receptor-mediated
responses resemble mechanisms identified in cultured cells. These experiments
will unequivocally establish that appropriate receptor systems to
neurotransmitters exist in glial cells from the different regions of the brain
and that these receptors correspond to the neurotransmitters used by neurons in
that region. Such a conclusion will show that glial cells in situ can respond
to neuronal activity but leave the question of whether such signalling occurs
unanswered.

(ii) Studies on brain slices (James T. Russell)

The brain slice preparation will be used to study glial cells in situ using the
Noran confocal microscope. The experiments planned will take the course of
previously published work on organotypic cultures of rat hippocampal slices. The
goal of the studies on brain slices is to find answers to a number of questions:
(1) does neuronal stimulation result in glial cell responses in nerve terminal
fields in the intact brain slice? (2) How far are these signals propagated? (3)
Are there special glial cell phenotypes involved in this form of glial signalling
in the brain? (4) are the gap junctional coupling between glia modulated
depending upon neuronal activity? (5) What is the status of coupling between
neurones and glial cells?

This project will be initiated only after the establishment of the confocal
microscope in the laboratory after its initial calibration experiments have been
completed. We expect this to be completed by December 1995 at the latest. This
involves minor modifications to Room 5C-28 in building 49 and installing black
out curtains. I expect to install the microscope and carry out initial
calibrations during the fall of 1995. In the meantime, we have begun processes
to recruit a postdoctoral fellow with extensive experience in brain slice
methodology. The project would be begun when the new fellow arrives in
collaboration with Dr. Chris McBain (LCMN) and a number of experiments are
planned. We are hiring the postdoctoral scientist together and would seek funds
from external funding agencies for the research project for fellowship support.

To investigate the initial question of whether neuronal activity results in glial

ZOl HD 1206 02 LCMN

cell signals we need to establish methods to monitor [Ca^*]. in glial cells in
a brain slice. We intend to use hippocampal slices because of the anatomical
uniqueness and because the preparation is well established. We intend to take
advantage of Dr. Chris McBain' s extensive experience in the slice technique. A
number of difficulties are expected, including satisfactory loading of glial
cells with reporter dyes. Since reporter dye loading using acetoxymethyl ester
forms is expected to degrade signals due to light scattering during optical
measurements, we intend developing techniques to inject cells with dyes.
Stimulating electrodes will be placed in the region of the dentate gyrus to
stimulate mossy fiber tracts and glial cells in the CA3 region where the mossy
fiber terminal fields exist will be imaged to monitor [Ca^*]j changes. In design
the experiments are identical to the ones carried out by the Smith group in
organotypic cultures of the hippocampus. If signals such as the ones measured
by these authors are recorded, control experiments will be carried out to
establish that the [Ca^*]- signals result from neuronal activity. One of the
most interesting possible outcomes of these experiments would be the
understanding of the extent of glial cell signalling in response to neuronal
activity. This knowledge will be crucial in order to reveal the functional
significance of such signalling. Two other projects are planned and would follow
these experiments. One involves studies of [Ca^*]. changes in dendritic spines
in hippocampal interneurons and the second a fluorescence recovery after
photobleaching (FRAP) study to investigate the extent of coupling and its
regulation in glial cell networks in the brain slice preparation.

C. VIP receptors on rat pineal cells

We have described a novel signal transduction mechanism for the VIP receptors in
rat pineal cells. Our studies showed that VIP receptors evoke a [Ca^*]- signal
in pineal cells by activation of guanylyl cyclase which gates a retinal rod-type
cyclic nucleotide gated channel on the plasma membrane. A number of questions
remain to be answered, namely (1) What is the coupling mechanism between VIP
receptors and guanylyl cyclase? (2) What is the molecular identity of the cyclic
nucleotide gated channel? (3) What are the ionic channels that underlie cellular
excitability in pineal cells? (3) how does this mechanism interact with the
adrenergic activation of pineal cells and melatonin production? We propose to
characterize the putative cyclic nucleotide gated cation channel and establish
its molecular identity. In addition, we plan to describe in detail the ionic
channels that regulate cellular excitability in pineal cells. Thirdly, the
signal transduction mechanism for the PACAP induced calcium signals will be
investigated. Dr. Nissim Darvish who joined us in December of 1994 is involved
in this study. He has established the necessary instrumentation for simultaneous
recording of electrical and fluorescence signals from cells. A number of
experiments are underway.

Identification and characterization of the channels mediating the VIP-
induced [Ca^*]j signal
(Nissim Darvish)

Patch clamp techniques in conjunction with [Ca^*]. measurements will be used to
study the channel and its receptor coupling which will allow for biophysical
characterization of the channel. VIP receptors are found in high density in
pinealocytes and we plan to use cells acutely isolated by mechanical dispersion
as well as by papain treatment. [Ca^*]- measurements will be carried out both
together with electrical measurements and in parallel to monitor functional
coupling of the channel.

Perforated patch experiments will be carried out to establish if VIP induce

3 J

ZOl HD 1206 02 LCMN

membrane currents. Under voltage clamp conditions, VIP evoked membrane currents
will be measured to rule out the possibility that VIP responses are mediated by
membrane depolarization. Dose response relationship curves will be constructed
for these VIP induced currents under voltage clamp conditions. Secondly, the
depolarization induced by VIP-will be measured in current clamp. In order to
investigate if the VIP currents are due to cyclic GMP generation, currents
induced by 8-bromo-cyclic GMP will be measured in perforated patch experiments.
The pharmacological characteristics of the currents induced by both VIP and
cyclic GMP will be studied and compared. 1-cis-diltiazem will be used as the
antagonist to verify our earlier identification of the channel.

Macroscopic current recordings will establish kinetics of channel gating and
inactivation if any. Using the perforated patch technique with the input
resistance of the cell being high (>1.0 cn ), and since the pineal cells are
small, single channel currents will be recorded under steady state conditions.
Channels will be activated using cyclic GMP or VIP. From this data, we will
analyze the dose dependent activation of the single channel and the Hill
coefficient of activation. Furthermore, kinetic parameters such as mean open
time, mean close time and open probability will be measured. Channel mean open
time will be measured at various times after addition of cyclic GMP to establish
the kinetics of inactivation of the channel. Using this paradigm, we will
investigate the mechanism of inactivation. Ion selectivity for channel
permeability will be established using I/V plots constructed in ion replacement
experiments and calculating permeability ratios. Sensitivity to 1-cis-diltiazem
of the channel will be determined.

2. Ion channels that regulate pineal cell excitability (Nissim Darvish)

In preliminary experiments, we have identified two distinct K* currents in rat
pineal cells. Previous studies by different laboratories have shown that K*
currents in pineal cells may be important in the regulation of pineal function
which is controlled by adrenergic input from the superior cervical ganglion. One
of these currents is an outwardly directed large conductance channel and the
other is an inward current which has not been previously observed. We propose
to characterize both of these channels in detail.

(i) Outward K* channel

Whole cell patch experiments and single channel conductance measurements will be
utilized to investigate the biophysical characteristics of this channel.
Initially the single channel conductance and ion selectivity will be established
from I/V plots derived from experiments under different ionic conditions.
Voltage dependence of the channel will then be investigated using conventional
electrophysiooogical methods. From this data channel gating kinetics will be
derived. Since this channel is outwardly rectifying as shown in our preliminary
experiments at the single channel level, we intend to explore the mechanism of
rectification (ie. Ca^", Na*, H* etc.). Since this channel is not blocked by
conventional K* channel blockers other than cobalt we would explore other
candidate blockers. Finally, based on our preliminary results potential
modulation of the channel by Ca^*, cyclic AMP and cyclic GMP will be
investigated. Cyclic nucleotide dependent effect will be investigated to
differentiate direct effects from effects mediated via phosphorylation of the
channel protein through protein kinase activation.

(ii) Inward K* channel

Biophysical characterization of the inward K* channel will be carried out in

ZOl HD 1206 02 LCMN

detail in whole cell and isolated patch experiments. These include single
channel conductance and ion selectivity from I/V relationships measured under
different ionic conditions. Channel open probability and its relationship to
membrane potential will be investigated as well as the dependence on K*
concentration out side. Pharmacological studies will characterize the blocking
action by Ba^* ions. The mechanism of inward rectification of the channel will
be studied and candidate agents such as Mg^* and polyamines will be tested.
Finally channel modulation by putative modulators such as ATP and Ca^* will be
examined.

Secretion of melatonin by pineal cells is tightly regulated by adrenergic input
from the superior cervical ganglion. Activation of adrenergic receptors is
coupled to increase in cellular Ca^*, cyclic AMP, and cyclic GMP. The cellular
physiology of coupling these second messengers to melatonin production and
secretion is currently unknown. It is likely that the two K* channels described
here play a crucial role in not only controlling cellular excitability but also
in regulating the circadian control of melatonin production and secretion.
Experiments to study modulation by adrenergic agents and their second messengers
are expected to reveal underlying mechanisms that control pineal function.

3. PACAP-induced [Ca^*]j signals in pineal cells (Nissim Darvish)

We have established that PACAP causes increase in [Ca ]■ in pineal cells by
release from intracellular thapsigargin-sensitive stores. The mechanism of
coupling receptor activation to [Ca^*]- elevation, however, is not known. It
appears that although PACAP shares significant sequence homology with VIP, it
acts through a different mechanism. Initially dose response relationships for
VIP and PACAP will be established using [Ca^*]. measurements to ensure that they
are acting through different receptor systems. Experiments will be carried out
to establish the second messenger system that mediates PACAP-induced [Ca^*]j
signals. In other cellular systems, in addition to increasing cyclic AMP, PACAP
has been shown to activate phospholipase-C and InsP, production. Heparin
injection experiments will be attempted to verify if in the pineal cell PACAP
receptors act via the phospholipase pathway.

Publications;

Fleishman, LF, Holtzclaw, L, Russell, JT, Mavrothalassitis, G and Fisher, RJ.
ETS-1 in astrocytes: Expression and transmitter-evoked phosphorylation. Mol Cell
Biol 1995;15:925-931.

Gallo, V and Russell, JT. Excitatory amino acid receptors in glia: Different
subtypes for distinct function? J Neurosci Res 1995;42 (In Press).

Holtzclaw, L, Gallo, V and Russell, JT. AMPA receptors shape Ca^* responses in
cortical oligodendrocyte progenitors and CG-4 cells. J Neurosci Res 1995, -42 (In
Press) .

Roth, BJ, Yagodin, S, Holtzclaw, L and Russell, JT. A mathematical model of
agonist-induced propagation of calcium waves in astrocytes. Cell Calcium 1995;
17:53-64.

Schaad, NC, Vanacek, J, Rodriguez, IR, Klein, DC, Holtzclaw, L and Russell, JT.
Vasoactive intestinal peptide elevates pinealocyte [Ca ]■ by enhancing influx:
evidence for involvement of a cyclic GMP-dependent mechanism. Mol Pharmacol
1995;47:923-933.

S'i

ZOl HD 1206 02 LCMN

Yagodin, S, Holtzclaw, L and Russell, JT. Subcellular calcium oscillators and
calcium influx support agonist-induced calcium waves in cultured astrocytes. Mol
Cell Biochem 1995 (In Press).

u U

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBLIC HEALTH SERVICE

NOTICE OF INIKftMURAL RESBVI«H roDOBCr

PROJECT NUMBER

ZOl HD 02000 04

PERIOD COVERED

October 1, 1994 to September 30, 1995

TITLE OF PROJECT (80 characters or less. Title must fit on one tine between the borders.)
Neurotransmitter Receptors in Glia

PRINCIPAL INVESTIGATOR (List other professional personnel below the Principal Investigator.) (Name, title.

PI: Vittorio Gallo
Others: Mario Perde

Steve Scherer
Montse Itolne
Fei Huang
Li Jin Chew
Jia Min Zhou
Pobp-rt Hagan

Visiting Scientist, LCMN, NICHD 12 months
Visiting Fellow, I£MN, NICHD 12 months
IRTA Fellow, DCMN, NICHD 12 months
Visiting Fellow, LCMN, NICHD 7 months
IRTA Fellow, LCMN, NICHD 7 months
Visiting Fellcw, LCMN, NICHD 2 months
Biologist, LCMN, NICHD 11 months
amrmer StixJent, IjCMN, NICHD 2 Tnonths

COOPERATING UNITS (if any)

Dr. R. Armstrong, Dept. Anatomy & Cell Biology, USUHS, Bethesda, MD

LAB/BRANCH

Laboratory of Cellular and Molecular Neurophysiology

SECTION

Unit on Neurobiology

INSTITUTE AND LOCATION

NICHD, NIH, Bethesda, Maryland 20892

TOTAL STAFF YEARS:
5.2

PROFESSIONAL:
5.2

OTHER:
0.2

CHECK APPROPRIATE BOX(ES)

D (a) Human D (b) Human tissues S (c) Neither

D (al) Minors
D (a2) Interviews

SUMMARY OF WORK (Use standard unreduced type. Do not exceed the space provided.)

The molecular and functional properties of neurotransmitter ligand-gated ion
channels in glial cells were studied, in order to understand their regulation and
function in the brain. Studies were carried out to: A. characterize the
physiological role of glutamate-gated ion channels in glial development and B. to
clone a glutamate receptor gene and analyze its 5' regulatory region in order to
understand its transcriptional regulation in glia. A. Physiological properties and

functional role. Primary cultures of rat oligodendrocyte progenitors {0-2A) were
used as a model system. 0-2A cells develop into oligodendrocytes in vitro
similarly to 0-2A progenitors in vivo. Kainate- and AMPA-pref erring glutamate
receptors are co-expressed in 0-2A cells. AMPA receptor activation inhibits 0-2A
cell proliferation and prevents lineage progression at early developmental stages.
Intracellular Ca^* transients, due to transmembrane influx of the divalent cations,
and immediate early gene {NGFI-A and c-fos) transcription are regulated by AMPA
receptors. Immediate early gene induction occurs at the transcriptional level
through protein kinase C-mediated phosphorylation of the transcription factor CREB
Genes encoding separate kainate-preferring glutamate receptor subunits are co-
expressed in undifferentiated, nestin-positive multi-potential precursor cells of
the rat neural tube. B. Glutamate receptor genes and analysis of their 5_!.

regulatory region. Genes encoding kainate receptor subunits are highly expressed in
cells of the oligodendrocyte lineage. The cloning of the entire rat GRIK5 gene,
which encodes the kainate-preferring subunit KA2, was accomplished, and its intron-
exon organization was partially determined. Two kb of the 5" flanking region of the
rat GRIK5 were sequenced and found to comprise 3 independent transcription start
sites. The transcriptional potential of the 2kb region was analyzed in £
oligodendrocyte cell line as well as in PC12 cells.

ZOl HD 02000 04 LCMN

Project Description;

Objective;

To study the molecular and functional properties of glial
neurotransmitter-gated ion channels. To understand the
physiological role of glutamate-gated ion channels in glia. To
study the transcriptional regulation of glutamate receptor genes in
neural cell development.

Major Findings.

1. Functional role of glutamate receptors in oligodendrocyte
development

Oligodendrocyte development is controlled by several growth
factors, including platelet-derived growth factor (PDGF) and basic
fibroblast growth factor (bFGF) , which regulate proliferation and
differentiation of oligodendrocyte progenitors (0-2A) . 0-2A cells
co-express functional AMPA- and kainate-pref erring glutamate
receptors (GluRs) . We hypothesized that the neurotransmitter
glutamate may regulate oligodendrocyte development through the
activation of membrane receptors expressed early on during 0-2A
proliferation and migration. Vittorio Gallo and Jia Min Zhou used
in vitro bioassays to analyze potential functions of GluRs in
modulating proliferation, migration and differentiation of 0-2A
progenitor cells. 0-2A progenitor cells purified from the embryonic
rat cerebral cortex were cultured under different conditions. The
different culture conditions tested were as follows: PDGF, bFGF,
PDGF+bFGF, 30% B104 (neuroblastoma cell line) conditioned medium
(B104-CM) . All these culture conditions promoted 0-2A cell
proliferation, as measured either by ^H-Thymidine (%-Thy) or by
bromodeoxyuridine (BrdU) incorporation. Co-incubation with the GluR
agonists glutamate, kainate or AMPA, significantly inhibited H-Thy
or BrdU incorporation in 0-2A cells. The effects of the GluR
agonists were dose-dependent and, at all concentrations tested,
kainate was more efficacious than AMPA and glutamate. The IC^q
values for kainate were 32.8/iM (cells cultured with PDGF) and
23.3mm (cells cultured with PDGF+bFGF). IC50 values for AMPA were
28.1/iM (PDGF) and 10.3/iM (PDGF+bFGF). Under all culture conditions,
the selective non-NMDA receptor antagonist DNQX not only prevented
the inhibitory effects of kainate, but also significantly
stimulated 0-2A cell proliferation. This was due to block by DNQX
of the inhibitory effects of endogenous glutamate present in the
culture medium, which ranged between 9 (cells in bFGF) and 10.2/iM
(cells in PDGF) , as determined by HPLC analysis of the media after
24 hr in culture (in collaboration with Judy Harvey White, NINDS) .
Activation of other transmitter receptors (GABA and muscarinic) did
not modify 0-2A cell proliferation. The inhibitory effects of the
selective agonist AMPA on cell proliferation indicate that AMPA

'3,

mmsmsma^Kmmsm

ZOl HD 02000 04 LCMM

receptors are primarily involved. Low micromolar concentrations of
glutamate are sufficient to activate such receptors to elicit
biological effects, as demonstrated by the IC50 values for GluR
agonists and the stimulatory effects of the antagonist DNQX on 0-2A
cell proliferation.

The inhibitory effect of GluR agonists on 0-2A cell proliferation
was likely to be a direct effect on the cell cycle, since the GluR
agonists did not induce neurotoxicity or apoptosis. 0-2A cells that
were cultured with PDGF+bFGF and GluR agonists for 24hr, and then
■^H-Thy-pulsed for different lengths of time in a GluR agonist-free
medium containing PDGF+bFGF, re-entered the cell cycle with a
temporal pattern similar to cells that were never exposed to GluR
agonists. These experiments demonstrated that the inhibitory
effects of kainate and AMPA were reversible and that the 0-2A cells
were still viable after a 24 hr exposure to GluR agonists.

Since 0-2A cells proliferate much more vigorously than cells at
later stages of the oligodendrocyte lineage, inhibition of cell
proliferation could also result from promoting 0-2A cell
differentiation. Therefore, the effects of long-term GluR
activation on 0-2A lineage progression were investigated by using
cell type-specific antibodies to identify pre-0-2A cells (GD3"
/nestin"^) , 0-2A cells (GD3'^/nestin'^) , oligodendroblasts (OA"*") , and
finally oligodendrocytes (01+) . In cultures maintained with
PDGF+bFGF for 1 day, the cells maintained the pre-0-2A or 0-2A
phenotype, which was not significantly altered after a 22hr
exposure to kainate or AMPA. Similar results were also obtained in
cells cultured in PDGF or bFGF alone. Under culture conditions that
permit 0-2A lineage progression (i.e. Nl, PDGF or bFGF), AMPA
significantly inhibited acquisition of 04, while DNQX significantly
increased the percentage of 04"*^ oligodendroblasts. Importantly, in
cultures maintained in bFGF, the small but significant percentage
of dividing 04'*' oligodendroblasts was not modified by treatment
with AMPA or DNQX. In contrast, under conditions that prevented O-
2A cell differentiation (i.e. PDGF+bFGF or B104-CM) , neither AMPA
nor DNQX affected 0-2A cell lineage progression. These data
indicate that long-term activation of AMPA receptors not only
inhibited cell proliferation at the 0-2A stage, but also prevented
lineage progression to the 04'*' oligodendroblast stage.

In collaboration with Regina Armstrong (Dept. of Anatomy and Cell
Biology, USUHS) , we examined whether GluR agonists can act as
chemoattractants, or alter migration to PDGF, which is a known
chemoattractant for 0-2A cells. In a microchemotaxis chamber assay,
0-2A cells migrated toward PDGF in a dose-dependent manner. GluR
agonists or DNQX, at concentrations which altered proliferation and
differentiation, did not affect migration of 0-2A progenitors to
PDGF. Control experiments showed that GluR agonists and the
antagonist DNQX did not stimulate chemokinesis in 0-2A cells. These
data emphasize the viability of the 0-2A cells in the presence of

ZOl HD 02000 04 LCMN

GluR agonists and are consistent with the previous observation that
proliferation is not required for 0-2A cell migration.

2 , Glutamate receptors. Ca^"*" transients and stimulus-coupled gene
transcription in oligodendrocyte progenitor cells.

Intracellular calcium signals triggered by glutamate receptor
activation were studied in primary cortical oligodendrocyte lineage
cells and in the oligodendrocyte cell line CG-4 in collaboration
with Lynne Holtzclaw and James Russell. Glutamate, kainate and AMPA
increased [Ca^'*']^ in both types of cells at the stage of 0-2A or
oligodendroblasts (04"^) . The amplitude of Ca^"^ responses to GluR
agonists was significantly larger in cortical cells and, in all
cells analyzed, kainate was more efficacious than AMPA and
glutamate. The percentage of 0-2A or oligodendroblast cells
responding to glutamate was significantly lower in the CG-4 cell
line than in primary cultures. Cellular responses typical of
metabotropic glutamate receptor activation were observed only in
20% of the cortical 0-2A progenitors, but in none of the CG-4
cells. The AMPA-selective antagonist GYKI 52466 blocked kainate-
induced Ca^"^ responses in cortical 0-2A cells. The selective AMPA
receptor modulator cyclothiazide greatly potentiated the effects of
AMPA on rCa^"^]^ in cortical and CG-4 cells. Our findings indicate
that Ca^ responses in cells of the oligodendrocyte lineage are
primarily shaped by functional AMPA receptors.

Previous studies by Mario Pende demonstrated that activation of
AMPA receptors and transmembrane Ca^"^ influx trigger immediate
early gene (lEG) transcription in 0-2A progenitor cells. Mario
Pende performed nuclear run-off assays in nuclei isolated from 0-2A
progenitor cells after exposure to kainate and demonstrated that
AMPA receptor stimulation directly activates transcription of NGFI-
A and c-fos. The intracellular events leading to gene expression
after non-NMDA receptor stimulation were further analyzed in
primary 0-2A cells. It was demonstrated that activation of
glutamate receptors causes phosphorylation of the transcription
factor CREB and subsequent lEG transcription through a previously
unidentified pathway, i.e. through the activation of protein kinase
C (PKC) . Immediate early gene induction was caused also by
stimulation of 0-2A cells with the phorbol ester TPA, which
directly activates protein kinase C. Pre-treatment with the
selective PKC inhibitor bisindolylmalemide blocked not only the
increase in lEG RNA transcripts caused by TPA, but also the effects
of kainate. Down regulation of PKC induced by long-term treatment
with phorbol esters drastically decreased the effects of kainate
and TPA on lEG induction. The muscarinic receptor agonist carbachol
also stimulated lEG transcription in a bisindolylmalemide-sensitive
manner. Incubation with kainate, carbachol or TPA caused an
increase in the specific binding of ^H-phorbol ester to intact 0-2A
cells, indicating that these agents stimulated PKC translocation to
the membrane. Within 5-10 minutes after stimulation of non-NMDA

Esnm^nm

ZOl HD 02000 04 LCMN

receptors by kainate, the transcription factor CREB was
phosphorylated on serine-133, as assessed by immunoblot analysis
with an antiserum specific for the phosphorylated form of CREB.
TPA and the muscarinic receptor agonist carbachol had a similar
effect on CREB phosphorylation. Surprisingly, forskolin and the
growth factors PDGF and bFGF failed to induce CREB phosphorylation
in 0-2A progenitor cells. Bisindolylmaleimide completely
counteracted the effect of kainate, TPA and carbachol on CREB
phosphorylation. Immunoblot analysis with a pan anti-PKC antibody
of cellular fractions from 0-2A progenitors demonstrated PKC
translocation to the nucleus after phorbol esters treatment. These
results define a novel pathway for CREB activation in neural cells
that is dependent on PKC.

3 . Kainate receptor expression in precursor cells of the rat neural
tube.

Primary progenitors derived from ElO rat neural tube are capable of
differentiating along both the neuronal and glial pathways. Steve
Scherer showed that plating the neural tube precursor cells on
polyamines as opposed to fibronectin not only had a marked effect
on the morphology of the cells (more neuron-like) , but also
increased the percentage of neural vs. smooth muscle precursors.
Previous RT-PCR analysis of neural tube/neural crest cell RNA
demonstrated that these nestin positive neural progenitor cells
expressed transcripts for the kainate-pref erring subunits GluR6,
GluR7, KAl and KA2 , and that these transcripts were also expressed
in ElO embryos in vivo. The subunit GluR5 was not expressed in vivo
nor in culture. The subunits GluR6 and KA2 are capable of forming
functional hetero-oligomeric channels. RNase protection assays
showed that, in ElO embryos and in cultured neural progenitors,
both GluR6 and KA2 are expressed at levels 10 times lower than in
the brain at postnatal day 1. RT-PCR analysis of total RNA from
neural tube cultures and from ElO embryos demonstrated that, both
in the cultures and in vivo, GluR6 is almost exclusively (92%)
unedited in the transmembrane region TM2. Indirect
immunohistochemistry of the cultured cells with anti-GluR5/6/7 and
anti-KA2 antibodies showed expression of these subunits in
differentiated neurons as well as in immature neuroepithelial
precursor cells. These results indicate that heterooligomeric
kainate-preferring receptors may be assembled in neural progenitor
cells early on during embryonic development.

4. Transcriptional analysis of the GRIK5 5' flanking region.

The gene GRIK5 encodes the kainate-preferring glutamate receptor
subunit KA2. We have previously obtained a rat GRIK5 clone
comprising 2kb of 5' flanking region. To identify the transcription
initiation site(s) of the rat GRIK5 gene, Montse Molne performed
primer extension analysis of total RNAs derived from rat cerebral
cortex and cerebellum, as well as liver and of total RNAs derived

w^mammm^sBB^^m

ZOl HD 02000 04 LCMM

from primary cultures of rat 0-2A progenitors, from the glial cell
line CG4 and from PC12 cells. The primer extension analysis
revealed the presence of 3 transcription start sites in the tissue-
derived and cell-line derived total RNAs . Thus multiple
transcription start sites seem to be used for rat GRIK5 gene
transcription, a feature commonly found in genes lacking a TATA box
in their promoter region. Multiple transcription initiation sites
have also been described in the promoter regions of other glutamate
receptor subunit genes that also lack a TATA box. In order to
detect more distal initiation sites, the total RNA samples used for
the primer extension were additionally tested by Northern analysis
with genomic and cDNA derived probes. No signal was detected with
a genomic probe spanning a region upstream of the putative
transcription start sites. So far, we have not found evidence of
preferential transcription initiation site use in different cell
types or brain regions.

We have placed the bacterial reporter gene CAT under the
transcriptional control of the 2 Kb fragment and under the
transcriptional control of a smaller 600bp fragment derived from
it, containing all the detected GRIK5 transcription initiation
sites. We have measured the transcriptional potential of these
constructs after transient transfection in different neural and non
neural cell types. The 2Kb fragment is able to direct high
expression levels of the reporter CAT gene when transfected in the
oligodendrocyte cell line CG-4 and in the cell line PC12
differentiated to a neuronal phenotype with NGF. This construct
remains silent when transfected in undifferentiated PC12 cells or
primary cultures of astrocytes. The 600bp chimeric construct,
containing all the transcription initiation sites, is moderately
active in CG-4 cells and is not sufficient to drive detectable
levels of transcription in differentiated PC12 cells. This
construct remains also silent or not detectable in astrocytes and
in undifferentiated PC12 cells, and thus only partially reproduces
the neural cell type specific transcriptional behavior of the 2Kb
fragment driven constructs. These observations indicate that GRIK5
genomic sequences comprised within the 600 bp fragment are
sufficient to ensure detectable levels of transcription in neural
cell types. In oligodendrocyte cells, the transcriptional levels
can be highly increased when additional upstream sequences are
present. In PC12 cells, high transcription levels are only observed
when upstream sequences are present and the cells are fully
differentiated to a neuronal phenotype by NGF. Thus this minimal
600bp fragment can be activated by upstream regulatory elements
and/or in response to specific differentiation factors.

The 2Kb and the 600bp fragments are both sufficient to maintain
tissue specific expression, as determined after transfection in
non neural cell types (Hela, 3T3 and HEK 293 cells) . The 6G0bp
fragment retains tissue specific expression and is sufficient to
direct basal levels of expression in oligodendrocytes, behaving

MitlUitHlllIM

ZOl HD 02000 04 LCMN

typically as a promoter region, it can also be activated in a cell
specific manner by sequences located distally, suggesting that
sequences able to functionally assemble the transcriptional
machinery are present in this DNA region. The primer extension
analysis indicates that all transcripts are initiated within this
600 bp region in CG4 and PC12 cells, as well as in vivo.

The sequence analysis indicates the presence of several INR
sequences and Spl sites, and a high G content, but no canonical
TATA or CAAT boxes in this fragment. All of this features are
commonly observed in the promoter regions of housekeeping genes. In
order to determine if these INR and Spl consensus sites are equally
functional and to further localize the promoter region of the GRIK5
gene, we have obtained new chimeric GRIK5-CAT constructs in which
these putative promoter sites have been serially deleted in the
presence or absence of distal sequences of the GRIK5 gene and
tested these constructs after transient transfection in CG-4 cells
and PC12 cells, and in non neural Hela and 3T3 cells. Our analysis
indicates that the INR sequences (220 bp) located at the most 5'
end transcription start site are sufficient to assemble efficiently
the transcriptional machinery, and that INR and Spl sites located
downstream somehow negatively regulate this INR site. Constructs
that do not contain the most 5' INR, but contain all or some of the
downstream INR and Spl sites give very low, but detectable levels
of transcription in CG-4 cells. All the deletion constructs tested
retain tissue-specificity of expression and remain silent in non
neural cell types. Taken together, these results indicate that the
220 bp region that comprises the most 5' INR contains
transcriptional regulatory cis-elements sufficient to direct high
transcription levels in CG4 cells and to maintain tissue
specificity of expression in other cell types.

5. Molecular cloning of the entire rat qlutamate receptor gene
GRIK5 and analysis of its exon-intron organization.

The cloning of the entire rat GRIK5 gene was accomplished. Fei
Huang screened a rat genomic library with KA2 cDNA-derived and
genomic GRIK5-derived probes, and isolated 7 independent,
overlapping genomic clones encoding the complete GRIK5 gene.
Preliminary restriction analysis indicates that the gene spans SO-
SO kb of genomic DNA. In order to study the role of regulatory DNA
elements other than those comprised in 5' flanking region (e.g.
intronic sequences) , Fei Huang has begun to determine the exon-
intron organization of the rat GRIK5 gene. The complete sequence
and exon-intron organization of the most 5 ' -end genomic clone,
including 4kb of the 5' flanking region, of the rat GRIK5 was
obtained. Eleven small exons, ranging between 54 and 179 bp, and
11 introns, ranging between 0.1 and 3.4 kb, were identified. The
GRIK5 translation start site is comprised within the second exon
(129 bp) . The first and second exons are separated by a 3.4 kb
intron.

4;:'

WB^^BmniBiai^^^sifs

ZOl HD 02000 04 LCMN

Proposed Course of the Project;

l.We plan to continue the analysis of the role of glutamate
receptors in oligodendrocyte development. Experiments in vitro will
be continued on cultured cortical 0-2A progenitors, to clarify the
mechanism by which activation of GluRs inhibits 0-2A cell
proliferation and lineage progression. We also plan to elucidate,
by flow cytometric analysis, whether activation of such receptors
causes a GO or a Gl arrest, or increases the length of the cell
cycle. In parallel studies, we will analyze cyclin expression in O-
2A cells cultured with different growth factors and determine
whether GluR activation induces the expression of any cyclin-
dependent kinase inhibitor in 0-2A cells. A new approach will also
incorporate double in situ hybridization and immunohistochemical
studies, to determine whether AMPA- and kainate-pref erring
receptors are expressed by oligodendrocyte lineage cells in vivo.

2. We will further investigate the mechanism of GluR-induced
phosphorylation of the transcription factor CREB as well as its
importance in the context of GluR-induced gene transcription in O-
2A progenitor cells. We will establish whether PKC directly
phosphorylates CREB, or triggers a cascade of events that lead to
CREB phosphorylation. We will also determine which specific amino
acid residues of CREB are phosphorylated in response to GluR and
muscarinic receptor activation, as well as direct PKC activation by
phorbol esters. Further experiments will also elucidate the
molecular mechanism of CREB activation by PKC and determine whether
PKC-induced CREB phosphorylation affects its binding to CREB-
response elements (CREs) , or only promotes the interaction with the
basal transcriptional machinery. Finally, we will extend our
analysis of PKC-induced, CREB-mediated transcriptional activation
to other CREs (low and high affinity) .

3. The molecular and functional characterization of kainate-
preferring receptors in the rat neural tube progenitors will be
completed. Follow-up molecular and biochemical studies will
determine whether kainate-preferring receptor subunits are co-
expressed in the same neural progenitor cells both at the mRNA and
protein level, and whether such receptors are functional. We will
also analyze the transcriptional potential of the cloned GRIK5 5'
flanking region in undifferentiated neural tube progenitor cells,
to determine whether this region is sufficient to direct tissue-
specific expression before lineage specification has occurred.

4. The molecular analysis of the GRIK5 gene promoter will be
continued in order to identify DNA elements directly involved in
the binding of trans-acting factors. In vitro techniques, such as
gel retardation and DNAse I footprinting assays, will be used and
combined to transient transfection experiments performed in CG-4
and PC12 cells. Nuclear extracts will be prepared from different
neural cell types expressing GRIK5 and from CNS tissue at different

ama

ZOl HD 02000 04 LCMM

developmental stages. To determine the sequence specificity of the
footprints, the effects of mutations on nuclear protein binding
sites will also be examined in vitro as well as in transient
transfection assays.

5. We will complete the analysis of the intron-exon organization of
the rat gene GRIK5 . The intron-exon arrangement of this gene will
be compared to that of other GluR genes and will provide a
framework for future studies. We plan to determine whether: i) the
5' flanking region further upstream of the previously analyzed 2Kb
(comprised in a newly-isolated genomic clone) is involved in
regulating GRIK5 transcription; ii) the first 3.4Kb intron
comprised between the transcription start sites and the ATG codon
is involved in regulating GRIK5 expression in different neural cell
types. The transcriptional analysis of these DNA elements will be
performed in cultured glial cells and neurons, and in transgenic
mice.

Significance to Biomedical Research and Program of the
Institute;

It is well established that glial cells perform a pivotal role in
brain function and express a large variety of voltage-dependent and
ligand-gated channels. This finding greatly increases the
complexity of the cellular elements that can respond to classical
neurotransmitters, because the ratio between glial cells and
neurons in the mammalian central nervous system is approximately
10:1. Glutamate receptors are abundantly expressed in different
glial cell types and they can mediate drastic changes in glial
membrane potential. The physiological role of glutamate-gated
channels in glia is not understood, but the close anatomical
relationship between neurons and glia both in the developing and in
the mature mammalian central nervous system indicates that also
glial cells are very likely to be directly affected by glutamate in
the intact brain. The appearance of glutamate receptors during
early glial development and the regulation of glial proliferation
and differentiation by glutamate indicate that, in addition to his
role as mediator of information transfer at synaptic junctions,
glutamate plays a role in shaping the development of the central
nervous system. This has very important consequences for the
physiology of the mature brain, in which glial cells perform
crucial functions, such as myelination and metabolic support of
neurons. Glutamate receptors in glial cells may also play an
important role in the pathophysiology of stroke and cerebral
anoxia, as already established for neurons. In addition,
periventricular white matter injury, the principal variety of brain
injury of the human premature infant, cause cell death of
differentiating oligodendroglia and appears to involve glutamate
receptors.

ZOl HD 02000 04 LCMM

Publications;

Gallo V, Armstrong R . Developmental and growth factor-induced
regulation of nestin in oligodendrocyte lineage cells, J
Neurosci 1995;15:394-406.

Gallo V, Russell JT. Excitatory amino acid receptors in glia:
different subtypes for distinct functions? J Neurosci Res (In
Press) .

Holtzclaw L, Gallo V, Russell JT. AMPA receptors shape Ca^"^
responses in cortical oligodendrocyte progenitors and CG-4
cells. J Neurosci Res (In Press).

Szpirer C, Molne M, Antonacci R, Jenkins NA, Finelli P,
Szpirer J, Riviere M, Rocchi M, Gilbert DJ, Copeland NG, Gallo
V. The genes encoding the glutamate receptor subunits KAl and
KA2 (GRIK4 and GRIK5) are located on separate chromosomes in
human, mouse and rat. Proc Natl Acad Sci (USA) 1994 ; 91 : 11849-
11853.

4^j

1,-pM

J-

DEPARTMENT OF HEALTH AND HUMAN SERVICES • PUBUC HEALTH SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD 01301-13 LDMI

PERIOD COVERED

October 1. 1994 to September 30, 1995

TITLE OF PROJECT 180 characters or less. Title must fit on one line between the tiorders.l

Human Immune Response to Polysacchan de-Protein Conjugate Vaccines

PRINCIPAL INVESTIGATOR (List other professional personnel below the Principal Investigator! {Name, title, laboratory, and institute affiliation)

PI: R. Schneerson
See Attached

Research Medical Officer

LDMI, NICHD

COOPERATING UNITS /if anyl

LCDB, NIDDK(J. Shiloach): SUNY Downstate Medical Center.
N.Y. (G. Schiffman): Carolinas Medical Center. Charlotte, N.C.. (J.C. Parke.
Jr.); National University Hospital. Iceland (J. Jonsdottir). LCI. NIAID (J.

LAB/BRANCH

Laboratory of Developmental and Molecular Immunity

SECTION

Section on Bacterial Disease Pathogenesis and Immunity

INSTITUTE AND LOCATION

IICHD. NIH. Bethesda. Maryland, 20892-2720

TOTAL STAFF YEARS:

5.8

PROFESSIONAL:

2.3

OTHER:

3.5

CHECK APPROPRIATE BOXIESI

H (a) Human subjects
D (a1 ) Minors
D (a2) Interviews

D (b) Human tissues D (c) Neither

SUMMARY OF WORK (Use standartj unreduced type. Do not exceed the space provided. I

The surface polysaccharides of bacterial pathogens, which include capsular
polysaccharides and lipopolysaccharides, serve as protective antigens. The
immunologic properties of these bacterial polysaccharides, namely their age-
related and T-cell independent immunogenicity. limit their use as vaccines.
Covalently attachment to medically-useful proteins to form conjugates, both
increases their immunogenicity and confers T-cell dependent properties to these
polysaccharides. The capsular polysaccharides of Streptococcus pneumococcus type
6B. Staphylococcus aureus types 5 and 8. Group B streptococcus type 3 have been
bound to several proteins and evaluated clinically. S. aureus type 5-rEPA was
evaluated in end stage renal disease patients: type 5 antibodies of the three
major Ig classes rose significantly though to a lesser degree than in healthy
volunteers, no booster response to reinjection at 6 weeks was found. These
antibodies had opsonophagocytic activities. Pn6B-TT was evaluated in patients
with sickle cell disease, healthy infants at 3. 4 and 6 months of age or at 7 and
9 months of age. Type specific antibodies of the three Ig classes, with booster
responses, were induced. The magnitude of these responses was lesser than of
Hib-TT. GBSIII-TT was evaluated in females of child bearing age. IgG antitype
III rose similarly to the response to the polysaccharide alone. Technical
problems with this lot were identified. All conjugates were safe, with only
minor local reaction. The LPS of shigellae was detoxified, their 0-specific
polysaccharides bound to bacterial toxoids and their immunogennicity in mice
found to be satisfactory. In Phase 1 and Phase 2 studies, these conjugates of
the 0-specif1c polysaccharides were safe and immunogenic: LPS antibody levels
elicited by the investigational conjugates were similar to those in recruits
convalescent from shigellosis. In preliminary studies, a S. sonnei - rEPA
conjugate protected against shigellosis caused by this pathogen. A more
extensive study showed protection of 75^

PHS 6040 (Rev. 5/921

ZOl HD 01301-13 LDMI

Others:

J.B. Robbins

Chief

LDMI

NICHD

V. Pozsgay

Visiting Scientist

LDMI

NICHD

D. Pavliakova

Visiting Associate

LDMI

NICHD

D.A. Bryla

Statistician

LDMI

NICHD

N.W. Tolson

Biologist

LDMI

NICHD

H. Ao

Adjunct Technician

LDMI

NICHD

O.L. Barrera

Adjunct Technician

LDMI

NICHD

Cooperating Units: Bennett): LMB. NCI (D. Fitzgerald): WRAIR, Washington, D.C,
(D.N, Taylor): IDF, Israel (D. Cohen): Sheba Medical Center, Israel (J,
Passwell): Children's Hospital of Michigan, Detroit, MI (S. Sarnaik)

■ii'iniwmwiiiii'iiHMwiwiiiiHiniiiir

ZOl HD 01301-13 LDMI
Pro.ject Description:

The surface polysaccharides of bacterial pathogens that cause systemic Infections
are both essential virulence factors and protective antigens. Primary pathogens
cause systemic infections in otherwise healthy individuals. Their surface
polysaccharides may be capsular polysaccharides or lipopolysaccharides. Most
opportunistic pathogens also have surface polysaccharides that may be protective
antigens. The immunologic properties of polysaccharides limit their usefulness
as vaccines. For capsular polysaccharides there is: 1) Age-related serum
antibody response. Immunization of infants elicits few responses of mostly IgM
and of short duration. "Protective" levels start to be achieved in two year-olds
and an adult response develops at ~6 - 12 years of age: 2) T-cell Independent.
Most polysaccharides do not elicit a booster response at any age. The situation
for lipopolysaccharides (LPS) is slightly different. LPSs are composed of three
domains: 0-specific polysaccharide - Core - Lipid A. The latter is attached to
the outer membrane and the pharmacologic properties of the bacteria are mediated
by Lipid A (pyrogenic, inflammatory, vasomotor activity [shock] and effect upon
WBCs and platelets). Accordingly, the lipid A alone or as part of LPS is not
suitable in a vaccine: and 3) the 0-specific polysaccharide is the region in
contact with the host and is a protective antigen but acts as a hapten when
purified.

Ob.lectives:

Conjugates are more immunogenic than systemic infections: Vaccines more
immunogenic than disease. When bound to foreign proteins, especially inactivated
and pun tied vaccine components (tetanus and diphtheria toxoids) , the immunologic
properties of polysaccharides improve. One of our objectives is to improve the
immunogenicity of polysaccharide-based conjugates. One important variable has
been the molecular weight of the polysaccharide component. This has been
controversial, with some suggesting an "optimal" size between high molecular
weight, perhaps about 40,000D, and "oligosaccharides." Our laboratory and others
have shown that higher molecular weight polysaccharides elicit higher levels of
antibodies. There is, however, the technical problem that high molecular weight
polysaccharides tend to form more insoluble and technically unsatisfactory
conjugates.

Another problem is the poor immunogenicity of the polysaccharides such as the
capsular polysaccharides of pneumococcus type 6 and the poly a(2^8)NeupNAc of
Group B meningococci and E. coll Kl: both of these capsules are associated with
important and common pathogens. As conjugates, these polysaccharides have proven
to be lesser immunogens, than other polysaccharide conjugates.

Pertussis toxin will certainly be a component, if not the sole component of the
new acellular pertussis vaccines. Since it is anticipated that additional
carrier proteins will be required for a multivalent polysaccharide conjugate
formulation, pertussis toxin should be a useful candidate. Preliminary
experiments showed that polyamons, including medically useful capsular
polysaccharides, combine and precipitate with pertussis toxin. Pertussis toxin
itself is not soluble at pH from about four to nine making it difficult to use
for conjugation. Further, the neutral polysaccharide, pneumococcus type 14. is
also from a common and important pathogen for children. Conjugates formed

^HiJ^HH^

ZOl HD 01301-13 LDMI

between pneumococcus type 14 and pertussis were synthesized and were not toxic
in vitro. The type 14-pertussis toxin conjugate elicited both type 14 and
pertussis neutralizing antibodies in mice. The clinical safety and efficacy of
such a conjugate must now be evaluated.

A heterofunctional . fully protected tetrasaccharide was synthesized which
corresponds to a complete repeating unit of the 0-specific polysaccharide of S.
dysenteriae type 1 and was used as a building block for the preparation of
higher-order saccharides. A dimer. a trimer and tetramer of this repeating unit
were assembled, corresponding to octa-, dodeca- and hexadeca-saccharide fragments
of the 0-specific.

Lastly, the LDMI has demonstrated good immunogenicity and preliminary efficacy
of Shigella sonnei 0-specific polysacchari de-protein conjugates in armed forces
recruits in the Israel Defense Force. The LDMI now began evaluating these
investigational vaccines in children, the age group that suffers the highest
incidence, morbidity and mortality caused by shigellae and other enteric
pathogens.

Methods Employed:

The capsular polysaccharides of pneumococcus type 6B and Group B streptococci
were purified and characterized according to methods established in our
laboratory. The LPS was extracted from Plesiomonas shigelloides (identical to
S. sonnei ) and S. flexneri type 2a purified by hot phenol extraction and its 0-
specific polysaccharide purified by acid hydrolysis. The final products were
purified by gel filtration and their purity and biosafety assayed by established
methods .

Clinical evaluation of conjugate vaccines at the Clinical Center. NIH, the Walter
Reed Army Institute of Research and the Israel institutions of Israel Defense
Force and the Sheba Medical Center proceeded after approval by the NIH. FDA. IRBs
of the above institutions.

Serum antibodies to capsular polysaccharides, 0-specific polysaccharides and to
the carrier protein are measured by a standardized ELISA.

In order to avoid the potential toxicity of pertussis toxin as a conjugate, a
non-toxic protein will be used. A genetically engineered strain of B. pertussis
synthesizing a cross-reacting non-toxic mutant pertussis toxin has been
cultivated and initial studies of it purification and methods for improving its
solubility are underway. Since the mutation has been created in the
enzymatically active site, the techniques of affinity chromatography, used
successfully for the native toxin are not applicable.

Ma.ior Findings:

In a preliminary study recruits of the Israel Defense Forces, the S. sonnei -rEPA
conjugate vaccine was shown to confer type-specific protection against diarrhea
and fever. A more extensive placebo-controlled, randomized blinded study showed
an efficacy of about lb%. A phase 2 study (safety and immunogenicity in children
and then, if the results are satisfactory, in infants) has been approved and

ZOl HD 01301-13 LDMI

started by the Pediatrics Department, Sheba Medical Center. Israel. Prelminary
epidemiologic analysis of enteric infections has started in a rural area of
Vietnam in order to provide information about Shigella types that are important
but not common in the U.S. or in countries like Israel.

S. aureus type 5-rEPA injected into 17 end-stage-renal disease patients induced
type 5 antibody rises in the three major Ig classes, mostly in IgG. Maximal
levels at six weeks, were not boosted by an additional injection at this time.
Fold increases and GM levels were about one-half those of normal volunteers, but
at six months, the GM IgG levels were 10- fold higher than the preimmunization
levels.

Pn6B-TT and GBS III-TT were found to be injected at about half the intended dose
due to the conjugates binding to the vials, GBS III-TT was also shown to break
down over a two year period. Pn6B-TT injected into adult and chronic obstructive
pulmonary disease individuals induced antibodies of isotypes and opsonic
activities similar to those induced by the 23 valent polysaccharide vaccine.
This Pn6B conjugate lot was injected into infants at 3. 4 and 6 months of age
(n=21) or at 7 and 9 months of age (n=19). In the younger infants only two
achieved the estimated protective level after the primary immunization, in the
older infants ten reached this level. Booster injections at 18 months induced
high antibody levels (5-15-fold the protective level) inmost children. This lot
was also injected into 17 infants with sickle cell disease. The antibody
responses of these infants was similar to those of the healthy infants.

GBS III-TT injected into child-bearing age females induced GM IgG anti-type III
levels similar to the reported for the polysaccharide alone, with a decline of
one-third at one year.

Proposed Course:

Methods to improve the immunogenicity of conjugate vaccines, especially of lesser
immunogenic polysaccharides will be investigated. Clinical lots of Haemophilus
influenzae type a and meningococcus group B will be prepared.

Significance to Biomedical Research and the Program of the Institute:

The above diseases are major causes of morbidity and mortality in the U.S. and
worldwide. In particular, shigellosis is a major cause of stunted growth of
children throughout the world. In addition to providing an effective vaccine for
preventing these infections, the LDMI research program is attempting to confirm
that vaccine-induced serum antibodies, especially of the IgG class, can be a
protective mechanism against enteric infections including both bacterial and
viral pathogens.

Publications:

Robbins JB. Schneerson R, Szu SC. Hypothesis: How licensed vaccine confer
protective immunity. J Jpn Soc Pediatr Infect Dis 1994:7:39-50.

Robbins JB. Schneerson R, Szu SC. Perspective: Hypothesis: Serum IgG antibody
is sufficient to confer protection against infectious diseases by inactivating

eeemiasaauaasBi

ZOl HD 01301-13 LDMI
the inoculum. J Infect D1s 1995:171:1387-98,

Robbins JB, Schneerson R, Vann WF, Bryla DA. Fattom A. Prevention of systemic
infections caused by group B Streptococcus and Staphylococcus aureus by
multivalent polysaccharide-protein conjugate vaccines. In: Williams JC,
Goldenthal KL, Burns DL. Lewis BP, Jr. eds. Combined vaccines and simultaneous
administration: Current issues and perspectives. New York: New York Academy of
Sciences, 1995:68-82.

Claesson BA, Trollfors B, Anderson PW. Johansson J. Taranger J. Schneerson R,
Robbins JB. Serum antibodies in 6-year-old children vaccinated in infancy with
a Haemophilus influenzae type b-tetanus toxoid conjugate vaccine. Pediatr Infect
Dis J (accepted for publication).

Welch PG, Fattom A. Moore J. Jr. Schneerson R. Shiloach J. Bryla DA, Li X.
Robbins JB. Safety and immunogemcity of Staphylococcus aureus type 5 capsular
polysaccharide-Pseudomonas aeruginosa recombinant exoprotein A conjugate vaccine
(SA5-rEPA) in patients on hemodialysis. J Am Soc Nephrol (accepted for
publication) .

mmmmmmm

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBUC HEALTH SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD 01308-12 LDMI

PERIOD COVERED

nrt.nhpr 1. 1994 to September 30. 1995

TITLE OF PROJECT (80 characters or less. Title must fit on one line between the borders!

rnniuQate- Induced Polysaccharide Antibodies

PRINCIPAL INVESTIGATOR /List other professional personnel below the Principal Investigator) {Name, title, laboraton/, and institute affiliation!

PI: S.C. Szu
See Attached

Research Chemist

LDMI, NICHD

COOPERATING UNITS (if any!

Walter Reed Army Institute of Research, Washington, U.C.
(D.N. Taylor): LCDB. NIDDK (J. Shiloach): Carolinas Medical Center. Charlotte,
N.C. (J.C. Parke, Jr.)

LA8/BRANCH

laboratory of Developmental and Molecular Immunity

SECTION

Unit on Molecular Biophysics and Biochemistry

INSTITUTE AND LOCATION

IICHD NTH Bethesda. Maryland 20892-2720

TOTAL STAFF YEARS:

5 4

PROFESSIONAL:

4.2

1.2

CHECK APPROPRIATE BOX(ES)

IS (a) Hunnan subjects D (b) Human tissues D (c) Neither
n (a1 ) Minors
D (a2) Interviews

SUMMARY OF WORK lUse Standard unreduced type. Do not exceed the space provided.)

Serum antibodies to the surface polysaccharides of Gram-negative enteric bacteria
confer protective immunity against these pathogens. Vi capsular polysaccharide
of Salmonella typhi was conjugated with protein to improve its immunogemcity.
Phase 1 studies confirmed the superior immunity of recombinant exoprotein A of
Pseudomonas aeruginosa (rEPA) bound to Vi . Phase 2 and phase 3 studies with this
investigational vaccine are planned for Vietnam, an area with a high attack rate
of typhoid fever. A semi -synthetic Vi antigen using plant polysaccharide,
pectin, was prepared and showed identical antigenicity as the native Vi . A
conjugate, composed of a di -0-acetylated pectin as a vaccine for typhoid fever,
IS being prepared for clinical evaluation. Escherichia coli 0157, considered as
an important emerging pathogen in the U.S.. causes hemolytic uremic syndrome in
young children. Phase 1 study of the detoxified 0-specific polysaccharide and
covalently bound to rEPA elicited 100-fold antibody rise in healthy adults after
four weeks and the antibodies were bactericidal. We plan a phase 2 study for
therapeutic usage of hyperimmune plasma in children. Salmonella paratyphi A
accounts for lOX of enteric fever in Southeast Asia. A vaccine prepare in the
similar fashion as E. co1i 0-157 was prepared and elicited bactericidal
antibodies in mice. We plan to use this conjugate as a control in our phase 2
study in Vietnam. Poly(dA) has similar charge distribution as polya(2-8) sialic
acid, the capsular polysaccharide of group B meningococci. We conjugate
poly(dA) with a carrier protein and mice Immunized with poly(dA)-BSA conjugate
elicited anti -sialic acid antibody. We showed for this a highly charged,
flexible polymer, the charge distribution is the dominant factor for specificity.

PHS 6040 (Rev. 5/921

MLlkUJWUJU

^maauB

ZOl HD 01308-12 LDMI

Others:

J.B. Robbins
S. Bystricky
E.Y. Konadu
Z. Kossaczka
D.A. Bryla
D.W. Towne
H.T. Iran

Chief

LDMI,

NICHD

Visiting Scientist

LDMI.

NICHD

Visiting Fellow

LDMI.

NICHD

Visiting Fellow

LDMI.

NICHD

Medical Statistician

LDMI.

NICHD

Chemist

LDMI.

NICHD

Biologist

LDMI.

NICHD

iimiiMiMiimiiiiiimiiiiinim

ZOl HD 01308-12 LDMI
Pro.iect Description:

Enteric bacterial pathogens are a major cause of mortality and morbidity,
especially in infants and children, throughout the world. In the United States,
enteric bacterial pathogens continues to cause outbreaks of diseases and
admissions to hospitals. The LDMI has developed experimental data to show that
the surface polysaccharides of these bacteria are both essential virulence
factors and protective antigens. Most importantly, the LDMI has shown that
vaccine-induced serum antibodies to these polysaccharides can exert protection
and a mechanism by which this protection may occur has been proposed. Lastly.
the LDMI has extended their expertise to develop polysaccharide-proteins
conjugate vaccines which can easily be standardized and can be administered to
infants as part of their routine schedule of immunization.

Objectives:

Enteric fevers are common and serious worldwide: the most important of which is
typhoid fever. In Southeast Asia, more than 70% isolates are antibiotic
resistant. The capsular polysaccharide (Vi) is both an essential virulence
factor and a protective antigen of Salmonella typhi . Field trials in Nepal and
in the Republic of South Africa showed that a single injection of Vi conferred
about 70^ protection in children s5 years-old and in adults. Its protective
action is to elicit a critical level of serum antibodies. Vi was conjugated to
proteins to increase its immunogenicity and make it suitable for routine
immunization of infants against typhoid. A Phase 1 clinical trial in the U.S.
showed that Vi conjugates elicited significantly higher levels of serum
antibodies than the Vi alone. LDMI is planning to clinical trials of Vi
conjugates in children and infants of endemic areas.

The Vi is a linear homopolymer of (1^4)-Q!-D-GalpANAc. variably 0-acetylated at
C3. Synthesis of Vi conjugates posed several problems. First, the high M, of
Vi C2 X 10^ kD) causes conjugates to be poorly soluble. Second, standardization
has been hindered by a lack of a col ori metric methods for Vi . Pectin, a common
polysaccharide of plants, is a copolymer composed mainly of poly (l-*4)-Q!-D-GalfiA.
Treatment of pectin with acetic anhydride results in C-acetylation of Cj and C3
(OAcPec). The resultant derivative differs from the Vi only in that its C^ is
0- rather than /V-acetylated. As a result, OAcPec can be measured by the
carbazole reaction. Pectin does not react with Vi antiserum and it does not
elicit Vi antibodies in mice. OAcPec. in contrast, precipitates with Vi
antiserum. Unlike Vi . OAcPec is not immunogenic in mice probably due to its
comparatively low M, ('400 kD). We conjugated OAcPec to proteins according to
the scheme used for Vi and other polysaccharides containing an aminohexuronic
acid. OAcPec has some advantages because it can be measured by standardized
calonmetric assays and because it forms more soluble conjugates with proteins
than Vi . One disadvantage is that its glycosidic bond is not as stable as that
of Vi . We have demonstrated in mice and guinea pigs that C-acetylated pectin
protein conjugates antibodies against Vi . We plan to compare di-OAc pectin with
Vi conjugates in infants.

In order to understand the difference between OAcPec and Vi . the conformational
stability of these two molecules are compared with energy minimization
calculation and molecular dynamic simulation. Polysaccharides, unlike proteins.

ZOl HD 01308-12 LDMI

do not fold into globular conformation and usually are in extended helical forms.
Replacing the /V-acetyl group at the C2 position with the 0-acetyl also remove the
positive charge and the possible proton donor to form hydrogen bond with the
neighboring carboxyl group. The distance and angles of the glycosidic bonds are
calculated both in vacuum and in water to compare the solvent effect.

The relation between polysaccharide structure and its immunologic properties is
important for designing an effective vaccine or in searching for cross-reactive
antigens. The aforementioned pectin-V1 study serves as a good example. One
other highly charged polysaccharide, the a(2^8) linked polysialic acid is the
surface antigen virulence factors of E. coli Kl and group B Neisseria
meningococcus and appears on the surface of mammalian neural cells. Its
immunologic properties has been found to be dependent on its carboxyl ic groups.
For example antibodies against poly q!(2^8) sialic acid cross react with
polyanions with similar charge distributions, such as poly(dA). To study the
immunogenic cross reactivity of two polymers with similar charge distribution but
different molecular structure, we prepared conjugates of poly(dA) and evaluated
the serum response to these immunogens in mice.

The second most common enteric fever in Southeast Asia is Salmonella paratyphi
A which account -15^ of all the cases. As with S. typhi . S. paratyphi A is a
habitant of and a pathogen for humans only and can be considered as a "clone".
However, unlike S. typhi , paratyphi A does not contain a capsule. We have
demonstrated that antiserum against Its 0-specific chain of LPS is bactericidal.
The importance of the O-acetyl to the structure and immunologic properties of the
0-specific polysaccharide was studied. We feel that the bactericidal antibodies
elicited by our investigational 0-spec1fic polysaccharide-protein conjugate of
S. paratyphi A is a clinical correlate of protection and clinical trials in
endemic area are planned.

Escherichia coli 0157 is considered as an "emerging pathogen." Disease caused
by this organism exerts a high morbidity and mortality despite effective
antimicrobials and supportive therapy. Incomplete data suggests that
antimicrobial therapy of the hemorrhagic colitis caused by E. coli 0157 increases
the incidence of the hemolytic uremic syndrome. The pathogenesis of this
complication is unknown but it likely involves the action of the shigella-like
toxins of this organism. We prepared 0157 0-spec1fic polysaccharide protein
conjugate prevention of infections caused by this pathogen. Vaccines prepared
with different methods were evaluated in adult volunteers at the North Carolina
Medical Center. LDMI Is also preparing conjugates using the shigella-like toxin
as a carrier for E. col1 0157 0-specific polysaccharide conjugates In order to
prepare therapeutic globulin for treatment of hemorrhagic colitis caused by E.
coli 0157 and related organisms such as E. coli 0124.

A related pathogen. E. coli 0111. is also an important cause of enteric
infections Including newborn nurseries. 0111 strains of varying H types and with
different virulence factors cause enteritis: this 0 type is found rarely in
healthy individuals. Prevention by active immunization and treatment by passive
immunization of this pathogen is planned. 0-specific polysaccharide conjugates
of this pathogen have been synthesized and found to induce bactericidal
antibodies in mice. Due to the acid labile nature of colitose. a dideoxysugar
in the 0-specific polysaccharide, a non-denaturing process has been developed to

mmmm

ZOl HD 01308-12 LDMI
detoxify the polysaccharide.

Lastly, this laboratory has reinvestigated vaccine preparation against endotoxic
shock caused by Gram-negative bacteria. Endotoxic shock is a major cause of
nosocomial infection in U.S. and has the highest mortality rate. There is still
no effective means for treatment and prevention. Most Gram-negatives share
similar lipid A and core regions in their lipopolysaccharide. Therefore a vaccine
that induces antibodies against this region(s) would cover a broad range of Gram-
negatives. Both the core and the lipid region are short and presented as poor
immunogens. LDMI had prepared conjugates using hydrazine-treated LPS from E.
coli J5, a rough mutant of E. coli 0111. Our results showed only species-
specific protection against E. coli in animals. We plan to study new methods for
conjugating short polysaccharides to carrier proteins.

Methods Employed:

Carboxyls on Vi were characterized by potentiometric titration, circular
dichroism and kinetic analysis of the reaction of Ca" and Na* salts with 2-nitro-
phenyl hydrazine.

Conformation energy calculations were performed with Program Random Molecular
Mechanics (RAMM) and molecular dynamics simulations was done in the Consistent
Valence Force Field (CVFF).

Vi and the 0-specific polysaccharides were purified and derivatized by methods
devised or used by the LDMI. Polysaccharides were studied by NMR. PAGE for
purity. Hestrin reaction for 0-acetyl and 2-nitro-phenyl hydrazine for carboxyl
groups, immunodiffusion and bioassays for antigenicity. Carrier proteins were
purified chromatography and their purity and safety assayed by physical chemical
methods. Conjugates of these polysaccharides and proteins were synthesized by
methods devised at the LDMI and prepared under good manufacturing practices to
achieve conditions suitable for FDA approval and clinical use.

The new method for Vi conjugation was to derivatize the carrier protein with
adipic dihydrazide and then bound to Vi in the presence of water soluble
carbodiimide. The new method for paratyphi A conjugation was to activate
polysaccharide with l-cyano-4-dimethylamino pyridinium tetrafluoroborate (CDAP)
at pH 7 to 8 with or without adipic dihydrazide: the former will form a conjugate
directly bound to the protein.

Evaluation of the experimental vaccines in mice used clinically relevant
protocols designed by the LDMI. Serum antibody responses induced in humans was
assayed by quantitative and standardized ELISA. Functional assays, including the
vibriocidal assay for V. cholerae. used in vitro and in vivo methods.
Determination of the safety of these conjugates used methods outlined in the U.S.
CFR. Clinical studies in humans were enacted after approval by the appropriate
NIH, FDA and local IRB.

Major Findings:

The 0-specific polysaccharide of E. coli 0157 LPS was detoxified with acetic acid
or hydrazine and derivatized with adipic acid dihydrazide and bound to P^

MIIIIIIIIIIIUMIHIIIIIIIIIIIII

ZOl HD 01308-12 LDMI

aeruginosa recombinant exotoxin A. In collaboration with Dr. J.C. Parke, Jr.,
87 healthy volunteers from the Carolinas Medical Center were injected with one
of the conjugate. All three conjugates elicited serum antibodies against LPS
(>100 fold rise) with bactericidal activity to the 0157 LPS. The highest rise
of LPS antibodies was of the IgG class. The antibody levels declined slightly
(to -80 fold rise) after six months.

The 0111 0-specific polysaccharide is composed of a pentasaccharide repeat with
colitoses bound to the C3 and to the C6 of glucose in a trisaccharide backbone.
The 0111 0-specific polysaccharide was prepared by treatment of its LPS with
acetic acid (0-SP) or the organic base hydrazine (DeA-LPS). The 0-SP had reduced
levels of colitose. These products were derivatized with adipic acid dihydrazide
(ADH) or thiolated with N-succinimidyl-3(2-pyridyldithio) propionate (SPDP) . The
four derivatives were covalently bound to tetanus toxoid (TT) by carbodiimide-
mediated condensation or with SPDP to form conjugates. Immunization of BALB/c
and general purpose mice showed that DeA-LPS-TT^^^H' of the four conjugates,
elicited the highest level of LPS antibodies. This differential immunogenicity
is best explained by the intact colitose content and by the superiority of the
adipic acid hydrazide as a linker.

We improved our synthesis of Vi conjugates. The improved method uses ADH
derivatized protein and elicited higher antibody levels in mice than the
conjugates synthesized previously with SPDP. We are planning a phase II study
with both conjugates in children and in infants in Vietnam.

Structure/function relations have been studied in order to understand the
immunologic properties of Vi . The degree of 0-acetylation is critical for the
immunogenicity of Vi . 0-acetyls blocks the binding of cations with carboxyl
groups on Vi . We showed by energy minimization calculation that there is only
one minimum for the glycosidic linkage. (0,(/? ) = (110, -86). This energy minimum
is shallower for Vi than for OAcPec. However, the effect of water is stronger
for OAcPec (+19.2 ev) than for Vi (3.6 ev). From molecular dynamics simulations
we have found that both Vi and OAcPec are stable with low probability of
conformational transition.

The use of a plant polysaccharide, pectin, provides a novel approach as an
immunogen for prevention of typhoid fever. OAcPec conjugated to tetanus toxoid
(TT) elicited Vi antibodies in mice and guinea pigs and reinjection elicited a
booster response. Clinical studies with OAcPec are planned.

0-specific poysaccharide conjugates from 5. paratyphi A with tetanus toxoid are
synthesized by several schemes: 1) CNBr activation and ADH as a linker for both
acetic acid and hydrazine-treated saccharides: 2) CDAP activation at neutral pH
with or without ADH as a linker with acid or hydrazine treated saccharides. Mice
immunized with these conjugates had similar anti-LPS responses for conjugates
using acetic acid-treated LPS. Hydrazine-treacted polysaccharide conjugates, in
contrast, did not elicit LPS antibodies. Chemical analysis showed that
hydrazinolysis had removed all 0-acetyls from the polysaccharide. "C NMR showed
that the polysaccharide is 0-acetylated at two sites rather than only the
rhamnose site as reported. The essential role of 0-acetyl in the immuogenicity
explains the failure of A component of TAB vaccine to confer protection against
infection with S. paratyphi A. The use of CDAP was more effective in retaining

ZOl HD 01308-12 LDMI

C-acetyls than CNBr.

In collaboration with Dr. Y.C. Lee, Johns Hopkins University, we started various
ways to purify, detoxify and conjugate oligosaccharide from mutant strains: E.
coli J5 and Salmonellae. We have shown that mild hydrazinolysis detoxified the
LPS and retained the diglucosamine structure of the inner core.

Proposed Course:

Clinical evaluation active immunization with the above conjugates in sequential
studies in adults, children and then infants is planned. Trials of the 0157
conjugates in cattle are planned in order to evaluate the potential for these
vaccines to eliminate carriage of E. coli 0157 in this reservoir. A phase II
studied is planned with the Carolinas Medical Center for 2-years-old children.
Plasmapheresis of sera from high responders in adult volunteers are planned for
therapeutic usage of this vaccine. The preparation of conjugates using non-
toxic shigella-like toxins I and II, subunit B for therapeutic globulin as a
treatment of hemorrhagic colitis caused by E. coli 0157 is planned.

Vi conjugates will be evaluated in Vietnam. The attack rate of typhoid fever in
Vietnam is about 1% and most strains (70^) are antibiotic resistant. 5.
paratyphi A is the second most common cause of enteric fever in Vietnam.
Conjugates using CDAP with and without a linker will be evaluated. Both phase
I and II studies of these conjugates for Salmonella will be performed in this
high endemic area.

We demonstrated the dominance of 0-acetyls in the immunogenicity of S. paratyphi
A, The reported structure of the 0-specific side chain of A. paratyphi A does
not explain our immunologic findings. In the reported structure, 0-acetyl is
linked to the rhamnose of the backbone the which sterically is not as prominent
as compared to the branch sugar, paratose. In addition, from NMR we have
observed two acetyl peaks. From model building and base hydrolysis, we propose
that at least one of the 0-acetyl group is linked to paratose and together they
form the immunodominant site.

Vaccines for endotoxic shock will be prepared from two mutant strains: J5 from
E. coli and Salmonella minnesota R595. At least two different methods for
conjugation of these oligosaccharides will be studied. One will utilize the
carboxylic on the KDO residue on the core. The other will be to remove the N-
acyl linked lipid with hydrazine at lOOoC and then link residual disaccharide
region of the lipid A to polyacryl amide as a backbone. Both methods should
preserve the diphosphate region of the lipid A which is believed to be
immunodominant.

Significance to Biomedical Research and the Program of the Institute:

Prevention of enteric bacteria! infections, especially of infants and children,
will be an important achievement of the NICHD. The development of immunotherapy
against the complications of E. coli 0157, 0111 as well as other enteroinvasive
pathogens, may establish a new approach for treating these serious disease caused
by an emerging pathogen.

ZOl HD 01308-12 LDMI

The major cause of nosocomial death is the endotoxic shock caused by Gram-
negative organisms. There is yet no available treatment for this serious
infection. From our earlier study and clinical trials of monoclonal antibody
study in other laboratories, the antibodies against the diphosphate di -
glucosamine region will be protective. Both the prevention and therapeutic
approach are taken here.

The relation between the polysaccharide structure and its immunologic properties
are better understood by the energy minimization calculation and molecular
dynamic simulation of Vi and its analogue, the 0-acetylated pectin. Through the
circular dichmism and intrinsic viscousity measurement of poly q:(2-8) sialic
acid, it is clear that this polysaccharide is extremely flexible and its binding
property with antibodies are mainly enthalpy driven. The cross-reaction of poly
a(2-8) sialic acid with other negatively charged polymer was further demonstrated
by animal immunized with the poly(dA)-BSA conjugate. We conclude that it is
possible to generate cross -reactive antibodies from two structurally different
molecules with similar charge distribution.

Publications:

Szu SC. Bystricky S. Hinojosa-Ahumada M. Egan W. Robbins JB. Synthesis and some
immunologic properties of an 0-acetyl pectin (poly alpha-[l->4]-D-GalpA)-protein
conjugate as a vaccine for typhoid fever. Infect Immun 1994:62:5545-9.

Bystricky S, Szu SC. Gotoh M, Kovac P. Circular dichroism of the 0-specific
polysaccharide of Vibrio cholerae 01 and some related derivates. Carbohydr Res
1995:270:115-22.

Gupta RK. Egan W, Bryla DA, Robbins JB. Szu SC. Comparative immunogenicity of
conjugates of Escherichia coli 0111 0-specific polysaccharide, prepared by
treatment with acetic acid or hydrazine, bound to tetanus toxoid by two synthetic
schemes. Infect Immun 1995:63:2805-10.

Liao J. Nickerson KG. Bystricky S. Robbins JB. Schneerson R. Szu SC, Kabat EA.
Characterization of a human monoclonal IgM antibody (IgMBEN) specific for the Vi
capsular polysaccharide of Salmonella typhi . Infect Immun (in press).

DEPARTME^^■ OF HEALTH AND HUMAN SERVICES - PUBUC HEALTH SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD 01311-02 LDMI

PERIOD COVERED

October 1, 1994 to September 30. 1995

TITLE OF PROJECT 180 characters or less. Title must fit on one line between r/ie borders.!

A Polysaccharide Vaccine for Mycobacterium tuberculosis

PRINCIPAL INVESTIGATOR lUst other professional personnel below the Pnncipal Investigator.! {Name, title, laboratory, and institute affiliation!

PI:

See Attached

J.B. Robbins

Chief

LDMI, NICHD

COOPERATING UNITS (if any!

LCDB. NIDDK (J. Shiloach): CBER, FDA (S. Morris. J. Muller;
Merieux Serum et Vaccins. Lyon. France (D. Schulz).

Pasteur-

LAB/BRANCH

Laboratory of Developmental and Molecular Immunity

SECTION

Section on Bacterial Disease Pathogenesis and Immunity

INSTITUTE AND LOCATION

ilCHD. NIH. Bethesda. Maryland. 20892-2720

TOTAL STAFF YEARS:

5.5

PROFESSIONAL

4.3

OTHER;

1.2

CHECK APPROPRIATE BOXIES)

D (a) Human subjects
n (a1 ) Minors
D (a2) Interviews

H (b) Human tissues D (c) Neither

SUMMARY OF WORK lUse standard unreduced type. Do not exceed the space provided.!

Tuberculosis remains a serious and common disease worldwide. In the U.S.
emergence of multi -antibiotic resistant strains, due to the rapid passage of
Mycobacterium tuberculosis in patients with AIDS now poses a public health
problem. Although the most frequently used vaccine in the world, there is no
scientifically-based evidence that BCG prevents primary pulmonary tuberculosis
and its protective effect in children has not been related with a protective
antigen or a host immune component. Our research into a new vaccine for
tuberculosis is based upon the similarity of primary infection caused by M.
tuberculosis with that of capsulated bacterial respiratory pathogens, vii; 1)
epidemiologic data which show that tuberculous meningitis has a similar age
distribution as capsulated bacteria: 2) the presence of a capsular polysaccharide
on M. tuberculosis and other mycobacteria In vitro and in vivo: 3) that BCG and
protein components of this and wild-type strains, without exception, at best have
prolonged but never conferred protection against challenge with wild-type M.
tuberculosis in laboratory animals: and 4) published data by F. Seibert that
protective immunity in rabbits was correlated with the level of serum
precipitating antibodies to M. tuberculosis polysaccharide II proposed to be a
linear homopolymer of poly a(1^2)-D-Glu. This a(1^2) glucose moiety is
identified in extracts of M. tuberculosis, but not in BCG, by its reactivity with
pneumococcal type 12F typing antisera. The reactivity in mycobacteria extracted
previously with saline and then with \% Triton. The residual organisms are
extracted by the hot phenol method used for LPS of Gram-negatives. The resultant
product precipitates with pneumococcal type 12F antisera giving an identity
reaction with dextran 1299 which has kojibiose residues. Animal and human
antisera reactive with pneumococcus type 12 and dextran 1299 elicit complement-
dependent killing of M. tuberculosis strains Erdmann and wild-type M123. The
cidal reaction is inhibited by a synthetic kojipentose but not by an Q!(l-^4)-D-Gal
trimer

PHS 6040 (Rev. 5/92)

mmmmm^rnKm

others:

ZOl HD 01311-02 LDMI

Schneerson

Research Medical Officer

LDMI

NICHD

Pozsgay

Visiting Scientist

LDMI

NICHD

Da1

Visiting Fellow

LDMI

NICHD

Dubois

Visiting Fellow

LDMI

NICHD

W. Tolson

Biologist

LDMI

NICHD

HiH

ZOl HD 01311-02 LDMI
Pro.iect Description:

Despite Its discovery over one century ago, there Is yet no scientifically-based
evidence for BCG or any other vaccine for prevention of primary pulmonary
tuberculosis. Tuberculosis continues to be a major cause of morbidity and
mortality throughout the world and, in the U.S., mult1 antibiotic resistant
strains of M. tuberculosis pose an important public health problem.

There Is good evidence that administration of BCG In Infants prevents tuberculous
meningitis In young children and Its administration Into the bladder Is an
effective therapy for bladder carcinoma. But the protective antigen and the host
immune response elicited by BCG are not characterized. The skin test for
tuberculosis Infection, detected by PPD or a comparable product. Is mediated by
T-cells but. is not related to protective Immunity to tuberculosis.

There Is published experimental evidence that specific serum antibodies should
be reinvestigated as a protective Immune mechanism to tuberculosis, vTz: 1) the
age Incidence of tuberculous meningitis, as an example of a systemic complication
of a respiratory Infection, is similar to other capsulated bacterial pathogens:
2) similar to other capsulated pathogens, polymorphonuclear white blood cells are
the first to gather at the site of an inoculum of M. tuberculosis. Although WBC
engulf and kill some organisms, these cells are replaced with mononuclear cells
after 24 hours (please note that serum capsular polysaccharide antibodies
facilitate opsonophagocytosis and intracellular killing of pneumococci ) : 3) M.
tuberculosis and other pathoenic mycobacteria are surrounded by a clear zone,
referred to as the "ETZ", shown to be a polysaccharide: and 4) preliminary
evidence that serum antibodies, induced in some rabbits by multiple intravenous
injections of inactivated mycobacteria, were correlated with protein against
challenge. One polysaccharide, identified as Polysaccharide II by its
discoverer. Dr. Florence Seibert, seems to be a reasonable candidate to
Investigate as a protective antigen.

Ob.iectives:

To establish an in vitro sopsonphagocytosis and killing assay for M. tubercuosis
with human antibodies, complement and white blood cells. To isolate the
"Polysaccharide 11" fraction of M. tuberculosis and to confirm its structre by
modern techniques. To synthesize oligosaccharides of a(1^2)-D-glucose and to
characterize their immunochemical and biological properties upon
opsonphagocytosis of M. tuberculosis.

Methods Employed:

Pneumococcal type 12F polysaccharide, a component of the 23 valent vaccine
contains an Q!(l-*2)-D-glucobiose moiety:

^4)-a-D-ManpNAcA-(l-4)-/?-/.-FucpNAc(l-3)-iS-D-GalpNAc(l-
3 3

t t

1 1

a-/}-Glcp(1^2)-a-Z?-61c a-O-Gal

ZOl HD 01311-02 LDMI

Streptocococcus pneumoniae type 12F (Leontein, et al. . 1981).

Typing antisera to this pneumococcal type was used to identify this unusual
glucan in dextrans: one dextran denoted as 1299. in particular, has been studied
extensively because of its unusual reactivity with type 12 pneumococcal
antiserum. Either rabbit or horse pneumococcal type 12F antiserum has been used
to detect the Q!(1^2)-D-glucobiose moiety by double immunodiffusion. Curiously,
two courses of multiple intravenous injections of formalin-treated M.
tuberculosis, strain H37R, failed to elicit antibodies to pneumococcus type 12F
or to dextran 1299. Immune sera from these animals, however, precipitated with
another polysacchride from M. tuberculosis known as lipoarabinomannan.

We established an in vitro assay for M. tuberculosis copied after that system
used for assaying bactericidal antibodies to Group B streptococci . Animal typing
antisera. murine monoclonal antibodies reactive with dextrans and those elicited
by immunization of mice with tubercle bacilli were evaluated for their reactivity
with the Q!(1^2)-D-glucose moiety.

Electron microscopy of M. tuberculosis incubated with pneumococcal type 12F
antisera showed intense surface strain no observed with murine antibodies to a
surface protein.

Human antibodies to pneumococcal type 12 were purified from the plasmapheresis
of an adult vaccination previously with a pneumococcal tpe 12F-diphtheria toxoid
conjugate. Purified human antibodies were prepared by affinity chromatography
with a type 12F polysaccharide-4B amino-hexyl Sepharose matrix.

Major Findings:

A saccharide-like structure, reactive with the anti -Q!(1^2)-D-glucose moiety of
rabbit typing, was purified from M. tuberculosis strain M123 from a patient
diagnosed with pulmonary tuberculosis. The preliminary isolation was done with
strains grown on Long's media at CBER, FDA and at the Pasteur-Merieux Institut.
Extraction was accomplished by the hot -phenol method for preparing LPS from
enterbacteriae. The Q!(1^2)-D-glucose was more definitely established by gas-
chromatography and mass spectroscopy of methylated derivative of the saccharide.

Dextran 1299, conjugated to chicken serum abumin, even when injected along with
complete Freund's adjuvant in mice, did not elicit anti -Q!(1^2)-D-glucose. Mice
so immunized elicited precipitating antibodies to the Q!(1^4) and Q!(1^3)-D-glucose
bonds of other dextrans.

Rabbit, horse and human anti -pneumococcus type 12 antibodies killed M.
tuberculosis strains Erdmann and M123 in the presence of complement and human
WBCs.

The antibody mediated, complement -dependent killing was inhibited by a
kojipentaose but, not with an Q!(1^4)-D-Gal -trimer.

Significance to Biomedical Research and the Program of the Institute:

The development of a vaccine based upon induction of serum antibodies to a

ZOl HD 01311-02 LDMI

surface polysaccharide of M. tuberculosis would provide an Important new public
health measure as well as a conceptual tool for understanding Immunity to
mycobacteria.

Publications:

None.

^^^mwffliy.iyiiiiiyjL iiiii,iiiiLiii,i|iiiHiiiii

DEPARTMENT OF HEALTH AND HUMAN SERVICES ■ PUBUC HEALTH SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD 01312-01 LDMI

PERIOD COVERED

rirtnhPr 1 1994 t.n Sppt.pmher 30. 1995

TITLE OF PROJECT 180 characters or less. Title must fir on one lirje between the borders.l

Tc;nlRtinn ^nd Purl firatinn of Subunit B of Shiga Toxin

PRINCIPAL INVESTIGATOR lUst other professional personnel below the Principal Investigator! (Name, title, laboratory, and institute affiliation)

PI:
Others:

V. Pozsgay

J.B. Robbins
L.B. Tnnh

Visiting Scientist

Chief
Chemist

LDMI, NICHD

LDMI. NICHD
LDMI. NICHD

COOPERATING UNITS lif any}

LCDB. NIDDK (J. Shiloach); Tufts University School of Veterinary
Medicine. N. Grafton. MA (A. Donahue-Rolfe) ; Massachusetts General Hospital.
Rnc;tnn MA (S R Calderwood)

LAB/BRANCH

I ahnratnry nf Developmental and Molecular Immunity

SECTION

^prtmn nn Rarterjal Disease Pathogenesis and Immunity

INSTITUTE AND LOCATION

MTPHn NTH Rpthp^da Maryland. 20892-2720

TOTAL STAFF YEARS:

iLH

PROFESSIONAL:

0.1

OTHER:

0,2

CHECK APPROPRIATE BOX(ESI

n (a) Human subjects D (b) Human tissues H (c) Neither
n (a1 ) Minors
D (a2) Interviews

SUMMARY OF WORK iUse standard unreduced type. Do not exceed the space provided. I

The Shiga toxins are multimeric proteins consisting of a single enzymatic subunit
(subunit A mw 32 kDa) and a symmetrically arranged, pentameric subunit B (mw 7.7
kDa) The latter mediates the binding of the holotoxin to cell -surface
glycolipids as the first step of its endocytosis. Galabiose is the minimum
binding receptor that recognizes the toxin, placed either terminally or
internally in glycolipids. The binding is stronger with P blood group
determinant tri saccharides, such as the P^ antigen and the Pj antigen. Specific
adsorbents of Shiga (Shiga-like) toxins may aid the development of serologic

tests for rapid diagnosis of
oligosaccharides that inhibit
receptors can be candidates
non-toxic subunit B may be
poly saccharide -protein
carbohydrate-specificities

Escherichia coli infections, and receptor-analog

the binding of these toxins to their cell -surface

for therapeutics against such diseases. The

a candidate as a carrier in the synthesis of

conjugates. Based on the reported

of subunit B, we developed a simple and efficient

we
isolation

using a

method for the preparative scale isolation of this toxin fragment
receptor-analog affinity sorbent. Specifically, we synthesized the Pi antigenic
tri saccharide equipped with an anchor at its reducing end. The Pj tri saccharide
was covalently attached to a solid support through this anchor. The resulting
affinity material was used over twenty cycles to isolate subunit B of Shiga toxin
from the fermentation fluid of a non-virulet Vibrio cholera strain that containd
the gene for this protein, in a semi -automated fashion.

PHS 6040 (Rev. 5/92)

ZOl HD 01312-01 LDMI

Project Description:

The proper selection of the carrier protein In the design of carbohydrate-protein
conjugates Is crucially Important and the search for proteins that can be
alternative to those already established continues. Non-tox1c subunlts of
bacterial toxins may have the potential of being immunogenic carriers and subunit
B of the Shiga and Shiga-like family of secreted toxins can be a candidate for
this purpose. The gene of this subunit has been expressed in a Vibrio cholera
strain which lacks the gene for subunit A that is responsible for the toxicity
of the holotoxin. The carbohydrate specificity of subunit B has been established
and a procedure was reported for the isolation of this subunit using
receptor-analog affinity chromatography. The very low amount of the carbohydrate
hapten in the glycoprotein from the hydatid fluid that was used in the published
protocol precludes scale-up. Also, the method may be objectionable as the
subunit to be purified is intended for human and animal experiments. Our goal
was to design a method for the isolation and purification of subunit B of Shiga
toxin.

Objectives:

To synthesize a receptor-analog affinity sorbent for the preparative scale
isolation of subunit B of Shiga toxin and to use it in a semi -automated fashion
to recover this protein from the fermentation fluid of V. cholera (Strain
0395-Nl).

Methods Employed:

The Pi trisacchande containing D-galactose and N-acetyl-D-glucosamine residues

a-D-GalE-(1^4)-^-D-Galp-(1^4)-/?-D-GlcENAc

was assembled from suitably protected monosaccharides in stereoselective manner,
using thioglycosides and 2-(tr1methylsilyl )ethyl glycosides as major
intermediates. The tri saccharide was attached to a heterobi functional spacer
that contained a reactive, aldehyde group at its terminus. Through this group
the hapten-spacer assembly was coupled to a hydrazino group-containing solid
support, using the method of reductive ami nation.

Subunit B of Shiga toxin was produced by Dr. J. Shi loach (LCDB. NIDDK) by
fermentation of a non-virulent V. cholera strain (0395-Nl) obtained from Drs. A.
Donahue-Rolfe and S.B. Calderwood (Tufts University School of Veterinary Medicine
and Massachusetts General Hospital). The periplasmic extract was loaded directly
on the affinity column from which subunit B could be eluted selectively in high
purity. In each cycle approximately 10 mg of pure subunit B was obtained. Until
now the column was used over twenty cycles with unchanged capacity. The identity
and purity of subunit B so prepared was assessed by SDS-PAGE electrophoresis and
its antigenicity was demonstrated by its reaction with a monoclonal antibody
directed against subunit B.

Major Findings:

We established that subunit B of Shiga toxin can be isolated from the

WMI^^smmBSsmsaaa

ZOl HD 01312-01 LDMI

fermentation fluid of a non-virulent V. cholera strain containing the gene for
subunit B in high purity using a synthetic trisaccharide-based affinity material .

Proposed Course:

The potentials of subunit B of Shiga toxin as a carrier in polysaccharide-protein
conjugates will be evaluated.

Significance to Biomedical Research and the Program of the Institute:

The availability of novel protein carriers is important for the development of
carbohydrate-protein conjugate vaccines of Improved efficacy.

Publications:

Pozsgay V. Tnnh L. Shiloach J. Robbins JB, Donahue-Rolfe A. Calderwood S.
Purification of subunit B of Shiga toxin using a synthetic trisaccharide-based
affinity matrix. Bioconjug Chem (in press).

HWBginaili|||||l|i|'lliii|i|||iii||lilli I

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBUC HEALTH SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD 01313-01 LDMI

PERIOD COVERED

nrt.nher 1. 1994 to September 30. 1995

TITLE OF PROJECT (80 characters or less. Title must fit on arte lire between the borders.!

Synthetic Vaccine Against Shigellosis

PRINCIPAL INVESTIGATOR lUst other professional personnel below the Principal Investigator.! (Name, title, laboratory, and institute affiliation!

PI: V. Pozsgay
See Attached

Visiting Scientist

LDMI, NICHD

COOPERATING UNITS lif any!

LMC, NIDDK (C.P.J. Glaudemans:
Gaithersburg. MD (B. Coxon)

National Institute of Standards and Technology,

LAB/BRANCH

laboratory of Developmental and Molecular Immunity

SECTION

Section on Bacteria! Disease Pathogenesis and Immunity

INSTITUTE AND LOCATION

NlfHn NIH. Bethesda. Maryland. 20892-2720

TOTAL STAFF YEARS:

PROFESSIONAL:

0.5

OTHER:

0.5

CHECK APPROPRIATE BOXIES)

D (a) Human subjects
D (a1) Minors
D (a2) Interviews

D (b) Human tissues

SUMMARY OF WORK (Use standard unreduced type. Do not exceed the space provided.!

Shi gel 1 osi s caused by Shigella dysenteriae type 1 conti nues to be a major enten c
disease worldwide. Because of the multiple-antibiotic resistance of many strains
of Shigellae, the treatment of shigellosis is difficult. Although the need for
vaccines to control this disease has been documented by the WHO, there is no
licenced vaccine against shigellosis. Following the discovery by Robbins and
coworkers that serum antibodies against the 0-specific polysaccharide (OSP) of
S dysenteriae type 1 confer protective immunity in humans we hypothesized that
extended fragments of the OSP may also be suitable for the induction of
protective antibodies when coupled to immunogenic proteins, provided that the
conformational ensemble of such saccharides approaches that of the conformational
determinant of the native polysaccharide. The use of synthetic saccharides of
defined structure instead of polysaccharides of complex architecture is likely
to offer advantages including enhanced uniformity of conjugates and elimination
of the analytical difficulties associated with the established,
polysaccharide-protein vaccines. Based on this hypothesis we are developing
synthetic oligo- and poly-saccharide-based immunogens of well-defined
characteristics. We designed a strategy to prepare fragments of the OSP of S.
dysenteriae type 1. The OSP consists of a tetrasaccharide repeating unit that
IS composed of D-galactose, N-acetyl -D-glucosamine, and L-rhamnose. Starting
from monosaccharide building blocks that carry orthogonal protecting and
activating groups, a tetrasaccharide donor/acceptor molecule was assembled.
Iterative combination of this building block afforded di-, tri- and tetramers of
the repeating unit corresponding to octa-. dodeca- and hexadeca-saccharides.
Nuclear magnetic resonance spectroscopy indicated that the dodeca- and the
hexadeca-saccharides exhibit a high degree of conformational similarity to the
native O-SP which we believe is a prerequisite for protective antibody induction.
In vitro experiments showed that the synthetic saccharides inhibit the binding
nf thP nsp to homologous monoclonal antibodies.

PHS 6040 (Rev. 5/92)

^Hi

wamBmmsms^mmm^^

ZOl HD 01313-01 LDMI

others:

J.B. Robblns

Chief

LDMI

NICHD

R. Schneerson

Research Medical Officer

LDMI

NICHD

A.B. Karpas

Microbiologist

LDMI

NICHD

C.W. Broome

Chemist

LDMI

NICHD

H. Ao

Adjunct Technician

LDMI

NICHD

■■

ZOl HD 01313-01 LDMI
Pro.iect Description:

Shigella dvsenterlae type 1 is the causative organism of epidemic and endemic
diarrhea and dysentery In may parts of the world with high mortality and
morbidity. It Is dysentery (fever, mucous and blood in the stool) which is the
leading cause of growth retardation 1n the word. An approach to combat this
disease can be vaccine development. In spite of the fact, that the causative
organisms have been identified a century ago. there is no licenced vaccine
against shigellosis. The need for such vaccines Is documented by a World Health
Organization document which accords priority to the development of
ant1-sh1gellos1s vaccines. Although shigellosis can be Induced In animals, the
animal model for shigellosis has limitations which is highlighted by the fact,
that the challenge dose for monkeys is 10^^ organisms which is 10^ times higher
than the dose required to Induce shigellosis in healthy humans. An essential
virulence factor of Shigellae strains is the 0-specific polysaccharide component
of their respective lipopolysaccharides. Only smooth strains, having fully
expressed 0-SPs are virulent, whereas rough strains that are devoid of these
structures, are not. The essential role of the O-specific polysaccharides of
Shigellae and also other enteropathogenic bacteria led to the hypothesis of
Robblns and co-workers, that serum IgG antibodies to the OSPs confer protective
immunity to the host. The validity of this hypothesis was substantiated by
experiments with conjugate vaccines containing purified, O-specific
polysaccharides covalently linked to an immunogenic protein. Evidence Indicates,
that anti -0-specific polysaccharide antibodies in vivo interact with the
lipopolysaccharides on the bacterial cell surface. However, little is known
about the mechanism of such Interactions and about the underlying factors at the
molecular level, such as the importance of the hapten geometry, and hapten
density in our current conjugate vaccines, concerning either antibody production
or antibody recognition.

Ob.iectives:

We surmise that extended fragments of the O-specific polysaccharides may be
suitable for the induction of protective antibodies when coupled to immunogenic
proteins, provided that the conformational ensemble of such saccharides
approaches that of the conformational determinant of the native polysaccharide.
Therefore we have been studying chemical synthetic approaches to such
saccharides. We are also studying the conformational properties of the synthetic
oligosaccharides in relation to the conformation of the native 0-specific
polysaccharide. A scheme has been devised to covalently attach the saccharides
to proteins. The evaluation of the immunologic properties of the synthetic
oligosaccharides up to a hexadecasaccharide is currently under study.

Methods Employed:

The 0-polysacchar1de of S. dvsenterlae type 1 consists of the linear
tetrasaccharide repeating unit, containing a-linked N-acetyl -D-glucosam1ne.
D-galactose and L-rhamnose as the monosaccharide constituents.

[3)-a-L-Rhae-(1^2)-a-D-Gale-(l-3)-a-D-GlceNAc-(1^3)-a-L-RhaE-(l^],

Our plan was to construct a repeating unit block that can serve both as a

mmmm^^

ZOl HD 01313-01 LDMI

glycosyl donor and a glycosyl acceptor in a stepwise fashion from monosaccharide
synthons and use It In an iterative manner. A key tetrasaccharide intermediate
was selected for this purpose. A major feature of this building block is that
after its initial coupling with the aglycon under mild conditions that do not
effect the functional /protecting groups, the site of the chain elongation can be
unmasked by a minimal protecting group manipulation involving the chemoselective
removal of one only protecting group. The monosaccharide synthons were prepared
from D-galactose, D-mannose and L-rhamnose, using the methods of synthetic
organic chemistry and were combined in stereoselective, stepwise manner. The
structures of all the synthetic intermediates and the target saccharides 1-4
were confirmed by analytical and spectroscopic methods including various mass
spectroscopic protocols and ^H and C nuclear magnetic resonance spectroscopy.

[3)-a-L-RhaE-(l-2)-a-D-GalE-(l-3)-a-D-GlceNAc-(l-*3)-a-L-Rhae-(l^],-OCioH

1 n=l

2 n=2

3 n=3

4 n=4

Particular support was provided by the ^^C NMR spectra of the dodeca-(3) and
hexadeca-saccharides (4) in the latter of which the anomeric resonances for the
four Gal residues. 3 and 4 Rha residues, and 3 GlcNAc residues, respectively,
coincide with the corresponding resonances of the native. 0-polysaccharide of S.
dysenteriae 1. This coincidence not only establishes the stereochemical
integrity of the interglycosidic linkages but also indicates that compounds 3 and
4 express a high degree of conformational similarity to the native
polysaccharide.

Ma.ior Findings:

We established that preparative-scale quantities of extended fragments of the
native 0-specific polysaccharide of S. dysenteriae type 1 can be synthesized by
iterative combination of a tetrasaccharide building block. Nuclear magnetic
resonance studies indicate that an internal portion of the octasaccharide
fragment, corresponding to two contiguous repeating units, already expresses
conformational similarity to the native 0-specific polysaccharide. A higher
degree of conformational similarity is expressed by the dodeca- and the
hexadeca-saccharides which correspond to three and four repeating units,
respectively. Preliminary experiments indicate that the synthetic saccharides
inhibit the binding of 0-specific polysaccharide specific monoclonal antibodies
raised against heat-killed S. dysenteriae type 1.

Proposed Course:

The antigenic properties of the saccharide fragments of the 0-specific
polysaccharide of S. dysenteriae type 1 will be evaluated using several
0-specific polysaccharide-specific monoclonal antibodies. Covalent conjugates
of the synthetic saccharides and immunogenic proteins will be prepared and their
efficacy to induce anti 0-specific polysaccharide antibodies will be studied.
The protective effect of these antibodies will be evaluated. The capacity of the
synthetic conjugates to induce antibody will be correlated to their molecular

mmmmmm^mm^^^^mi

ZOl HD 01313-01 LDMI

characteristics Including hapten size and hapten density.

Significance to Biomedical Research and the Program of the Institute:

Shigellosis is a continuing public health problem in both the developing and
industrialized countries and the development of an effective vaccine will be an
important step toward eliminating this disease.

Publications:

Pozsgay V. Synthesis of a hexadecasaccharide fragment of the 0-polysaccharide
of Shigella dvsenteriae type 1. J Am Chem Soc 1995:117:6673-81.

Pozsgay V, Coxon B. Stereoselective preparation of alkyl glycosides of
2-acetamido-2-deoxy-a-D-glucopyranose by non-classical halide-ion catalysis and
synthesis and NMR spectroscopy of Q!-D-Gale-(1^3)-Q!-D-GlceNAc-0Me. Carbohydr Res
(in press) .

is:

DEPARTMENT OF HEALTH AND HUMAN SERVICES ■ PUBUC HEALTH SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD 01314-01 LDMI

PERIOD COVERED

October 1, 1994 to September 30, 1995

TITLE OF PROJECT (80 characters or /ess. Title must fit on one fine between the borders.)

Analysis and Synthesis of Carbohydrate Antigens of Mycobacterium tuberculosis

PRINCIPAL INVESTIGATOR (List other professions! personnel below the Principal Investigator.) (Name, title, laboratory, and institute afiiliationi

PI:

See Attached

V. Pozsgay

Visiting Scientist

LDMI. NICHD

COOPERATING UNITS (if any!

LAC. NIDDK (N. Whittaker):
Gaithersburg. MD (B. Coxon)

National Institute of Standards and Technology.

LAB/BRANCH

Laboratory of Developmental and Molecular Immunity

SECTION

Section on Bacterial Disease Pathogenesis and Immunity

INSTITUTE AND LOCATION

IICHD. NIH. Bethesda. Maryland. 20892-2720

TOTAL STAFF YEARS:

1.9

PROFESSIONAL:

1.4

OTHER:

0.5

CHECK APPROPRIATE BOX(ES)

D (a) Human subjects D (b) Human tissues E] (c) Neither
D (a1 ) Minors
D (a2) Interviews

SUMMARY OF WORK (Use standard unreduced type. Do not exceed the space provided.)

Mvcobacteri um tubercul osi s . the causative organism of tuberculosis is responsible
for approximately three million death cases annually and the mortality rate of
infections caused by multiple drug-resistant strains of this bacterium exceed
80^. M. tuberculosis is coated by a number of polysaccharides of unusual
structure. A unique component termed polysaccharide II is composed of alpha.
l,2-1inked D-glucose residues. Schemes were designed to isolate this
polysaccharide from polysaccharide-containing extracts of M. tuberculosis. We
demonstrated that the major monosaccharide components in soluble polysaccharide
fractions from Mycobacteria are arabinose. mannose and galactose. A
polysaccharide fraction was also identified which contains mannose and glucose
in approximately equal amounts. We demonstrated the presence of a glycogen-like
polysaccharide as one of the extracellular polysaccharides of Mycobacteria. We
developed a method for the chemical synthesis of oligosaccharides corresponding
to the putative polysaccharide II. We showed that for such compounds the
stepwise synthesis Is the method of choice.

PHS 6040 IRev. 5/921

IBMBI

ugammammm

ZOl HD 01314-01 LDMI

Others; J.B. Robbins Chief LDMI, NICHD

R. Schneerson Research Medical Officer LDMI. NICHD

E. Dubois Visiting Fellow LDMI. NICHD

C.W. Broome Chemist LDMI. NICHD

sanaie^a^saam^^^Ka^iaiBtia^

ZOl HD 01314-01 LDMI

Pro.ject Description:

Tuberculosis caused by Mycobacterium tuberculosis continues to be a major cause
of death worldwide and the fatility rate of this disease when caused by multiple
antibiotic-resistant strains in immunocompromised patients is higher than 80%.
There is increasing evidence that antibodies against mycobacterial surface
antigens may play a role in protection.

The major component of the extracellular fraction of M. tuberculosis termed
polysaccharide II by Siebert was shown to be a linear of al. 2-1 inked D-glucose
residues. Experiments were designed to isolate and identify this unique
polysaccharide for studies of its possible role as a protective antigen. As an
alternative approach to this polysaccharide, chemical synthetic approaches were
also planned.

Ob.iectives:

To identify glucose-containing polysaccharide(s) among the carbohydrate fractions
of the cell -surface polysaccharides of M. tuberculosis. To synthesize
oligosaccharides containing Q!l,2-linked D-glucose residues, corresponding to
polysaccharide II and to covalently attach them to high molecular weight
carriers.

Methods Employed:

(a) Polysaccharide fractions of M. tuberculosis were analyzed by a combination
of chemical and instrumental methods. These included chemical conversion of the
saccharides into (partially methylated) alditol acetates which were analyzed by
gas chromatography combined with mass spectroscopy. The intact saccharide
fractions were also analyzed by high-resolution nuclear magnetic resonance
spectroscopy.

(b) The putative glucan antigen of M. tuberculosis termed Polysaccharide II is
a linear polysaccharide containing a- (1^2) -linked D-glucopyranose residues.
Retrosynthetic analysis of this polysaccharide indicated that a single building
block, equipped with a temporary, non-participating group at 0-2 may be
sufficient for its construction. We have chosen the glucosyl donor
3.4,6-tri -O-acetyl-2-O-benzyl -/9-D-glucopyranosyl bromide for this purpose which
was obtained stereoselectively from the corresponding, oj-phenylthio
glucopyranoside. The donor afforded a-linked glucosides upon solvolysis with
primary alcohols in a completely stereoselective manner. Subsequent, iterative
chain extension involving chemoselective liberation of the elongation site
followed by silver salt-catalyzed glucosidation with the donor/acceptor building
block afforded fully protected kojibiose to kojipentaose derivatives from which
the protecting groups were removed to provide di - to penta-saccharide analogs of
the elusive Polysaccharide II. Because of the limitations of the monosaccharide
donor-based chain extension we also investigated several donor/acceptor
derivatives of the disaccharide kojibiose. The rationale behind this approach
is that if successful, the chain is increased by two residues after each
glycosidation. Reaction of these donors with partially protected glucosides
having their HO-2 free afforded kojitriose derivatives in high stereoselectivity.
On the other hand, attempted coupling of the kojibiose donors with kojibiose or

ws^m^^^^Hmm^^^a^ai^^a^^^mssaiusmaBaa^a^^^^^^mi^maim

ZOl HD 01314-01 LDMI

kojitriose derivatives having a free HO-2 at their non-reducing end termini
failed. Major products in these reactions included cyclo-kojibiose and
trehalose-type tetrasaccharides.

Ma.ior Findings:

(a) We demonstrated that the prevalent monosaccharide components in
polysaccharide fractions from Mycobacteria are arabinose, galactose and mannose.
We identified a polysaccharide fraction which contains an mannose and glucose in
approximately equal amounts together with a smaller proportion of arabinose. We
also demonstrated the presence of a glycogen-like polysaccharide as one of the
extracellular polysaccharides of Mycobacteria.

(b) We developed a method for the chemical synthesis of koji -oligosaccharides.
We showed that the stepwise synthesis is the method of choice for the synthesis
of such compounds. The failure to use [2+n (n>l)] block-synthetic approaches
indicate extensive clashes between the acceptor and the donor. The
immunochemical properties of the synthetic kojipentaose have been studied as
described in Dr. John B. Robbins' report (ZOl HD 01311-02).

Proposed Course:

(a) The structure of the glucose-enriched fraction of M. tuberculosis will be
studied using specific degradation instrumental analytical methods.

(b) Higher-order oligosaccharides corresponding to Polysaccharide II will be
synthesized in a form that allows their covalent attachment to macromolecules.
The conformation of the synthetic koji -oligosaccharides will be studied. These
studies may reveal a conformational determinant of Polysaccharide II and may
provide a better understanding of the molecular basis of the unique properties
of this polysaccharide.

Publications:

Pozsgay V, Robbins JB. Synthesis of a pentasaccharide fragment of polysaccharide
II of Mycobacterium tuberculosis. Carbohydr Res (in press).

-™-°''°°~"

DEPARTMENT OF HEALTH AND HUMAN SERVICES ■ PUBUC HEALTH SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD 01315-01 LDMI

PERIOD COVERED

nrtnhpr 1 1994 to September 30. 1995

TITLE OF PROJECT 180 characters or less. Title must fit on ore line between the borders. I

Mnrliilatmn nf Protein and Cell Functions by Heparin/Heparan Sulfates

PRINCIPAL INVESTIGATOR lUst other professional personnel below the Principal Investigator.! IName, title, laboratory, and irstitute affiliatlor)

PI:

A.L. Stone

Chemist

LDMI, NICHD

COOPERATING UNITS lif ary)

LCM NIMH (G E. Deibler): Harvard Medical School. Cambridge. MA (R.P. Junghans)
LDDRD. NCI (J.B. McMahon): BEIP. NCRR (M.S. Lewis)

LAB/BRANCH

ahoratory nf Developmental and Molecular Immunity

SECTION

Sertion nn Ractenal Disease Pathogenesis and Immunity

INSTITUTE AND LOCATION

ITCHD NIH Rpthesda. Maryland. 20892-2720

TOTAL STAFF YEARS:

PROFESSIONAL:

1.0

OTHER:

0.87

CHECK APPROPRIATE BOX(ES)

D (a) Human subjects H (b) Human tissues D (c) Neither
n (a1 ) Minors
D (a2) Interviews

SUMMARY OF WORK lUse stardard unreduced type. Do not exceed the space provided.!

Heparin\heparan sulfate and heparin-mimetic sulfated oligosaccharides (S-oligoS)
were studied in vitro to elucidate the structural basis of their specific
modulations of protein and cell membrane functions, using physi -cal . biochemical,
and biological methods. In one such study, strategy was based on our view that
heparin-like molecules play a role in the molecular mechanisms underlying the
infectivity of human immunodeficiency virus (HIV-l): If the known inhibition of
the cytotoxic (CT) and syncy-tium-formimng (SF) infectiousness of HIV-l by
S-oligoS were governed by a structural specificity, then rationale for clinical
usefulness of a putative S-oligoS agent against AIDS would be enhanced.

Here first, structural specificity in the inhibition of HIV-l by S-oligoS was
demonstrated, using an HIV-inhibitory anticoagulant pharmaceu-tical which
comprised a mixture of S-oligoS components (Cps) from sulfated xylan. Highly
active (HA) Cp were then purified and this specificity was separable from
anticoagulantly active molecules.

Now 1) The minimum-sized HA-Cp. CpF, (EC50 vs SF and CT= 100 and 200 ng/ml )
was purified. Molecular mass of Cps was determined by S-E ultra-centrifugation
of monodansyl derivatives. Minimum mass of anti -HIV-l S-oligoS was -4500 and a
smaller S-oligoS. HA vs CT but not SF. had mass of -nonomer. 2) Cp contain S-a
1.2 D-GlcA monomers on the S-xylose chain. Molecular modeling revealed that such
GlcA could impart local structures resembling those in heparin. Analysis for
GlcA by o-nitrophenylhydrazide reaction revealed unexpectedly that Cps had
multiple GlcA, supporting a view that their heparin-mimetic action redounds from
GlcA-containing structures. 3) Anti -HIV-l activity appears to reside in ~ a
tetradecaglu-curonoxyloside containing 2 or 3 GlcA. 4) Methods upscaling
preparation of CpF were developed and ^re being modified for clinical
application.

PHS 6040 (Rev. 5/92)

ffisa

ZOl HD 01315-01 LDMI

Pro.iect Description:

This project studies the structure/function relations (st/fn) of heparins and
heparan sulfates (H/HS) (glycosamino polysaccharides of cells and the
extracellular matrix) to elucidate the molecular basis of their capacity to bind
to and modulate the biological activity of proteins and cell membranes (e.g. , in
development, blood coagulation, cytokine regulation, tumor growth, lymphocyte
homing, and viral infectivity) [ZOl MH 013605-12 LNB]. The methodological
approaches in these studies are physical . biochemical and biological . Except for
Rosenberg's elucidation of the structural specificity of H/HS in anticoagulation,
st/fn of H/HS remain undetermined. We provided a model of structural elements
in H that may be applicable to the diverse specificities of H/HS and heparin-
mimetic oligosaccharides (S-oligoS). Current studies include H/HS and S-oligoS
in vitro inhibition of HIV-l. as ligands for CD4 lymphocytes, and conformational
modulation of human basic fibroblast growth factor (bFGF). Understanding such
structural specificity would enhance the development of new H/HS-mimetic
pharmceuticals as well as generate understanding of the basic structural modes
of action of the H/HS class of biopolymers. Other collaborative projects study
structural aspects of the biology of the soluble T-activation antigen chain of
IL2R and myelin basic protein.

Objectives:

1. AL Stone with JB McMahon, FCRDC. and MC Lewis, NCRR.

To study specificity in st/fn of heparin-mimetic S-oligoxylans, S-oligoS
components (Cps)of SP54 were purified by our analytical -scale methods (LPLC
through 195 cm, i.d. 0.6 cm, BioGel PIO in 0.5 M NH4HCO3, using quantitative
spectroscopy for detection and quantitation. Cps were bioassayed for activity
vs cytoxicity (anti-CT) (soluble formazan method) and vs the syncytium- forming
infectivity of HIV-l (anti-SF) (by quantitation of fusion between infected and
uninfected CD4 cells) [ZOl MH 02593 02-3 LBG] . We found that antiviral
capacities of the S-oligoS were governed by a structural specificity (evidenced
by their differential and bi phasic potencies). This enhanced a rationale for
potential usefulness of an S-oligoS against AIDS and offered an explanation for
previous clinical failures using mixtures. Molecular modeling suggested that the
heparin-mimetic capacity of such S-oligoxylans might obtain by virtue of
monomeric a-linked D-glucuronic acid branches which occur randomly on the average
of one in 10 xyloses in the native xylan chain. Three types of anti-AIDS virus
Cps were identified among 19 which ranged from -30,000 to <2000 in apparent
molecular mass: 1) CpFC, relatively high In molecular mass, ~ an eicoS, highly
active (HA) against CT and SF (EC5o=50-100 ng/ml) and anticoagulantly active in
vitro (up to 25X the potency of heparin); 2) CpF, a HA -tetradecS having EC5o=250
vs CT and 50 ng/ml vs SF, but low activity vs thrombin: and 3) CpC, an -nonoS,
HA vs CT but not vs SF and anticoagulantly inactive [1994 project report].

Antiviral Cps were characterized by: 1) refinement of the analytical -scale LPLC
system to analytical use in identifying CpFC, CpF and CpC, based on elution
volume: 2) quantitation of maximum metachromatic reactivity of Cp-methylene blue
(MB) complexes (based on minimum extinction coefficient obtained at the a
absorption band of MB {E^^^ J , which yielded characteristic E^,^ _, among Cp: 3)

- ■»«--._»— ~~-^i-~-.

ZOl HD 01315-01 LDMI

establishment of a quantitative colorimetric analysis of GlcA content for these
sulfated oligoS through application of the o-nitrophenyl hydrazlde method of
determination of activated uronic acids [GlcA standard curve was linear in the
ug range: sample response was linear in the range 5 to 40 ug GlcA]: 4) determina-
tion of molecular mass by sedimentation-equilibrium analysis of uv-absorbing,
monodansylated S-oligoS using absorption optics, our calculated n (0.475) and an
equation we derived to correct the mass^bserved by an "effective charge" factor: 5)
quantitation of antiviral potency as above: and 6) measurement of APTT by an
automated clinical assay and evaluated relative to heparin.

Characteristics of CpF showed that this Cp contained the minimum-sized HA struc-
ture vs HIV-1. Moreover, the structure was ~S-tetradecaS, containing 2 or 3
GlcA. These data showed that the heparin-mimetic activity of sulfated xylan was
due to glucuronoxyloside degradation products rather than associated with a
"sulfated pentosan". Since CpF might yield an agent suitable for testing against
the AIDS virus In vivo, we next developed up-scaled preparations of CpF.

a) We reasoned that CpFC region through 190 cm x 2.3 cm BioGel P6 in 0.5 M
NH4HCO3 would be separated near the void peak, while the CpF region would
be better fractionated. Using reference Cp as chromatographic markers and
lambda -carrageenan to determine the void volume, three 150-200-fold
scaled-up preparations yielded the CPFC. CpF, and CpC regions. Yield of
CpF was -10 percent. This product was further purified. Under modified
good manufacturing practices, the CpF (-600 mg) was further purified by
the modified methods (through 176 cm x 2.3 cm), yielding three fractions
of which aliquots were prepared for anti-CT and ant1-SF bioassays and vs
HIV-1 in the hollow fiber assay in mice.

b) A 25-fold, direct upscale of a modified analytical system is in progess:
and

c) Multiple 40-fold upscaled preparations using the modified P6 system are
underway as a means towards obtaining sufficient CpF for potential
clinical application. CpF regions have been identified for ongoing
combinations and procedures.

The findings provide a basis for: 1) potential development of an inexpensive
adjunct to AIDS therapy (CpF): 2) study of molecular mechanisms involving S-
oligoS in the effects of HIV-1 that kill immune cells: and 3) improvement of
clinical usefulness of S-oligoS in anticoagulation therapy, in addition to
elucidating st/fn of H/HS and S-oligoS. Studies will be continued. A manuscript
has been submitted. A manuscript, "Differential inhibitory potency of sulfated
xylan oligosaccharides in the inhibition of HIV-1 cytotoxicity and syncyti urn-
forming infectivity in vitro" , by AL Stone, DJ Melton, and JB Mc Mahon, is in
progress.

2) AL Stone, RP Junghans collaboration with MS Lewis.

To determine the molecular form of the soluble T activation antigen (Tac) chain
of the interleukin 2 receptor(IL2R) , a purified recombinant Tac was examined by
size exclusion chromatography, velocity-sedimentation and sedimenation-

ZOl HD 01315-01 LDMI

equilibrium ultracentrifugation analysis and low ultraviolet circular dichroism
spectroscopy. High affinity binding of the cytokine. IL-2, to IL2R during
lymphocyte activation requires cooperation among three known receptor polypeptide
chains. One of these. Tac. is absent in resting lymphocytes and is strongly
upregulated upon cell activation to generate the high affinity IL2R. After the
acivation. Tac is released from the membrane in a soluble form. There was
controversy about whether this highly regulated chain was dimeric or monomeric
In the native form, which clouded all proposed molecular mechanisms of action.
Seiving HPLC indicated oligomeric. mixed or dimeric structure, while
immunological and electrophoretic evidence indicated a monomeric form. We now
show that HPLC chromatography led to artifacts probably induced by solute-column
interactions and that the Lewis ultracentrigfugal methods are critically
superior. Spectroscopy showed that variation in forms due to raising or lowering
the solvent salt concentration were not seen absent chromatographic matrix.
Moreover, molecular mass determined by sedimentation ultracentrifugal analysis
was that of a monomeric form. Studies of renal filtration and a one-epitope
stoichiometry of antibody-antigen titrations indicated that Tac was monomer.
These findings resolve the contoversy (above) and point out that the biologically
relevant form is a momomeric protein. Knowledge of the molecular form of soluble
Tac is essential to understanding the regulation and process of lymphocyte
activation in vivo. A manuscript is in preparation.

3) AL Stone with MR Ranson. Christie Hospital. Manchester, UK.

To study the molecular basis of the modulation of bFGF function by H/HS (low
affinity bFGF receptor) and suramin (S) (an anti -cancer, growth factor
antagonist), the structure of human recombinant bFGF was examined by low uv
circular dichroism (CD) and by static fluorescence (F) and anisotropy
spectroscopy in the absence and presence of the modulators (work completed
previously [ZOl MM 02593-02-3 LBG]), In brief, we showed that S was bound within
0.9nm of bFGF cell -receptor site W. H (from CD analysis) also altered the
peptide conformations around the W in the cell -receptor and concomitantly led to
purturbation of the W and oligomerlzation of bFGF (by anisotropy of F analysis).
Findings were important for understanding molecular events that subserve
modulation of bFGF function by H/HS: they support our idea that heparin modulates
protein functions by exerting a "conformational catalysis" which enables a
specific transition to a new stable state. A manuscript this year was revised
for resubmission.

4) AL Stone, GE Deibler collaboration, LCM. NIMH.

Myelin basic protein (MBP)is the encephalitogenic antigen in the experimental
autoimmune model for study of multiple sclerosis. Brain myelin contains several
isoforms, in particular a phosphorylated form which appears to be coupled to a
kinase and phosphatase in the membrane and to have a rapid rate of phosphate
turnover. The role(s) of MBP isoforms in myelin is unclear. We proposed that
MBP has a conformational adaptability which might be subserved by the isoforms.
To study this model in human MBP, threonine-phosphorylated 18.5 kDa isoform
(PT96) and nonphosphorylated isoforms were purified by Dei bier's new method of
selective thrombi c digestion and structural properties of isoforms were compared
by CD spectroscopy [1994 project report]. Results showed that ijn vivo
phosphorylation to PT96 produced an increase in stable folding of 18.5 kDa hMBP

'.' ^

■"-—"'-■"■

iiiiiMiiiiiniiiiiiiiiiiiiini|iiiiiiii|iiii

ZOl HD 01315-01 LDMI

up to -30^ (similar to that predicted from its amino acid sequence), which was
likely transition to a class B type II beta turn and beta-structure. These
findings supported our suggestion that conformational adaptability associated
with isoforms of hMBP may subserve a structural integrity of myelin.

Significance to Biomedical Research and the Program of the Institute:

Discovery of biologically active components among the polymeric sulfated
saccharides has potential for providing probes for cell -surface HIV-l
interactions as well as for other human proteins such as basic myelin protein.
Clinically acceptable fragments of sulfated inulin are being evaluated for their
activity in HIV-l infected CD4 human lymphocytes.

Publications:

Stone AL, McMahon JB, Melton DJ, Lewis MS. Sulfated
glucuronoxylyloligosaccharides as potent inhibitors of the cytotoxocity and
syncytium-forming infectivity of the human immunodeficiency virus (HIV-l) in
vitro. Glycoconjug J 1995:12:586.

Dei bier GE. Burl in TV, Stone AL. Three isoforms of human myelin basic protein:
Purification and structure. J Neurosci Res (in press). . ,>

«ll

lemiaa^mmssB^^msiiaaBa^^^mBim^s^^

IDfJ-F

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBLIC HEALTH SERVICE
NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

Z01 HD 00047-26 LDN

PERIOD COVERED

October 1, 1994 to September 30, 1995

TITLE OF PROJECT (80 ctiafBCter^ or less Title must tit on one line between the borders.)

Biochemical Studies of Neurons and Other Cell Types

PRINCIPAL INVESTIGATOR (Ust other professional personnel below the Principal Investigator.) (r^me. title, laboratory, and institute attiliatkyi)

PI: D. Brenneman

Others: J. HiU

G. Glazner

G. Gibney

I. Gozes,R. Alvero

T. Dibbem, J. Wu

D. Warren

Head
Expert
IRTA Fellow
IRTA FeUow
Guest Researcher
Guest Researcher
Bio. Lab. Tech.

LDN,NICHD
LDN,NICHD
LDN,NICHD
LDN,NICHD
LDN,NICHD
LDN,NICHD
LDN,NICHD

COOPERATING UNITS (it any)

Dept. CUn. Biochem. Tel Aviv Univ., Israel (1. Gozes); Dept. Organic Chem., Weizmann Inst. (M.
Fridkin); Dept. Neurol., Univ. Kansas (B. Festoff); Dept. of Immunol. George Washington Univ.
(T. Phillips); Univ. Queensland, Australia (Alice Cavanaugh)

LAB/BRANCH

Laboratory of Developmental Neurobiology

SECTION

SpTtinn nn Ppvplnprnental and Molecular Pharmacology

INSTrrUTE AND LOCATION

NICHD. NIH, Bethesda, Mandand 20892

TOTAL MAN-YEARS: PROFESSIONAL:

8.1

5.3

OTHER

2.8

CHECK APPROPRIATE BOX(ES)

D (a) Human subjects
D (a1) Minors
D (a2) Interviews

n (b) Human tissues (^ (c) Neither

SUMMARY OF WORK (Use standard unreduced type. Do not exceed tt)e space provided.)

Structure-activity studies of a peptide derived from Activity Dependent Neurotrophic Factor
(ADNF), a novel protein released from ghal cells by vasoactive intestinal peptide (VIP), revealed
that complete survival-promoting activity was retained with a nine amino acid peptide (ADNF-
9). A panel of thirty test peptides was utilized to establish the loci and identity of critical amino
acid residues. In dissociated cell cultures from the rat cerebral cortex, ADNF-9 was found to

exhibit neuroprotective actions to prevent neuronal cell death associated with excitotoxicity and
gpl20, the envelope protein from the human immunodeficiency virus. Further structural studies
on peptides from protease digests of intact ADNF indicated close homology to a recognized
intracellular stress protein.

Radiolabeled gpl20 administered to pregnant rats was recovered in the brains of fetuses. Both
intact gpl20 and its neurotoxic fragments were found to preferentially accumulate in fetal brain
two days after injection. Treatment of whole cultured mouse embryos with gpl20 produced a
dose-dependent inhibition of growth as measured by somite number and image analysis of
whole body size. The growth deficits produced by gpl20 were prevented by co-treatment with
either VIP or a VIP analogue, peptide T.

A novel VIP agonist (SNV) was synthesized and shown to be neuroprotective against the
neuronal cell killing action of beta amyloid peptide, a toxic substance associated with
Alzheimer's disease. SNV was shown to preferentially interact with VIP binding sites that were
insensitive to GTP in the developing rodent brain and that were independent of cAMP action.
SNV, like VIP, produced significant increases in growth in early post-implantation embryos.
VIP-mediated increases in mitosis in embryos was associated with increased cyclin A expression
as measured by reverse transcriptase polymerase chain reaction.

PHS 6040 (Rev 1/84)

cpo e t 4-9t e

■MllHTtlfWltlll*MHllnwt»TWTnmn«ij»niiviB;i»

ZOl HD 00047-26 LDN

Project Description:
Objectives:

1) To investigate the structure and functions of activity dependent neurotrophic factor (ADNF),
a survival-promoting substance that is released by vasoactive intestinal peptide (VIP).

2) To study the relationship between VIP and glia-derived cytokines.

3) To extend the investigation of VIP, VIP analogues and ADNF neurotrophism to embryonic
development in utero.

4) To develop and characterize new agonists and antagonists to VIP receptors in the central
nervous system (CNS).

5) To study the neurotoxicity of the HIV envelope protein in developing animals and to devise
VIP-related protective agents against HlV-related viral products.

Methods Employed:

Dissociated cultures derived from the central nervous system of fetal rodent were used for
studies of VIP, cytokines and growth factors that increase neuronal survival. In particular, the
development of cerebral cortical cultures from term rats was emphasized. This preparation
allowed for a system homologous to that used for the isolation of ADNF. Computerized image
analysis, calcein uptake and ouabain binding were used to assess neuronal populations in
culture. For specific cellular phenotype assessment, immunocytochemical methods for neuron
specific enolase, glial fibrillary acid protein, methionine-enkephalin and VIP were performed.
Quantitative Northern blot hybridizations and RT-PCR were conducted for the analysis of
neuropeptide gene expression. The foUowdng biochemical techniques were used:
radioimmunoassay for peptides and cyclic nucleotides, capillary electrophoresis combined with
limiinescence-based ELISA assays for cytokines, one- and two-dimensional polyacrylamide gel
electrophoresis, assays for choline acetyltransferase and glutamic acid decarboxylase and
radioligand binding. Standard chromatographic (anion exchange, sizing, and reverse phase)
separations were conducted to purify ADNF and to measure VIP degradation products.
Autoradiography of VIP receptor binding was conducted on frozen thin sections of brain as was
in situ hybridization. Whole embryo cultures and embryos obtained in utero were used to study
the growth promoting actions of VIP and related drugs. E9.5 mouse embryos were incubated
for up to 24 hours in the presence of fresh rat servmi. Embryos were evaluated by embryonic
volume, protein and DNA content, receptor distribution and by mitotic assay of
bromodeoxyuridine incorporation into nuclei as detected by immunocytochemistry. For analysis
of dendritic arbor, Golgi impregnation was utilized with quantitative image analyses.

Major Findings:

1. Structure-activity studies of thirty peptides indicated that the survival-promoting action of
ADNF (Activity Dependent Neurotrophic Factor) could be retained in a nine amino acid

HMaHH,-{3ca«:

ZOl HD 00047-26 LDN

peptide (ADNF-9). This finding of a peptide that mimics an intact growth factor is
unprecedented.

2. In cerebral cortical cultures, ADNF-9 was found to be neuroprotective against the neuronal
cell death produced by beta amyloid peptide, a substance associated with neural degeneration
in Alzheimer's disease.

3. Neuronal cell death produced by the envelope protein of the human immunodeficiency
virus was prevented by treatment with ADNF-9 at femtomolar concentrations in cerebral cortical
cultures.

4. The amino acid sequence of five additional peptides derived from protease digests of
ADNF were determined. Four of the five peptides had sequence homology to a recognized
intracellular stress protein.

5. Radiolabeled and biologically active gpl20 was recovered in the brains of mouse fetuses
after systemic administration of gpl20 to the maternal circulation. Toxic fragments of gpl20
were also found in fetal brain as assessed by cerebral cortical cultures treated with brain
homogenates.

6. Inhibition of somite growth was observed in whole mouse embryo cultures treated with
gpl20. The growth deficits produced by gpl20 were prevented by VIP or the VIP analogue,
peptide T.

7. A hpophUic VIP analogue (stearyl, norleucinei7VIP, SNV), discovered and synthesized by
our Israeli collaborators (Drs. Gozes and Fridkin) was found to increase somite growth of early
post-implantation embryos.

8. In vitro autoradiographic studies revealed that SNV displaced VIP binding from those sites
that were insensitive to GTP and not linked to cAMP accumulation.

9. SNV provided neuroprotection from neuronal cell death produced by beta amyloid peptide
in dissociated cerebral cortical cultvires.

10. PACAP38, a peptide that shares 50% homology to VIP, was shown to increase the release
of cytokines from astroglia, albeit 60-807o less than that observed with VIP. Peptide T,
cytoprotective VIP analogue currently in phase two clinical trials for the treatment of
NeuroAIDS, also increased cytokine secretion, but was 3-5 fold less efficacious than VIP.

11. Stimulation of microglia with 0.1 nM VIP produced no detectable release of the eight
cytokines observed wdth astrocytes. However, micromolar amounts of VIP did produce
significant release of interleukin-1 beta.

Significance to Biomedical Research and the Program of the Institute.

Investigations of the neurotrophic properties of VIP have now been extended to drug
development. The fundamental concept which provides the rationale for this extension is that

mmaaa

ZOl HD 00047-26 LDN

molecules important to regtilation of neuronal survival during development may also be
important in the regulation of repair mechanisms after nervous system injury. Having
established that VIP exhibits potent neuronal-survival promoting activity in cell culture, the
demonstration that VIP and VIP analogues are neuroprotective against agents associated with
human neuro-degenerative diseases sustains the development/injury association. The
development of lead compounds for the treatment of Alzheimer's disease is certainly suggested
by our results which indicate that our lipophilic VIP analogue is effective in preventing cell
death associated with the beta amyloid peptide. Similarly, our studies with the protective
properties of VIP on gpl20-induced neurotoxicity and growth retardation are strongly suggestive
as useful strategies to treat the problems associated with growth retardation and neurological
impairment associated with infection with the human immunodeficiency virus during pregnancy.
With the discovery that a nine amino acid peptide that exhibits neuroprotection at femtomolar
concentrations, the therapeutic potential of ADNF is greatly increased. One of the major
dilemmas in the use of growth factors in the treatment of neurodegenerative diseases lies in the
challenges of getting these proteins into the brain and protecting against potential autoimmune
effects. With the discovery that ADNF can be mimicked by a nine amino acid peptide, both CNS
entry and drug development may be realized for the treatment of the neurological symptoms
associated with HIV infection, Alzheimer's disease and potentially other prevalent
neurodegenerative diseases.

Proposed Course:

Based on the discovery of ADNF-9, peptidemimetics will be developed and tested for
neuroprotection. Investigations of ADNF-9 and related analogues will also be conducted for
their ability to cross the blood-brain barrier.

Antisera against ADNF will be tested for effects on the development of mammalian embryos.
If the anti-serum produces growth retardation, detailed neuroanatomical and biochemical studies
will be conducted. Investigations of developmental milestones and learning behavior will also
be tested in neonates from mothers treated with anti-ADNF.

In collaboration with our colleagues in Israel, the molecular cloning of ADNF will continue to
be pursued. Both our lab and Dr. Gozes's group will employ separate tactics to obtain the
cDNA of ADNF. Expression cloning methods will be undertaken with our newly-acquired anti-
ADNF-15; in addition, strategies involving PCR cloning will be conducted. More amino acid
sequence of ADNF digests may also be required.

The relationships between VIP, ADNF and chaparonin-10 will also be explored during mid-
gestational pregnancy. Collaborations with Alice Cavanaugh (Univ. of Queensland, Australia)
VfiW focus on the role of early pregnancy factor and its potential role in regulating embryonic
development.

Effort will intensify on developing a biochemical/immunological assay for ADNF. With the
production of large amounts of rabbit anti-ADNF and the procurement of capillary
electrophoresis equipment, this goal should see progress.

The lipophilic analogues of VIP will be used to further investigate the relationship between VIP
and cytokines release from astrocytes.

mimmMmmmmm

ZOl HD 00047-26 LDN

Protocols:

Animal

93-025

93-026

94-021
95-003

Brenneman The role of vasoactive intestinal peptide and related

peptides in neurobehavior development and the
morphology and neurochemistry of the developing
nervous system

Brenneman The role of vasoactive intestinal peptide and related
peptides on behavior and the morphology and
neurochemistry of the brain

Brenneman Glial-derived trophic factors

Brenneman Influence of VIP and antagonists on developing
neuroepitheUum in mouse embryos

Publications:

Arimura A, Somogyyvari-Vigh A, Weill C, Fiore C, Tatsuno 1, Bay V, Brenneman DE. PACAP
functions as a neurotrophic factor. NY Acad Sci 1994;739:228-43.

Brenneman DE. Neurotrophism, mitogenesis and nerve regeneration as models for neuropeptide
mediation. Ann NY Acad Sci 1994;739:226-7.

Brenneman DE, Hill JM, Glazner G, Gozes 1, PhilUps TM. lnterleukin-1 alpha and vasoactive
intestinal peptide: enigmatic regulation of neuronal survival. InU J Dev Neurosci. 1995;13:187-
200.

Brenneman DE, McCune S, Mervis RF, Hill JM. Gpl20 as an etiologic agent for neuroAlDS:
neurotoxicity and model systems. Adv. Neuroimmunol. 1994;4:157-65.

Gozes I, Brenneman DE, LilUng G, Davidson A, Moody TW. Neuropeptide regvilation of mitosis.
NY Acad Sci 1994;739:253-61.

Gozes 1, Fridkin M, Brenneman DE. Stearyl-Nle-VIP: a non-invasive impotence drug and a
potent agent of neuroprotection. Drugs of the Future. In press.

Gozes 1, Fridkin M, Brenneman DE. A VIP hybrid antagonist: from neurotrophism to clinical
applications. Cell Mol Neurobiol. In press.

Gozes 1, Fridkin M, Westphal H, Glowa ], Hill JM, Reshef A, Zhukovsky S, Waner T, Niska A,
Rubinrout S, Lilling G, Davidson A, Glazer-Steiner R, Moody TW, Rostene W, Brenneman DE.
Neuronal VIP: from gene to sexual behavior, memory and clinical applications. In: Proceedings

ffllBBilWWWW)WVWimMHWtffmtmWittMlttHI<f^^

ZOl HD 00047-26 LDN

Intl Symposium on VIP, PACAP and Related Regulatory Peptides, Rosselin G, ed., Singapore:
World Scientific Press, pp. 314-24, 1994.

Gozes I, Lilling G, Glazer R, Ticher A, Ashkenazi E, Davidson A, Rubinraut R, Fridkin M,
Brenneman DE. Superactive lipophilic peptides discriminate multiple vasoactive intestinal
peptide receptors. J Pharmacol Exp Therap 1995;273:161-7.

Gressens P, Hill JM, Paindaveine B, Gozes I, Fridkin J, Brenneman DE. Severe microcephaly
induced by blockade of vasoactive intestinal peptide function in the primitive neuroepithelium
of the mouse. J Clin Invest 1994;94:2020-27.

HiU JM, Mervis RF, PoliH J, McCune SK, Gozes I, Fridkin M, Brenneman DE. Blockade of VIP
during neonatal development induces neuronal damage and increases VIP and VIP receptors in
brain. NY Acad Sci 1994;739:211-25.

Lilling G, Wollman Y, Goldstein MN, Rubinraut S, Fridkin M, Brenneman DE, Gozes I. Inhibition
of human neuroblastoma growth by a specific VIP antagonist. J Mol Neurosci. In press.

McCune SK, Hill JM. Ontogenic expression of two alpha-1 adrenergic receptor subtypes in the
rat brain. J Mol Neurosci. In press.

Olah Z, Lehel C, Anderson WB, Brenneman DE, Agoston DV. Subnanomolar concentration of
VIP induces the nuclear translocation of protein kinase C in neonatal rat cortical astrocytes. J
Neurosci Res 1994;39:355-63.

Sone M, Smith M, Gozes I, Brenneman DE, Fridkin M, Ghatei MA, Bloom SR. Pituitary
adenylate cyclase-activating polypeptide (PACAP)/vasoactive intestinal peptide (VIP) receptor
subtypes in rat tissues: investigation of receptor binding and molecular identification by chemical
cross linking. Biomed Res 1994;15:145-53.

1-.;^ »\:..

mmmammmmms^

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBLIC HEALTH SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD 00056-20 LDN

PERIOD COVERED

_i223-

TITLE OF PROJECT (80 characters or less. Title must fit on one line t)elween the txxders)

RinHynthpai Hr prorPBFsing & aprrphinn of npiirnpppi-. i dPH K pi 1-ii i 1-.ary pppt- i rtp hnrmnnpg

PRINCIPAL INVESTIGATOR (List other professional personnel betow the Pnncipallnvesligator ) (Name, title, laboratory, and institute afTiliation}

PI: Y. Peng Loh

Others: Le-Ping Pu

Theodore Friedman
AnaMaria Bamberger
David Cool
Niamh Cawley
Diane Louie

Section Chief LDN, NICHD

Visiting Fellow LDN, NICHD ChunFa Zhang Courtesy Associate LDN, NICHD

Med Staff Fellow LDN, NICHD Vicki Olsen Guest Researcher LDN, NICHD

Visiting Fellow LDN, NICHD

IRTA LDN, NICHD

Pre-IRTA LDN, NICHD

COOPERATING UNITS (If any)

ERRB(H-C Chen); U of Auckland, New Zealand (N. Birch); Brown U (E. Nillni, I.
Jackson) ;Sandoz Inst. Med Res. London (C. Snell);NCI (A. Bird) ;NIDDK (L. Parnell)

UB/BRANCH

T,ahnra1-nry nf DpvpI npment.al Npiirnhinl

iia¥_

SECTION

Sprt: 1 on on Tplliilar Npiirnhi n1 ngy

INSTITUTE AND LOCATION

NTCHn,. NTH, Ret-hesda,

m ZQflaZ.

TOTAL STAFF YEARS:

fi-Q

PROFESSIONAL:

OTHER:

CHECK APPROPRIATE BOX(ES)

D (a) Human subjects
D (a1) Minors
n (a?) Intsrviftws

D (b) Human tissues E (c) Neither

SUMMARY OF WORK (Use standard unreduced type Do not exceed the space provided)

The sorting signal motif for targeting pro-opiomelanocortin (POMC, pro-ACTH/endorphin)
to the regulated secretory pathway has been identified as a 13 amino acid amphipathic
loop located at the N-terminus of POMC. An amphipathic loop sorting signal has also
been identified for pro-enkephalin, indicating the generality of these motifs as
sorting signals for the regulated secretory pathway. Recently, a sorting receptor has
been identified in bovine pituitary Golgi and secretory granule membranes. It is a -47
kD protein and is highly specific for the N-POMC sorting signal. The POMC sortinc
signal bound to this receptor optimally at pH 5.5-6.5, consistent with the pH of the
trans Golgi network where sorting begins. Our findings support a receptor-mediated
mechanism for sorting prohormones to the regulated secretory pathway. A novel class
of prohormone processing aspartic proteases was studied. One of these, yeast aspartic
protease 3 (YAP3), has been purified and shown to cleave prohormones at a pair of
basic residues; and a pair or mono-basic residue with an additional basic residue
upstream or downstream from the cleavage site. The Km and kcat of YAP3 for various
prohormone substrates showed that the catalytic efficiency was enhanced by increasing
the number of basic residues flanking the cleavage site. The specificity and ph
optimum of YAP3 were similar to the -70 kD mammalian aspartic protease, pro
opiomelanocortin converting enzyme (PCE). YAP3 antibody cross-reacted with PCE or
Western blots, suggesting that YAP3 and PCE are homologues. Western blot analysis
showed the presence of two YAP3-like proteins, -70 kD and -90 kD in mouse brain and
pituitary. Immunocytochemistry using anti-YAP3 in combination with In sitv
hybridization for neuropeptide cDNAs revealed colocalization of YAP3-like
immunoreactivity with CCK mRNA in hippocampal neurons and vasopressin mRNA ir
supraoptic nucleus neurons. Thus YAP3-related processing enzymes likely play a role
in pro-hormone processing in the CNS. YAP3-related cDNA clones from a mouse brair
library are currently being analyzed.

PHS 6040 (Rev. 5«2)

BaagKUWIWJUBIHiWWMBail^^

ZOl HD 00056-20 LD

Project Description

Objectives:

The ACTH-endorphin family of peptides play many physiological roles (e.g. in stress, fetal
development, osmoregulation and adrenal regeneration). These peptides are derived from a single
multivalent prohormone (pro-opiomelanocortin, POMC). This prohormone is proteolytically
processed at paired and tetrabasic residue sites to yield various active peptide products (a-MSH,
ACTH, p -endorphin, P-LPH, N-POMC,^,, Y3-MSH) which, in turn, are modified at the C- and/or
N-terminus. Differential processing occurs in different brain and pituitary regions. Proteolytic
processing of POMC occurs in specific cellular compartments, beginning in the trans-Golgi network
and continues within the secretory granule (vesicle). A highly specific set of proteases are involved
in the various cleavages of this prohormone. The prohormone is differentially processed in various
brain and pituitary regions, giving rise to a large diversity of secreted end-products with different
biological actions. Moreover, the processing also appears to be modifiable by physiological inputs
to the tissue. Our long-term goal is to understand the molecular mechanisms involved in the
intracellular trafficking and sorting of POMC to the secretory granule for processing and the
regulation of tissue- and stimulus-specific processing of this prohormone by different processing
enzymes. Such cellular mechanisms uncovered for POMC will likely be extendable to other
prohormones and pro-neuropeptides. Elucidating the molecular mechanisms which regulate the
intracellular trafficking, biosynthesis and secretion of the POMC-derived peptides will facilitate the
overall understanding of the regulation of the hypothalamic-pituitary-adrenal (HP A) axis in normal
and disease states. We have chosen yeast cells, the frog and mouse brain, anterior and intermediate
pituitary and bovine pituitary secretory vesicles as models and biological preparations, which are
especially usefial for addressing these issues. Our immediate major objectives are (i) to investigate
the signals and mechanisms involved in the intracellular targeting and sorting of POMC, into the
regulated secretory pathway of endocrine and neuronal cells; (ii) to identify, purify, characterize and
clone the enzymes involved in the processing of pro-opiomelanocortin to their respective hormones
and to determine how these enzymes regulate tissue-specific processing; (iii) to study the regulation
of biosynthesis (at the transcriptional, translational and post-translational level), turnover and
secretion of peptides that regulate the HP A axis such as ACTH, a-MSH, endorphin, vasopressin and
pro-TRH-derived pepfides, in the pituitary and CNS, during various types of pertubations (e.g. stress,
salt-loading) and clinical paradigms.

Methods Employed:

Acid-urea and SDS gel electrophoresis, high performance, and fast performance liquid
chromatography (HPLC, FPLC), radioimmunoassays (RIA), immunochemistry, enzyme assays,
column chromatography, antibody production, gene cloning techniques, in situ hybridization, electron
microscopy and tissue culture.

Z01 HD 00056-20 LDN

Major findings:

Identification of prohormone sorting signals for the regulated secretory pathway.

The search for molecular sorting signals to the intracellular regulated secretory pathway has been a
major challenge in cell biology. Dr. David Cool and Winnie Tarn have continued to investigate the
signal and mechanisms involved in the targeting and sorting of POMC to the regulated secretory
pathway. Previous studies used different POMC/CAT fusion genes containing different fi-agments
of the POMC cDNA fused to the CAT gene, transfected into AtT-20 cells, to define the molecular
domain of POMC that is sufficient and necessary for targeting the prohormone into secretory
granules. Those studies indicated that the targeting signal for the regulated pathway resides in the
first 26 amino acids of POMC. This region of POMC is highly conserved across species and contains
a unique structural motif of a hairpin loop stabilized by two disulfide bridges located at Cys2-Cys24,
and CySg-Cys2o.

To further define the sorting signal motif, two sets of mutations were made to the N-terminal region
of full-length POMC. In the first set, three deletion mutations were made; i) amino acids 2-26 were
deleted (A 2-26); ii) amino acids 8-20 were deleted to remove the loop (Cys8-Cys20 Del); iii) 78
amino acids from Lys24to Arg,o, were deleted (78 aa DEL). A second set of three mutations were
made to disrupt one or both disulfide bridges by substituting a serine residue for a cysteine residue
at positions 2, 8 or both 2 and 8 (CS2; CSS; CS2,8, respectively), to determine whether the disulfide
bridges were necessary for sorting, and if so, which bridge was more important. The constructs were
transfected into the mouse neuroblastoma cell line, Neuro-2a, which does not synthesize endogenous
POMC, but does have both a regulated and a constitutive secretory pathway. Wild type POMC and
the CS2 and 78 aaDEL mutants, when expressed in these cells, showed punctate ACTH
immunopositive secretory granules and stimulated secretion. However, when A 2-26, Cys8-Cys20
Del, CSS or CS2,S mutants were expressed in the Neuro-2a cells, the ACTH; was not found in
punctate granules but rather in the Golgi and perinuclear region. Only constitutive secretion of
unprocessed POMC was found for Neuro-2a cells expressing these mutants. This data showed that
a motif containing residues 8-20 and one disulfide bridge (CySg-CySjo) was not only sufficient but
necessary for the sorting of POMC to the regulated secretory pathway. In collaboration with Dr.
Christopher Snell (Sandoz Institute, London), this hairpin loop region of N-POMC (Cys, to CySjo)
was modeled and identified as an amphipathic loop with two lobes containing four amino acid
residues (Asp,o Leun Glui4 LeUjg) which are highly conserved across species. This, 13 amino acid
amphipathic loop structure is stabilized by the disulfide bridge, CySg-CySjo- This model has recently
been confirmed by NMR studies with the N-POMCi.jg peptide, done in collaboration with Dr.
Andrew Bird (NCI) and Dr. Lewis Pamell (NIDDK).

We proposed the hypothesis that the mechanism of sorting POMC to the regulated secretory pathway
involves the binding of the sorting signal to a membrane receptor at the trans Golgi network which
then pinches off to form a secretory granule. Dr. Cool assayed for a sorting receptor using iodinated
N-POMC ,.26 which contains the sorting signal as a ligand, and membranes from lysed Golgi and
secretory granules derived bovine intermediate lobe as a source of sorting receptors. He found pH-
dependant binding of N-POMCi.26 to granule membranes which was optimal between pH 5.5-6.5 and

Z01 HD 00056-20 LDN

was trypsin-sensitive. Unlabeled N-POMC1.26 inhibited the binding to secretory granule membranes
with an IC5o=65 |iM at pH 5.5. The Kd was 130 |iM and Bmax was 550 |imoles/mg protein. Ca^,
Na^, K*, Mg^"^ and EDTA had no effect on binding. Recombinant full length [^'S] POMC expressed
using the baculovirus/sf9 cell system by Dr. ChunFa Zhang also bound the secretory granule
membranes optimally at an acidic pH of 6.5 and the binding was inhibited by N-POMC,.26. In
contrast, recombinant [^'S] labeled A2-26-POMC lacking the sorting signal showed essentially no
binding to the same membranes. The specificity of the binding of '^'l N-POMCj.zg was fiirther
confirmed by the lack of inhibition by other parts of the POMC molecule such as ACTH and P-
endorphin. In addition ['"'I] N-POMC,.26 did not bind to intact secretory granules showing that the
receptor is localized only on the luminal and not to the cytoplasmic side of the granules. Finally,
cross-linking experiments revealed binding of a N-POMC1.26 to a -64 kD protein in lysed secretory
granule membranes which was inhibited by unlabeled N-POMC,.26. Thus all the data indicate the
existence of a specific sorting receptor for the N-POMC1.26 sorting signal, supporting a receptor-
mediated mechanism for sorting POMC to the regulated secretory pathway. Work is now in progress
to purify and clone the sorting receptor.

To determine the generality of an amphipathic loop motif as a sorting signal for the regulated
secretory pathway in other prohormones, Dr. Bamberger investigated the sorting of pro-enkephalin
(pro-ENK). Molecular modeling studies in collaboration with Dr. Snell revealed an amphipathic loop
motif in the N-terminal 3 1 amino acids of this prohormone. Expression of pro-ENKi.3,-CAT fusion
protein in Neuro2a cells showed that the pro-ENK,.,, motif is sufficient to act as a sorting signal to
target the CAT bacterial protein to the regulated secretory pathway. Thus an amphipathic loop
sorting signal motif for the regulated secretory pathway appears to be more general. Dr. Snell has
now modeled numerous prohormones using the POMC sorting signal as a template and found similar
putative amphipathic loop sorting signals in those molecules.

Characterization of the Yeast Aspartic Protease 3 Proprotein Processing Enzyme

Niamh Cawley has continued to study prohormone converting enzymes focusing on the novel aspartic
protease family which has specificity for mono- and paired basic residues. He purified and
characterized the only member of this family that has been cloned, the yeast aspartic protease 3
(YAP3p) which processes pro-a-mating factor in KEX-2 deficient mutants. YAP3p is a membrane-
associated enzyme which is attached to the membrane via a glycosyl phosphatidyl inositol (GPI)
anchoring site at the C-terminus of the protein. Truncation of the YAP3 gene to remove the GPI
binding site resulted in secretion of YAP3p. YAP3p has a pH optimum and an isoelectric point of
4.0-4.5. Analysis of the temperature stability of YAP3p showed that the enzyme retained 100% of
its activity at 37°C, while at 50°C there was 80% loss of activity after Ih. The dependence of activity
on temperature demonstrated a calculated Qio of 1 .95. Secreted, purified YAP3p was found in two
forms of -180 and -90 kD due to differential glycosylation. Removal of the sugars resulted in one
apparent band of -65 kD on SDS gels. However, upon N-terminal amino acid sequencing (in
collaboration with Dr. H-C Chen), two forms of YAP3 were found. One sequence had an N-terminal
beginning with Ala^ and another with Asp'''^ These two sequences led to the discovery that YAP3p
is synthesized as a pro-enzyme and is cleaved on the carboxyl side of the Lys**-Arg*^ pair to remove
the pro-region. Subsequently this active form is cleaved between Asn'^^-Asp"" to yield two subunits

HMMUIHWUH

Z01 HD 00056-20 LDN

which presumably associate to form a heterodimer, similar to the Cathepsin D aspartic protease. Dr.
Zhang has expressed full length YAP3p and a form with the pro-region deleted using the baculovirus
expression system. Full length YAP3p containing the proregion was inactive but upon incubating the
enzyme at pH 4.0, the pro-region was autocatalytically cleaved to yield an active enzyme. However,
the expressed YAP3p enzyme lacking the pro-region was not active at pH 4.0. This indicates that
the pro-region is essential for the proper folding of the protein so as to yield an active enzyme. The
specificity of YAP3p for 15 different potential paired/monobasic cleavage sites were examined in
various substrates, including ACTH, anglerfish pro-somatostatin I and II, pro-insulin, dynorphin,
amidorphin and cholescytokinin 33 (CCK 33). YAP3p recognized three specific motifs. One of these
is the classical paired basic residue motif The other two are either a paired or a mono-basic residue
site with an additional basic residue upstream at P4-P6, or downstream at P4'-P6' relative to the
cleavage site at the PI position. The Km and kcat values were determined for the cleavages. Km
values were between 10'^-10"'M for the various substrates with the highest affinity exhibited for the
tetrabasic site of ACTH''^' (1.8 x lO'^M). The relative catalytic efficiency of YAP3p indicates that
the tetra-basic residue site of ACTH^"^' was cleaved with the highest relative efficiency (kcat/Km =
1.6 X 10* M''s'') while that of the mono-basic site of CCK 13-33 and the paired basic site of pro-
insulin B-C junction were cleaved less efficiently at 4.2 x 10' M"'s'' and 1.7 x 10' M"'s"' respectively.
Comparison of the primary sequence around the cleavage site of the substrates with the relative
eflBciency of cleavage also suggested that the greater the number of basic residues in the P4-P6 and
P2'-P6' positions, the more efficient the cleavage. Denaturation of pro-insulin decreased the efficiency
of cleavage of the B-C junction indicating that in addition to the primary sequence, the efficiency is
influenced by the secondary and tertiary structure of the prohormone.

Large scale production of recombinant, C-terminally truncated, (minus the GPI binding site) secreted
YAP3p, with the pro-region intact (inactive), and the activated form, are underway for crystallization
and X-ray diffiaction studies to be done in collaboration with Dr. T. Blundell (England). Knowledge
of the specificity and the best substrates for this model enzyme in conjunction with the X-ray crystal
structure will lead to the understanding of the mechanism of action of basic-residue specific aspartic
proteases.

The specificity of YAP3p also indicates that it will be a very useful enzyme commercially for cleaving
a large number of recombinant proproteins including prohormones, pro-neuropeptides, growth factors
and serum proteins.

Mammalian homologies of yeast aspartic protease 3

Pro-opiomelanocortin converting enzyme (PCE) is an aspartic protease which we have purified from
bovine pituitary secretory granules and characterized as a paired basic residue specific prohormone
processing enzyme. Recently, Dr. Niamh Cawley showed by Western blotting that an antibody
against YAP3 cross-reacted with one 70 kDa protein, the size of PCE, from secretory vesicle
extracts, as well as a purified preparation of PCE fi"om bovine intermediate lobe. In addition. Western
blots of mouse anterior pituitary showed a -70 kD and a -90 kD YAP3p-like immunoreactive band,
while mouse hypothalamus contained only the -90 kD form. Preimmune antiserum did not show any
immunostaining.

iwiBWWBiwwBtfwiiMmmiwfffimi

Z01 HD 00056-20 LDN

Immunocytochemistry was performed by Dr. L.-P. Pu on mouse brain sections using the YAP3
antibody. The results showed immunoreactive YAP3p staining specifically in the granular cell layer
of the dentate gyrus and the pyramidal cells in the hippocampus, both neuropeptide-rich regions.
Immunostaining was also observed in the arcuate and paraventricular nuclei where POMC and
vasopressin are synthesized, respectively. In addition, the YAP3p antibody immunostained cells in
the bovine pituitary intermediate lobe and a subset of cells characteristic of the distribution of
corticotrophs in the bovine anterior pituitary. No staining was observed with preimmune antiserum
in all studies.

Preliminary results fi'om immunocytochemistry using anti-YAP3 in combination with in situ
hybridization for various neuropeptide cDNAs revealed colocalization of the YAP3-like
immunoreactivity with CCK mRNA in hippocampal neurons and vasopressin mRNA in supraoptic
neurons. These data suggest that YAP3 -related processing enzymes are colocalized with peptidergic
neurons and may play a role in pro-hormone processing in the CNS.

Previous studies showing the similarity in specificity, pH optimum and immunological identity of PCE
with YAP3p indicates that PCE is a mammalian homologue of YAP3p. Moreover, there appears to
be two forms of YAP3 -related mammalian homologues in brain and pituitary. We have recently
screened a mouse brain XgtlO library using the fiill length YAP3 sequence as a probe. Two positive
clones have been characterized with respect to restriction sites. We now have partial sequences for
both clones. Expression libraries are also being screened with the antibody against YAP3p to try and
clone the family of mammalian homologues of YAP3p.

Processing of pro-TRH by prohormone convertases PCI and PC2.

Pro-TRH is a 26 kD precursor molecule which in the rat contains 5 copies of TRH (pGlu-His-Pro-
NHj) as well as other cryptic peptides. One of the cryptic peptides, proTRHi7g.,99, has been found to
be a corticotropin releasing inhibitory factor (CRIF), and plays a role in the regulation of the HPA
axis. The enzymes that process proTRH have not been identified. Dr. Friedman, in collaboration
with Drs. Eduardo Nillni and Ivor Jackson at Brown University, studied the processing of pro-TRH
by the subtilisin-like prohormone processing enzymes, PCI and PC2. Recombinant PCI (provided
by Dr. N. Birch, New Zealand) and intermediate lobe secretory vesicle membranes which contain PCI
and PC2 were used. Purified, ^H-radiolabeled pro-TRH was incubated with recombinant PCI or
intermediate lobe secretory vesicle membranes and antibodies specific for various cryptic peptides of
pro-TRH were used for immunoprecipitation to identify the cleavage products. Recombinant PCI
was found to cleave labeled pro-TRH at either position 128-129 or 134-135 to generate the 15 kD
N-terminal (pro-TRH,.i27) and the 10 kD C-terminal (pro-TRHi36.23i) products. The 15 kD N-
terminal product was then cleaved to a 6 kD (pro-TRH,.5o) and a 3.8 kD (pro-TRH59.8g) product. The
6 kD product was fijrther cleaved to a 4 kD product (pro-TRH,.25). The 10 kD C-terminal product
was cleaved to a 5.4 kD (pro-TRH,84.23i) product. These cleavages were inhibited by Zn"^ and had
a pH optimum of 5.5 consistent with the characteristics of these subtilisin-like enzymes.

The vesicle membrane preparation cleaved pro-TRH at basic residues in a similar manner to
recombinant PCI. In addition a 16.5 kD C-terminal product was generated. The pH profile for the

±(^

■:

Z01 HD 00056-20 LDN

appearance of both C-terminal and N-terminal propeptides showed a bimodal distribution with optima
at both 5.5 and 7.5. The cleavage of pro-TRH was enhanced by Ca** and partially inhibited by ZrC* .
Immunodepletion studies with antisera specific for PC 1 or PC2 demonstrated that both these enzymes
contribute to the processing of pro-TRH by the membrane preparation. However, the difference
between PCI and PC2 is the generation of the 16.5 kD C-terminal (pro-TRHji.jj,) product by PC2
only, while PCI generated the 6 kD (pro-TRH,.5o), 4 kD (pro-TRHi.j,) and 3.8 kD (pro-TRH59.g8)
N-terminal products. Thus, both PCI and PC2 can process pro-TRH and there appears to be
enzyme-specific processing of this precursor.

In order to determine if these two enzymes are candidates for pro-TRH processing in v/vo, co-
localization studies were carried out in rat brain by Dr. Pu. Double in situ hybridization using
radiolabeled riboprobes for PC 1 and PC2 mRNAs and digoxigenin-labeled riboprobes for pro-TRH
mRNA were used in this study to determine the cellular co-localization. Analysis of rat brain showed
overlapping, as well as differential co-localization of PCI and PC2 mRNA with pro-TRH mRNA in
various brain regions, supporting the role of PCI and PC2 in pro-TRH processing in vivo. Many
neurons in the hypothalamus were double labeled for mRNAs of pro-TRH and PC 1 and PC2,
particularly in the paraventricular nucleus. In contrast, pro-TRH neurons in the amygdala, glomerular
layer and external plexiform layer of the olfactory bulb and the basal lateral hypothalamus showed co-
existence with PC2 mRNA, but not PCI mRNA. Cellular co-localization of mRNAs for pro-TRH
and PCI was evident in the neurons of the tenia tecta but PC2 mRNA was absent. However, in the
thalamic reticular nucleus, pro-TRH mRNA containing neurons did not contain PC 1 or PC2 mRNA.
The differential distribution of PCI and PC2 within TRH neurons may account for the differential
post-translational processing of pro-TRH in various brain regions reported. Interestingly, in the
thalamic reticular nucleus where there was a lack of co-expression of pro-TRH mRNA with PCI and
PC2, large processed forms of TRH but no TRH tri-peptide were found. It is possible that
prohormone convertases other than PCI and PC2 may participate in the processing of pro-TRH in
this brain region.

Another line of evidence which can support a role of PCI and PC2 in processing pro-TRH in vivo
is the demonstration of co-ordinate regulation of synthesis of these enzymes with pro-TRH. Pro-
TRH mRNA has been shown to increase in the dentate gyrus of rats after electrical kindling, which
results in seizure. Dr. Pu examined the expression of PCI, PC2 and pro-TRH mRNAs in electrically
kindled rats using quantitative in situ hybridization. A statistically significant increase of PCI (1.1
fold), PC2 (1.4 fold), and pro-TRH (3 fold) mRNAs was observed in the granule cells of the dentate
gyrus 4h following seizure; compared to non-kindled sham controls. This coordinate regulation of
PCI and PC2 mRNA with pro-TRH mRNA in the dentate gyrus provides fijrther evidence that PCI
and PC2 play a role in the processing of this prohormone in vivo. Interestingly, the increase in PC2
mRNA expression was greater than PCI, suggesting that there may be changes in the pattern of pro-
TRH processing in the dentate gyrus after kindled seizure, since PCI and PC2 process pro-TRH in
a differential manner.

J- o

■s

Z01 HD 00056-20 LDN
Significance of Biomedical Research and the Program of the Institute:

The pro-opiomelanocortin family of peptides (ACTH, a-MSH and P-endorphin) which are
synthesized both in brain and the pituitary, have been implicated in higher brain functions (e.g.,
avoidance learning, analgesia and stress response) and the regulation of the hypothalamo-pituitary-
adrenal axis. In addition a-MSH has been shown to have a role in neural regeneration, neuronal
differentiation and osmoregulation. Studies on the regulation of the biosynthesis, intracellular
transport post-translational modification and secretion of these pro-opiomelanocortin-related peptides
in the nervous system of adult and developing animal models will ultimately provide insights into the
integrative and regulatory roles these peptidergic neurons may play during development, as well as
genetic diseases associated with defects in these cellular processes.

Proposed Course:

In the coming year. Dr. Cool will purify and attempt to clone the sorting receptor for the regulated
secretory pathway. A new postdoctoral fellow will join the Section in October to study the role of
intermolecular association of POMC as part of the mechanism involved in sorting pro-
opiomelanocortin to the regulated secretory pathway. Winnie Tam will study the role of the four
highly conserved amino acids (Di0L„E,4Lig) in the POMC sorting signal motif using site-directed
mutagenesis. Dr. Cawley will focus on the expression and characterization of the YAPS -related
mammalian cDNA clones and the enzymes they encode. This will be facilitated by Dr. Pu who will
study the distribution of these enzymes in different tissues and brain and their co-expression with
various prohormones and proneuropeptides using in situ hybridization and immunocytochemistry.
Vicki Olsen, a graduate student fi'om the University of Copenhagen will study the in vivo biosynthesis
and processing of YAP3p using various yeast genetic mutants that are defective in certain intracellular
transport steps.

Publications

Azaryan AV, Friedman TC, Cawley NX, Loh YP. Characteristics of YAPS, a new prohormone
processing aspartic protease fi'om S. cerevisiae. In: Takahashi K, ed. Aspartic Proteinases. New
York: Plenum Press, 1995;569-572.

Bamberger AM, Bamberger CM, Pu L-P, Puy LA, Loh YP, Asa SL. Expression of Pit- 1 mRNA and
protein in the human placenta. J Clin Endo and Metab 1995, 80:2021-2026.

Bamberger AM, Bamberger CM, Puy LA, Asa SL. Regulation of the GH gene family in placenta and
pituitary. In: De Bellis A, ed. New prespectives in endocrinology, Rome, Italy: Serono Symposia
Publications 1995, in press.

Bamberger AM, Pu L-P, Cool DR, Loh YP. The neuro-2A cell line expresses [met]-enkephalin and
vasopressin mRNA and peptide. Mol and Cell Endocrinol 1995, in press.

!■.:

Z01 HD 00056-20 LDN

Birch NP, Hakes DJ, Dixon JE, Mezey E. Distribution and regulation of the candidate prohormone
processing enzymes SPC2 and SPC3 in adult rat brain. Neuropeptides 1994;27;307-322.

Cawley NX, Wong M, Pu L-P, Tam W, Loh YP. Secretion of yeast aspartic protease 3 is regulated
by its carboxy-terminus tail: Characterization of secreted YAP3 p. Biochemistry 1995;34:7430-7437.

Cool DR, Fenger M, Snell CR, Loh YP. Identification of the sorting signal motif within pro-
opiomelanocortin for the regulated secretory pathway. J Biol Chem 1995; 270:8723-8729.

Cool DR, Loh YP. Identification of a sorting signal for the regulated secretory pathway at the N-
terminus of pro-opiomelanocortin. Biochimie 1994,76:265-270.

Friedman TC, Garcia-Borreguero D, Hardwick D, Akuete CN, Doppman JL, Dom LD, Yanovski
JA, Chrousos GP. Decreased delta sleep and plasma delta sleep-inducing peptide concentrations in
patients with Cushing Syndrome. Neuroendocrinology 1994;60:626-634.

Friedman TC, Gordon VM, Leppla SH, Klimpel KR, Birch NP, Loh YP. In vitro processing of
anthrax toxin protective antigen by recombinant PCI (SPC3) and bovine intermediate lobe secretory
vesicle membranes. Arch Biochem & Biophys 1995;3 16:5-13.

Friedman TC, Loh YP, Birch NP. /// vitro processing of pro-opiomelanocortin by recombinant PCI
(SPC3). Endocrinology 1995;135:854-862.

Friedman TC, Loh YP, Cawley NX, Birch NP, Huang SS, Jackson IMD, Nillni EA. Processing of
pro-TRH by bovine intermediate lobe secretory vesicle membrane PCI and PC2 enzymes.
Endocrinology 1995, in press.

Friedman TC, Yanovski JA. Morning plasma free Cortisol: Inability to distinguish patients with mild
Cushing syndrome fi-om patients with pseudo-cushing states. J Endocrinol Invest 1995, in press.

Friedman TC, Yanovski JA, Jayasvasti V, Yanovski SZ, Koenig RJ, Wilk S. Pyroglutamyl peptidase
II ("Thyroliberinase") Activity in Human Serum: Influence of Weight and Thyroid Status. J Clin
Endo andMetab 1995; 80:1086-1089.

Hill RM, Ledgerwood EC, Brennan SO, Pu L-P, Loh YP, Christie DL, Birch NP. Comparison of
the molecular forms of the Kex2/subtilisin-like serine proteases SPC2, SPC3 and flirin in
neuroendocrine secretory vesicles reveals differences in carboxyl terminus truncation and membrane
association. JNeurochem 1995, in press.

Loh YP, Cawley NX. Strategies and techniques for endoproteolytic processing enzyme purification
and characterization. In: Beinfeld, MC, ed., series "Processing of Neuropeptides" in Neuroprotocols,
San Diego: Academic Press, 1994;5:122-129.

nam

Z01 HD 00056-20 LDN

Loh YP, Cawley NX. Processing enzymes of the aspartic protease family: Yeast aspartic protease
3 and pro-opiomelanocortin converting enzyme. In: Barrett, AJ, ed. Methods in Enzymology 1995;
248:136-145.

Loh YP, Cawley NX, Friedman TC, Pu L-P. Yeast and mammalian, basic residue-specific aspartic
proteases in prohormone conversion. In: Takahashi K, ed. Aspartic Proteinases. New York: Plenum
Press, 1995;5 19-527.

NUIni EA, Friedman TC, Todd RB, Birch NP, Loh YP, Jackson IMD. ProTRH processing by PCI .
J Neurochem 1995, in press.

Pu LP, Hayes WP, Mill JF, Ghose S, Friedman TC, Loh YP. Frog prohormone convertase PC2
mRNA has a mammalian-like expression pattern in the CNS and is co-localized with a subset of
thyrotropin-releasing hormone expression neurons. J Comp Neurology 1994;354:71-86.

Pu L-P, Van Leeuwen FW, Tracer HL, Sormemans MAP, Loh YP. Localization of Vasopressin
mRNA and immunoreactivity in pituicytes of pituitary stalk-transected rats after osmotic stimulation.
Proc. Natl. Acad. Sci. USA. 1995, in press.

Pu L-P, Ma W, Barker JL, Loh YP. Differential co-expression of genes encoding pro-TRH and
prohormone convertases (PCI and PC2) in rat brain neurons: Implications for differential processing
of pro-TRH. Endocrinology 1995, in press.

^MHWrnfr***^"*^^^ "'"'"

man

nFPARTMFNT OF HFAI TU ANH HUMAN <;FRVir.FR - PIIRI 10 HFAI TH SFRVICF

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

Z01 HD 00064-19 LDN

PERIOD COVERED

October 1. 1994 through September 30, 1995

TTTLE OF PROJECT (80 chanden or has. Title must fit on one line between borden.)

Neurobiologic Studies of Neurons and Glia in Cell Culture

PRINCIPAL INVESTIGATOR (List other professional personnel below the Principal Investigator.) (Name, title, laboratory, and institute affiliation)

PI: P.G.Nelson Chief LDN,NICHD

Others: R.Davenport NRC FeUow LDN,NICHD

M. Li Snr NRC FeUow LDN,NICHD

M. Jia Visiting Associate LDN,N1CHD

P. Latham IPA LDN,NICHD

D. Bisant Adj Post Doc LDN,N1CHD
V. Dunlap Biol Lab Tech LDN,N1CHD

E. Thies Technician LDN,NICHD
A. Parfitt Expert LDN,NICHD

COOPERATING UNITS fif any)

LAB/BRANCH

Laboratory of Developmental Neurobiology

SECTION

Section on Neurobiology

INSTITUTE AND LOCATION

NICHD, NIH, Bethesda, MD 20892

TOTAL STAFF YEARS:

6.9

PROFESSIONAL:

4.9

OTHER:

2.0

CHECK APPROPRIATE BOX(ES)

n Ja) Human subjects D (b) Human tissuesH (c) Neither
(a1) Minors
(a2) Interviews

SUMIMARY OF WORK (Use etandaid unreduced type. Do not exceed the space provided.)

We have previously shown that chronic (>24 hr.) stimulation of sensory neurons in vitro produces
an alteration in growth cone sensitivity to stimulation and a down regulation of voltage-sensitive
Ca++ channels (VSCC). We now show specificity in this effect both with regard to the type of
VSCC affected and for different patterns of stimulation. The low voltage activated VSCC (T
current) is extremely sensitive to stimulation; single stimuU at a steady rate of 0.5 Hz completely
abolishes the T current while not affecting the high voltage activated (N and L) VSCC. The same
number of impulses organized into different patterns of tetanic Cbursty') or steady stimulations
have differential effects on the N and L currents. Binding of a radio-labelled VSCC Ugand, PN200-
110 is also down regulated by chronic stimulation confirming that the number of Ca++ channels is
affected by the stimulation. Activation of cultured skeletal muscle cells w^ith acetylcholine produces
an increase in prothrombin mRNA, which is consistent wath the role we have proposed for thrombin
in the process of activity dependent synapse elimination. We fail to demonstrate a modulation of
the serine protease inhibitor Protease Nexin 1 in muscle, however. If this endogenous inhibitor does
play a role in synapse elimination it may be because of its presence (and modulation) in some other
cell type, either nerve or gUa. Evidence for PN Ts presence and modulation in glia cells has indeed
been obtained by Brenneman's group in the LDN.

To investigate the cellular localization of activity-independent guidance cues in the developing
visual system, we have established retinal and tectal cultures in modified 3-compartment chambers.
Using embryonic chick we have localized a previously identified aversive component to tectal
neurons. We have discovered a possibly novel, positive guidance component on radial glia. Both of
these guidance mechanisms are restricted to one portion of the tectiun during development.

PHS 6040 (Rev. 5/92)

ran

Sia^HK^Bamm

ZOl HD 00064-19 LDN

Project Description

Objectives

We continue our interest in the role of electrical activity in regulating development of the
nervous system, in particular its synaptic circuitry. We are also studying those molecular and
cellular cues w^hich are involved in neuronal interactions involved in synaptogenesis which may
be independent of electrical activity. The relationship between these classes of phenomena
should be important and we are using vertebrate cell culture systems to allow cellular and
molecular imderstanding of them.

Methods Employed

A variety of dissociated cell culture preparations from the mammalian central nervous system
are utilized, with emphasis on the fetal mouse spinal cord and sensory neuron culture system.
The electrophysiological techniques utilized include intracellular and patch electrode recording
and voltage-clamping. Pharmacological agents are applied by local perfusion or bath
application.

Calcium imaging was performed with calcivmn-sensitive dyes and a video/computer analytic
system.

A culture system is being utilized which incorporates three chambers separated from one another
by barriers which allow the growth of axons between chambers, but retain neuronal cell bodies
within the chambers in which they are plated. Various markers or stains can be used to
demonstrate axons that project from one chamber to the other, and the development of synaptic
interactions between specific cell types can be followed morphologically and physiologically.

To assess developmental guidance mechanisms in the optic pathway of embryonic chick and
mammals, we have dissociated retinal and tectal cells from animals at embryonic ages (E6-11 in
chick; El 3-1 6 in mouse) when retinal axons are just innervating their target nuclei m vivo. Target
cells are dispersed usually in the center compartment and retinal cells or explant strips are
placed in the side compartments of a redesigned Campenot chamber. We have replaced the
teflon chamber barriers which separate compartments with glass coverslips to reduce the
distance retinal axons must extend to reach the center compartment by approximately 10-fold.
Cells are kept in culture for 2-10 days and assessed with time-lapse video microscopy or whole
cell patch pipette recording.

Semiquantitative and quantitative polymerase chain reaction technology is used to measure a
variety of mRNAs of interest.

Major Findings

Our previous findings had indicated that chronic (>24 hours) electrical stimulation of sensory
neurons produced a down regulation of the voltage-sensitive calcium channels expressed by
these cells. We have examined the recovery time course following termination of stimulation.

_LO

HIIiHU

ZOl HD 00064-19 LDN

and the observation that 24 hoxirs or more is required for full recovery indicates that probably
more than functional alteration of the Ca** channels is involved in the response to chronic
activation. Further analysis has shown that the consequence of activation of these cells shows
specificity, both in terms of the type of Ca** channel affected and in the stimulus pattern
specificity of the effect. The low voltage activated Ca** current (T current) is substantially the
most sensitive species of channel, being essentially eliminated by single pulse stimulation at 0.5
Hz. Its recovery had not begun at 24 hours after cessation of stimulation, but was complete at
120 hours. The peak and steady high voltage activated currents (N and L currents) were
somewhat less sensitive to stimulation and showed some differential responsiveness to different
patterns of chronic stimulation. Thirty Hz. stimulation delivered for 2 sec every 10 sec produced
less response than 10 Hz stimulation delivered for 0.5 sec every two sec, even though the total
number of stimuli delivered was the same in each case. Steady stimulation at 0.5 Hz produced
effects intermediate between the two types of 'burst' stimulation. The electrophysiological
experiments were confirmed by binding studies using a radio-labelled Ca** channel ligand,
PN200-110. Specific binding was decreased by about 50% by chronic stimulation of the sensory
neurons, with a time course very similar to that demonstrated in the electrophysiological
experiments.

Previous results had indicated that activity-dependent synapse elirrunation(ADSE) in a tissue
culture model of the cholinergic neuron to muscle connection was mediated, at least in part by
the serine protease, thrombin. The process can be completely blocked by an endogenous serine
protease inhibitor. Protease Nexin 1 (PN 1), and we hypothesized that a proper balance between
protease activity and PN 1 might be crucial for optimal development and stabilization of
synapses. Thrombin secretion and prothrombin mRNA are modulated by cholinergic stimulation
of cultured skeletal muscle, but we find no significant effect of activation on PN 1 message
levels. If PN 1 modulation by activity plays a role in ADSE, then some cell other than muscle
cells must be involved. Indeed, Dr. Brenneman's group has shown that stimulation of gUal cells
can increase the output of PN 1. Such a differential cellular disposition of protease and inhibitor
provides a basis for a model of selective elimination and stabilization of synapses during
development. That thrombin may be involved in central nervous system development is
suggested by preliminary data indicating that hirudin, a specific blocker of thrombin, increases
neuronal survival imder certain conditions in CNS cultures.

During development of the nervous system both activity-dependent and -independent
mechanisms are utilized to bring about the ordered connections. Our focus has been on
projections in the PNS and the critical role electrical activity serves. We have begun to
investigate model systems in embryonic chick and mouse based on the retinal projection to its
target nuclei which wiU enable us to evaluate both activity-dependent and -independent
mechanisms during CNS development. The map formed by retinal ganglion cells on their target
nuclei has served widely as a model system to investigate mechanisms underlying the highly
precise and stereotypic connectivity of the nervous system. Activity-independent guidance
components are known to initially affect rudimentary map formation; activity-dependent events
refine these maps. While it is clear that activity-dependent mechanisms must involve post-
synaptic target neurons, the substratum for activity-independent guidance remains unknown.
To investigate cellular localization of activity-independent guidance cues in the developing
retinotectal system, we modified our standard (Campenot) 3-compartment chamber and created
areas where cultured embryonic retinal and tectal cells could encounter one another and

RHns^Hnn

ZOl HD 00064-19 LDN

guidance behavior be readily assessed and electrical activity imposed. Preservation of relevant
guidance information on living cell populations enabled separate evaluation of retinal growth
cone behavior upon encounter with tectal neurons and radial gUa. Using embryonic chick, we
localized a previously defined, aversive guidance component to neurons and discovered a
possibly novel guidance component on radial gUa, both restricted to cells from one region of the
optic tecta during development. Together these differential interactions suggested a novel
hypothesis for the development of retinotectal topography that more clearly defines how
development of ordered connections critically depends on ceU-specific cues distributed selectively
on target neurons and gha.

We have learned that retinal ganglion axons are electrically excitable under our culttire
conditions. Intracellular calcium levels rise with imposed activity using a redesigned
compartmentalized chamber. Furthermore, their cocultured target cells are not only electrically
excitable, but possess a full complement of voltage-dependent channels. These channels seem
to be expressed with a time course in culture that requires several days, thus allowing
examination of the functional consequences of voltage-dependent channel expression.
Spontaneous synaptic activity occurs in these cultures, but also may require several days to
develop.

Based on our previous experiments which suggest a functional role for protease/protease
inhibitors in synaptic plasticity and remodeling of the neuromuscular junction in culture, we are
beginning to investigate whether these same molecules serve a similar role m vivo in the CNS.
In the developing chick, a transient ipsilateral projection occurs from retinae to tecta and
subsequently lost spontaneously and completely. Initial experiments performed in collaboration
with Dr. S McLoon (Univ Minn) suggest that hirudin and possibly the less specific protease
inhibitor leupeptin can reduce the loss of the ipsilateral projection in vivo. Though preliminary,
these results suggests that protease inhibitors may serve a similar role in natural CNS synaptic
reduction as we have been studying at the neuromuscular junction.

Significance to Biomedical Research and the Program of the Institute

Assembling appropriate and functionally effective neural circuits is a critical developmental
process of fundamental importance to the successful adaptation of the organism. Understanding
the mechanisms involved in the process is clearly important for neuroscience and the NICHD,
which has a lead role in dealing with problems of mental retardation and more subtle aspects
of inadequate cognitive development.

Proposed Course

Further studies of the mechanism of the ADSE will be carried out in the nerve-muscle
preparation. A peptide is available that partially mimics a receptor mediated component of
thrombin's activity. We will test if this peptide can produce synapse loss as does thrombin itself.
If it does not it would suggest that thrombin is acting through some other target. We need to
examine a variety of stimulus patterns to see if synapse elimination shows any specificity in this
regard. The tetanic stimuli used to date produce a non-Hebbian loss of both stimulated and
non-stimulated connections, and it may be that other stimulus configurations may have different
effects. Possible retrograde signals coupling postsynaptic function have been postulated and we

ZOl HD 00064-19 LDN

will exploit our system in looking for such factors. The nitric oxide and arachidonic acid systems
are attractive candidates for such studies. Anatomical and physiological studies of the acute and
chronic sequelae of synapse activation wiU be continued.

We will further investigate the localization and characterization of the activity-independent
guidance components differentially distributed across retinal target nuclei in both the embryonic
chick and mouse. We will examine the spontaneous and evoked calcium responses of target
cells to evaluate activity-dependent regulation of the topographic development of the retinotectal
projection. We will further characterize the development of voltage-dependent channels in tectal
cells and the electrical, synaptic input from projecting retinal axons. We intend to foUow up the
preliminary studies on the effect of protease/protease inhibitors in embryonic chick CNS
projection in vivo and to further evaluate the natviral abundance and timing of these molecules
during development.

Protocols:

Animals:

94-013 Nelson Developmental neurobiology in culture systems

Publications:

Liu Y, Fields RD, Festoff BW, Nelson PG. Proteolytic action of thrombin is required for electrical
activity-dependent synapse reduction. Proc Natl Acad Sci 1994;91:10300-04.

Nelson PG, Fields RD. Developmental plasticity of the brain: problems and solutions in dynamic
imaging of the synaptic bouton. Proceedings of 2nd Hamamatsu Intl Symposivim on
Biomolecular Mechanisms and Photonics: Cell-Cell Communications 1995 (in press).

Nelson PG, Fields RD, Liu Y. Neural activity, neuron-glia relations and synapse development.
Persp Dev Neurobiol 1995;2:399-407.

Nelson PG, Neale EA. Spinal cord cell culture. Encyclopedia of Neurosci 1995 (in press).

Sheng HZ, Lin P-X, Nelson PG. Analysis of multiple heterogeneous mRNAs in single cells.
Analytical Biochem 1994;222:123-130.

Sheng HZ, Lin P-X, Nelson PG. Combinatorial expression of immediate early genes in single
neurons. Mol Brain Res 1995;30:196-202.

Singer HS, Chiu AY, Meiri KF, Morell P, Nelson PG, Tennekoon G. Advances in understanding
the development of the nervous system. Curr Opin Neurol Neurosurg 1994;7:153-9.

*"•■

r-..

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBLIC HEALTH SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD 00094-25 LDN

PERIOD COVERED
October 1, 1994

September 30, 1995

TITLE OF PROJECT (80 characters or less Title must fit on one line between the borders )

Pineal Regulation: Environmental and Physiology Factors

PRINCIPAL INVESTIGATOR (List other professional personnel below the Principal Investigator) (Name, title, laboratory, and institute affiliation)
PI: David C. Klein Head LDN, NICHD

Others: Steven Coon NRC Fellow LDN, NICHD

COOPERATING UNITS (if any)

Georgetown University (M.A.A. Namboodiri) ; University of Geneva (N. Schaad)

University of Giessen (H. Korf)

LAB/BRANCH

Laboratory of Developmental Neurobiology

SECTION

Section on Neuroendocrinology

INSTITUTE AND LOCATION
NICHD, NIH, Bethesda,

MD 20892

TOTAL STAFF YEARS
1 .0

PROFESSIONAL:
1 . 0

OTHER:
0 . 0

CHECK APPROPRIATE BOX(ES)

D (a) Human subjects

D (a1) Minors
n (a?) Inffirvifiws

D (b) Human tissues Kl (c) Neither

SUMMARY OF V\ORK (Use standard unreduced type. Do not exceed the space provided )

The regulation of the pineal gland by environmental and physiological factors is
analyzed, exclusive of transmembrane and intracellular regulatory mechanisms (see ZOl
HD 00095-25 LDN) . The pineal gland is part of the melatonin rhythm generating system,
a neural circuit which includes a circadian clock in the suprachiasmatic nucleus
(SCN) ; the SCN is reset and entrained by light acting through the eye. An area of
special current interest is proteins involved in transmembrane signal processing. The
most exciting finding is that there is a daily rhythm in the abundance of mRNA
encoding the a- iB-adreneraic receptor. This increase seems to be driven by the SCN,
and is blocked by light acting through the eyes. Studies in organ culture indicate
the receptor is regulated by cAMP . This second messenger is controlled by several
receptors, two of which have been found to mediate increases in the abundance of mRNA
encoding the a-lB-adrenergic receptor. This work is important because it provides an
interesting mechanism through which one receptor can regulate another.

PHS 6040 {Rev. 5/92)

f^ (~^

»ra

iUHU

mmmiaiimtiim^mmitmmmm^mAmmimiMmii'tiiM^^ii^

ZOl HD 00094-25 LDN
Project Description

Objectives:

To determine how the pineal gland is influenced by physiological, and pharmacological factors,
and to study the mechanisms involved in the development and regulation of neural stimulation
of the pineal gland.

Methods Employed:

The function of the pineal gland is studied by measuring the production of melatonin.
Melatonin is measured directly by high performance liquid chromatography-electrochemical
detection. In addition, the major metabolite of melatonin, 6-hydroxymelatonin, is measured
in urine using a mass chromatography method. Pineal enzymes are measured by radiochemical
techniques. Proteins are resolved by polyacrylamide electrophoresis and electroblotted to
membranes for immunological analysis (see ZOl HD 00095-25 LDN for further details). PCR
technology, in situ hybridization and Northern blot technology are used to analyze niRNA
expression.

Major Findings:

The major new finding in the program was the discovery that there is a high abundance of
mRNA encoding the a,B-adrenergic receptor subtype, and that this appears to be a primary
member of the a, -adrenergic receptor family which is expressed in the pineal gland. Other
studies established that the abundance of mRNA encoding this receptor is higher in the pineal
gland than in other tissues, and that there is a 2- to 4-fold daily rhythm in the abundance of
this receptor. Organ culture studies have indicated that this receptor is regulated by cAMP and
agonists which elevate cAMP.

Developmental studies have indicated that the major developmental increase in the abundance
of this receptor occurs during the first few days of life in the rat.

Significance to Biomedical Research and the Program of the Institute:

The a,B-adrenergic receptor plays a key role in signal processing regulating the production of
the hormone of the pineal gland, melatonin. Accordingly, studies on this receptor in the pineal
gland provide valuable information in two areas:

Control of melatonin: The program yields new information about the control of melatonin
production. This compound is thought to influence reproduction and mood in humans.
Therefore, any advance made by this program could eventually have direct influence in clinical
practice. In addition, pioneering work done within this program may lead to tests of pineal
function in humans.

UmiMHJWI

ZOl HD 00094-25 LDN

Neural processing and development: The program yields new information about neural
processing and development. This provides investigators an opportunity to learn how a single
neural system develops and how it functions during development. In addition, it might provide
some insight into how environmental factors alter development.

Proposed Course:

The relationship of changes in mRNA encoding the am-adrenergic receptor and in total
receptor number will be studied using agents which elevate the receptor without interacting
with it directly.

Protocols:

93-016 Klein Neuroendocrine regulation: Rat pineal gland

Publications:

Coon, S.L., McCune, S.K., Sugden, D. and Klein, D.C.: Day/night rhythm in pineal a,B-
adrenergic receptor mRNA: Regulation by a P-adrenergic -^ cyclic AMP mechanism
Endocrinology; in press.

2^i

Mill*

mmmm

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBLIC HEALTW SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD 00095-25 LDN

PERIOD COVERED
October 1, 1994

September 30, 1995

TITLE OF PROJECT (80 characters or less Title must fit on one line between the borders }

Pineal Regulation: Transsynaptic and Intracellular Mechanisms

PRINCIPAL INVESTIGATOR (List other professional personnel below the Principal Investigator) (Name, title, laboratory, and institute affiliation)

PI:

C. Klein

Others:

L. Waller

Roseboom

Coon

Baler

Bernard

Head

Chemist

Staff Fellow

IRTA Fellow

Biotechnology Fellow

Visiting Fellow

LDN, NICHD

COOPERATING UNITS (if any)

Georgetown Univ (M.A.A. Namboodiri)
LDN, NICHD (J. Russell)

Univ of Giessen (H. Korf); NIMH (D. Jacobowitz)

L<^B/BRANCH

Laboratory of Developmental Neurobiology

SECTION

Section on Neuroendocrinology

INSTITUTE AND LOCATION

NICHD, NIH, Bethesda, MP 20892

TOTAL STAFF YEARS
8 . 0

PROFESSIONAL
6 .9

OTHER

1 . 1

CHECK APPROPRIATE BOX(ES)

D (a) Human subjects

D (a1) Minors
n (a?) Intfirvifiws

£] (b) Human tissues D (c) Neither

SUMMARY OF WORK (Use standard unreduced type Do not exceed the space provided.)

Mechanisms involved in the transduction of neural signals and the control of tissue
specific gene expression are studied. The pineal and pituitary glands are used as
model systems. The details of the chemical and ionic components of transmembrane
signal processing and of neural and tissue specific regulation of gene expression are
analyzed. Signal transduction in the pineal gland is of special interest because cAMP
and cGMP are regulated by dual receptor mechanisms which appear to interact at the
level of regulation of adenvlvl and guanylyl cyclases. One leg of these pathways
activates these enzymes via GTP binding regulatory proteins, similar to Gsa .

In the area of the neural and developmental control of gene expression, a major
advance has been made by the cloning of the gene encoding serotonin N-
acetyltransf erase . This will make it possible to study how neural signal regulate
expression of this gene in vertebrates and how the activity of this enzyme is
controlled .

The major hormonal product of the pineal gland is melatonin. Melatonin has been
found to block GnRH induced increase in [Ca++] i and to block GnRH-induced
depolarization . Melatonin appears to act on a subpopulation of GnRH- sensitive cells.
IT is clear from associated studies that melatonin acts through an action on the
calcium . not cylcic AMP, and that the effect of melatonin on calcium is secodary to an
effect on membrane potential.

PHS 6040 (Rev 5/92)

,', 0

Z01HD00095-25LDN
Project Description

Objectives:

To describe the transsynaptic and intracellular mechanisms through which neuronal activity
regulates the biochemical activity of a neural tissue, the pineal gland.

Methods Employed:

General: Tissue is obtained from normal animals or surgically prepared animals. Cultures of pineal
tissue are used to study biochemical activity under highly defined conditions. Pineal glands and
dispersed suspension cultures of pineal cells are used. Tissues are treated with chemicals of interest
and the activity of enzymes and concentrations of cell chemicals are measured by a variety of
techniques. To study development, pineal glands are removed from rats at sequential stages of
development and enzyme and metabolic chemicals are measured. Enzymes are purified by a variety
of methods. Antisera are produced in rabbits; antisera are used to identify proteins. cDNA libraries
are prepared using a A.gtl 1 vector and screened with synthetic oligonucleotides and cDNA probes.

Pineal tissue was obtained from normal animals or surgically operated animals. Pineal enzymes are
measured by radiochemical techniques. Proteins are resolved by polyacrylamide gel electrophoresis
and electroblotted to membranes for immunological analysis. Total RNA is isolated by a single step
method using acid guanidinium thiocyanate-phenol-chloroform extraction; resolved on an agarose
gel and transferred onto positively charged membranes for message level analysis. In addition,
polymerase chain reaction technology is used to confirm the results obtained by Northern analysis.

Cyclic nucleotide regulation: Pineal glands from rats are cultured, and treated with various agents
to stimulate cAMP and cGMP. Cyclic nucleotides from both glands and culture medium are
measured by radioimmunoassay (RIA). Guanylate cyclase activity was measured by in vitro
reaction at 37°C in the presence of Mn*\ GTP, and phosphodiesterase inhibitor, then subsequent
RIA for cGMP. Proteins are analyzed by SDS-PAGE and Western blot. Immunoprecipitations are
performed using antisera against Gso.

Melatonin mechanism of action: The mechanism of intracellular transduction of melatonin effect
is investigated using primary cell culture from neonatal rat pituitary glands. LH-release is
determined by radioimmunoassay, and intracellular Ca"^^ concentrations are determined from
fluorescence of a suspension of cells loaded with calcium sensitive dye Fluo-3. A similar approach
is used to measure membrane potential. In addition, single cell analysis of neonatal pituitary cells
has been initiated.

Transcription factors: Analysis is through use of highly specific antiserum, combined with
labelled oligonucleotide probes in gel shift assays. In addition, PCR is used to detect mRNA
transcripts.

IMBHOMBinwat

Z01HD00095-25LDN

Major findings:

Guanylate cyclase and cGMP: The role of NO in the regulation of guanylyl cyclase was examined
using several approaches. First, it was found that NO synthase, the enzyme which generates NO and
citrulline from arginine is present in the pineal gland. Second, it was found that this enzyme is
calcium and calmodulin dependent. Third, it was found that inhibitors of this enzyme inhibited the
norepinephrine stimulation of cGMP production. Fourth, it was found that norepinephrine
stimulation of cGMP was accompanied by an increase in the production of citrulline and of a
diffusible factor with NO-like characteristics. These findings indicate that NO plays a key role in
the adrenergic stimulation of cGMP in the pineal gland. In addition, it was found that the sensitivity
of the cGMP system to NO is regulated by a GjO mechanism, providing clear indication that
interaction of receptor input can occur at this level.

An important effect of cGMP on pineal signal processing was discovered. It was found that cGMP
activates influx of calcium, and that this reflects the interaction of cGMP with a cGMP gated
channel, which was found to be identical to the channel found in the retina, accordingly to
polymerase chain reaction (PCR) analysis. The presence of this channel is important because is
appears to mediate effects of vasoactive intestinal peptide (VIP) on calcium.

HIOMT: It was found that HIOMT is expressed at very low levels in the retina. In addition, it was
found that HIOMT levels in Y-79 cells are strongly dependent upon a serum factor, and that
removal of serum results in a rapid loss of enzyme activity and mRNA encoding HIOMT. Other
studies indicate that HIOMT can be regulated by retinoic acid, suggesting that this might be
involved in the regulation of HIOMT.

Melatonin mechanism of action: Studies using a single cell analysis technique indicated that
melatonin acts on a subpopulation of GnRH sensitive gonadotrophs. Melatonin appears to act by
causing hyperpolarization which then alters calcium currents through voltage sensitive channels.
Melatonin acts through a G-protein linked receptor mechanism which is antagonized by pertussis
toxin. Preliminary studies suggest that this action involves a phosphatase and that the melatonin
receptor is a member of a family of receptors which act on membrane potential through a pertussis
toxin sensitive mechanism involving regulation of the activity of a phosphatase.

N-Acetyltransferase; An expresison cloning technique was used to identify a clone encoding
serotonin N-aceltyltransferase. The identity of the clone was verified substrate specificity analysis.
This was extended by the cloning of cDNA encoding the human , rat and chicken forms of the
enzyme. cDNA probes were used to determine there was a large increase in mRNA encoding this
enzyme in the rat pineal gland, , a smaller increase in the chicken and not increase in sheep. Tissue
distribution studies revealed that mRNA encoding this enzyme is not detectable in any species in
any endocrine or internal organ but is consistently detectable in retina. In addition, it is detectable
in sheep brain regions.

9
i~^ I

Z01HD00095-25LDN

Adrenergically induced proteins: The most prominent adrenergically-stimulated protein in
norepinephrine-treated glands, as judged by incorporation of ^'S-methionine incorporation into
proteins has a molecular weight of about 37 kDa and a pi of about 6. This protein was partially
purified and resolved by two-dimensional gel electrophoresis. It was identified by microsequencing
to be an isoform of malate dehydrogenase. This was confirmed by immunochemical techniques,
using antisera raised against synthetic malate dehydrogenase peptides. It was also determined that
the appearance of this protein reflects two processes, new synthesis of protein and cotranslational
phosphorylation.

Transcription factors: The cAMP-responsive factor CREB has been identified in the pineal gland
and it has been found that it is phosphorylated following norepinephrine stimulation. The time
course and dose dependency of this response has been characterized.

The presence in the pineal gland of the newly identified inducible inhibitory transcription factor
ICE-CREM has been confirmed by PCR. The regulation and expression of this factor is being
analyzed. Studies on c-fos revealed that norepinephrine does not alter c-fos expression significantly,
but does cause massive induction of a 42/46 kDa fos related antigen (fra). This has been
documented both by Western blot and Northern blot analysis. The protein exhibits a marked
circadian rhythm, which appears to be due to neural stimulation of the gland by the suprachiasmatic
nucleus. The gene has been cloned and the coding region has been fully sequenced, revealing this
protein to be the rat homolog of 42/46 kDa fra.

Significance to Biomedical Research and the Program of the Institute:

The pineal gland offers an attractive model system for the study of the neurochemical transduction
from the receptor to the gene. This is a fundamental aspect of brain function and further
understanding will lead to better treatment of neurological and mental diseases, and a better
understanding of problems of mental retardation.

Proposed Course:

N-acetyltransferase: The promoter will be identified and the factors regulating expression of this
gene will be determined, including transcription factors. Special attention will be paid to the role
ofCREB, ICERandFra-2.

Factors regulating the activity of the enzyme will be analyzed using expression systems, with special
attention paid to the role of phosphorylation of two putative protein kinase A sites.

Antiserum is being raised against expressed protein, against three peptides and against two
phosphopeptides. These will be used to monitor the total amount of enzyme protein and the
phosphorylation state of the protein, and to determine how this relates to enzyme activity.

.iij

■■»™««™''WiBTmwwg«niimiiiiiM«Hwmiyiiyiwi|iiiWiiii'r''ffl<'N^^

ZOIHD 00095-25LDN

HIOMT: Rat HIOMT: Upon determination of the DNA sequence of the rat pineal protein
hydroxyindole-0-methyltransferase, PCR will be used to determine the expression of this gene
at various stages in the light/dark cycle, after pharmacological manipulation and at various
stages of development. In addition, oligonucleotides will be used as probes for Northern blot
analysis and in situ hybridization studies to characterize the tissue distribution in the rat.

Human HIOMT: Isolation of the promoter region of the human HIOMT gene - The 5'-
flanking region of the HIOMT gene will be isolated and subcloned in a vector that contains
a reporter gene, i.e. chloramphenicol acetyltransferase (CAT). The inserted DNA will then
undergo nested deletions from the 5' end. CAT activity will be assayed following transfection
of Y-79 cells (retinoblastoma cell line). Y-79 cells have been shown to express HIOMT
activity and synthesize melatonin in a cAMP-dependent manner. By assaying for CAT activity,
regions of DNA important for transcription will be identified. The sequence of the region that
activates transcription will then be analyzed by computer to determine whether known promoter
and/or enhancer sequences are present.

The three mRNA splice variants will be expressed and the expressed protein will be analyzed
for enzyme activity and immunogenicity.

Y-79 HIOMT: The factor regulation HIOMT activity in Y-79 cells will be purified and
identified.

Protocols:

93-016 Klein Neuroendocrine regulation: Rat pineal gland

Publications:

Baler, R. and Klein, D.C. Physiological regulation of oncogene FRA-2: Circadian — >
adrenergic -^ cyclic AMP control in the rat pineal gland . J Biol Chem. ( in press).

Bernard, M., Donohue, S.J., and Klein, D.C. Hydroxyindole-o-methyltransferase in the human
pineal gland and retina and in Y79 retinoblastoma cells. Brain Res. (in press).

Detera-Wadleigh, S.D., Hsieh, W-T, Berrettini, W.H., Goldin, L.R., Rollins, D.Y., Muniec, D.,
Grewal, R., Guroff, J. J., Turner, G., Coffman, Barrick, J., Mills, K., Murray, J., Donohue, S.J.,
Klein, D.C, et al. Genetic linkage mapping for a susceptibility locus to bipolar illness:
Chromosomes 2,3,4,7,9,10p, Up, 22 and Xpter. Am J Med Genetics 1994; 54:206-218.

Lin, A.M.-Y., Schaad, N.C., Schulz, P.E. Coon, S.L., and Klein, D.C. Pineal nitric oxide
synthase: Characteristics, adrenergic regulation and function. Brain Research 1994;651:160-
168.

ZOIHD 00095-25LDN

Roseboom, P.H., Weller, J.L., Babila, T., Aiken, A., Sellers, L.A., Moffett, J.R., Namboodiri,
M.A.A. and Klein, D.C. Cloning and characterization of the c and C, isoforms of the rat 14-3-3
proteins. DNA and Cell Biology. 1994;13:629-640.

Roseboom, P.H. and Klein, D.C. Norepinephrine stimulation of pineal cyclic AMP response
element-binding protein phosphorylation: Primary role of a p-adrenergic receptor/cyclic AMP
mechanism. Molecular Pharmacology 1995; 147:439-449.

Schaad, N.C., Vanecek, J., Rodriguez, I.R., Klein, D.C, Holtzclaw, L., and Russell, J.T.
Vasoactive intestinal peptide elevates pinealocyte intracellular calcium concentrations by
enhancing influx: Evidence for involvement of a cyclic GMP dependent mechanism.
Molecular Pharmacology 1995; 47:923-933.

Vanecek, J. and Klein, D.C. Mechanism of melatonin action on neonatal rat gonadotrophs. In
Advances in Pineal Research, Moller, M., Pevet, P. and Reiter, R.J. (eds), John Libbey
(London), 1994; Vol. 8, pp. 363-370.

Vanecek, J. and Klein, D.C. Melatonin inhibition of GnRH-induced LH release from neonatal
rat gonadotroph: involvement of Ca"" not cAMP. Am. J. Physiol. 1995;269:E85-E90.

White, B. and Klein, D.C. Stimulation of cyclic GMP accumulation by sodium nitroprusside
is potentiated via a G^ mechanism in intact pinealocytes. Journal of Neurochemistry 1995;
64:711-717.

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBLIC HEALTH SERVICE
NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD 00704-11 LDN

PERIOD COVERED

October 1, 1994 - September 30, 1995

TITLE OF PROJECT (80 characmrs or leis Tnie must tit on one line tietween ttie tyorOers)

Tetanus Toxin Effects and Localization in Neurons

PRINCIPAL INVESTIGATOR (Ust other p/vteistonal personnel tielow the Pnncipal Invetrtgalor j (Name, title, laboratorf. and institute atliliation)

PI:

E.A. Neale

Head

LDN, NICHD

L.C. Williamson

Sr. Staff Fellow

LDN, NICHD

Others:

L.M. Bowers

Biologist

LDN, NICHD

S.C. Fitzgerald

Biologist

LDN, NICHD

V. Dunlap

Biol. Lab. Tech.

LDN, NICHD

COOPERATING UNITS (if any)

Div.Bacterial Prod., Ctr Biol. Eval. Res., FDA (JH Halpem, WH Habig); NIDR (SH Leppla);
U.S. Army Med. Res. Inst, of Infectious Diseases (JE Brown); Dipartimento di Science, Univ.
of Padova (C. Montecucco); Dept. of Biology, Georgetown U. (M.A.A. Namboodiri)

LAB/BRANCH

Laboratory of Developmental Neurobiology

SECTION

Section on Cell Biology

INSTITUTE AND LOCATION

MTPHn NTTH_ RpfhpgH^_ Maryland 'P0P.97

TOTAL MAN-YEARS:

2.1

OFESSIONAL:

1.4

i OTHER:

0.7

CHECK APPROPRIATE B0X(ES)

n (a) Human subjects
D (a1) Minors
D (a2) Interviews

D (b) Human tissues S) (c) Neither

SUMMARY OF WORK (Use standard unreduced type Do not exceed the space provided.)

Blockade of evoked neurotransmitter release by tetanus and botulinum neurotoxins is
correlated with proteolysis of their respective intracellular substrates, VAMP/synaptobrevin,
SNAP-25 and syntaxin in intact neurons. Botulinum neurotoxin C cleaves syntaxin and in
addition, acts directly or indirectly on SNAP-25. BoNT C is the only clostridial neurotoxin
that acts on two of the three synaptic proteins that form the core complex for vesicle fusion
and is cytotoxic for both young and mature neurons. Toxin blockade of synaptic vesicle
exocytosis is coincident with a block in synaptic vesicle membrane retrieval, except with
botulinum neurotoxin A. Synaptic vesicle endocytosis, visualized by activity-dependent
uptake of FMl-43 or of horseradish peroxidase, occurs with stimulation even when vesicular
release is blocked totally by botulinum neurotoxin A.

The rank order of potency of the seven serotypes of botulinum in blocking neurotransmitter
release in spinal cord cell cultures is D>A>B>C>G>E>F. Competitive binding studies
indicate that A, B and E have different high affinity receptors.

A chimeric protein consisting of the transmembrane domain of anthrax toxin and the
catalytic domain of tetanus toxin inhibits (in the presence of the anthrax binding factor)
endocytosis and exocytosis in nonneuronal cells, suggesting that this chimeric protein affects
multiple membrane trafficking pathways.

PHS 6040 (Rev 1/84)

GPO 01 4*91 6

Z)l HD 00704-11 LDN

Project Description:

Objectives:

To analyze the presynaptic action of the clostridial neurotoxins at the biochemical and
ultrastructural levels.

Methods Employed:

Primary neuronal cell culture; morphologic techniques including Ught, video, fluorescence,
and electron microscopy; immunohistochemistry and immunoelectron microscopy; computer-
assisted image analysis and morphometry; biochemical techniques including scintillation
spectroscopy, thin-layer chromatography, and gel electrophoresis and immunoblot.

Major Findings:

Clostridial neurotoxins (CNTs) include tetanus toxin (TeNT) and the seven serotypes of
botulinum neurotoxin (BoNT). The CNTs are zinc endopeptidases which block
neurotransmission and have been shown to cleave, in vitro, specific proteins involved in
synaptic vesicle docking and/or fusion. TeNT and BoNT B, D, F and G cleave
VAMP/synaptobrevin. SNAP-25 is cleaved by BoNT A and E, and syntaxin is cleaved by
BoNT C. We have examined the alterations in toxin substrates, as visualized by
immunohistochemistry and immunoblotting, in intact neurons under conditions of toxin-
induced blockade of neurotransmitter release. In TeNT-exposed spinal cord cell cultures,
VAMP is no longer detected in synaptic terminals identified by synaptophysin
immunoreactivity. BoNT A and C have no effect on VAMP staining. In BoNT C-exposed
cultures, although syntaxin remains detectable by immunohistochemistry, immunoblots
demonstrate syntaxin cleavage but not total degradation. In BoNT A-treated cultures,
staining of the carboxyl terminus of SNAP-25 is abolished, consistent with its proteolytic
action at this site. Unexpectedly, immunostaining of the carboxyl terminus of SNAP-25 in
cultures blocked by either TeNT or BoNT C differs from controls. When VAMP is cleaved by
TeNT, staining of the carboxyl terminus of SNAP-25 is enhanced suggesting that VAMP
binds within the region of SNAP-25 that is cleaved by BoNT A. In contrast, in BoNT C-
treated cultures, SNAP-25 (CCXDH-terminus) immunoreactivity is greatly diminished and
immunoblots using antibodies against the amino terminus of SNAP-25 show two reactive
bands. Thus, in addition to cleavage of syntaxin, BoNT C has a secondary action on the
carboxyl terminus of SNAP-25. The loss of SNAP-25 immunoreactivity in BoNT C-exposed
cultures occurs over a slower time course than syntaxin cleavage. The loss of the carboxyl
terminus of SNAP-25 is not likely due to contamination of BoNT C with BoNT A since
immunoblots of each toxin show no reactivity with antibodies against the other toxin.
Furthermore, the action of BoNT C was blocked totally by preincubation with antibodies
against BoNT C, whereas preincubation with antibodies against BoNT A had no effect on
BoNT C cleavage of syntaxin or on the loss of the carboxyl terminus of SNAP-25. In
addition, the cleavage sites for BoNTs A and C on SNAP-25 appear slightly different.
Whereas, BoNT A cleaves the last 9 amino acids from the carboxyl terminus of SNAP-25, we
believe BoNT C cleaves the last 5 or 6 amino acids. This study is the first to demonstrate in

o i~

Bmn

■1"

5ifMWl^^BEHifH0Qul

Z)l HD 00704-11 LDN

vivo that the physiologic response to these toxins is correlated with the proteolysis of their
respective substrates. Furthermore, we demonstrate for the first time that BoNT C also acts,
directly or indirectly, to cleave SNAP-25. BoNT C is the only CNT that acts on two of the
three synaptic proteins that form the core complex for vesicle fusion. A paper describing
these findings has been submitted.

UnUke the other CNTs, we have discovered that BoNT C is severely cytotoxic for both young
and mature neurons. In newly-plated spinal cord cultures, neurons do not survive beyond
five days foUovsrtng treatment with BoNT C. In mature cultures, synaptic terminals become
enlarged wdthin 24 hours of toxin exposure. During the next four days, axons degenerate
and then dendrites and cell bodies are lost. Electron microscopy after two days of BoNT C
exposure reveals synaptic terminals packed with vesicular elements (100-200 nm) and
membranous dense bodies indicative of degeneration. Synaptic release sites remain relatively
intact with synaptic vesicles accumulated at most active zones. This neurotoxicity is unique
to BoNT C. AU the clostridial neurotoxins block neurotransmitter release but only BoNT C
causes overt neuronal degeneration. These observations suggest that syntaxin, alone or in
combination vdth SNAP-25, may be necessary for neuronal survival. A manuscript
describing BoNT C neurotoxicity is in preparation.

We have correlated further the synaptic blockade induced by either TeNT or BoNT A vdth
immimo-histochemistry of their respective protein substrates, with fine structure of the
synaptic release site, with activity-dependent uptake of the dye FMl-43, and wdth stimulated
uptake of horseradish peroxidase (HRP) into newly recycled synaptic vesicles. Potassium-
evoked release of both an inhibitory (glycine) and an excitatory (glutamate) neurotransmitter
is blocked completely in toxin-treated cultures. As mentioned above, immunoreactivity for
VAMP and for SNAP-25 is abolished in cultures exposed to TeNT or BoNT A, respectively.
Electron microscopy shows that approximately twice the number of synaptic vesicles are
"docked" (located within 10 nm of the membrane) at the active zones of toxin-blocked
culttires as are seen in either spontaneously active or Kcl-stimulated control cultures.
Consistent with the block in synaptic activity, TeNT- (and BoNT B-, C-, or D-) treated
culttires show no activity-dependent uptake of FMl-43. In contrast, cultures treated with
BoNT A (up to 300 ng/ml) continue to show FMl-43 loading, indicating the persistence of
synaptic vesicle endocytosis in the absence of exocytosis. These findings are confirmed by
Kcl stimulation in the presence of HRP. TeNT-blocked synapses show HRP-labehng of only
occasional synaptic vesicles, whereas a large proportion of synaptic vesicles are labeled in
BoNT A-blocked terminals. The labeled vesicles appear to form via conventional clathrin-
coated intermediate structures. These observations suggest that BoNT A may be used to
uncouple the cycle of triggered exocytosis-endocytosis that occurs at the synaptic terminal,
allowing discrete analysis of the endocytosis component. They further suggest that VAMP
(or another as yet imidenttfied substrate of tetanus toxin) participates in the process of
synaptic vesicle membrane retrieval. This work has been presented in abstract form and a
manuscript is in preparation.

While the proteolytic action of the clostridial neurotoxins on their respective substrates has
been characterized, very little is known about their receptors or internalization. We have
shown that spinal cord neurons in cell culture display a differential sensitivity to the action
of the various serotypes on inhibitory and excitatory neurotransmitter release. The rank

gjlll

MBUMinnnwi a i nomwuniBn

Z)l HD 00704-11 LDN

order of potency is D>A>B>C>G>E >F. BoNT D is as potent as TeNT. Differential
sensitivities of these culttires to BoNTs may reflect differences in receptor binding and
internalization. Competitive inhibition of [1251] BoNT A binding was examined using 100-
fold excess of tmlabeled BoNTs B or E. BoNT B offers no competitive inhibition whereas
BoNT E is 70% as effective a unlabeled BoNT A, suggesting some homology in the binding
domains of these two toxins. However, BoNT E does not inhibit the action of BoNT A in
blocking neurotransmitter release. These data indicate that BoNT E may bind to the same
low affinity ganglioside receptors as BoNT A, but does not bind to the same high affinity
(functional) receptors. Additionally, BoNT B does not bind to the same low or high affinity
receptors as BoNT A. These data further suggest that BoNTs A, B, and E have different high
affinity receptors. These studies provide the first comparative analysis of all seven serotypes
of BoNTs in an intact cell culture system and demonstrate the potential of this system for
addressing questions of toxin binding, internalization and action. This work has been
presented in abstract form and a manuscript is in preparation.

In collaboration with Drs. Naveen Arora and Stephen Leppla (National Institute of Dental
Research) and Dr. Jane Halpem (Food and Drug Administration), we have characterized a
chimeric protein consisting of anthrax lethal factor and TeNT light chain (LF-LC) (Arora et
al., 1994). The hght chain of TeNT inhibits neurotransmitter release by cleavage of VAMP. A
homologue of VAMP, cellubrevin, has been identified in nonneuronal cells and may be
important in membrane trafficking and vesicle fusion. The function of cellubrevin in
nonneuronal cells (which lack receptors for tetanus toxin) was examined using LF-LC, which
allows the uptake of TeNT light chain into cells by receptor-mediated endocytosis via anthrax
toxin receptor. We have demonstrated previously that LF-LC is cytotoxic to several
nonneuronal cell lines. In order to elucidate the mechanism of this cytotoxic effect, we have
studied the effects of LF-LC on cellubrevin levels and vesicle trafficking in Chinese Hamster
Ovary (CHO) cells. At a concentration of LF-LC that completely inhibits cell growth, there is
no consistent detectable cleavage of cellubrevin as measured by immunoblot or
immvmofluorescence. At the same concentration of LF-LC, transferrin receptor recycling and
fluid phase pinocytosis are significantly inhibited. These data suggest that LF-LC can affect
multiple membrane trafficking pathways. We are further characterizing LF-LC action under
a variety of conditions and in CHO cell variants to determine if its cytotoxicity results from
cellubrevin cleavage. This work has been presented in abstract form and a manuscript is in
preparation.

With our collaborators, Drs. Mark Dertzbaugh and Michael West (U.S. Army Medical
Research Institute of Infectious Diseases), we have examined BoNT A for segments that,
when injected into mice, are effective in stimulating the production of protective antibodies.
PCR was used to amplify and clone fragments of the BoNT A gene, resulting in ten
overlapping fragments. The peptides encoded by these fragments were expressed in E. coli
and purified. The abUity of each peptide to induce an antibody response and to protect
against a lethal challenge of BoNT A was assessed in mice and in murine spinal cord cell
cultures. All of the truncated peptides were immunogenic except one segment that encoded
amino acid residues H1078-1220 of BoNT A. Only two of the peptides were shown to confer
protection against BoNT A intoxication in mice. Both of these peptides are derived from
segments of the heavy chain, H455-661 and HI 150-1289, of BoNT A. Antibodies against each
of these peptides were premixed with BoNT A and appUed to spinal cord cell cultures.

3-^

■aniHi»wwwnwiwwnHiiiiiiiiBiiiBhMiiiiiiiiiiiiiiif>iiiiiii>iiiiiiiiii¥iiiii>iiiiiiiiiiiiiiiiMi/Miiimw

Z)l HD00704-n LDN

Blockade of evoked neurotransmitter release by BoNT A was inhibited only with antibodies
against the peptide derived from segment H455-661. These data suggest that the segment
H455-661 may be a useful candidate for incorporation into a vaccine for BoNT A. A
manuscript describing these findings has been submitted.

We have collaborated with Dr. M.A.A. Namboodiri (Georgetown University) in assisting his
graduate student, V. Sung, with studies on the ultrastructural localization of the excitotoxin
quinolate in human peripheral blood monocytes /macrophages. A manuscript describing
these findings has been submitted.

Significance to Biomedical Research and the Program of the Institute:

Tetanus is a leading cause of neonatal death worldwide and studies of its effects on young
neurons and on the possibility of recovery might have clinical implications. Botulinum
neurotoxins are used therapeutically to alleviate certain pain and to correct an increasing
number of spasmodic disorders including torticollis, spinal dystonia, and facial, voice and
limb disorders. Characterization of similarities and differences among the BoNTs provides
significant support to clinicians in terms of their understanding and controlling toxin
treatment for these disorders. We recently demonstrated that BoNT C acts on two substrates
and is cytotoxic to neurons. If corroborated by in vivo studies, these findings are critical
since it may or may not be desirable to totally eliminate a population of neurons with toxin
therapy.

Proposed Course

We will continue to use the clostridial neurotoxins to analyze the molecular mechanisms
responsible for vesicle targeting, fusion and recycling. We will further characterize BoNT C
neurotoxicity. We wall take advantage of botuUnum neurotoxin A for uncoupling synaptic
vesicle endocytosis from exocytosis to study the regulation of synaptic vesicle membrane
retrieval and the limitations of the nerve terminal plasmalemma to provide synaptic vesicle
membrane. We will continue studies of the internalization, transport, and translocation of
tetanus toxin from the endosome into the cytoplasm. Additionally, we plan to compare,
biochemically and morphologically, the internalization and transport of tetanus toxin and
botulinum neurotoxin in ventral horn-muscle co-cultures in Campenot multicompartment
chambers. These latter studies promise to provide basic information on the cell biology of
protein targeting, much as the toxins have contributed to our vmderstanding of membrane
trafficking.

We will continue collaborative efforts to characterize the proteolytic action of clostridial
neurotoxins in nonneuronal cells using fusion proteins of anthrax lethal factor and toxin light
chains. Studies have been initiated to characterize the action of a chimeric protein consisting
of BoNT A light chain and anthrax lethal factor in neuronal and nonneuronal cells.

Protocol:

94-012 Cell culture studies of neuronal development

Oc?

[CTinnii«HnBmmmnfP»*''™''»'~*g*<ifiii)«HmiiiimmTyr

Z)l HD 00704-11 LDN

Publications:

Arora N, Williamson LC, Leppla SH, Halpem JL. Cytotoxic effects of a chimeric protein
consisting of tetanus toxin light chain and anthrax toxin lethal factor in non-neuronal cells, J
BiolChem 1994;269:26165-71.

Halpem JL, Neale EA. Neurospecific binding, internalization, and retrograde axonal

transport. In: Montecucco C, ed. Clostridial neurotoxins. Curr Top Microbiol Immunol.
Berlin: Springer-Verlag, 1995, in press.

WUliamson LC, Neale EA. Bafilomycin Al inhibits the action of tetanus toxin in spinal cord
neurons in cell culture, J Neurochem 1994;63:2342-5.

X
X

nFPARTMFNT OF HFAI TH AND HUMAN SFRVICFS . PIIRI iC. HFAI TH SFRVirF

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

Z01 HD 00708-11 LDN

PEROD COVERED

October 1, 1994 through September 30, 1995

TfTLE OF PROJECT (80 chtncttn or tes*. Tnle must fit on one line between borders.)

Morphological Studies of Neuronal and Non-neuronal cells in CNS Cell Cultures

PRINCIPAL INVESTIGATOR (List other professional personnel below the Principal Investigator.) (Name, title, laboratory, and institute affiliation)

PI:
Others:

E.A. Neale
A.G. Parfitt
L.C. Williamson
L.M. Bowers

Head

Expert

Snr Staff Fellow

Biologist

LDN,NICHD
LDN,NICHD
LDN,NICHD
LDN,NICHD

COOPERATING UMTS pf '"y)

Lab. of Neurophysiol., NINDS, NIH (T.J. Smith, Jr.); Div. of Bacterial Products, FDA(W.H. Habig,
J.L. Halpern); Lab. of Cell Biology. NIMH (L. Eiden); Dept of Biochem., USUHS (S. Chung); Dept
Physiol. & Biophysics, Case Western Reserve Univ. (M. McEnery).

LAB/BRANCH

Laboratory of E)evelopmental Neurobiology

SECTKM

Section on Neurobiology

INSTITUTE AND LOCATION

NICHD, NIH, Bethesda, MD 20892

TOTAL STAFF YEARS:

0.7

PROFESSIONAL:

0.3

OTHER:

0.4

CHECK APPROPRIATE BOX<ES)

CD (a) Human subijects [H (b) Human tissuosTxl (c) Neither
M (a1) Minors
□ (a2) Interviews

SUMMARY OF WORK (Use standard unreduced type. Do not exceed the space provided.)

Instrumentation is in place to capture images for morphometry of neurons developing in the absence
of electrical activity. Redesign of multicompartment chambers for the study of activity effects on
synapse stabilization/elimination improves optical resolution and axonal growth. Antibodies
against the vesicular acetylcholine transporter label a subpopulation of neurons and synaptic
terminals, and may allow the immunohistochemical identification of cholinergic neurons as early as
one week after plating. Antibodies against the homeobox gene (HowaT) product label nem^onal and
not glial nuclei.

PHS 604O (Rev. 5/92)

r? 7

ZOl HD 00708-11 LDN

Project Description:

Objectives:

To identify classes of cells, in cultures of nervous tissue, based on cell type specific proteins,
neurotransmitters, neuropeptides, or specific receptors, in order to understand their
interrelationships during development and to examine the morphologic development of
neurons and phenomena related to neurite outgrowth and formation of early contacts with
target cells.

Methods Employed:

Primary neuronal cell cultures; extracellular electrical stimulation; intracellular injections;
morphologic techniques including light, video, fluorescence, and electron microscopy;
computer-assisted image processing and analysis; immunohistochemistry; radioautography.

Major Findings:

Preliminary studies of early neuronal development indicated that, in the absence of electrical
activity, axon outgrowth and dendrite branching patterns appear normal after one week in
culture, but that the number of synapses formed is almost twice that found in control,
electrically active preparations. Progress on this study was limited by the image quality
obtained with the available video camera. Dr. Parfitt has obtained a high sensitivity cooled
CCD camera with a dynamic range that will accommodate wide variations in fluorescence
intensity without loss of spatial detail. This camera will allow us to capture images showing
delicate, and dimly stained, structural detail, and will be used for experiments involving
either immunohistochemistry or intracellular injection.

In order to visualize clearly fluorescently labeled components of the neuromuscular junction
in living cells, we have had to develop culture conditions to meet the needs of cultures
grown on glass coversHps. We have therefore continued to experiment with surface
treatments for glass to identify substrates that are capable of supporting the growth of both
neurons and muscle cells. The most successful treatment to date comprises the sequential
application of PEDA (a proprietary amine containing sUane from SAIC Corporation) and L-
lamintn. PEDA can be replaced with poly-L-lysine with acceptable efficacy although
contracting myotubes are not as well retained, possibly because poly-L-lysine is adsorbed
rather than covalently linked to the surface of the glass and is thus less able to withstand the
mechanical stress of muscle contraction. The Campenot chamber also has been redefined to
facilitate a more luxuriant growth of axons into the field of myotubes. The compartment
sizes have been reduced and the teflon barrier section between the side wells and the center
slot has been replaced by a removable glass barrier. Substantial experimentation established
that an effective barrier comprises two thin sections of glass, each having their lower edge
ground optically flat, separated by a thin spacer. This arrangement prevents cell leakage
between compartments without blocking the outgrowth of neurites from ventral horn cells
into the field of myotubes when the barrier is removed. Since fluorescence visuaUzation will
be achieved using the newly acquired IK by IK cooled CCD camera, appropriate microscope

/>minnhiiiiMiiiiiniiiiiiiiiiHiiiiiMiiiiMyiiiiiiii>ii

ZOl HD 00708-11 LDN

and computer display modifications have been made to allow fuU chip image display on a
high resolution monitor.

Immunohistochemistry for ChAT (choline acetyltransferase) and GFAP (glial fibrillary acidic
protein) in spinal cord cultures prepared from the trisomic 16 mouse (accepted as a model for
Down's syndrome) has revealed specific deficiencies in the number of cholinergic neurons
and in the number of non-neuronal cells which stain for GFAP. This system is described by
Dr. Nelson in Project ZOl HD 00064-18.

Collaborative studies (with Dr. Lee Eiden) have characterized an antibody against the
vesicular acetylcholine transporter (a-VACh) using immimohistochemistry for comparison
with antibodies against choline acetyltransferase. In mature spinal cord cell cultures, staining
patterns with both antibodies are similar. However, it appears that VACh can be detected
several weeks prior to the appearance of ChAT. VACh immunoreactivity is present in
neuronal cell bodies as early as six days in culture. By two weeks in vitro, VACh staining is
found in varicosities presumed to be synaptic terminals, as well as in cell bodies. After three
weeks, the cultures exhibit extensive networks of immunostained varicosities. Thus, a-VACh
might provide us with an additional means of identifying cholinergic neurons and terminals.
Anti-VACh staining in spinal cord cultures during development has been included in a
review chapter submitted by Dr. Eiden.

In a collaborative study with Dr. Su Chung, we demonstrated that antibodies against the
homeobox gene, Hoxa7, stains the nuclei of neurons and not of gUa in mixed cultures. This
is a question not answered readily in tissue sections. The staining persists for several weeks
after plating. A manuscript which includes this finding has been submitted.

In collaborative studies with Dr. Maureen McEnery, we are in the process of assessing
antibodies against a number of neuron-specific proteins. Antibodies against sodium-
potassium ATPase specifically stained neurons in spinal cord cultures; immunolabel was
localized throughout the neuronal plasma membrane. Antibodies against w-conotoxin
receptor promises to be more valuable. Omega conotoxin GVIA binds to voltage-gated N-
type calcium channels. Whereas we could demonstrate substantial specific binding of w-
conotoxin in the spinal cord cultures, immunohistochemistry to localize the N-type channel
has been hampered by the lack of an affinity-purified antibody.

Sienificance to Biomedical Research and the Program of the Institute:

Ongoing electrical activity is critical for neuronal survival and for the formation and
maintenance of appropriate neuronal circuits during early development. Activity effects are
likely mediated by alterations in synaptic function. These studies, which allow the control of
activity levels and the identification of synaptic structures, should increase our understanding
of the relationship between neuronal structure and function, and provide insight into
developmental disorders which result in loss of nervous function.

Proposed Course:

The quantitative analysis of neuronal structures as affected by electrical activity wdll be

61)

X
X

■■™™»»"mHiniBmM^^— iw^i^—wiiiiiMiiiii 1 1 iiiiiiiiimiiiiinn

ZOl HD00708-n LDN

carried out using images recorded by the high resolution CCD camera. We will further
examine electrically inactive preparations with the electron microscope, with particular
attention to the fine structure of synaptic terminals that have never been active. We also will
monitor preparations treated chemically to induce an increase in electrical activity. Finally,
we will repeat these studies using cultvires prepared from ventral horn instead of whole
spinal cord, since previous data indicate that cholinergic neurons are particularly sensitive to
activity blockade early in development.

We will apply quantitative methodologies to the analysis of synaptic terminals in cultures
incubated with protease inhibitors. These experiments relate to the phenomenon of synapse
elimination and are described in detail under Dr. Nelson's Project # ZOl HD 00064-18.

In order to define the role of the N-type calcium channel in synaptic function, we wall
examine the effect of w-conotoxin on neurotransmitter release. Further, we will affinity-
purify antibodies against the w-conotoxin receptor in order to localize this calcium channel in
relation to other synaptic proteins by double-label immunohistochemistry.

We wdll continue to monitor morphologic changes that accompany synapse elimination from
multiply innervated muscle fibers. At the light microscope level, this will involve monitoring
four-color fluorescence. Innervating neurites originating from opposite sides of the modified
Campenot chamber will be stained with Dil and DiO. This will be accomplished either by
adsorbing microcrystals onto cotton or glass fibers which will be placed gently into contact
with emergent neurites, or by allowing dye crystals to settle onto the neurites that pass
through the fluid compartment between each of the glass coverslips that comprise the
removable chamber barrier. Post-synaptic cholinergic receptors wiU be labelled with Lucifer
Yellow-bungarotoxin. Presynaptic activity in neurites labelled with Dil will be monitored
using FMl-43. Synapse elimination from multiply innervated muscle fibers wUl be elicited
by unilateral electrical stimulation of neurites entering the field of myotubes. Stimulation
frequency and duration have yet to be optimized. Morphologic analysis will focus initially on
the appearance of the neuromuscular junction at two time points only, before and after the
experimentally elicited loss of functional bilateral innervation. We hope eventually to label
other postsynaptic markers (e.g., utrophin) and components of the synaptic cleft (e.g., s-
laminin, agrin) to characterize further the process of junctional dissolution. An important
extension of our understanding both of the dynamics of synapse elimination and of the
factors which elicit it will come from analysis of time-lapse video taken during the process of
elimination.

We presently are exploring the co-localization of VACh with CHAT or with synaptophysin
by double-label immunohistochemistry. Further studies are in progress to examine, under a
number of experimental circumstances, the co-expression of VACh and ChAT.

Protocol:

94-012 Cell culture studies of neuronal development

i:
c

ZOl HD00708-n LDN

Publications:

Erickson J, Weihe E, Shaefer MK-M, Neale E, Williamson L, Bonner Tl, Tao-Cheng SJ-H,
Eiden LE. The VACh/ChAT "cholinergic gene locus": new aspects of genetic and vesicular
regulation of cholinergic function. In: Cuello CA, ed. Progr Brain Res, Amsterdam:
Elsevier, in press.

'i±

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBLIC HEALTH SERVICE
NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

Z01 HD 00710-07 LDN

PERIOD COVERED

October 1, 1994 to September 30, 1995

TITLE OF PROJECT (80 characters or less Title must lit on one line between the txirders.)

Molecular Characterization of Glutamate Receptor Expression in Brain

PRINCIPAL INVESTIGATOR (List other prolessional personnel below the Pnncipal Investigator) (Name, title, laboratory, and institute affiliation)

PI:
Others:

A. Buonanno
M. Ozaki
M. Sasner
J. Cheng

Head

IRTA Fellow
IRTA Fellowr
Biologist

LDN,NICHD
LDN,NICHD
LDN,NICHD
LDN,NICHD

COOPERATING UNITS (if any)

LAB/BRANCH

Laboratory of Developmental Neurobiology

SECTION

Unit on Molecular Neurobiology

INSTrrUTE AND LOCATION

NICHD, NIH, Bethesda, Maryland 20892

TOTAL MAN-YEARS;
2.2

PROFESSIONAL:

OTHER:

0.4

CHECK APPROPRIATE 80X(ES)

D (a) Human subjects
D (a1) Minors
D (a2) Interviews

D (b) Human tissues H (c) Neither

SUMMARY OF WORK (Use starKlard unreduced type. Do not exceed the space provided.)

Activity-dependent neural plasticity during development and in the adult is largely mediated by
N-methyl-D-aspartate receptor (NR) activation; the natural ligand for this excitatory amino acid receptor
is glutamate. The effects of NMDA receptor activation on neural plasticity are largely mediated by the
calcium influx, which needs to be regulated since excessive Ca^* entry kills neurons. For this reason, the
regulation of NR channel kinetic properties has important consequences on the signals transduced by these
receptors. There are changes in NR function during development in response to activity. We have found
that the expression patterns of NR 2B repression and NR 2C activation closely follow the spatial-temporal
pattern of granule cell innervation. The down-regulation of NR 2B mRNAs, which occurs after granule
cells have migrated into the IGL where they receive afferent inputs, may result from synaptic activity.
Thus, different NR subunits may not only function to distinctly modulate synaptic connections in response
to activity, but their expression patterns may also be responsive to epigenetic factors. Transcriptional
control is a common mechanism directing cell-type specific expression of genes during development and
in the adult. In order to begin understanding the complex mechanisms that direct regional-specific
transcription of NR genes during neurogenesis and modulate their levels in response to synaptic activity,
we have begun to investigate the mechanisms that control expression of the NR2 subunits at the
transcriptional level. We have found that transcription of NR 2B transcripts is initiated from different sites,
however differential promoter use cannot account for the patterns of NR 2B expression during
development since the patterns of initiation were the same in cerebellum and forebrain. Analysis of
regulatory regions in transgenic mice reveal that distinct DNA regulatory sequences are required for the
neural-specific and developmental regulation of the gene. We have identified an 800bp upstream region
that directs neural-specific transcription in transgenic mice but fails to impart the proper developmental
down-regulation of the gene in cerebellum. Additional cis-acting sequences residing downstream of the
major initiation sites, possibly located within the first intron, were shown to be necessary to repress NR
2B expression during cerebellar development.

PHS 6040 (Rev 1/84)

GPO 91 4-9ia

Project Description

The changes in synapse efficacy are dependent on NR activity, and are accompanied by changes
in the electrophysiological and pharmacological properties of the receptor. The molecular basis
for these changes are presently unknown. However, there has been considerable progress in
understanding the basis for changes in receptor properties in the cerebellum. During
development, there is a switch in the expression of NR subunits where the 2B subunit is
expressed early in migrating granule cells but is repressed, as the cells mature, to be replaced
by the 2C subunit. This switch in subunit expression is followed by changes in the single charmel
properties of the receptor. Interestingly, in the forebrain expression of the 2B gene peaks at P14
during maximal synaptogenesis but continues to be expressed thereafter. Since the receptor
composition can influence the amount of calcium entering the ceU and the downstream signaling
that leads to neural plasticity, we are interested in understanding the molecular mechanisms that
regulate the expression of the 2B and 2C subunits during cerebellar development. In particular,
we are interested in the effects of innervation (and activity) on 23 down-regulation specifically
in cerebellum and the up-regulation of the 2C subunit. As described below, we found that the
5' non-coding region of the NR 2B gene is composed of two exons with transcription starting at
multiple sites. The down-regulation of the 2B gene can not be accounted by differential
transcription from distinct promoters, since similar utilization was observed in forebrain and
cerebellum of 7-day-old mice. Using transgenic mice, we have mapped two different sites in the
2B gene. An upstream 800 nucleotide region, including the first exon, is sufficient to direct
neural-specific expression in the brain. However, additional sequences residing in the first intron
or second exon are necessary to direct the proper developmental repression of the gene in the
cerebellum.

Objectives:

1) The aim has been to isolate and map the promoter regions of the mouse NR 2B gene, in order
to determine if the transcription initiation sites are differentially used in distinct regions of the
CNS.

2) To characterize the regulatory regions that confer neural and developmental-specific
transcription to the NR 2B gene. A special emphasis will be directed at identifying the sequences
that may differentially direct forebrain Vs cerebellum transcription during development. In order
to perform delineation analysis of the NR 2B gene regulatory regions we have used transgenic
mice.

3) To isolate and characterize the regulatory regions of the NR 2C gene to identify the regulatory
sequences that confine its expression to granule cells during cerebellar maturation.

4) To inactivate the mouse NR 2C gene locus by homologous recombination to determine the
functional role of this receptor during cerebellar development.

Methods:

In situ hybridization histochemistry: Newborn and post-natal rats were used for the
experiments. The arumals were sacrificed, and brain and cerebella were rapidly dissected and
freeze-embedded in Tissue-Tek media using dry ice. Sections on gelatin-coated slides were fixed
in paraformaldehyde, acetylated in acidified triethanolamine, and delipidated in an ethanol series
and chloroform. Oligonucleotides were end-tailed with terminal transferase using standard

4<3

ZOl HD 00710-07 LDN

methods. One hundred microHters of the probe (1.5x106 cpm) were pipetted onto 6-8 sections
per slide, covered with parafihn strips and incubated in humidified dishes at 37°C for 18-20
hours. Slides were washed at high stringency, dried, exposed to film for one week, and then
dipped in emulsion and stored in the dark for 4-5 weeks at 4°C. They were then processed,
stained in 0.2% methyl green, dehydrated and cover slipped.

Cloning and characterization of NMDA receptor genes; Cloning of the mouse NR 2B gene were
described in last year's report. The NR 2C gene was isolated from the same 1 DASH-1 library.
The genomic inserts were mapped on Southern blots using specific oligonucleotides. DNA
sequence of subcloned fragments was obtained by the Sanger dideoxy-termination technique
using oligonucleotides as primers. The technique of Rapid Amplification of cDNA Ends,
designated RACE, and RNase protection assays were used to map the transcription initiation
sites of the 2B gene. RNase protection was performed on forebrain and cerebellar RNA isolated
from 129/SV mice of different ages to map transcription initiation sites and to quantitate the
relative contributions of the different initiation sites. Because the results of RACE cloning
suggested that NR 2B gene transcription initiates at multiple sites over a large area, two probes
were used to identify start sites. Probes were transcribed with T7 RNA polymerase and a-32P
UTP (800Ci/mmol, NEN). Conditions were optimized and checked for linearity of assay by
varying amount of RNA used, concentration of RNase(s), and comparing RNase A+Tl mix with
RNase Tl alone using unleveled sense riboprobes as positive controls. Quantitation was
performed on a Molecular Dynamics phosphorimager system.

Generation of transgenic mice and analyses: Transgenic mice were generated by injection of
(C57BL/6 X SJDFl or FVB/N pronuclei with linearized DNA containing mouse 2B and 2C gene
upstream sequences linked either to the CAT or 6-gal reporter genes. The construct 0.8-CAT is
a 806bp region extending from the Hindlll to the SacII site which contains the first exon and
upstream sequence. 0.8int-CAT harbors a fragment between the same Hindlll site and the end
of the second exon. The region -552 to +255 (0.8-CAT) and the region -552 to +1631 (0.8int-CAT)
were cloned into the Hindlll-Xbal sites upstream of the CAT gene in the pCAT-Basic vector
(Promega). Three CAT-expressing lines (#6184, 6188, and 6190) for the 0.8-CAT construct and
two CAT-expressing lines (#6944, 6948) for the 0.8int-CAT construct were characterized. Fl
animals were tested for incorporation of the transgene by slot blotting DNA prepared from tail
clips. Slot blots were also used to determine transgene copy number. The tissue-specificity
conferred by the regulatory sequences was determined by assaying for CAT activity in extracts
made from different tissues or staiiung for S-gal activity using X-gal as the chromophore. A time
course was performed to show that these conditions are in the linear range of the assay.
Quantitation was performed on a Molecular Dynamics phosphorimager system.

Major Findings

Although there has been considerable progress in understanding the molecular mechanisms
regulating tissue and developmental expression of genes in non-neuronal cells, this advancement
has not been significantly evident in the nervous system. The bewildering complexity of neural
and glial cell types in the nervous system, the intricate connections between cells and the added
disadvantage that genetic material is difficult to introduce into the post-mitotic neurons, greatly

ZOl HD 00710-07 LDN

accounts for the lack of systems suitable for studying genetic regulation in neurons. The cloning
of cDNAs coding for NR subunits represented the first step in understanding the molecular
composition of native receptors. Recent studies strongly support the notion that
hetero-oligomerization, between the widely expressed NMDARl subunits and members of the
NR 2A-D family, is necessary to form receptors with appropriate electrophysiological properties.
The composition of native receptors largely depends on the regional and temporal expression
of the NR2 subunits. The restricted expression of NR2 subunits may contribute to the changes
in receptor function associated with modulation of neural function during development and in
the adult. Considering that the various combinations of NR2 subunits impart different channel
properties and modulate the extent of magnesium block to the receptor, comprehending how
NR2 subunit expression is developmentally regulated would contribute to our understanding
of the molecular mechanisms underlying neural plasticity. Our goal is to elucidate the molecular
mechanisms regtilating NR2 gene expression.

Regulation of NR channel properties during development may play an important role in
modulating downstream events that occur after channel activation and the entry of Ca^* into the
cell. Thus, the kinetic properties of the channel and its regulation play an important functional
role during development and in the adult. The duration of NR-mediated currents have been
shown to decrease during development in the superior colliculus and layer IV neurons in the
visual cortex. Dark-rearing interferes with the time-course of receptor change, indicating that
activity signals these changes. However, in these cases the molecular basis for the changes in
receptor properties have not yet been identified, but could include post-translational
modifications, differential splicing of subunits or the differential control of genes coding for NR
subunits. Given that these changes in receptor properties are long-lived, the changes in gene
transcription would seem a plausible mechanism. Perhaps the best case supporting changes in
subunit gene expression can be made for the developmental switch that occurs in cerebellar
granule cells. The kinetic properties of NRs in cerebellar granule cells change during
development; migrating granule cells (prior to innervation) express receptors that have higher
conductance, longer mean open channel times and burst lengths than the post-migratory granule
cells residing in the IGL of P22 rats (which probably have received afferent inputs). Granule
cells present in the EGL and those that have migrated into the IGL express NRl and NR 2B,
whereas by PI 4 most granule have begun to express the NR 2C transcripts and down-regulate
NR 2B expression. The single channel properties of the pre-migratory and maturing granule cells
closely ressemble those of in vitro synthesize dimeric receptors NRl/NR 2B and NRl/NR 2C,
respectively . As in the case of the visual system, the NR-mediated currents in the pre-migratory
receptors on granule cells are longer. Assuming that the properties of these receptors are
manifested at synapses, the earlier receptors could function to accomodate inputs that have less
coincident activity. A switch to receptors with lower conductance and/or mean open channel
times during development would presumably lower Ca^* influx. The change in NR subunit
expression is reminiscent of the t to d switch that occurs at the neuromuscular junction. In this
case, activity plays an important role in down-regulating transcription of the x subunit gene. As
discussed below, we and others have evidence that the NR 25 gene is also down-regulated by
innervation. (Dur work is presently focused in identifying the regulatory interactions and DNA
sequences that mediate the NR 25 to NR 2C switch in the developing cerebellum, and that lead
to the changes in NMDA receptor function.

: U

ZOl HD 00710-07 LDN

Gradients of NMDA receptor expression in developing cerebellar granule cells. In situ
hybridization histochemistry was used by Dr. Buonanno to analyze the developmental regulation
of NR2 subunits in the developing cerebellum. This analysis reveals that the expression of NR
2B and 2C transcripts during development differs markedly in the external and internal granule
cells layers (EGL and IGL, respectively). Expression of 23 transcripts is observed in
pre-migratory granule cells in the EGL and after migration into the IGL. In contrast, 2C
expression is restricted to the IGL. Furthermore, the NR 2C gene is expressed in a gradient
during development. Begin at P7 expression progresses from the posterior to the anterior lobes,
and within a lobe transcripts accumulate in a medial to lateral pattern. This gradient cannot be
simply accounted by the migratory pattern of granule cells, since the expression of NRl and 2B
in the IGL significantly precedes 2C expression. Similar to the acetylcholine receptor x to subimit
switch that occurs during skeletal muscle development, our results suggest that the initial NR
2B expression in granule neurons is cell-autonomous whereas innervation may provide the
signals that activate NR 2C expression and repress NR 2B. In order to decipher the molecular
mecharusms directing the 2B to 2C switch, we have generated transgenic mouse lines harboring
different regulatory regions of these genes. The results are described below.

Multiple transcription start sites are used by the NR 2B gene. A transcriptionally regulatory
mechanism frequently utilized for the differential expression of a gene in various tissues or at
distinct times in development, is to initiate transcription from different promoters. Since
expression of the NR 2B gene is differentially regulated in the brain cortex and cerebellum of
1 -week-old and adult rodents, we were interested in mapping transcription initiation sites using
the two independent methods of RNase protection and RACE . The cDNA products obtained
by RACE originate from using an "anchored primer" located downstream of the translation
initiation site and extending to the 5'-most end of the RNA. Dr. Mike Sasner found that discrete
products of different sizes were obtained by RACE and that these originated from NR 2B
transcripts of different lengths. He used RNase protection as a second independent method to
map the transcription initiation sites. Two probes were used for the RNase protection assays,
hybridization products from both probes show multiple bands, indicative of multiple
transcription start sites. Caution was taken to circumvent conditions that may result in the
generation of false RNase protected bands due to "probe breathing", frequently seen with probes
containing stretches of adenosines and thymidines, which could be misinterpreted as
representing initiation sites.

Genes containing multiple transcription sites frequently alternate their use during in different
tissues or during development. However, we foimd no evidence for the differential use of
promoters in the NR 2B gene during development, since the patterns of protected products were
similar when forebrain or cerebellar RNAs from 1 -week-old and adult mice were used. From
the calculated sizes of the protected products, transcription start sites were mapped to the
genomic sequence. The sites obtained by RNase protection and RACE mapped to similar
positions in the gene. No consensus TATA or CAAT boxes are evident upstream of the
transcription start sites, but consensus initiator elements are present. Given that there are neither
multiple genes nor differential use of promoters that can account for the distinct expression of
NR 2B mRNAs in the cerebellum and forebrain during development, next we investigated if the
cloned genomic DNA had the potential to properly regulate transcription of the bacterial reporter
gene chloramphenicol acetyl transferase (CAT) when driven by NR 2B upstream sequences.

ZOl HD 00710-07 LDN

A 0.8kB upstream region is sufficient to confer tissue-specificity to a reporter construct.
Two CAT reporter constructs with identical 5' ends were used to assay transcriptional regulation.
The first, 0.8-CAT, harbors a fragment that extends from 550 bp upstream of the 5'-most start
site down to the first non-coding exon (see Methods). The second, 0.8int-CAT, originates from
the same upstream site and extends to the second non-coding exon; it was used to test the
potential role of the first intron and second exon in regulating transcription. Both constructs
conferred reporter activity specifically in the brain and failed to be expressed in the other 5
non-neural tissues tested, including excitable tissues such as cardiac and skeletal muscle.
Analysis of the same tissues from one-week old mice also showed that CAT expression was
restricted specifically to the brain. Non- transgenic littermates displayed no detectable reporter
levels in any of the tissues tested. Interestingly, expression driven by the upstream promoter
region alone was much higher than that driven by constructs containing sequences downstream
of the first exon; a situation that may result from the presence of repressor sequences in the
O.Sint-CAT construct. The number of integrated copies did not correlate with the levels of CAT
expression. These results indicate that regulatory cis-acting elements residing in the upstream
550 bp or in the first exon are sufficient to speciJFically direct NR 2B expression in the brain.

Downstream sequence is necessary to confer developmental down-regulation of the NR 2B gene
in the cerebellum. Dr. Sasner quantitated the levels of CAT activity in extracts made from the
cortex and cerebellum of transgenic mice of different ages, to test if these sequences were also
sufficient to confer proper developmental regulation. Interestingly, he found that although mice
harboring the 0.8 construct expressed CAT specifically in the brain, these sequences failed to
repress expression of the reporter in the cerebellum of adult mice. Moreover, levels of CAT
activity in mice from 3 independent 0.8-CAT lines were higher in the 1 -week-old and adult
cerebellum than in the cortex. In stark contrast, expression driven by the O.Sint-CAT construct
was down-regulated in the cerebellum after 1 week of age. Both of the O.Sint-CAT lines had
negligible levels of CAT expression in the cerebellum as adults, although there was significant
cerebellar expression at one week of age, as occurs with the endogenous NR 2B gene. To test
the time course of cerebellar down-regulation of the reporter, tissue from 10-day old and 2-week
old animals were assayed. Cerebellar expression in the 10-day old was decreased relative to
7-day and was lost in 2-week old animals; again, this is similar to the developmental
down-regulation of the endogenous gene. These results demonstrate that sequences in the first
intron and /or exon are necessary to recapitulate the proper pattern of NR 2B expression. The
fact that CAT reporter levels in mice harboring the O.Sint-CAT construct are considerably lower
than those obtained with the intron-less construct suggest that this region may contain important
repressive elements.

In simimary, we have identified a genomic region sufficient to confer tissue specific expression
to a reporter gene in transgenic mice, and a distinct intronic region that is responsible for
developmental dowTi-regulation in the cerebellum. Sequence elements have been identified in
these regions that have previously been shown to bind brain-specific transcription factors.
Expression of the reporter gene driven by the larger of the constructs described shows the same
temporal and spatial expression as the endogenous gene. These constructs will allow a detailed
investigation of the molecular mechanisms that control both the tissue-specificity of the gene and
the developmental down-regulation of the NR 2B gene in the cerebellum. This will include
analysis of the potential role synaptogenesis and/or neuronal activity may play in the control
of subunit composition of the receptor during development. (A manuscript describing this work

BBKBBHSSiSa

Cii

ZOl HD 00710-07 LDN

has been submitted for publication).

Characterization of the NMDA receptor NR 2C gene. As disucussed above, there is a NR 2B to
NR 2C switch during maturation of granule cells as they receive afferent inputs from ascending
mossy fibers. We are interested in developing molecular markers to analyze how the NR 2B
gene is repressed by granule cell innervation, and replaced by expression of the NR 2C. To this
end. Dr. Miwako Ozaki, who recently joined the laboratory, has cloned and characterized the
mouse NR 2C gene. The upstrem region of the NR 2C gene is extremely complex. Initial results
by Dr. Sasner had shown that splicing at the 5' noncoding sequence of the NR 2C mRNA was
complex; recent results show that there are four 5' non-coding exons that can be differentially
spliced. Transgenic lines harboring fi-gal constructs driven by different 5'-falnking sequences
have been generated. Preliminary resutls demonstrate that more than the 5'-flanking six
kilobases are required for the expression of the NR 2C gene in cerebellum. Experiments are in
progress to identify the sequences that direct the cerebellar expression of the NR 2C gene during
granule cell maturation.

Significance to Biomedical Research and the Program of the Institute:

Glutamate receptors are known to mediate excitatory postsynaptic transmission in the brain, and
are the most abundant excitatory receptors in the vertebrate CNS. Glutamate and its analogs are
potent neurotoxins. In addition, experimental animal studies have suggested that GluRs maybe
implicated in some degenerative phenomena in the CNS, such as Huntington's disease. In fact,
intrastriatal injections of kainate produce a pattern of neuronal degeneration and biochemical
changes similar to those observed in postmortem tissue from Huntington's disease patients. An
important step in understanding the basis of neurotoxicity, and the expression of GluRs during
development, is to identify the genes coding for these receptors. The analysis of how GluRs are
expressed and regiilated during cerebellar development at a genetic level will set some of the
foundations for the understanding of the developmental biology of glutamatergic synapses; sites
for the action of several psychoactive drugs.

On the other hand, the delineation of regulatory sequences that temporally and regionally direct
gene expression in the brain is of extreme importance for their future use in gene therapy.
Presently, there is a notable absence of characterized promoters directing neural expression in
specific regions of the brain. Current experiments on the use of gene therapy in the CNS
generally utilize expression coristructs that are driven by viral promoters; these promoters have
the disadvantage that they lack specificity. The present approaches available to introduce the
expression constructs into the CNS rely on the stereotaxic delivery of recombinant viral vectors
(i.e. herpes simplex virus or adenovirus) or genetically modified neuronal cells harboring the
expression constructs; the former strategy has the advantage of avoiding cellular immunological
responses. However, the level of specificity provided stereotaxic injections of either viruses or
cells is minimal. The use of developmental- and regional-specific neural promoters in the
expression constructs would greatly improve the specificity of these approaches. An obvious
requirement in the use of neural-specific promoters to target expression is that the viral vectors
have a size limitation. For this reason, it will be imperative to delineate the shortest DNA
regulatory sequences providing specificity and quantitative expression. Our work on the
characterization of NR regulatory sequences provide an initial step toward this goal.

Ci!

fissBaaaifflimniffliuiflytiBiaais

ZOl HD 00710-07 LDN

Proposed Course:

Dr. Sasner will continue to analyze the regulatory sequences that impart neural- and
developmental-specific expression of the NR 2B gene in cultured cells and transgenic mice. We
are presently generating transgenic lines that harbor constructs containing deletions in the NR
2B gene in order to locate the cis-elements conferring down-regulation of the gene during
development. In collaboration with Dr. P. Nelson's group we vdll determine if NR transcription
is regulated by electrical activity, if so, we wUl begin mapping the cis-acting elements conferring
activity-dependence.

Drs. Ozaki , Buonanno and Ms. Jen Chung are working to understand what aspects of granule
cell maturation are necessary for the down-regulation of NR 2B and the up-regulation of the NR
2C gene. Dr. Ozaki will continue to identify the sequences that activate the NR 2C gene during
innervation.

Protocols:

Animal:

93-029

Buonanno

93-030

Buonanno

Publications:

Targeted DNA delivery into tissues
Production of transgenic mice

None

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBLIC HEALTH SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

Z01 HD 00711-06 LDN

PERIOD COVERED

October 1. 1994 to September 30. 1995

TITLE OF PROJECT (80 charactan or less. Title must lit on one line Oetween the borders.)

Transcriotipnal Regulation of Skeletal Muscle-Specific Genes by Electrical Activity

PRINCIPAL irfVESTIGATOR (t/st other prolassional personnel be/ow the Prhapal Investigator.) (Name, Vtle. laboratory, and instituti

PI:

A. Buonanno

Others:

S. Calvo

M.Nakayama

J. Stauffer

J. Cheng

P. Roseboom

Head

Adjunct Scientist

IRTA

Adjunct ScientistVIRTA

Biologist

Staff Fellow

I institute attillation)

LDN, NICHD
LDN, NICHD
LDN, NICHD
LDN, NICHD
LDN, NICHD
LDN, NICHD

COOPERATING UNITS (it any)

University of Aachem, Germany (J. Weis)

Laboratory of Developmental Neurobiology

SECTION

Unit on Molecular Neurobiology

INSTITUTE AND LOCATION

NICHD. NIH. Bethesda. Maryland 20982

TOTAL MAN-YEARS:

2.0

PROFESSIONAL:

1.6

OTHER:

0.4

CHECK APPROPRIATE BOX(ES)

n (a) Human subjects
D (a1) Minors
n (a2) Interviews

D (b) Human tissues H (c) Neither

SUMMARY OF WORK (Use star>dard unreduced type. Do not exceed the spaca provided.)

Skeletal muscle diversity and plasticity are dramatically influenced by motoneuron depolarization;
nerve-derived electrical activity regulates the contractile properties of slow- and fast-twitch muscles by
regulating transcription of spjecific genes. We have used transgenic mice to identify regulatory
sequences that direct the fiber-specific transcription of the slow and fast troponin I (Tnl) genes which
are specifically regulated by selective depolarization frequencies. Mice harboring different
chloramphenicol acetyltransferase (CAT) reporter constructs driven by progressive deletions of the rat
Tnl slow gene revealed that a 128 bp enhancer was sufficient to direct specific expression in slow
muscle when linked to the 500bp promoter sequence; mice harboring just the SOObp Tnl slow promoter
region failed to show transcriptional activity in any tissue. A 144 bp enhancer from the quail Tnl fast
gene was found to confer transcription specifically in fast muscles when linked to the same SOObp
promoter from the Tnl slow gene. Interestingly, the rat Tnl slow and quail fast fiber-specific enhancers
share similar core elements. These results suggest that related trans-acting factors, or the formation of
higher-order transcriptional complexes, are used to respond to selective patterns of muscle
depolarization. On the other hand, we have identified sequences in the myogenin promoter that cause
transcriptional repression in response to activity. Analyses in transgenic mice and muscles injected
with myoblasts engineered to express the CAT reporter have revealed a region of approximately O.Skb
involved in the down-regulation of myogenin expression in response to innervation. Factors expressed
specifically in denervated muscle were found to bind a 30bp element residing in the O.Skb fragment.
Ets factors have Ijeen implicated in coupling extrinsic signals to transcription. A novel ets-related
transcription factor, PEF, was cloned and characterized. We found that PEP interacts synergistically
with MyoD and myogenin to enhance transcription of muscle genes. Cotransfection with MEF2A
showed no cooperativity with PEF. Our results suggest the possible combinatorial interaction lietween
members of the ETS and myogenic BHLH families may mutually contribute to the stringency of
binding site selection by these two families resulting in the differential modulation of various muscle
specific genes.

PHS 6040 (Rev 1/84)

GPO SI 4-9ia

WaSBBBBBBISBm^^l^

ZOl HD 00711-06 LDN

Project Description:
Objectives:

1) To identify the transcription regulatory sequences in the slow and fast troponin I (Tnl) genes
that confine gene expression to slow- or fast-twitch muscle in response to specific frequencies of
nerve-derived electrical activity. Analysis was performed in muscles injected with DNA constructs
and transgenic mice, since cultured muscle cells do not manifest fiber-type specific properties.

2) To delineate sequences in the myogenin promoter that confer innervation-dependent regulation
in skeletal muscle, using transgenic mice and mature muscles injected with stably transfected
myoblasts.

3) To analyze if the myogenic factor MRF-4 is also regulated by denervation.

4) To characterize the function of PEF, an ets-related factor that we isolated from muscle. The
interaction of PEF with the skeletal muscle basic helix-loop-helix factors was analyzed.

5) To identify the trans-acting factors that mediated regulation by electrical activity using the
delineated cis-acting sequences as probes to screen expression libraries.

Methods:

Characterization of Tnl regulatory sites in transgenic mice: The cloning and initial characterization
of the rat Tnis gene was described in our previous Annual Report. Tnis sequences extending from
2.7 kb upstream of the transcription initiation site (position +1) down to position 4-50 located in
the first non-coding exon were sequenced. Nested deletions in sequences residing in the upstream
region were generated from both directions using ExoIII and SI nuclease. DNA sequencing was
performed using standard di-deoxynucleotide termination reactions. GeneWorks software
(IntelliGenetics, Inc.) was used for the sequence and alignment analysis.

Transgenic mice were generated essentially as described by Hogan et al. [Hogan, Beddington,
Constantini and Lacy (1994) Manipulating the Mouse Embryo, Cold Spring Harbor Press].
Transgenic mice were generated in (C57B1/6 X SJL) Fl or FVB/N embryos. A series of Tnl slow
constructs containing different amounts of the upstream sequences were used to generate transgenic
lines. Putative founders and their generations were screened by slot blot analysis of tail DNA using
a CAT probe. Adult transgenic mice were used to analyze tissue- and muscle-type-specific
expression of CAT activity. Tissue-specificity was determined by measuring levels of CAT activity
in extracts made from tissues that either express or do not express the endogenous Tnl slow gene.
In order to further determine if fiber-type-specific expression was conferred by the constructs,
muscle groups known to contain different proportions of fast and slow myofibers were analyzed
for CAT activity.

Characterization of myogenin gene transcription regulatory elements in muscle: The generation of
transgenic mice harboring chloramphenicol acetyltransferase (CAT) expression vectors driven by
the mouse myogenin gene upstream regulatory sequences was described on the previous Annual
Report. We analyzed mice harboring 3.7, 1.5 and l.Okb of upstream sequence. The tissue,

D^
D^
>'

ZOl HD 00711-06 LDN

developmental and innervation specificity conferred by the regulatory sequences was determined by
assaying for either CAT transcripts or enzyme activity. We analyzed for innervation responsive
elements in the construct by comparmg CAT mRNA levels in denervated and innervated hind limb
muscles; these were denervated by cutting the sciatic nerve at the upper thigh. The relative CAT
mRNA levels were quantitated by RNase protection and Northern blots using the Molecular
Dynamics phosphorimager.

Implantation of transfected myoblasts into mouse muscle: Myoblasts containing the stably
transfected myogenin-CAT constructs were injected directly into both gastrocnemius muscles of
adult mice, one leg was then denervated by transection of the sciatic nerve, and two weeks
post-injection both muscles were collected and assayed for CAT activity. As controls, we injected
other mice with myoblast stably transfected with CAT expression vectors driven by either the
rLAChR a subunit or myosin light chain enhancers. The fi-galactosidase activity was used to
normalize for viable injected cells. The CAT assays were performed with muscle extracts containing
equal amounts of protein or expressing the same fS-gal activity (these numbers were similar).

Cloning and analysis of muscle ets transcription factors: A cDNA library was constructed using
RNA isolated from rat extensor digitorium longus (EDL) muscle, which in rat is composed
predominantly of Type II fibers (>95%). An oligonucleotide, corresponding to the conserved ets
domain, was used to screen the soleus muscle cDNA library at moderate stringency. The screen
yielded approximately 30 positive clones, originating from 8 genes, that coded for ets-related factors.
One of the cDNAs coded for a unique and novel protein which we called PEF. In order to study
PEF function in transfected cells, the cDNA was cloned into the expression vector pSVT7, which
contains the SV40 promoter, enhancer, and poly-adenylation sites. Rat myogenm and MyoD were
expressed from the plasmid EMSV. The plasmid MLC1CAT920 consists of about 1.5 kb of the
myosin light chain 1 promoter region driving a CAT reporter flanked 3' by a 920 bp enhancer
region (a gift from N. Rosenthal) and the plasmid -UOMCK consists of the 110 bp enhancer region
of the muscle creatine kinase gene fused to -81 nt upstream of the MCK start site (a gift from H.
Weintraub). The reporter plasmid 4X PEA3 and 4X mB contain 4 copies of a wild type ETS
binding site or a double-point mutation that blocks ETS binding, respectively, adjacent to a TATA
box (a gift from J. Hassell). HeLa and COS7 cells were transfected with these vectors using the
standard calcium phosphate coprecipitation method. After 16-20 hours the media was removed and
the plates were washed two times each with 3 ml PBS and were then refed with 5 ml of growth
media. The cultures were allowed to grow an additional 48 hours before harvesting for CAT
assays.

Major Findings:

Plasticity of the skeletal muscle and neuronal phenotype results from the selective repression and
activation of gene expression in response to innervation patterns. In embryonic myotubes, the
distribution of receptors and the types of contractile protein isoforms expressed change during
development with motoneuron innervation. A selective distribution of nicotinic acetylcholine
receptors (nAChRs) is achieved after innervation by the repression of receptor expression in the
extra-junctional regions of the fiber, and the accumulation of "adult-type" receptors at the
neuromuscular junction. Recently, we and others have shown that a family of muscle-specific
transcription factors, which include MyoD and myogenin, regulate the expression of nAChRs.
Motoneurons, eliciting different patterns of depolarization, regulate the contractile properties of the

ii»>iii>iiiii»iiiiiyiiiitiiiiiiiiiiiiyiiiiiiiwiiiiiiiiiiiuiiii»'ihiiftMiimwiiiii)iiiiiiiyiiiit^

ZOl HD 00711-06 LDN

myofibers they innervate by selectively activating expression of specific fiber-type (fast vs slow)
contractile proteins. The plasticity of the muscle properties are attributed to innervation, because
removal of the nerve results in the re-expression of nAChR in extrajunctional regions of the fiber
and the replacement of "adult-type" myofibril proteins by their fetal counterparts; muscle re-attains
its adult properties when re-innervated. Furthermore, it was elegantly demonstrated by Eccles in the
1960's that a fast-twitch muscle adopts slow-twitch properties when it is re-innervated ectopically by
a nerve that normally innervates a slow muscle, and vice-versa. Conceivably, the nerve could
regulate muscle properties by either the release of chemical factors and/or the patterns of
depolarization it elicits. Recent experiments have shown that the electrical stimulation of
denervated adult muscle, with frequency and patterns that mimic the natural depolarization of slow
and fast muscle, are sufficient to down-regulate the expression of nAChR extrajunctional receptors
and modify contractile properties of muscle. These experiments suggest that there should be at least
two innervation-dependent regulatory cascades that differentially modulate muscle fiber-type. The
molecular mechanisms underlying the regional and fiber-type-specific expression of genes during
innervation are due, in large part, to the selective repression and stimulation of transcription. The
long-term goals of our laboratory are to elucidate the molecular mechanisms that respond to
innervation either by restricting gene expression to confined regions of the myofibril through
transcriptional repression, or by stimulating transcription differentially in response to selective
frequencies of electrical stimuli. Towards these goals, we have chosen as models to study the
fiber-type-specific activation of Tnl genes in either slow- or fast-twitch muscle, and the repression of
myogenin and MRF-4 transcription by activity.

Common core elements are shared by the Tnl enhancers conferring specific expression in either
slow and fast muscles: We have used the regulation of the Tnl genes as a model to elucidate the
mechanisms that generate fiber diversification in response to epigenetic factors. Three Tnl genes
code for the isoforms that in the adult are specifically expressed in slow (TnQ, fast (Tnl^ and
cardiac (TnlJ muscles. The troponins expressed in skeletal muscle form part of a complex, in
combination with troponins C and T, that is involved in the regulation of acto-myosin
calcium-mediated interactions during contraction. Transcription of Tnl genes is activated during the
myoblast terminal differentiation. During muscle maturation, the expression of the two Tnl
isoforms is confined to either slow or fast muscles in response to the type of moto-innervation the
myofibers receive. As we previously demonstrated, specific patterns of electrical activity that mimic
the endogenous frequencies of fast (100 Hz) and slow (10 Hz) nerve regulate Tnl gene expression
selectively. Because fiber-specification is not observed in cultured mammalian muscle, Dr. Manabu
Nakayama, Dr. Jim Stauffer and Ms. Jun Cheng have used transgenic mice to identify regulatory
sequences that direct the fiber-specific transcription of the slow and fast Tnl genes. Multiple lines of
transgenic mice were generated that harbored reporter constructs driven by systematic deletions of
the upstream regions of the rat Tnis gene. Mice harboring chloramphenicol acetyltransferase (CAT)
reporter constructs, driven by approximately 1900 and 950 bp of the rat Tnl slow gene upstream
sequence, are transcribed specifically in slow muscle, but not in fast muscles or other tissues such as
heart, liver or brain. In contrast, mice harboring constructs containing the upstream Tnl slow 250
or 500 bp failed to be expressed in embryonic or adult skeletal muscle. Fiber-type-specific
transcription in slow muscle was restored when a 128 bp enhancer, that is located between the 950
and 500 bp upstream region of the gene, was fused to the 500 bp fragment. In order to test if the
transcription specificity is conferred by the 128 bp enhancer and not the 500 bp region, we
generated transgenic mice with a construct containing the 500 bp promoter fused to an intronic
regulatory region derived from the quail Tn I fast gene. This 144 bp sequence was previously

■DBBSI

ZOl HD 00711-06 LDN

shown to function as an enhancer in differentiated cultured myotubes. Mice harboring the chimeric
Tnl fast/Tnl slow CAT construct specifically expressed the reporter gene in fast but not in slow
muscles, indicating that these 2 enhancers selectively direct fiber-type-specific transcription when
fused to a common Tnl slow 500 bp promoter region. Alignment of the rat Tnis and quail Tnlf
sequences showed that the 2 enhancers share similar core elements, namely a MEF-2 site, a
CCAC box, an E box, and a previously uncharacterized motif. These results suggest that
related trans-acting factors, or that higher order complexes, are necessary to generate
muscle diversity. (A manuscript describing this work has been submitted for publication).

Characterization of the myogenin gene regulatory sequences: Our previous work has strongly
suggested a causal link of nAChR gene regulation by the MyoD-related factors, specially the
innervation-dependent transcriptional regulation of receptor genes by myogenin at
extra-junctional nuclei (see Previous Reports). To elucidate the molecular pathways linking
the effects of muscle depolarization with the repression of gene transcription we have begun
to characterize the regulatory elements in the myogenin gene, since its regulation by
innervation is more proximal. To this end, transgenic mice harboring 3.7, 1.5 and l.Okb of
myogenin upstream sequence driving the expression of the CAT reporter gene were
generated and analyzed by Dr. Soledad Calvo, Dr. Andres Buonanno and Ms. Jun Cheng.
These myogenin upstream sequences were found to confer muscle-specific and to
down-regulate reporter levels during innervation. The down-regulation of the gene was
shown to result from the repressive effects of the nerve, and not simply occur as part of a
developmental program, because denervation of transgenic mice increased reporter mRNA
levels by at least 10-fold. We have implanted myoblasts stably transfected with
myogenin\CAT reporter constructs into adult mouse muscle to further delineate the myogenin
gene denervation-responsive region. The constructs present in these "chimeric muscles" are
then tested for their ability to respond to innervation, by comparing the levels of reporter in
the denervated vs the innervated muscles. Using this method. Dr. Gibney previously
uncovered a region conferring denervation-responsiveness that resided separately from that
imparting muscle-specificity. In collaboration with Dr. Pat Roseboom, band-shift assay s
were used to analyze regulatory elements residing in the 0.5kb region that imparts the
denervation response. Extracts from denervated muscle, but not from innervated muscle,
contain factors that interact with a 30bp element residing in the 0.5kb fragment.

MRF-4 is expressed by myotubes and not satellite cells after denervation. In collaboration
with Dr. Joaquim Weis, we have analyzed the expression of MRF-4 protein the innervated
and denervated muscles of adult rats. In contrast to the other myogenic factors, transcripts
coding for MRF-4 accumulate in skeletal muscle during maturation. Paradoxically,
denervation causes a further accumulation of MRF-4 mRNAs. In order to determine if the
accumulation in innervated muscles occurs selectively in different muscle types and if
denervation causes expression of MRF-4 in myofibrils or satellite cells, we developed a
specific antibody to the factor and analyzed its expression by immunohistochemistry.
Western blotting and fluorescence immunohistochemistry on fibroblasts transfected separately
with mammalian expression vectors for the 4 myogenic factors showed that the MRF-4

D'i

^B^BOBmsBmm^mn

aM«iMlt»JIKtlilHHIiii|l|i|IIIIIIIIIMIIIilllllllKi(ii

ZOl HD 00711-06 LDN

antisera was specific. No selective accumulation of this transcription factor was observed in
specific fiber-types. Electron microscopy of sections of adult hemi-denervated rat diaphragm
demonstrated that MRF-4 accumulates in the nuclei of denervated fibers and not in satellite
cells; results that are consistent with our prior Northern blot time-course analysis of
denervated muscles. Thus, we have shown that MRF-4 transcripts and protein accumulate in
the myofibrils of denervated muscle and would suggest that this factor, like MyoD and
myogenin, play a role in the re-activation of selective genes normally expressed in embryonic
muscle. (A manuscript describing this work has been submitted for publication).

Identification of trans-acting factors that interact with MyoD-related factors to synergistically
increase transcription of skeletal muscle-specific enhancers: As discussed above, experiments
performed in transgenic mice have demonstrated that transcriptional regulation plays an
essential role in confining myofibril expression to specific fiber types and conveying
responsiveness to innervation. The ets family of transcription factors, many of which are
involved in signal transduction, are recruited to the DNA template by interactions with other
transcription factors to form higher order complexes. Dr. Jimmy Stauffer has isolated and
characterized a novel ets-related factor that is similar to PEA-3 and ER81; it was named
PEP for PEA-3\ER81-like factor. The full-length sequence of the PEE cDNA was obtained
by the rapid amplification of cDNA ends (RACE). Analyses of PEE expression on Northern
blots containing mRNA isolated from a repertoire of tissues, including slow and fast
muscles, showed that these genes are transcribed in most tissues. However, PEE mRNA
splicing, or the expression of closely related genes, differs among tissues. (This work has
been submitted for publication).

Significance to Biomedical Research and the Program of the Instimte:

The work carried out in our Unit is of general interest to the program of NICHD, since it
seeks to understand the cellular and molecular mechanisms underlying muscle development.
The identification of regulatory elements that determine muscle specification during
development, is specially important for its use in targeted gene therapy. Our myoblast
implantation experiments demonstrate that these regulatory regions function in adult muscle,
and show that gene products are delivered to myofibers. In addition, the fact that we have
identified elements that confine expression to either fast- or slow-twitch fibers is extremely
pertinent to gene therapy, since autosomal mutations underlie diseases like Duchene's
Muscular Dystrophy which predominantly affect fast muscle.

The projects directed at understanding how nAChR expression is restricted to adult synapses
and regulated by innervation, are helpful for studies on Myasthenia gravis (MG). This
syndrome is caused by an autoimmune response of antibodies directed against skeletal muscle
nAChRs. Approximately 80% of myasthenic patients have circulating antibodies to the
receptor. Although experimental autoimmune MG has been induced in laboratory animals by
injection of nAChRs, the causes for the namral onset of disease are not known. It is
extremely important, therefore, to understand the mechanisms that regulate the levels and

kJO

wmmmmmm

ZOl HD 00711-06 LDN

localization of nAChRs during development. The erroneous regulation of receptor levels, or
its expression outside of the neuromuscular junction in the adult, could possibly result in the
onset of autoimmune responses.

Proposed Course:

The major focus in the laboratory is to identify the cis- and trans-acting factors that confer
specific tissue, developmental and innervation-dependent regulation of muscle genes.
Although the myoblast implantation technique permits an initial analysis of innervation
responsive elements, our results suggest that this technique using C2C12 myoblasts will not
be useful to analyze fiber-type specific transcription. For this reason, the continued use of
transgenic mice will be imperative for our studies.

Dr. Nakayama has begun a mutational analysis of sites in the Tnl enhancers to identify which
factors are involved in mediating fiber-type-specific transcription of these genes. With the
assistance of Ms. Jun Cheng, transgenic mice are being generated for these studies. The
information provided by the analysis of the cis-acting sequences will be used by Drs. Stauffer
and Nakayama to identify the trans-acting factors which dictate fiber diversity.

Protocols

Animals

93-030 Buonanno Production of transgenic mice

Publications:

Banerjee-Basu S, Buonanno A. Isolation and structure of the rat gene coding troponin I
slow. Gene 1994,145:241-244.

Gibney G, Buonanno A. Analysis of neural-responsive myogenin upstream sequences by
myoblast implantation. Developmental Biology, 1995 (in press).

«JO

nmn

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBLIC HEALTH SERVICE
NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

Z01 HD 00712-04 LDN

PERIOD COVERED

(pctober 1, 1994 through September 30, 1995

TTTLE OF PROJECT (80 c/iaracfefs or less Title must lit on one line between ttie txirders.)

1 Regulation of Phenotypic Differentiation in the Developing MammaUan CNS

PRINCIPAL INVESTIGATOR (Ust other protessionaJ personnel below the Pnncipal Investigator) (Name, title, laboratory, and institute aftiliaOonj

PI: D.v. Agoston Head LDN,N1CHD

(Dthers:

A. Dobi
E. Santha
C. Palkovits
Ravi Lala
Marc Mahan

Visiting FeUow
Visiting Associate
Guest Worker
Summer Student
Summer Student

LDN,NICHD
LDN,NICHD
LDN,NICHD
LDN,NICHD
LDN<NICHD

COOPERATING UNITS (il any)

XB,NIMH (M. Palkovits); Dept. NeurobioL, Duke Univ. (M. Riggott, W.D. Matthew); LNS,
^JINDS (A. Donevan, M. O'Donovan); LN, NINDS (B. Andrews), LMB, NINDS (R. McKay),
^IDDK (B. Sauer), NIMH (U. Hochgeschwender) Dept. Chemistry, ComeU Univ. (S. Chandra)

LAB^RANCH ~

Laboratory of Developmental Neurobiology

SECTION

[Jnit nn Mnlenilar Cnr\\rc\ nf Npiirndiffprpntiatinn

mwiumt^^''^^'^^^' ^''^n^^s^^

INSTITUTE AND LOCATION

3.0

2.2

OTHER:

0.8

CHECK APPROPRIATE BOX(ES)

D (a) Human subjects
D (al) Minors
D (a2) Interviews

D (b) Human tissues S (c) Neither

SUMMARY OF WORK (Use standard unreduced type. Do not exceed the space provided)

The enkephalin (ENK) gene was used as a model to study the molecular mechanism of phenotypic
differentiation with special emphasis on the identification of novel DNA binding proteins and their
cis-acting elements and on their role in far-distant DNA-DNA interactions and nuclear organization during
neurodevelopment. Three nuclear proteins were identified that specifically bind to the novel octamer-like
motif (TTTGCAT=sept) of the ENK gene; stem-sept was expressed in multipotent neuroglial stem cells;
neuro-sept and glia-sept were found to be exclusive to proliferating neuronal and glial precursors

resf)ectively and needed to be phosphorylated to bind the DNA. Specific extent of DNA bending was
observed following these protein-DNA interactions. Nuclear protein(s) that binds to both a far upstream
(-2450 bp) AT-rich [(ATT),ol sequence and the proximal promoter / TATA region of the ENK gene was
identified by a magnetic bead based "looping assay". The formation of the protein-DNA complex was
found to be restricted to basal ganglia and cerebral cortex at the ages of P2 and P8 respectively and was
found to be negatively correlated to ENKmRNA levels. The protein(s) was found to be highly charged
and showed an unusually high affinity (-10^'^ M) to this AT rich region and also to (membrane)lipids.
DNA binding was found Ca2+-dependent and required at least partially phosphorylated protein(s). Four
proteins of NfW -200 kd, ~ 90 kd, - 40 kd and -14 kd were identified by using a combination of

UV-crosslinking, mobility shift assays and DNA affinity purification. The 14 kd protein which alone forms
a high mobility protein-DNA complex with the AT rich motif was selectively enriched in protein extracts
prepared from the nuclear matrix. Treatment of primary developing neuronal cortical cultures with
distamvcin resulted in an -15-fold increase in ENKmRNA levels as quantified by quantitative (mimic) RT
PCR. The mouse enkephalin gene was further characterized and several constructs with B-gal as reporter
I gene were made for transgenic studies using the Cre-lox based targeted insertion system. Efficient DNA
transfer into various primary developing neuronal cultures were developed in combination with
|immunomagnetic sorting of transfected cells and DNA molecular decoy for functional studies.

PMS 6040 (Rev 1/84)

CPO 91 4.SII

mmBHKmm^Biim

ZOl HD 00712-04 LDN

Project Description:
Objectives:

1) To understand the molecular basis of phenotypic differentiation in the developing
mammalian central nervous system using the ENK gene as a model.

2) To identify embryonic-specific ds-elements of the ENK gene and define their roles during
phenotypic differentiation.

3) To identify, isolate and characterize their DNA binding proteins.

4 ) To investigate the role of far distant DNA-DNA interactions, structural changes of the
DNA and the nuclear structure during neuronal differentiation.

Methods Emploved:

Mobility shift assays , competition and supershift assays were used to identify and characterize
embryonic- and brain-region-specific DNA-protein complexes and DNA bending using various
wild type and mutant DNA probes. Various techniques were used to isolate nuclear proteins
from our frozen "developmental brain bank" and nuclear matrix associated proteins from freshly
dissected embryonic brain regions. DNA footprinting techniques, including methylation
interference, DNAse 1, and osmium tetroxide footprinting were used to identify the specific
CIS-elements and to assess structural changes in the DNA. A biotinylated magnetic bead based
assay system was used to study protein based far distant DNA-DNA interactions and for DNA
affinity purification of nuclear proteins. Different protein purification methods including
preparative isoelectrofocussing were used to purify DNA binding proteins. Routine recombinant
DNA techniques were employed to generate various transgenic constructs. Primary developing
neuronal cultures derived from fetuses at different developmental stages of cerebral cortex and
striatum were established. Proliferating multipotent neuroglial stem cell cultures, which can be
differentiated into glial and neuronal lineages, were used. Various physical DNA transfer
methods for transient transfection and DNA molecular decoy were developed in combination
with immunomagnetic isolation of transfected cells. Quantitative (mimic) RT PCR was
established to quantify ENKmRNA and other transcripts.

Major Findings:

1. Three novel nuclear proteins were identified that bind the truncated octamer

[(A/Genk)TTTGCAT=sepf] motif located at -542 relative to the transcriptional start site of the rat
ENK gene. Stem-sept is expressed in all brain regions between ElO and E14, but the temporal
expression of this protein strictly follows the maturation gradient of the various brain regions
(spinal cord first, cerebellum last). Stem-sept is down regulated at El 6 which is paralleled with
the expression of two distinct binding proteins, neuro-sept and glia-sept whose appearance is
paralleled wdth the emergence of neuronal and glial lineages which also follows the maturation
gradient of the different brain regions.

IIMMHM"""

3D.
3D

ZOl HD 00712-04 LDN

This observation was further supported by experiments done in collaboration with Dr. McKay.
We established that multipotent neuroglial stem cells grown in culture under proliferating
conditions express both the neuro- as well as the gUa-sept. However, following changes in culture
conditions that induce differentiation into the glial lineage, as indicated by the expression of the
gUa specific marker GFAP, neuro-sqjt expression is down-regulated and only glia-sept expression
remained detectable. Also, only glia-sqjt but not neuro-sept is expressed in primary cultures of
cortical astroglia.

Stem-sept does not require to be phosphorylated as suggested by different dephosphorylation
experiments and it can be competed only with the septamer motif but not with the octamer one.
By contrast, both neuro-sept and glia-sept require to be phosphorylated to bind the DNA and they
can be competed equally with both septamer and octamer motifs. None of the proteins are
recognized by any of the existing antibodies raised against various octamer-binding proteins
suggesting that these proteins are novel.

The binding of the various sepf-proteins to the DNA results in a characteristic DNA bending of
seemingly different degrees.

The temporo-spatial expression pattern of proteins suggests that they may contribute to early
neurodifferentiation events of neuronal and glial lineages which precede phenotypic
differentiation, and the observed DNA bending may play an important role in this process by
contributing to the assembly of 3D nuclear/DNA structure of developing neurons.

2. A protein complex that binds in a strict spatio-temporal manner to both to the

far-upstream (-2453 bp) AT-rich region [(ATT)19 repeat] and to the proximal regulatory /TATA
region of the ENK gene has been identified. The formed DNA-protein complex is resistant
against 2.5 M urea, 2 % NP-40, 14 mM B-mercapto ethanol, 10 mM BAPTA and 1.2 M KCl and
to enter PAGE gels. Treatment with 0.001% SDS eliminates DNA binding, and the complex also
binds with high affinity to cationic lipids. The DNA binding was found Ca^*-dependent and
required at least partially phosphorylated protein(s). Competitive mobility shift assays with
various mutants of the ATT motif indicated that the protein complex requires a specific DNA
structure characterized by AT rich sequences characteristic to matrix attachment regions (MARs)
rather than specific nucleotide sequences characteristic to conventional DNA binding
proteins /transcriptional factors. The protein complex seems to bind to the DNA in an all-or-
nothing fashion as evidenced by protein titrations, suggesting cooperative binding between
proteins or between the protein subunits. The length of the repeat was another factor in the
strength of the bond. As the length of the oligonucleotide became shorter, the protein became
less tightly bound to the DNA, implying a possible binding multimer that is cooperatively
bound, but separated by a non-bound DNA section. This is a common pattern for nuclear
matrix proteins.

Dialysis analysis indicated that the size of the protein complex is approximately 350 kd. UV
crossUnking and DNA affinity purification showed that the complex is formed by four proteins
of MW -200 kd, ~ 90 kd, ~ 40 kd and -14 kd. The 14 kd protein is selectively enhanced in
nuclear matrix preparations and forms a high mobility protein-DNA complex with the AT rich
motif. However, most abundant protein of the complex is the -40 kd protein as indicated by
high enrichment following DNA affinity purification.

ZOl HD 00712-04 LDN

The very characteristic the spatio-temporal expression pattern of this complex in the developing
brain suggested a neuronal cell type- and developmental-specific repressor role. This hypothesis
was supported by in vivo experiments using the drug distamycin that specifically binds to the
minor groove and thereby perturbs minor groove-protein interactions typical of nuclear matrix
proteins. At the age of E20 when the complex is being expressed distamycin treatment of
primary cortical neuronal cultures that normally express low levels of ENK mRNA resulted in
a ~15-fold increase in ENKmRNA. This finding further supports the proposed repressor role for
the complex and the first time indicates the involvement of the nuclear matrix in phenotypic
differentiation of the CNS.

3. In collaboration with Drs. Hochgeschwender and Sauer we have made the first three
transgenic constructs using the Cre-lox based targeted insertion system containing B-gal as
reporter gene.

4. We further studied the di-nucleotide repeat (TG/AC)n, located between -674 -618 of the
rENK gene. Analyzing the migration patterns of the (AC) and (TG) strands following chemical
degradation we have established that single stranded (TG) and (AC) cannot form stable double
stranded DNA because the overall conformation of the two strands are different. This structural
feature of the repeat may be responsible for the Ca^*-induced DNA sliding with consequence on
the 3D structure of the DNA.

Significance to Biomedical Research and the Program of the Institute:

The unparalleled and precisely balanced diversity of neuronal phenotypes in the mammalian
central nervous system is the cellular prerequisite of normal neuronal functions. This diversity
is the result of a tightly controlled developmental process. Much is known about early
developmental events but late molecular events in neurodevelopment such as phenotypic
differentiation is almost completely unknown at the molecular level. Also, the complexity of the
mammalian CNS requires many levels of gene regulation and likely a coordinated and combined
regulatory mechanism is required to achieve precise regulation. Understanding the regulation
of phenotypic differentiation has many implications for future studies on the general mechanism
that regvilate neurodevelopment and differentiation. Equally if not more importantly it will yield
critical information to understand the molecular basis of neurological and psychiatric disorders,
many of them are believed to be the resulted from abnormal development. The involvement of
the enkephalin phenotype has been suggested to contribute to symptoms of various disorders
including autism, Huntington's disease, epilepsy and suspected in drug abuse. Very likely, more
and more neuropsychiatric diseases and disturbances wall be proven to be at least partly caused
by faulty neurodevelopmental process. As "timing" seems to be one of the "master controller"
during neurodifferentiation, markers that indicate developmental stages can be of great use. Our
approach ("sample and probe") has already been identified significant number of novel,
development-specific DNA binding proteins and these can be used as markers that precisely
indicate the stage of differentiation in the developing mammalian central nervous system.

bl'

IIHIIIIIIIIIHIUIIIPIIIIMmillllllllllMIIIIIIHI

ZOl HD 00712-04 LDN

Proposed Course:

Our highest priority is to understand the in vivo role for the identified DNA motifs during the
various stages of phenotypic differentiation as well as to isolate and characterize their specific
DNA binding proteins.

1) Our mobility shift assays and DNA footprinting have led to the identification of three
development-specific novel elements on the ENK gene that bind proteins in spatio-temporal
manner.

We wall study the in vivo role of these motifs in both transgenic models and in primary
developing neuronal cultures using transient transfection and DNA molecular decoy.

a) Transgenic studies

We vdll determine ds-requirements for correct development-, cell- and brain region-
specific expression of the ENK gene using the first construct (pMENK/B-gal/lox5.0) that has
been made in Brian Bauer's CRE-lox vector (pBS392). The basal and induced expressions of the
reporter gene under control of the 5 kb 5' regulatory region of the ENK gene were successfully
tested in a cell culture model. In the likely case of correct expression pattern in
PMENK/B-gal/lox5.0 animals, in the next generation of transgenes we will attempt to perturb
the developmental- and cell-specific expression pattern of the ENK gene by mutating aU three
elements [(TG/AOjg; septamer andCATT),,] simultaneously. In the likely case of perturbed
expression pattern of the reporter gene the following generation of transgenic animals will bear
the constructs that are mutated in only one motif. This way both the possible combinatorial effect
of the three motifs as weU as the role of individual motifs can be studied.

b) Primary neuronal culture project: Parallel vdth the transgenic effort we wdll perform
both transient transfection assays as well as DNA molecular decoy using various primary
neuronal culture systems and constructs identical to those used for transgenes as well as double-
stranded decoy DNA molecules. IL2R co-transfected cells will be immimomagnetically sorted
(MACS) and ENKmRNA levels wUl be quantified by mimic RT PCR. In collaboration with Dr.
MacKay's laboratory, we plan to use his neuroglial stem cell culture system as a cell culture
model for both for transient transfection as well as for DNA decoy to elucidate the fvmction of
the septamer motif.

2) We will purify the DNA binding proteins of the (ATT)19 repeat by the newly developed
single step DNA affinity chromatography, microsequencing the proteins, and using degenerate
oligonucleotides we will clone the proteins. Because both the septamer and (TG)28 binding
proteins bind as monomers to their motifs, their isolation can be attempted by expression cloning
using the DNA motif for screening. To identify trans-acting factors which do not bind directly
to DNA but act on transcriptional regulation through protein-protein interaction, we wUl use a
two-hybrid system which has been developed to identify cDNA clones encoding proteins which
specifically interact with the protein of interest or subdomain of the protein. This two-hybrid
system is commercially available in the form of existing and custom-made systems by Clontech
(Matchmaker).

Protocols:

Animals:

ZOl HD 00712-04 LDN

95-008

Molecular and cell biological studies of neuronal differentiation

Publications:

Agoston Dv, Komoly S, PaUcovits, M. blockade of neuronal activity is a cofactor in axotomy:
selective up-regulation of galanin expression in septo-hippocampal cholinergic neurons. Exp
Neurol 1994;126:247-55.

Agoston Dv, Palkovits CG, Fitzgerald SC, Brenneman DE. Regulation of c-fos and its
responsiveness in developing mouse spinal cord cultures. Dev Brain Res, in press.

Dobi AL, Matsumoto K, Santha E, Agoston Ov. Guanine specific chemical sequencing of DNA
by osmium tetroxide. Nucl Acid Res 1994;22:4846-7.

Dobi AL, Palkovits M, Palkovits CG, Santha E, Agoston Dv. Protein-DNA interactions during
phenotypic differentiation. Mol Neurobiol 1995;10:185-203.

Senna M, Bravo DT, Agoston Dv, Waschek JA. High conservation of upstream regulatory
sequences on the human and mouse vasoactive intestinal peptide (VIP) genes. J DNA Sequences
1994;5:25-9.

D
D

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBLIC HEALTH SERVICE
NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

Z01 HD 00713-01 LDN

^eRft)%?^f ^^94 to September 30, 1995

WLE OF P(^OJECTr(SQ chai^(Mr^ or hss.rnia must In og arw line betwaen me borders ), ■.-, , , , « .. ..

Regulation of GeneTranscnpnon and Neurite Outgrowth by Neural Impulse Activity

PRINCIPAL INVESTIGATOR (Ust ottier professx^al perscnnel below the Pnnapal Investigator) (Name. one. latxyatofy. and institute atlillation)

PI:

R.D. Fields

Others:

K. Itoh

B. Stevens

Head

Visiting Fellow
Biologist

LDN,NICHD

LDN,NICHD
LDN,N1CHD

<I9F!^I!ft^^SaY^(?f1'her, Depart. Neurobiology, Swiss Federal Institute of Technology, Zurich,
Switzerland (M. Schachner); Brain Research Institute, University of Zurich, Zurich Switzerland
(C. Bandtlow)

Laboratory of Developmental Neurobiology

SECTION

Unit on Neurocytology and Physiology

INSTITUTE AND LOCATION

NICHD, NIH, Bethesda, Maryland 20892

TOTAL MAN-YEARS;

3.0

PHGPESSIONAL-

2.0

0THS=l:

1.0

CHECK APPROPRIATE BOX(ES)

D (a) Human subjects
D (a1) Minors
D (a2) Interviews

D (b) Humein tissues S (c) Neither

SUMM^ny 05. WORK (Use stspdard unreduced type. Dq.npf exceed the tP'c&P^v'f^) ■ i i ti i i i j i.

Our studies show that expressicSn or the neurarceiradnesion molecule LI can be regulated by
action potential stimulation, by controUing levels of LI mRNA. Regulation is selective for
specific frequencies of stimulation, and for specific types of neural cell adhesion molecules
(NCAM is not affected). Adhesion of neuroblastoma cells to DRG axons is reduced after 5 days
electrical stimulation at the frequency that is effective in lowering LI expression, but no changes
in cell<ell adhesion are produced by stimulus patterns that do not alter LI levels. Association
of Schwann cells with axons is reduced for up to 4 days after stimulation at 0.1 Hz, but not after
1 Hz stimulation. Fasciculation of neurites can be regulated by impulse activity at an appropriate
frequency.

Intracellular signaling from neural impulses can be sensitive to temporal features of action
potential stimulation, rather than to the concentration of second messengers or products in
signaling reactions activated by the stimulus. Transcription of the c-fos gene can be more
strongly activated by action potential stimuli producing low amplitude intracellular calcium
transients repeated at short intervals, compared to action potential bursts that produce much
larger intracellular calcium transients, repeated at less frequent intervals. Phosphorylation of the
transcription factor CREB at Ser 133, does not correlate with c-fos transcription for some patterns
of stimulation, particularly those consisting of intense bursts repeated at longer intervals.

Accommodation of growth cones to electrically-induced collapse is associated with a decrease
in trans-membrane calcium currents, which results from removal of calcium channels from the
cell membrane after 24 hours of action potential stimulation at appropriate frequencies. Growth
cone responses to action potentials involve different intracellular calcium signaling pathways
than those activated by the receptor-mediated growth cone collapsing factor NI-35.

ZOl HD 00713-01 LDN

Project Description

Objectives

The Unit on Neurocytology and Physiology, directed by Douglas Fields, is investigating
intracellular signaling from the membrane to the nucleus in response to patterned neural impulse
activity. The current goal of the Unit is understanding how expression of genes controlling the
structural and functional properties of neurons is regulated by specific patterns of impulse
activity in developing neural circuits. This work has focused on activity-dependent regulation
of neural cell adhesion molecules, second messenger systems, phosphorylation of transcription
factors controlling immediate early gene transcription, and the influence of action potentials on
growth cone motility.

Methods Emploved

Dorsal root ganglion neurons dissociated from fetal mouse spinal cord are maintained in a multi-
compartment cell culture preparation equipped with platinum electrodes for chronic electrical
stimulation. Other cell culture preparations including Schwann cells dissociated from the sciatic
nerve of mice and neuroblastoma cells.

Electrophysiological recording, advanced imaging techniques, and molecular methods are
employed to study neuronal responses to action potential stimulation. Growth cone responses
to guidance factors and action potentials are studied by low-light level time-lapse video
microscopy. Ratiometric imaging of calcium-sensitive dyes is used to measure the temporal and
spatial dynamics of electrically-evoked calcium transients in neurons. Immunocytochemistry,
electron microscopy and confocal microscopy are used for morphological studies.

Specific mRNA transcripts in cultured neurons are measured by polymerase chain reaction,
competitive PCR, and Northern blots. Western blot, immunoprecipitation and kinase assays are
used to monitor expression of neural cell adhesion molecules, immediate early genes,
neurotrophin receptors, CaM Kinase and MAP kinase activity, and phosphorylation of the
transcription factors SRF and CREB.

Surgical implantation of miniperfusion pumps to infuse protease inhibitors into the brain are
being used in combination with trans-synaptic neuronal labeling to investigate the formation of
ocular dominance columns in visual cortex of kittens.

Major Findings

Regulation of Neural Cell Adhesion Molecules by Specific Patterns of Neural Impulses. Nervous
system structure is modified by electrical activity in developing neural circuits, but the molecular
mechanisms are not well understood. Neural cell adhesion molecules, such as LI and NCAM,
influence development of the nervous system by regulating cell adhesion, transmembrane
signaling, neurite outgrowth, fasciculation and myelination. Our recent research on mouse DRG
neurons in an in vitro preparation shows that expression of specific neural cell adhesion
molecules is regulated by neural impulse activity. Moreover, this regulation is sensitive to
specific temporal patterns of impulses.

6-i

■

sn^BB

mmoK

ZOl HD 00713-01 LDN

Kouichi Itoh has performed measurements, using competitive RT-PCR, which show that
stimulation of surprisingly low frequency (1 action potential every 10 seconds) decreases the
level of LI mRNA more than 10 fold in DRG neurons. Western blots confirm that this decrease
in message results in significantly lower expression of LI at the protein level. Interestingly, other
patterns of stimulation (1 action potential per second, for example) do not alter LI mRNA or
protein, and another neural cell adhesion molecule, NCAM, is not affected by the stimulus.
Regulation of neural cell adhesion molecules by specific patterns of neural impulses could be an
important mechanism regulating cellular interactions that coordinate structure and function of
the nervous system during development. The signaling mechanism responsible for sensitivity
to specific patterns of impulses in neurons is currently under study.

Effects of Impulse Activity on CeU Adhesion. Morphogenesis of the nervous system requires
recognition and adhesion of appropriate cells to form proper structural /functional complexes.
We have developed two assays of neural cell adhesion in multicompartment cell cultures to test
whether cellular adhesion is influenced by impulse activity of DRG neurons. Adhesion of test
cells (neuroblastoma N2a cells) plated on stimulated DRG neurites is reduced after stimulation
at the same frequency of impulses that is effective in lowering expression of LI (0.1 Hz). Higher
frequency stimulation or blockade of activity are without effect on adhesion or LI expression.
Pre-incubation of N2a cells with antibodies against LI inhibits their adhesion to DRG axons,
indicating the involvement of LI in this adhesion assay.

Schwann cells provide critical functions in development of peripheral nerves, and during
regeneration after injury, by promoting neurite outgrowth, providing neurotrophic support,
consolidating unmyelinated axons into bundles, and myelinating large diameter axons to enable
rapid long-distance conduction. Beth Stevens has conducted experiments in which Schwann
cells, purified from the sciatic nerve of mice, are plated onto DRG axons that have received
different frequencies of electrical stimulation for 5 days. This work shows that association of
Schwann cells to DRG axons is highly sensitive to the pattern of impulse activity in the axons.
The effects on Schwann cell association with axons are observed only in response to specific
frequencies of impulse activity (0.1 Hz, but not 1 Hz), and the effects persist for up to 4 days
after the stimulus treatment. The persistent reduction in number of Schwann cells on DRG
axons 4 days after 0.1 Hz stimulation may result from a combination of processes, including
adhesion, survivaL or proliferation of the Schwann cells.

Effects of Impulse Activity on Axon Fasciculation. EXiring normal development, DRG neurons
pass through a progression of distinct developmental phases that include neurite outgrowth,
defasciculation of axon terminals to innervate peripheral targets, fasciculation and consolidation
of axon terminals innervating fvmctional sensory structures, and synaptogenesis vdth motor
neurons in the spinal cord. Each of these developmental phases is accompanied by distinct
patterns of spontaneous neural impulse activity, suggesting that functional activity within these
developing neural circuits may provide an important regulatory influence on these
developmental processes. We have tested whether the association between distinct patterns of
impulse activity and defasciculation and fasciculation of DRG axons are functionally related.
Using quantitative digital morphometry, this work shows that axons stimulated at the specific
frequency that is effective in lowering LI expression produces significant defasciculation of DRG
neurites in the multicompartmental culture preparation. Several molecules may participate in

fflt^HMmW*"*'^™^'"— ^—""

HBinH

ZOl HD 00713-01 LDN

activity-dependent regulation of axon fasciculation, but as with LI expression, 0.1 Hz stimulation
for 5 days leads to significant defasciculation, whereas 1 Hz stimulation or the absence of
impulse activity produces no changes in fasciculation or LI expression. Stimulus patterns that
are effective in lowering LI expression, and the associated functional effects we observe, both
parallel endogenous changes in impulse patterns that fire spontaneously in mouse DRG neurons
in utero. These changes in firing patterns have long been correlated with distinct developmental
phases, but prior to these observations concerning LI expression, a possible molecular
mechanism for a causative relationship between these firing patterns and axon fasciculation has
not been available.

Intracellular Signaling from Patterned Neural Impulses. A major area of current research
concerns is how extracellular signals regulate expression of genes that are necessary for adaptive
responses of cells to changing environmental conditions. Considerable activity is being directed
toward identifying the sequence of reactions linking extracellular stimulation to the processes
regvilating gene expression, but how these signaling reactions might extract and transmit
information from temporally varying stimulus patterns is not known. This question is
particularly important in the nervous system, where information is conveyed in the temporal
pattern of neural impulses.

We are addressing this question by using an in vitro preparation of neurons dissociated form the
DRG of mice, stimulated via electrical impulses of different patterns. Using this method of
stimulation in combination with RT-PCR, ratio fluorescence calcium imaging,
immunocytochemistry, and protein phosphorylation assays, we are pursuing our previous
findings that expression of immediate early (IE) genes can be activated by specific patterns of
impulse activity. Several stimulus patterns have been identified in which expression of the IE
gene cannot be understood as a simple response to increasing intensity of stimulation (i.e.,
increasing frequencies of action potentials), but appears to discriminate differences in temporal
features of the stimulus. By monitoring the temporal dynamics of the activation and inactivation
of known reactions in the signaling cascades, we hope to begin to understand whether signaling
cascades have resonant properties, such that trains of impulses at specific frequencies or patterns
can activate signaling pathways in a selective manner.

Calciimi Signaling and Regulation of Gene Transcription in Neurons. Calcium is critical in
mediating intracellular signaling and in regulating gene expression in neurons. Imaging
methods that allow calcium transients to be studied in living neurons in response to electrical
stimulation are being used to investigate whether discrimination of different impulse patterns
may result from differences in calcium influx. Changes in resting calcium concentration,
amplitude, and dynamics of calcium increase and recovery after stimulation have been
quantified in neuronal cell bodies, and subcellular domains, including axons, dendrites, the
nucleus, and growth cones. These measurements are correlated with levels of transcription of
the immediate early gene c-fos, using RT-PCR in DRG neurons in culture. We find that
transcription of c-fos is largely independent of the number of action potentials delivered in the
stimulus, and that for a number of stimulus patterns the amount of intracellular calcium
generated by impulse activity fails to correlate with levels of c-fos transcription. In the case of
repeated bursts of impulses, a better correlation is found between c-fos transcription and the
interval between successive bursts of action potentials, rather than the amplitude of the
intracellular calcium or duration of each stimulus burst. For example, pulse trains of

IP """ UMIUI"!^—

tmm

ZOl HD 00713-01 LDN

comparatively long duration (10 Hz, for 9 sec), which produces greater increase in intracellular
calcium [Ca**]; (735 nM), fails to activate c-fos expression when repeated at 5 min intervals, but
brief pulse trains (10 Hz, 1.8 sec), which induce less increase in [Ca^^^l^ (485 nM), sigruficantly
increases expression of c-fos when delivered at more frequent intervals (every 1 minutes). This
evidence of the importance of temporal aspects of second messenger generation is consistent
with previous work showing that even a single action potential, which produces the smallest
possible increase in [Ca*""]; (about 20 nM), can stimulate gene transcription if delivered at
appropriate intervals (every 10 sec for the c-fos).

Dynamics of CREB Phosphorylation in Response to Action Potentials. Nuclear transcription
factors are key substrates for calcium dependent protein kinases which help convey action
potential stimulation to the nucleus to control expression of genes regulating adaptive responses
of neurons. Phosphorylation of the nuclear transcription factor CREB by cAMP or calciimi has
been shown to be critical in regulating transcription of the c-fos gene in neurons, and in
consoUdating long term memory in mice and invertebrates. Using an antibody specific for CREB
phosphorylated at Ser 133, we have studied the kinetics of phosphorylation of this transcription
factor in response to action potential stimulation of specific frequencies, and correlated this wdth
c-fos expression and the changes in intracellular calcium produced by the stimulus in DRG
neurons. Our work shows that phosphorylation in response to action potentials at a frequency
of 10 Hz is rapid, reaching maximal values in less than 10 sec, but dephosphorylation of CREB
at the critical site of Ser 133 proceeds with a much slower time course, decUning by 1/2 within
about 5 minutes after a brief stimulus burst. Phosphorylation of CREB at Ser 133 parallels the
rapid rate of increase in [Ca^^^lj, but we find that [Ca**]; recovers to normal levels within several
seconds; much faster than CREB dephosphorylation. This work has revealed a number of
stimulus patterns in which c-fos expression is not consistent with levels of CREB phosphorylation
at Ser 133. For example, high levels of [Ca**];, if repeated infrequently, e.g. 9 sec 10 Hz
stimulation every 5 min, produces large increases in phosphorylation of CREB at Ser 133, yet this
stimulus induces minimal c-fos transcription. This discrepancy between CREB phosphorylation
and transcription of c-fos, indicates that CREB phosphorylation is not necessarily sufficient for
transcription of this gene in response to action potentials, and that the dynamic responses of
parallel signaling reactions regulating gene transcription need to be considered to better
understand how transcription is regulated by patterns of impulses in the nervous system.

Growth Cone Responses to Action Potentials. (Dur previous research has shown that action
potentials induce growth cone collapse, similar to responses mediated by receptors on the
growth cone for specific guidance molecules in the mammalian brain. This research showed that
growth cone collapse was associated with the large increase in intracellular calcium that
accompanies action potential depolarization. However, after 24 hours exposure to this electrical
stimulus, growth cones no longer collapse in response to action potentials. Ratio fluorescence
imaging showed that although high levels of intracellular calcium were reached in
accommodated growth cones, the kinetics of calcium increase were slowed significantly. This
suggests that a slower rate of increase in intracellular calcium, will faU to activate mechanisms
inducing the collapse response. Either changes in calcium buffering and removal mechanisms
or decreased calcium influx through voltage-sensitive calcium channels could slow electrically-
induced calcium transients in these neurons. Using calcium imaging, we have found that
chronic impulse activity also leads to slower electricaUy-tnduced calcium accumulation in the cell
body of DRG neurons after chronic stimulation. This allowed the use of electrophysiological

ranM

^nmHRin

ZOl HD 00713-01 LDN

methods to measure differences in calcium influx through voltage-sensitive calcium channels in
DRG neurons after chronic electrical stimulation. Experiments performed in collaboration with
Dr. Nelson and colleagues, using whole-ceU patch clamp recording of calcium currents, reveal
that voltage-sensitive calcium currents are reduced after chronic stimulation of DRG neurons.
Receptor binding studies indicate that this reduction is accompanied by loss of L-type calcium
channels from the plasma membrane. This adaptive regulation appears to contribute to the
resistance of growth cones to electrically-induced collapse after a period of chronic stimulation.
In addition, activity-dependent changes in calcium signaling could regulate a vsdde variety of
other calcium-dependent processes in neurons, possibly including gene expression or synaptic
plasticity.

Signal Transduction Pathways in Growth Cone Responses to Action Potentials and Substrate-
Bound Factors. Recent work in collaboration with Dr. Christine Bandtlow of the University of
Zurich, is exploring differences in signaling reactions mediating growth cone collapse in response
to a protein isolated from CNS myelin (Nl-35), and collapse induced by action potentials. Both
these stimuli are associated with an increase in intracellular calcium in the growth cone, and
each is dependent on influx of calcium through voltage-sensitive calcium channels. Calcium
imaging and voltage clamp methods showed that calcium fluxes become attenuated in
accommodated growth cones. We therefore tested whether the membrane protein isolated from
central nervous system myelin (Nl-35) would also induce attenuated calcium responses in
growth cones accommodated to electrically-induced coUapse. This research indicates that Nl-35
stimulates large increases in intracellular calcium in cell bodies of DRG neurons that are not
significantly different before or after chronic electrical stimulation. This suggests that divergent
signal transduction mechanisms, involving partially independent calcium-signaling cascades, are
activated by the two growth cone collapsing stimuli.

Significance to Biomedical Research and the Program of the Institute

Nervous system development and recovery from nervous system injury are critically dependent
upon selective ceU-ceU adhesion, fasciculation, and growth cone guidance. The regulation of
these and other developmental processes by extracellular signals is essential for coordinating the
complex structural /functional interrelations necessary for development and regeneration of the
nervous system. Elucidating the responses of signaling reactions that link action potential
stimulation to gene transcription, and that in turn activates specific adaptive responses in
neurons, vdU ultimately enable better interventions to prevent or correct developmental deficits
and promote recovery from nervous system injury.

Proposed Course

We will continue to study the responses of signaling reactions to temporally varying stimulation
in DRG neurons, and seek to understand the cellular basis for the temporal specificity to
different frequencies of stimulation. Future work will emphasize spatial heterogeneity in calcium
transients in the nucleus and cytoplasm of neurons, the temporal dynamics of activation of
kinases that preferentially phosphorylate transcription factors CREB and SRF, and calmodulin
activation by action potentials in growth cones and DRG neurons.

i)d

wnnmamm

ZOl HD 00713-01 LDN

Protocols:

Animals:
94-013
94-016
Publications

Nelson Developmental neurobiology in culture systems

Fields Effects of proteases on ocular dominance columns of visual cortex

Fields RD. Regulation of neurite outgrowth and immediate early gene expression by patterned
electrical stimulation. Prog in Brain Res 1944;102:125-36.

Fields, R.D. and P.G. Nelson (1994) Role of electrical activity in synapse formation. In: Stenger
DA, McKenna TM, eds. Enabling Technologies for Cultured Neural Networks. San Diego:
Academic Press Inc, 1994, pp. 237-60.

Nelson PG, Fields RD. Developmental plasticity of the brain: Problems and solutions in dynamic
imaging of the synaptic bouton. In: T. Hiruma, ed. Biomolecular Mechanisms and Photonics:
Cell-Cell Communication (in press).

nmn

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBLIC HEALTH SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD 01202-7 LDN

PERIOD COVERED

nrl-nhPr ^, 1994 tn SpptPmhpr .-^0, ^99H

TITLE OF PROJECT (80 characters or less. Title must fit on one line between the borders)

Rpqiil atinn of pxprpaaion and f iinrh i nn nf npiirnpppf.i dPH diirintj ripvpl npmpnt-.

PRINCIPAL INVESTIGATOR (List other professional personnel below the Principal Investigator.) (Name, title, laboratory, and institute affiliation)

PI: Y. Peng Loh Section Chief LDN, NICHD

Others: William P. Hayes IRTA LDN, NICHD

COOPERATING UNITS (if any)

Lab. Biol. Chem. , NCI (J. Battey), Lab. Neurochem. , NINDS (S. Wray, J. Mill and H.

Chin); Univ of Montreal (J. Chan); Lab. Cell. Mol. Neurophys., NICHD (V. Gallo)

LABfflRANCH

T.ahnrat-nry nf DpvpI nprnpnta 1 NpiirnhinT

OgjL.

SECTION

Sertinn on Cell n Tar Neiirnhi nl ntjy

INSTITUTE AND LOCATION

WTCHn,. NTH. Rethesda, MP ?089?

TOTAL STAFF YEARS:
1 -1

PROFESSIONAL:
1 ■ 1

OTHER:
0

CHECK APPROPRIATE BOX(ES)

D (a) Human subjects

D (a1) Minors
n (a?) Intfirvipw5;

D (b) Human tissues H (c) Neither

SUMMARY OF WORK (Use standard unreduced type Do not exceed the space provided)

Project has been terminated.

PHS 6040 (Rev. 5/92)

son

UUUHUUIHM

DEPARTMENT OF HEALTH AND HUMAN SERVICES ■ PUBLIC HEALTH SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD 01004-12 LMG

PERIOD COVERED

October 1, 1994 to September 30, 1995

TITLE OF PROJECT fSO characters or less. Title must fit on one line between the borders.)

Regulation of Amino Acid and Nucleotide Biosynthesis in Saccharomyces cerevisiae

PRINCIPAL INVESTIGATOR lUst other professional personnel below the Principal Investigator! IName, title, laboratory, and institute affiliation)

PI: A.G. Hinnebusch, Research Microbiologist, (All listed personnel LMG:NICHD)

Marton IRTA Fellow G. Pavitt Visiting Fellow
Romano IRTA Fellow J.
Rolfes IRTA Fellow E.

Natarajan Vis. Fellow
Zhang (Ebon)

Garcia
Vasguez
Yang
Jackson

Vis . Fellow M.
Vis. Fellow P.
Vis. Assoc. R.
Biologist K.

Pavitt

Anderson IRTA Fellow

Duenas Guest Researcher

Drysdale IRTA Fellow F.

Phan

Guest Researcher

COOPERATING UNITS lif any)

None

LAB/BRANCH

Laboratory of Molecular Genetics

SECTION

Section on Molecular Genetics of Lower Eukaryotes

INSTITUTE AND LOCATION

NICHD, NIH, Bethesda, Maryland 20892

TOTAL STAFF YEARS:
11.06

PROFESSIONAL:

10.06

OTHER:
1.0

CHECK APPROPRIATE BOXIES)

D (a) Human subjects

D (a1) Minors
D (a2) Interviews

D (b) Human tissues K] (c) Neither

SUMMARY OF WORK (Use standard unreduced type. Do not exceed the space provided.)

We are studying transcriptional and translational control mechanisms that regulate amino acid and purine biosynthetic
genes in yeast in response to nutrient availability. GCN4 is a transcriptional activator of amino acid biosynthetic
genes that is regulated at the translational level by the eIF-2a kinase GCN2. We have proposed that phosphorylation
of elF-2 by GCN2 under starvation conditions stimulates GCN4 translation by inhibiting guanine nucleotide exchange
on eIF-2 catalyzed by eIF-2B, thereby reducing the concentration of the ternary complex eIF-2/GTP/Met-tRNA'"". A
large number of point mutations were isolated in the GCN3, GCD7, and GCD2 subunits of eIF-2B that uncouple
GCN4 translation from eIF-2a phosphorylation without affecting eIF-2B catalytic activity. These mutations cluster in
two regions of sequence similarity shared among the three proteins which may define an interaction surface between
eIF-2B and eIF-2(aP). Many of the mutations affect residues conserved between yeast and mammalian eIF-2B, and
the rat homologue of GCN3 can partially substitute for GCN3 in yeast cells. GCD2, GCD7, and GCN3 can form an
eIF-2B subcomplex in vivo that reverses the effects of eIF-2a phosphorylation, probably by sequestering eIF-2(aP).
We have shown that GCDIO is a component of the 8-subunit eIF-3 complex, and that mutations in GCDIO are
suppressed by overexpressing tRNA'"'" . These results suggest that gcdlO mutations reduce binding of ternary
complexes to 40S ribosomal subunits. GCNl and GCN20 are components of a protein complex required in vivo for
phosphorylation of eIF-2a by GCN2. An N-terminal segment of GCN20 and a domain in GCNl related to translation
elongation factor 3 (EF-3) mediate complex formation between the two proteins. GCN20 contains ATP-binding
cassettes found in membrane transporters; however, these sequences are partially dispensable for GCN20 function, and
GCNl and GCN20 are distributed uniformly throughout the cytoplasm. Based on their homology to EF-3, we
propose that GCN1/GCN20 interact with ribosomes and facilitate binding of uncharged tRNA to GCN2. GCN2
contains a pseudo-kinase domain in its N-terminus that is conserved in the Neurospora homologue of GCN2 that is
also required for regulating kinase function in vivo. The transcriptional activation domain of GCN4 contains 7-8
subdomains, each consisting of 2-3 bulky hydrophobic amino acids surrounded by acidic residues. These subdomains
can cooperate in different combinations to activate transcription. An ADE5,7 promoter fragment sufficient for
adenine-repressible transcription contains three 6-lOnt elements that probably function as binding sites for the BASl,
BAS2, and ABFl transcription factors. Overexpressing BAS2 increases ADESJ expression under repressing
conditions, suggesting that adenine repression involves reducing the ability of ADE gene promoters to compete with
other yeast genes for binding limiting amounts of BAS2.

PHS 6040 (Rev. 5/92)

j^BsmmBmm^BiB'i

ZOl HD 01004-12 LMG

Project Description:

Objectives:

To understand at the molecular level how the yeast S. cerevisiae regulates its capacity for synthesizing
amino acids and nucleotides according to the availability of these nutrients in the environment. One
facet of this regulation, known as general amino acid control, involves increased expression of a large
number of amino acid biosynthetic genes in response to starvation for any amino acid. Synthesis of
GCN4 protein, the transcriptional activator in this system, is stimulated under starvation conditions by
a translational control mechanism involving short open reading frames (uORFs) in the GCN4 mRNA
leader, several general translation initiation factors, and a protein kinase known as GCN2. Genetic and
biochemical experiments are being conducted to understand how the uORFs regulate the flow of
scanning ribosomes to the GCN4 start codon according to the availability of amino acids. A
combination of genetic, molecular and biochemical approaches are being used to identify and
characterize the trans-acting factors, both positive and negative, that mediate the regulatory functions
of the uORFs. Some of these factors are subunits of the general translation initiation factors eIF-2 or
its guanine nucleotide exchange factor elF-2B; others appear to be components of previously
unidentified general initiation factors; still others, including the protein kinase GCN2 and its positive
effectors GCNl and GCN20, are dispensable regulatory factors dedicated to GCN4 control. Wevvish
to identify the biochemical functions of each factor and determine its position in the signal-transduction
pathway that detects uncharged tRNA in amino acid-starved cells and modifies the translational
machinery in a way that stimulates GCN4 expression. This regulatory mechanism involves
phosphorylation of the a subunit of eIF-2 by the protein kinase GCN2 and consequent reduction in the
ability of eIF-2B to recycle eIF-2. This same mechanism operates in mammalian cells to inhibit total
protein synthesis in response to various stress conditions, including amino acid starvation. By
combining the powerful genetics and molecular biology of yeast with the biochemical analysis of
translation initiation, we hope to provide a detailed molecular description of this highly conserved
mechanism for regulating protein synthesis. The human eIF-2a kinase DAI (double-stranded RNA-
activated inhibitor of translation) is an important regulator of cell growth and differentiation and a
critical component of the interferon response to viral infection in humans. We have shown that DAI
can functionally substitute for GCN2 in yeast cells and we are using the GCN4 genetic system to probe
the mechanism of DAI activation by dsRNA.

In addition to studying the regulation of GCN4 translation, we are conducting an in-depth mutational
analysis of the transcriptional activation domain of the GCN4 protein, with the goal of obtaining a better
understanding of the detailed structure of this domain and the identification of specific transcription
factors with which GCN4 interacts in stimulating the expression of its target genes. We have also
undertaken a study of the transcriptional control of adenine biosynthetic genes in response to exogenous
purines. We are defining the minimal cis-acting sequences at the ADE5 gene necessary for adenine
repression of this gene and studying the regulatory proteins that bind to these sites, including BASl,
BAS2 and GCN4. We hope to learn how the expression and activity of these regulatory proteins is
modulated by purine availability in the cell. Parallel studies of general amino acid control and the
regulation of nucleotide biosynthesis in yeast should provide an integrated view of how eukaryotic cells
control the levels of substrates (amino acids and nucleotides) and the machinery for protein and nucleic
acid synthesis in the face of a changing nutritional environment.

Bl^

HUWHIW

ZOl HD 01004-12 LMG
Major Findings

I. Mechanism of translational control of GCN4 expression.

A. Evidence that the efficiency of GCN4 translation is inversely coupled to the concentration of
eIF-2«GTP«Met-tRNA'^'"i ternary complexes in the cell.

Phosphorylation of the a subunit of translation initiation factor-2 (eIF-2a) on residue Serine-51 is a
prominent mechanism for regulating protein synthesis in mammalian cells. The phosphorylated form
of eIF-2 inhibits translation initiation by impairing the conversion of eIF-2 GDP to eIF-2 GTP by the
guanine-nucleotide exchange factor eIF-2B. In yeast, phosphorylation of eIF-2 by the protein kinase
GCN2 stimulates translation of GCN4 mRNA under conditions of amino acid starvation. According
to our model, essentially all ribosomes that bind to the 5' end of GCN4 mRNA translate uORFl, under
both starvation and nonstarvation conditions, and ca. 50% remain attached to the mRNA and resume
scanning downstream as 40S subunits. Under nonstarvation conditions, when the active form of elF-2
is abundant, these 40S subunits rapidly rebind the eIF-2»GTP»Met-tRNA'"^', ternary complex and regain
the ability to recognize an AUG codon as a translational start site. Consequently, most reinitiate at
uORFs 2, 3, or 4, and then dissociate from the mRNA, failing to reach the GCN4 start codon. Under
starvation conditions, elF-2a is phosphorylated by the protein kinase GCN2 and, by analogy with
mammalian systems, this reduces the level of elF-2»GTP»Met-tRNA'''" ternary complexes by inhibiting
the recycling factor elF-2B. Following translation of uORFl, many ribosomes now scan the entire
distance between uORFl and uORF4 without rebinding the ternary complex. Lacking the initiator
tRNA*^", they cannot recognize the AUG start codons at uORFs 2, 3, and 4 and continue scanning
downstream. While traversing the leader segment between uORF4 and GCN4, most of these ribosomes
re-bind the ternary complex and reinitiate translation at GCN4.

One of the key tenets of our model is that increased translation of GCN4 is triggered by a reduction in
the concentration of elF-2»GTP»Met-tRNA'^^'i ternary complexes. In support of this idea, we showed
that reducing the number of chromosomal genes encoding initiator tRNA'''^' mimics the effect of eIF-2
phosphorylation in causing derepression of GCN4 translation in the absence of amino acid starvation
or GCN2 function. In addition, overexpression of all three subunits of elF-2, which leads to a ~10-fold
increase in the heterotrimeric elF-2 complex in vivo, prevents derepression of GCN4 in response to elF-
2 phosphorylation by GCN2. It also partially suppresses the slow-growth phenotype associated with
hyperphosphorylation of eIF-2 by GCN2' enzymes, and this suppression is augmented when tRNA'^'^'j
is co-overexpressed with the elF-2 complex. These results provide strong support for the idea that
phosphorylation of eIF-2 inhibits general translation and specifically stimulates GCN4 expression by
reducing the levels of the ternary complex in the cell.

B. Saturation mutagenesis of the regulatory domains in the GCD2, GCD7 and GCN3 subunits
of eIF-2B.

Another important piece of evidence supporting our model is that mutations can be isolated in subunits
of eIF-2B that suppress the phenotypes associated with eIF-2a hyperphosphorylation by hyperactivated
GCN2' protein kinases. The suppressor mutations include point mutations and deletions in GCN3 a
nonessential subunit, and point mutations in GCD7 and GCD2 two of the four essential subunits of the
yeast eIF-2B complex. These mutations could decrease the affinity of eIF-2B for phosphorylated eIF-2
and thus prevent sequestering of eIF-2B in an inactive state; alternatively, they could allow eIF-2B to
catalyze nucleotide exchange on eIF-2. The C-terminal half of GCD2, GCD7 and GCN3, are very
similar in amino acid sequence, suggesting that the interaction of each protein with eIF-2 involves a
structural feature that is common to all three. To test this idea, we set out to identify all of the amino

■IIIIIIIIMIIIIIIII

ZOl HD 01004-12 LMG

acids in GCD2, GCD7 and GCN3 that are critically required for negative regulation of eIF-2B by
phosphorylated eIF-2. Towards this end, we have isolated and characterized a large number of
dominant or semi-dominant point mutations in GCD2 GCD7, and GCA^iwith the suppressor phenotype
described above. For both GCD2 and GCD7, we have obtained point mutations that render
phosphorylated eIF-2 completely ineffective in down-regulating general protein synthesis and in
stimulating GCN4 translation (Gen' phenotype). These mutations are actually more effective than a
deletion of GCN3 in suppressing the toxic effect of eIF-2 hyperphosphorylation by PKR. Thus, it
appears that all three proteins make important contributions to the inhibitory effects of eIF-2(aP) on
eIF-2B activity.

The results obtained thus far identify two clusters of Gen" mutations which affect residues located in
regions of homology between GCD2, GCD7 and GCN3. One of the two clusters in each protein is
located at the extreme C-terminus, the domain of greatest sequence similarity among the three. A
second cluster occurs within a region of high similarity near theN-termini of GCN3 and GCD7 and in
the center of GCD2 (which contains a large N-terminal domain not shared with GCD7 and GCN3).
It is noteworthy that all of the GCD2 mutations fall within the region of the protein that is related in
sequence to GCD7 and GCN3. In a few cases, amino acids occupying identical positions in the
sequence alignments have been mutated in two of the three proteins. Together, these results suggest that
homologous segments in GCD2, GCD7 and GCN3 are devoted to the regulatory interactions between
eIF-2B and phosphorylated eIF-2. These homologous segments might be juxtaposed on the surface of
eIF-2B and make direct contact with different residues in the phosphorylated N-terminal domain of elF-
2a, as suggested previously. Alternatively, one of the two regulatory segments in each protein might
interact with eIF-2a while the other could mediate a conformational change in eIF-2B that distorts the
active site when phosphorylated eIF-2 is bound. The segments located between the two clusters of Gen'
mutations in each protein are also very similar in sequence. Perhaps these segments perform a structural
role in mediating subunit interactions between GCD2, GCD7, and GCN3.

C. Suppression of eIF-2(aP) toxicity by overexpression of eIF-2B subcomplexes.

A second approach we have taken in probing the structure and function of eIF-2B is to overexpress
different combinations of the eIF-2B subunits. We showed previously that overexpressing all 5
subunits, or all 4 subunits except GCN3, suppresses the toxicity of eIF-2a phosphorylation in GCNl"
strains. More recently, we observed the same phenotype when only GCD2, GCD7 and GCN3 were
overexpressed from high copy-number plasmids. In fact, overexpressing just GCD7 and GCD2
conferred nearly the same suppression as did the combination of all three proteins. In contrast,
overexpressing each of the individual subunits and many other combinations of two or three subunits
of eIF-2B conferred no suppression of GCN2' mutations. One interpretation of these results is that
GCD7, GCD2 and GCN3, or just GCD7 and GCD2 alone, can form stable subcomplexes that sequester
phosphorylated eIF-2 and neutralize its inhibitory effect on the native eIF-2B complex. In support of
this interpretation, we succeeded in co-immunoprecipitating nearly all of the excess GCD7 and much
of the excess GCN3 with anti-GCD2 antibodies from cells overexpressing GCD2, GCD7 and GCN3.
Thus, we now have biochemical evidence for subcomplex formation by these proteins in vivo. An
alternative explanation for our results is that subcomplexes containing GCD2, GCD7 and GCN3 have
nucleotide exchange activity and thus can functionally replace native eIF-2B. This explanation seems
unlikely because overexpressing the subcomplex does not complement the growth defects of mutations
in the GCDl subunit that are suppressed when eIF-2 is overexpressed.

mwwMTwiimmwwinmwiaiiMiiMHMiiNiiiiiaiimytiuiiyiwnBnMiim^^

ZOl HD 01004-12 LMG

D. In vivo analysis of the mechanism for inhibiting eIF-2B by phosphoryiated eIF-2.

Overexpression of eIF-2 in strains containing an activated GCN2' kinase leads to the production of
much higher levels of the phosphoryiated form of the protein than are present in the same strain
containing wild-type amounts of eIF-2; yet, we found that the toxicity of GCN2' proteins is suppressed
by overexpressing elF-2. This apparent paradox can be explained by noting that the ratio of
phosphoryiated to nonphosphorylated eIF-2 was reduced in the strain overexpressing elF-2 versus the
parental strain. If we propose that the mechanism of inhibition of eIF-2B by eIF-2(aP) is competitive
inhibition, rather than irreversible binding of a non-competitive inhibitor, we can explain why the degree
of inhibition depends more on the ratio of eIF-2(aP):eIF-2 rather than the absolute amount of
phosphoryiated eIF-2. We can eliminate the trivial possibility that overexpression of eIF-2 simply
decreases the requirement for elF-2B, perhaps as a result of spontaneous nucleotide exchange on elF-2,
because several mutations in subunits of elF-2B are not complemented by overexpression of eIF-2.
Therefore, our results indicate that elF-2(aP) acts as a competitive inhibitor of eIF-2B rather than
forming an extremely stable inactive complex with it.

E. GCDIO is the RNA binding subunit of eIF-3.

Mammalian elF-3 is an 8-subunit complex that stimulates several steps in the initiation pathway in cell-
free translation systems. Yeast contains a structurally similar complex that can functionally replace
mammalian elF-3 in an in vitro translation system containing all the initiation factors from Hela cells
except for eIF-3. In collaboration with Mercedes Tamame's and John Hershey's laboratories, we have
now obtained strong biochemical evidence that GCDIO encodes the 54.6kDa RNA-binding subunit of
yeast elF-3. Mutations in GCDIO lead to constitutive derepression oi GCN4 translation in the absence
of GCN2 and GCN3, and to temperature-sensitive growth on rich medium, the same phenotypes
associated with gcd mutations affecting subunits of eIF-2 or eIF-2B. We cloned GCDIO and found that
a deletion of the gene is lethal. In addition, we observed dissociation of polysomes and accumulation
of SOS subunits after shifting gcdlO temperature-sensitive mutants to the non-permissive temperature,
consistent with a defect in an essential translation initiation factor. After producing antibodies against
GCDIO, we showed that the protein is present in a high molecular weight complex, and that a fraction
of GCDIO is physically associated with polysomes and ribosomal subunits. GCDIO co-purifies with
eIF-3 biochemical activity and other subunits of the eIF-3 complex through Superose-6 gel filtration
chromatography and MonoS ion-exchange chromatography, and is identical in electrophoretic mobility
to the 62 kDa subunit of eIF-3 described previously. This 62 kDa subunit binds RNA in vitro and we
showed that GCDIO has RNA binding activity by a Northwestern assay using radiolabeled globin
mRNA as the probe. We also discovered that our GCDIO antibodies cross-react with the RNA-binding
66kDa subunit of human eIF-3. Finally, we showed that GCDIO is specifically co-immunoprecipitated
with the -90 kDa subunit of eIF-3 encoded by PRTl

Biochemical studies on mammalian eIF-3 have implicated this factor in several steps of initiation,
including dissociation of SOS ribosomes into free 60S subunits and 46S preinitiation complexes
containing eIF-3 and 40S subunits, binding of eIF-2/GTP/Met-tRNA ^"i ternary complexes to form 43S
complexes, and binding of mRNA to 43S complexes containing both eIF-3 and the ternary complex.
We have proposed that gcdlO mutations reduce the ability of eIF-3 to rebind to 40S subunits and form
46S complexes following termination at uORFl. This would result in "naked" 40S subunits scanning
downstream from uORFl which have a reduced ability to rebind ternary complexes (Fig. S).
Alternatively, gcdlO mutations could decrease the ability of eIF-3 to stimulate binding of elF-
2/GTP/Met-tRNA'^"i ternary complexes to 46S complexes rather than delaying the formation of these
complexes. Either of these last two defects would decrease the rate at which ternary complexes can

BBiraH

■DUBBBBnan

ZOl HD 01004-12 LMG

rebind to 40S subunits scanning downstream from uORFl, explaining why gcdlO mutations cause
ribosomes to ignore uORFs 2-4 and reinitiate at GCN4 in the absence of eIF-2 phosphorylation by
GCN2.

It is thought that eIF-3 functions at multiple steps of the initiation pathway and interacts with most of
the other known initiation factors, including elF-lA, eIF-2A, eIF-3A, eIF-4, eIF-4A, eIF-4B, and eIF-5.
It could thus be imagined that eIF-3 provides a platform on the 40S subunit which facilitates binding
of these factors in the correct orientation with respect to one another and with the decoding sites on the
ribosome. One way of testing this idea is to employ genetic suppressor analysis to identify specific
interactions between particular subunits of eIF-3 and the components of other initiation factors or
ribosomal proteins. Towards this end, we have isolated high copy piasmids from wild-type yeast
genomic libraries that complement temperature-sensitive gcdlO mutations. Five of the six piasmids
partially suppress the derepression of GCN4 (Gcd' phenotype) seen in gcdi. gcdl and gcdlS mutants
in addition to gcdlO mutants; however, none suppresses the temperature-sensitive phenotypes of these
other gcd mutations, or of prtl mutations, suggesting a specific interaction with GCD 10. Hybridization
of the cloned fragments to the Olson-Riles ordered genomic library shows that the suppressor genes
derive from seven genomic loci. Five of these loci contain one of the structural genes for initiator
tRNA^^' and in one case, we have shown that disruption of this gene abolishes suppressor activity.
Thus, it appears that overexpression of initiator tRNA"^'"' can compensate for the general initiation defect
in gc^/Omutants. The simplest explanation for this finding is that binding of the ternary complex to
small ribosomal subunits is defective in the gcdlO mutants. This provides in vivo evidence that eIF-3
stimulates binding of ternary complexes to 40S subunits and supports the idea that gcdJO mutations are
specifically defective for this aspect of eIF-3 function. The fact that multi-copy tRNA*^"! genes do not
suppress mutations in PRT] encoding another subunit of eIF-3, seems to indicate that PRTl mutations
alter this or some other function of eIF-3 in a way that cannot be overcome by simply increasing the
concentration of ternary complexes.

F. Analysis of novel GCD factors

Mutations in the GCDI 3 and GCD14 genes have the same phenotypes as mutations in any of the four
GCD genes encoding subunits of eIF-2B. Thus, GCD13 and GCD14 could be additional subunits of
eIF-3 or components of another factor that influences the production or utilization of the ternary
complex in translation initiation. In accordance with this idea, we found that GCZ)7imutations lead to
polysome dissociation in vivo when mutants are incubated at the nonpermissive temperature. Efforts
to clone GCDI 3 by conventional approaches have failed; therefore, we mapped GCDI 3 to the telomeric
region of the left arm of chromosome XV. Using this information we are attempting to isolate the gene
from ordered lambda clones or cosmids containing this portion of XV.

We confirmed that GCD14 mutations lead to constitutive derepression of GCN4 expression in the
absence of GCN2 and GCN3. The GCD14gene has been isolated and found to encode a protein of ca.
40 kDa that contains a degenerate RNA recognition motif (RRM), and thus could be an RNA binding
protein. Based on peptide sequencing done in John Hershey's lab, GCD 14 does not appear to be the
39kDa subunit of eIF-3.

G. GCNl and GCN20 are components of a heteromeric protein complex that stimulates the
phosphorylation of eIF-2a by GCN2.

GCNl and GCN20 are positive regulators of GCN4 translation that function by stimulating the ability
of GCN2 to phosphorylate eIF-2 in vivo under amino acid starvation conditions. We have shown that
GCNl and GCN20 interact in vitro by co-immunoprecipitation experiments and by using the yeast two-

fssmm

ZOl HD 01004-12 LMG

hybrid system; thus, they are components of a protein complex that is required in vivo to couple GCN2
kinase activity to amino acid availability. Interestingly, ca. 800 amino acids of GCNl shows sequence
similarity to the fungal-specific translation elongation factor EF-3. It is thought that EF-3 fiinctions
in stimulating release of uncharged tRNA from the E (exit) site of the ribosome and thereby stimulates
binding of charged tRNA to the A site. Remarkably, GCN20 also shows strong sequence similarity to
a portion of EF-3 containing two ATP-binding cassettes characteristic of the "ABC" family of
transporter proteins. GCNl, in contrast, does not contain the signature sequences of ABC domains and
is most similar to EF-3 in a region N-terminal to its ABC domains. Thus, GCN20 and GCNl are
similar to different portions of EF-3.

The great majority of ABC proteins are membrane transporters that use the energy of ATP hydrolysis
to pump substrates against a concentration gradient. The typical ABC transporter, typified by the
multiple-drug resistance protein MDR, consists of four membrane-associated domains. Two of these
domains contain six hydrophobic a-helical membrane-spanning segments that form the pore through
which substrates cross the membrane. The other two domains are peripherally located on the
cytoplasmic face of the membrane, bind ATP and couple ATP hydrolysis to the transport process.
GCN20 seems to lack regions sufficiently hydrophobic to function as transmembrane domains; however,
in prokaryotes at least, the individual domains of the transporter can be expressed as separate
polypeptides. It is thus conceivable that GCN20 contains the two ATP-binding domains and interacts
with one or more transmembrane proteins in carrying out a transport function. One possibility would
be that GCN20 interacts with GCNl, which does contain numerous hydrophobic regions, to comprise
one of the transporters identified in yeast that delivers amino acids from the cytoplasm to the vacuole.

A few members of the large family of ABC proteins, such as EF-3, are not associated with membrane
transport events. Therefore, GCN20 and GCNl could be components of an ABC complex with no role
in membrane transport that acts more directly to stimulate GCN2 kinase activity by uncharged tRNA.
One possibility, motivated by their sequence similarity to segments of EF-3, would be that GCNl and
GCN20 interact with the ribosome and facilitate an interaction between uncharged tRNA bound at the
ribosomal A site and the HisRS-related domain of GCN2. This activity could be akin to the proposed
function of EF-3 in stimulating release of uncharged tRNA from the ribosomal E site. We have
identified proteins from diverse eukaryotic organisms, including C. elegans, rice, Arabadopsis, and
humans, that are more closely related to GCN20 than to any other known ABC proteins, and contain
sequence similarity in regions flanking the ABC domains that are not conserved among diverse
members of the ABC family. These sequence similarities could indicate that GCN20 belongs to a new
subfamily of ABC proteins with a biological function conserved from yeast to mammals. If so, higher
eukaryotes may contain a mechanism for coupling the rate of translation initiation to the level of
uncharged tRNA similar to that uncovered in S. cerevisiae involving GCN2 and its positive effectors
GCNl and GCN20.

We have used our antibodies against GCNl to localize the protein in yeast cells by indirect
immunoflourescence. Our results to date indicate that the protein is uniformly distributed throughout
the cytoplasm. Its distribution appears distinctly different from that of a known vacuolar membrane
protein, VAT2, that was analyzed in parallel. These results do not support the model that the
GCN1/GCN20 complex are components of a membrane transporter. We have also been

investigating the importance of the ABC domains in GCN20 for its regulatory function by making
deletions and site-directed substitutions of conserved residues in the Walker A and B motifs of the ATP-
binding cassettes. Surprisingly, mutations in these motifs reduce but don't eliminate GCN20 function.
In fact, a deletion allele lacking all 634 residues C-terminal to amino acid 118, which includes both

ZOl HD 01004-12 LMG

ABC domains, retains substantial complementing function. In contrast, a deletion of residues 4-118 that
leaves both ABC domains intact completely abolishes GCN20 function. These mutations do not
significantly lower GCN20 protein levels. The N-terminal 118 residues is sufficient for interaction with
the C-terminal two-thirds of GCNl in the yeast two-hybrid system. In addition, a GCN20-lacZ fusion
protein containing only the N-terminal 118 residues of GCN20 can be co-immunoprecipitated with
GCNl, whereas a nearly full-length GCN20-lacZ fusion lacking only residues 4-118 fails to interact
with GCNl. Thus, the N-terminal domain of GCN20 is both necessary and sufficient for its function
as a positive regulator of GCN2 and for its physical interaction with GCNl. The fact that the ABC
domains of GCN20 are partly dispensable for its positive regulatory function also seems to be at odds
with the idea that GCNl and GCN20 form an amino acid transporter because ATP hydrolysis is
essential for the function of ABC transporter proteins. However, we found a predicted ABC protein
highly similar to GCN20 (Yer036), in the yeast genome database that could conceivably substitute for
the ABC domains of GCN20 in a gcn20 mutant that retains only the N-terminal 1 18 residues.

Using the yeast two-hybrid system, we showed that the N-terminal 1 18 amino acids of GCN20 could
interact with the C-terminal two-thirds of GCNl. We have been extending this analysis by testing
different overlapping segments of GCNl for interaction with the N-terminus of GCN20. Our results
thus far indicate that all GCNl fragments that interact with GCN20 contain the EF-3-related segment
of GCNl. The smallest of these segments contains 345 amino acids and is coincident with the region
most highly related to EF-3.

H. GCN2-independent derepression of GCN4 translation.

The regulatory mechanism described above applies to the derepression of GCN4 translation that occurs
under conditions of prolonged amino acid-limited growth. There is evidence that a different pathway
is responsible for the transient derepression of GCN4 that accompanies a "shiftdown" from amino acid-
rich to minimal medium. We uncovered another instance of GCA^2-independent induction of GCN4
translation in strains overexpressing tRNAs under conditions in which the excess tRNA cannot be
aminoacylated. The most complete derepression occurred in strains overexpressing a mutant
tRNA^"'(AAC) with a mutation in the acceptor stem that is expected to abolish aminoacylation.
Overexpression of the mutant tRNA^'' derepresses GCN4 translation under nonstarvation conditions in
a manner dependent on the uORFs and independent of GCN2 and Ser-51 on eIF-2a. Overexpression
of the mutant tRNA^"' did not affect cellular growth under nonstarvation conditions in an otherwise
wild-type strain; however, it exacerbated the slow-growth phenotype of a GCNl" mutant. These
findings suggest that the presence of excess unchargable tRNA^'' lowers the level of eIF-2 activity by
a mechanism that does not involve eIF-2a phosphorylation. A weaker GCyV2-independent derepression
was observed in gcn2A strains overexpressing wild-type tRNA"" under histidine starvation conditions,
indicating that the GCA^2-independent mechanism can be activated by an excess of uncharged wild-type
tRNA.

II. Regulation of the mammalian eIF-2a kinase DAI in yeast.

In mammalian cells, stimulation of the latent protein kinase activity of the double-stranded-RNA-
activated inhibitor of translation, DAI, occurs in response to viral infection. The N-terminus of DAI
contains two copies of a sequence motif found in several dsRNA-binding proteins and numerous
deletion and point mutations have been introduced into these sequences that affect dsRNA binding in
vitro. We showed that DAI expressed at low levels in yeast mimics the function of GCN2 in
stimulating translation of GCN4, and inhibits total protein synthesis when expressed at high levels.
Exploiting these findings, we went on to demonstrate that the dsRNA-binding domain is required for

imiMIUIUIMlHI

ZOl HD 01004-12 LMG

DAI kinase function in yeast cells and that the dsRNA-binding motifs are involved in the activation
mechanism. Secondly, we find that the N-terminal copy of the dsRNA-binding domain plays a greater
role than the C-terminal copy in activating kinase function in yeast ceils. Thus, the requirements for
dsRNA binding in vitro and for kinase activation in vivo closely coincide, providing direct evidence that
dsRNA binding to the repeated motifs stimulates DAI kinase activity in vivo.

We have been attempting to identify sites of autophosphorylation in PKR that are important for its
activation by dsRNA in vivo. Mike Mathew's lab have identified several tryptic peptides from the N-
terminal regulatory domain of PKR that are autophosphorylated in vitro by PKR purified from cultured
human cells. In collaboration with their group, we have analyzed the effects of making alanine
substitutions in the various Ser and Thr residues in these peptides on PKR function in yeast. Our results
thus far indicate that substitutions at a combination of these potential phosphorylation sites partially
impairs PKR activation. This result implies that, although these sites may be important for activation,
there are probably additional sites of autophosphorylation in the catalytic domain of PKR.
Accordingly, we have also been examining the importance of two Thr residues located in the insert
between kinase subdomains VII and VIII as potential autophosphorylation sites. The activity of several
protein kinases, including PKA, MAPK, MAPKK and CDK requires phosphorylation or
autophosphorylation of one or two closely spaced Ser or Thr residues in this segment. In the case of
CDK, this phosphorylation is thought to alter the conformation of a loop that protrudes into the active
site and prevents proper positioning of the protein substrate. It can also promote proper interactions
between the N-terminal and C-terminal lobes of the kinase domain, as in the case of PKA and MAPK.
We have preliminary evidence that Ala substitutions at Thr-445 and Thr-450 each decrease PKR kinase
function in vivo, and completely inactivate it when combined in the same protein. These mutant
proteins are expressed at high levels and migrate in SDS-PAGE with the same mobility as the
subdomain II K296R mutant protein; thus, the mutations at Thr-445 and Thr-450 appear to prevent
autophosphorylation of PKR on most, if not all of its phosphorylation sites.

III. Mechanism of transcriptional activation by GCN4.
A. Mutational analysis of the GCN4 activation domain.

Previous work by Struhl and his colleagues established that GCN4 contains a potent transcriptional
activation domain located in the center of the protein between residues 107-144. There were indications
that additional activation determinants were located N-terminal to this central acidic activation domain
(CAAD); however, they appeared to be insufficient for high level activation in the absence of a portion
of the CAAD. We set out to locate more precisely the activation determinants in the N-terminal portion
of GCN4 and to compare the relative efTiciencies of transcriptional activation by the CAAD versus the
N-terminal activation domain (NTAD) when GCN4 is expressed at physiological levels. Towards this
end, numerous deletion and point mutations were constructed in a single-copy GCN4 gene containing
the native promoter and translational control elements, and analyzed for their effects on gene expression
after inducing the mutant proteins by histidine starvation. The levels of all mutant proteins were
quantified by immunoblotting using antibodies raised against the DNA-binding and dimerization domain
of GCN4. Because activation domains in several well characterized mammalian activators, most notably
VP16 and yeast GAL4, contain hydrophobic residues critically required for the activation functions of
these proteins, we also wished to determine whether this structural feature extended to the CAAD or
NTAD of GCN4 protein.

Our analysis showed that GCN4 contains a potent activation domain in its N-terminal 100 amino acids
that confers nearly wild-type transcriptional activation of HISS or HIS4 in the complete absence of the
CAAD. Similarly, the CAAD can promote high-level transcription when the NTAD is missing. The

ZOl HD 01004-12 LMG

CAAD and NTAD are both strongly dependent on the ADA2 protein for activating transcription at HIS3
and HIS4. The CAAD was found to be critically dependent on two clusters of aromatic and bulky
hydrophobic residues: Met-107, Tyr-110, Leu-113 and Trp-120, Leu-123 and Phe-124. The latter
corresponds to a short stretch of amino acids (Trp-X-Ser-Leu-Phe), that is conserved between the
activation domains of GCN4 and its homologues in Neurospora (cpc-1) and Aspergillus (G. Braus,
personal communication). The NTAD was found to be dependent on a pair of Phe residues at positions
97-98. Consistent with the deletion analysis, point mutations in the hydrophobic and aromatic residues
in the CAAD impair GCN4 function only when combined with a deletion of the NTAD or with
substitutions at Phe-97, Phe-98. Likewise, deletions in the NTAD or point mutations at Phe-97 and
Phe-98 impair GCN4 function only when the CAAD is missing or bears mutations at the
aforementioned hydrophobic and aromatic residues located between Met-107 and Phe-124. Combining
together substitutions at all eight hydrophobic residues in full-length GCN4 essentially eliminated
activation of HISS and reduced activation of HIS4 to 20% of wild-type. These findings indicate that the
activation domain of GCN4 is more complex than was previously imagined and shares significant
structural similarities with activation domains found in other acidic activators.

In accordance with the results just described, a deletion that removes the entire CAAD and Phe-97 and
Phe-98 does not completely abolish activation of HIS4. Surprisingly, overexpression of this mutant
protein achieved by removing the uORFs is lethal under nonstarvation conditions. This lethality is
dependent on the GCN4 DNA binding domain, suggesting that it results from sequestration of one or
more general transcription factors in non-productive complexes by the excess mutant GCN4 protein.
Based on these observations, we set out to identify activation determinants in the NTAD located in the
region N-terminal to position 97. To accomplish this goal, PCR was employed in a random mutagenesis
of the NTAD between residues 17-100 in a GCN4 allele lacking the CAAD, and alleles with greatly
reduced GCN4 function were isolated and sequenced. Thirty out of 42 alleles analyzed contained a
mutation at one of 6 Phe residues present in the NTAD segment that we mutagenized (Phe-45, Phe-48,
Phe-67, Phe-69, Phe-97, and Phe-98). Using site-directed mutagenesis, we found that single alanine
substitutions of these Phe residues could account for the Gen" phenotype of all 30 alleles.

In one of the 12 mutant alleles in which a Phe residue was not altered, the mutant phenotype resulted
from a lysine substitution at Glu-88. This was the only mutation at an acidic residue that had a strong
effect on GCN4 function. Moreover, we found that an alanine substitution at Glu-88 has little effect
on activation, indicating that an acidic residue is not required even at this position. Interestingly, Glu-88
maps within the only short stretch of amino acids in the NTAD that is conserved between GCN4 and
its relatives in Neurospora and Aspergillus: Thr*^-X-Leu-X-X-Pro-X-Leu*', located 8 residues N-
terminal to the critical Phe-97, Phe-98 pair. Using site-directed mutagenesis, we found that Leu-89 is
the sole conserved residue in this segment that is required for activation by GCN4. Thus, the Lys
substitution at Glu-88 probably reduces activation by interfering with the function of the adjacent
residue, Leu-89. It is noteworthy that leucine has a bulky hydrophobic side-chain and can partially
substitute for certain critical Phe residues in the VP16 activation domains.

Two additional Phe residues are present in the N-terminal domain of GCN4 (Phe-9 and Phe-16) just
outside the region that we mutagenized by PCR. Using site-directed mutagenesis, we found that Phe-9
is also critical, whereas Phe-16 makes only a minor contribution to NTAD function. A methionine at
position 14 is also dispensable for activation by the NTAD. The latter results show that not all bulky
hydrophobic or aromatic residues in the NTAD are functionally important. This conclusion was
underscored by the fact that alanine substitutions at Leu-65 and Leu-71( located very close to the
critical Phe-67 Phe-69 pair) and at Leu-84, Val-93 and Val-94 (in the vicinity of Phe-97 Phe-98), have

ZOl HD 01004-12 LMG

no effect on NTAD function. To determine whether Phe is uniquely required at residue 45, we
randomized this codon and isolated alleles covering a wide range of GCN4 function. Sequencing these
alleles indicated the following functional hierarchy for different amino acids at position 45:
Trp, Phe > Leu > Tyr > He > Val,Ser > Thr, Pro, Glu, Gin, Cys, Ala, Lys, Gly,
with the first group conferring wild-type activation and the last showing the null phenotype. Thus,
either an aromatic or bulky hydrobphobic residue seems to be required at position 45. Finally, we
verified that alanine substitutions at the three new Phe clusters in the NTAD at positions 9/16, 45/48
and 67/ 69 reduce activation by full-length GCN4 only when combined with point mutations in the
CAAD at Met-107, Tyr-110, Leu-113 and Trp-120, Leu-123 and Phe-124. In addition, we used
immunoblot analysis to verify that these mutations do not reduce the level of GCN4 protein in the cell.

Taken together, our results indicate that GCN4 contains a minimum of 7 clusters containing 2-3
aromatic or bulky hydrophobic amino acids distributed throughout the N-terminal 150 amino acids that
comprise its two activation domains. We cannot determine from our data whether these residues are
required for the overall structure of the activation domains or for specific contacts they make with co-
activators or basic transcription factors. It is worth noting, however, that certain transcription factors
contain critical hydrophobic residues, including TBP and TAF||230, and it has been suggested that
hydrophobic interactions are important in stabilizing contacts between activators and their targets in the
transcriptional machinery.

IV. Genetic analysis of the regulation purine biosynthetic genes.

A. Identification of the minimal cis-acting elements in the ADE5, 7 promoter required for adenine-

repressible transcription.

The transcription of genes encoding enzymes of de novo purine nucleotide biosynthesis, is repressed
by addition of adenine to the medium. The BASl and BAS2 proteins are required for high-level
expression ofADE5,7, ADE8, ADEl and ADE2 under derepressing conditions (minimal medium lacking
adenine), and it is believed that adenine repression involves down-regulating the ability of BASl or
BAS2 to stimulate transcription. One way this could occur would be if a repressor protein binds to the
ADE genes when excess adenine is present in the medium. According to this model, a negative
regulatory site would be present in the promoters and its removal would lead to constitutively
derepressed expression. To test this possibility, we set out to identify the minimal sequence element
from ADE5,7 sufficient for adenine-regulated transcription. We found that a 74 bp fragment containing
twoTGACTC elements separated by 33 bp confers BASl/2-dependent, adenine-repressible transcription
on a CYCl-lacZ construct that contains a TATA box but lacks the native CYCl UAS elements. The
gene-proximal TGACTC sequence in this fragment is required for promoter function and purine
regulation, whereas the distal element augments expression that is fully dependent upon the proximal
site. Interestingly, the proximal TGACTC sequence is more critical than the distal sequence at ADEl
as well. Unexpectedly, deletion of the region between the two TGACTC sequences decreased
expression dramatically from the ADE5, 7-CYCl hybrid construct.

To map in greater detail the sequence requirements for this regulatory element, we are making clustered
substitutions in consecutive 3-bp intervals extending across a 67 bp fragment that contains only the
proximal TGACTC sequence and ca. 30 5'- and 3 '-flanking nucleotides. From the results obtained thus
far, we find that substitutions in the TGACTC core element and in the 6 nt immediately 3' to this
sequence (GTGTCC) eliminate both expression and adenine regulation of the hybrid promoter,
presumably due to inactivation of the BASl binding site. Changes in the sequence TAATAA located
10 nt 3' to the TGACTC site also greatly reduce expression and adenine regulation. These latter

mna

ZOl HD 01004-12 LMG

nucleotides might also be required for BASl binding in vivo; alternatively, they may comprise a BAS2
binding site. Consistent with a requirement for sequences located between the two TGACTC elements,
substitutions in the sequence CCGTCGGTAGTGACA located 24 nt 5' of the proximal TGACTC
sequence either reduce or eliminate both expression and adenine regulation. Inspection of this sequence
and consideration of the relative effects of mutations at different positions in the sequence strongly
suggests that it constitutes a binding site for ABFl protein. None of the ca. 50 clustered substitution
mutants we analyzed thus far have the derepressed phenotype expected from inactivation of a negative
control site where a repressor protein would bind. Therefore, our results are most consistent with the
idea that adenine repression of transcription involves antagonizing the function of the positive factors
BASl, BAS2 or ABFl, through a protein modification like phosphorylation or by a protein-protein
interaction with a repressor protein. BASl is the most likely target for regulation because both ABFl
and BAS2 interact with promoters that are presumably unresponsive to purine levels.

B. Genetic selections for trans-acting mutations that abolish adenine-mcdiated repression of
transcription.

We wish to to identify additional factors that might be involved in regulating BASl or BAS2 by
adenine and to delineate the regulatory domains in BASl or BAS2. Towards these ends, we have
constructed yeast strains in which defects in adenine repression lead to an easily detectable growth
phenotype. These strains contain a fusion of the ADE5, 7 promoter to the HIS3 coding region replacing
the wild-type HIS3 gene. These strains are His* on minimal medium but His" when grown on of
adenine, due to repression of the ADE5,7-HIS3 construct. By selecting for His* revertants on adenine-
containing medium, we hope to isolate mutations that abolish adenine repression of ADE5, 7. These
strains also contain an integrated ADE5, 7-lacZ construct to identify mutations that derepress ADE5, 7
in trans, an adel allele to eliminate adenine uptake mutants that are derepressed because of low
intracellular adenine pools, and a gcn4A allele to avoid isolation of gcd mutations. We have begun
using these strains to select three kinds of regulatory mutations.

First, we mutagenized a plasmid-bome copy of BASl by hydroxylamine and selected two dominant
alleles that allow growth of the test strain on +Ade -His medium. The products of these BASl alleles
may be insensitive to adenine repression or may have increased affinity for DNA or other proteins with
which BASl interacts to activate transcription. It is unlikely that they simply increase BASl protein
levels because wild-type BASl on a high copy plasmid does not confer a His* phenotype in the test
strain. These two alleles have a relatively weak phenotype; therefore, we are screening a second pool
of BASl plasmids mutagenized with an E. coli mutator strain to obtain alleles with a stronger His*
phenotype. In addition, we will sequence the two existing mutations and construct a double mutation.
We will also screen a pool of BASl plasmids mutagenized with the mutator strain for the same
phenotype.

In a second approach, we screened high copy plasmid yeast genomic libraries for dosage suppressors.
Among the His* transformants analyzed thus far, the only suppressor plasmid we obtained (that does
not contain HISS) encodes BAS2. As just mentioned, high copy-number BASl does not suppress the
His' phenotype of the tester strain (although we have not confirmed that BASl is overexpressed from
this plasmid). These findings could indicate that BAS2 is rate-limiting for activation of ADE5, 7
transcription, in accordance with the fact that BAS2 is involved in regulating other genes besides ADE
genes. It also suggests a model in which BASl must recruit BAS2 to an ADE promoter to activate
transcription, and this function of BASl is diminished by excess adenine. We are now determining
whether overexpression of BAS2 leads to constitutively derepressed ADE5,7-lacZ expression, as

lllllllllilllllilllllilllllMlllllllllilllllllilllllllilllllMlililiiililllli

ZOl HD 01004-12 LMG

predicted by the model just described, or whether ADE5,7 expression is elevated under derepressing
conditions but remains adenine-repressible even when BAS2 is overexpressed. The latter would indicate
that recruitment of BAS2 from other promoters is not the adenine-repressible parameter.

In a third approach, we have isolated chromosomal mutations that suppress the His" phenotype of the
test strain on +Ade -His medium. Thus far, we obtained 10 His* revertants in which expression of the
ADE5,7-lacZ fusion is derepressed 2-3 fold on medium containing adenine, indicating the presence of
a trans-acting mutation that impairs adenine repression. We are investigating whether this phenotype
is conferred by a single mutation in each revertant and determining the number of complementation
groups involved. Once this analysis is completed, we will clone the wild-type alleles of the suppressor
genes by complementing the His* phenotype of the mutants. It will also be worthwhile to carry out
insertional mutagenesis with the Snyder library using the tester strains described above and selecting
for His* mutants on +Ade -His medium. If the key regulatory factors are nonessential, this approach
should allow rapid identification and isolation of the genes involved.

Significance to Biomedical Research and the Program of the Institute:

Yeast is a eukaryotic organism that carries out many of the fundamental processes of eukaryotic cells
and offers a level of molecular genetic analysis that cannot be achieved with multicellular organisms.
General amino acid control and the regulation of purine biosynthetic genes involve a variey of strategies
for regulating gene expression. Investigating the molecular details of these two systems is providing
useful paradigms for coordinate transcriptional activation of unlinked genes, transcriptional repression,
gene-specific translational control, functional modifications of protein complexes involved in general
translation initiation, and the involvement of protein kinases and protein phosphatases in signal-
transduction pathways. The results of these inquiries can be expected to provide valuable insights into
how higher eukaryotic cells control the availability of amino acid and nucleotide substrates for protein
and nucleic acid synthesis in response to changes in the environment. In fact, GCN4 translational
control appears to be a gene-specific instance of a highly conserved mechanism that regulates protein
synthesis in mammalian cells under conditions of stress such as amino acid starvation. The regulatory
mechanisms under study in yeast are critical elements of the network of interrelated controls that couple
growth rate and division to environmental stimuli in all eukaryotic cells.

Proposed Course of the Project.

A. Investigation of the role of poly-A binding protein (PAB) in reinitiation on GCN4 mRNA.

Previous studies have suggested that inactivation of PAB impairs general translation initiation and that this
defect can be partially overcome by a variety of mutations that reduce 60S subunit biogenesis. This has led
to the idea that PAB is involved in some aspect of 40S-60S subunit joining; however, its precise role in
translation initiation is unknown. We propose to isolate mutations in PAB that perturb GCN4 translational
control as a means of increasing our understanding of PAB function in translation. There are three principal
ways that perturbing PAB function could affect GCN4 expression. First, it could decrease the ability of
ribosomes to remain attached to the mRNA, or to resume scanning and re-form an initiation complex, following
termination at uORFl. This would produce a Gen' phenotype. Second, it could decrease the dissociation of
ribosomes from the mRNA following termination at uORF4, causing a Gcd' phenotype. A third possibility is
that PAB could be required for efficient subunit joining during reinitiation. If PAB must rebind to ribosomes
as they scan downstream from uORFl, as postulated for the eIF-2GTP-Met-tRNA'"", ternary complex, then
reduced-function mutations in PAB should have a greater effect on reinitiation at uORF4 than at GCN4 and
produce a Gcd' phenotype. They would lead to increased GCN4 expression only when uORFl and uORF4 are

o
iS

ZOl HD 01004-12 LMG

both present in the leader, whereas a Gcd' phenotype arising from reduced ribosome dissociation following
uORF4 translation (the second mechanism above) would be seen with uORF4 alone.

To isolate mutations in PAB affecting GCN4 translation, we will mutagenize the cloned gene on a plasmid,
subclone various segments into a wild-type copy oi PAB and analyze the resulting plasmids in a pablA strain
as the only copy of PAB by plasmid shuffling. Mutant plasmids will be screened for altered sensitivity to 3-AT
or 5-FT to identify Gen' or Gcd" phenotypes, respectively. It will be particularly interesting if mutations having
different effects on GCN4 are found to map in different domains of PAB. This mutational analysis should also
be helpful in identifying portions of the protein that are specifically involved in translation initiation as opposed
to other potential functions of PAB in poly(A) tail metabolism and mRNA turnover. In addition, mutations with
a 3-AP phenotype could be used in suppressor analysis to identify proteins that interact with PAB in the
translation initiation pathway.

B. Molecular genetic analysis of eIF-2B.

1. In vitro analysis of protein-protein interactions in the GCD2/GCD7/GCN3 subcomplex.

From immunoprecipitation experiments, we obtained biochemical evidence that GCD2, GCD7 and GCN3 can
form a stable subcomplex in vivo. To test our prediction that the GCD2/GCD7/GCN3 subcomplex can interact
with eIF-2, we intend to purify the subcomplex from yeast, or reconstitute it in vitro from proteins
overexpressed in E. coli, and probe for interactions with eIF-2 purified from yeast. With the higher protein
concentrations that can be achieved with purified preparations, we hope to detect complexes that are not stable
enough to be detected by immunoprecipitations from cell extracts. We have constructed plasmids for
overexpression of GCD2, GCD7 and GCN3 in E. coli and established conditions for purification of these
proteins from bacteria. We can now purify 1-2 mg of glutathione-S-transferase (GST)-tagged versions of
GCD7 and GCN3, from which the GST moiety can be cleaved in vitro by thrombin, and a His-tagged version
of GCD2. We will attempt to form a trimeric complex with purified GCD2, GCN3 and GST-GCD7 fusion
proteins and isolate the complex on a glutathione column. If we find that GCD2 must be present in the
reactions in order to retain GCN3 on a glutathione column with GST-GCD7, this will be a good indication of
specific complex formation between all three proteins. One important application for this assay will be to
identify residues in GCD2 that are required for its interaction with GCD7 and GCN3. Initially, we will analyze
the effects of deleting different segments of GCD2 on its ability to be specifically retained by the GST-GCD7
protein on a glutathione column. It should not be necessary to purify the various mutant GCD2 proteins from
the bacterial extracts since their binding to the column can be assayed by immunoblot analysis. We should be
able to study interactions between GCD2 and GCD7 by this approach even if the trimeric complex with GCN3
cannot be formed, since we have in vivo evidence for a GCD2-GCD7 complex. Using the His-tagged version
of GCD2, we can also identify residues in GCD7 required for its interaction with GCD2, and it will be
interesting to learn whether regions of homology between the two proteins mediate their physical interactions
with one another.

If we succeed in reconstituting a GCD2-GCD7-GCN3 subcomplex, we will attempt to demonstrate binding of
purified eIF-2 to this complex and analyze the effects of phosphorylating eIF-2 on the interaction. Wild-type
eIF-2 will be purified according to our previously published procedures or as described below, and
phosphorylated in vitro using recombinant PKR as already described. The elF-2 will be applied to a nickel or
glutathione column containing bound GCD2-GCD7-GCN3 trimeric complex and its retention on the column
will be analyzed or by immunoblotting using antibodies against subunits of eIF-2. The K^ for the mammalian
eIF-2/eIF-2B complex has been estimated between 0.1-1.0 nM (S. Kimball, personal communication). Thus,
even if the Kp for interaction between eIF-2 and the GCD2-GCD7-GCN3 trimeric subcomplex is only 1 )aM,
we should be able to detect it using this technique if we bind ca. 25 |ig of the reconstituted trimeric complex
to ca. 50 |iL of resin. We will establish the specificity of an interaction between eIF-2 and the elF-2B

^^u^^ma^mai^^mmmmamammmim

M^n

ZOl HD 01004-12 LMG

subcomplex by showing that it depends on both GCD2 and GCD7 and that the affinity is higher when GCN3
is also present in the complex. We anticipate that phosphorylation of eIF-2 will increase its affinity for the
trimeric complex and that introduction of one or more Gen" mutations into the GCD2, GCD7 or GCN3 proteins
will either diminish the affinity for phosphorylated eIF-2 or increase the affinity for nonphosphorylated eIF-2.
An increase in affinity will be detected by retention of the same fraction of applied eIF-2 in the presence of less
trimeric complex bound to the column.

2. Studies on the regulation of mammalian eIF-2B by phosphorylated eIF-2.

There is a high degree of sequence similarity between GCD2, GCD7 and GCN3 and the corresponding subunits
of mammalian eIF-2B. Interestingly, a number of the Gen" alleles we isolated in these subunits alter residues
that are conserved between the yeast and mammalian proteins. We wish to determine whether these amino
acids are involved in regulating mammalian eIF-2B by eIF-2(aP) in the same way that they function in yeast.
To test this idea, we used site-directed mutagenesis to introduce the amino acids present in five Gen' alleles of
GCD2 at the corresponding positions of the rat eIF-2B5 cDNA. These mutant cDNAs are being inserted into
different vectors designed for expression in mammalian cells. We will examine the effects of expressing these
mutant subunits of eIF-2B on the response to eIF-2 phosphorylation in mammalian cells by the following two
assays.

It has been established that certain expression vectors activate PKR in COS cells and this leads to greatly
reduced expression of the DHFR gene contained on the same plasmid. This inhibition has been attributed to
localized activation of PKR by double-stranded RNAs formed between DHFR mRNA and antisense RNAs
made from the same plasmid. Overexpression of the non-phosphorylatable eIF-2a-S51A protein from a
different plasmid can rescue expression of DHFR from the PKR-sensitive vector. The overexpressed of eIF-2a-
S51A protein replaces wild-type eIF-2a in the trimeric eIF-2 complex and thereby reduces the amount of eIF-2
in the cell containing the phosphoryated form of the a subunit. We will determine whether overexpression of
the 5 subunit of rat eIF-2B bearing one or more Gen' mutations has the same effect on DHFR expression as
that described for eIF-2a(S51A). Each of the constructs containing wild-type or mutant rat eIF-2B5 cDNAs
inserted in the vector pMTVA- will be co-transfected into COS-1 cells with plasmid pD61 encoding the DHFR
reporter gene and expression of DHFR will be analyzed as described. As controls, the DHFR reporter plasmid
will be co-transfected with pMTVA- derivatives encoding wild-type eIF-2a or eIF-2a-S51A, provided by R.
Kaufman. If we find that expression of one or more mutant forms of eIF-2B5 leads to increased DHFR
expression relative to that given by the wild-type 5 subunit, this will imply that the mutation renders
mammalian eIF-2B less sensitive to the inhibitory effects of eIF-2(aP), just as occurs in in yeast. These
experiments should also reveal the extent to which eIF-2B mediates the regulatory effects of eIF-2(aP)in
mammalian cells.

In collaboration with Scot Kimball's lab the same mutant and wild-type rat eIF-2B5 cDNAs will be expressed
from the pBK-RSV vector in rat cells and analyzed for their effects on eIF-2B nucleotide exchange activity in
cell extracts. They have established procedures for increasing eIF-2a phosphorylation by treating rat GH3
pituitary tumor cells with Brefeldin A and shown that eIF-2B in the whole cell extracts exliibits reduced
nucleotide exchange activity for eIF-2f H]GDP binary complexes added as substrate. We will determine
whether overexpression of any of the mutant forms of eIF-2B5 in the rat cells can rescue eIF-2B exchange
activity. It is hoped that the various Gen" mutations will have similar relative effects on eIF-2B activity in
COS-1 cells, as measured indirectly by rescue of DHFR expression, and in the rat cells as measured directly
by assaying eIF-2B exchange activity.

ZOl HD 01004-12 LMG

C. Studies on GCDIO and the eIF-3 complex in GCN4 translational control and general translation
initiation.

1. Domain structure of GCDIO and identification of its RNA binding site.

To begin our analysis of the role of RNA binding by GCDIO in eIF-3 function we will identify the minima!
segment of GCDIO required for RNA binding in vitro and evaluate its importance for GCDIO function in vivo.
We will construct nested deletions progressing from the N- and C-termini of GCDIO in a His-tagged allele
tailored for expression in E. coli and for in vitro transcription/translation in rabbit reticulocytes. The mutant
proteins will be tested for RNA binding by Northwestern and by binding to RNA immobilized on Separose.
After identifying the minimal fragment of GCDIO required for RNA binding in vitro, we will analyze the in
vivo effects of deletions in this domain. Assuming that the RRM-like sequence will be required for RNA
binding in vitro, we will also examine effects of point mutations in the conserved RNP-1 and RNP-2 elements
on RNA-binding in vitro and GCDIO function in vivo. For all these experiments, a myc epitope-tagged GCDIO
allele will be mutagenized and introduced into yeast in place of wild-type GCDIO by plasmid shuffling. We
will determine whether the gcdlO mutations reduce growth rate or lead to derepression of GCN4 (Gcd'
phenotype). If the mutations are lethal, they will be tested for a dominant slow-growth or Gcd' phenotype. If
a dominant-negative phenotype is observed, we will determine whether the mutant protein competes with wild-
type GCDIO for incorporation into eIF-3 by immunoprecipitating with antibodies against PRTl and analyzing
the immune complexes with anti-myc antibodies. If so, mutagenesis of the dominant-negative allele will be
carried out to isolate mutations that abolish this phenotype without reducing the level of the mutant protein.
These mutations should identify residues in GCDIO required for stable complex formation with the other elF-3
subunits. This interpretation will be tested by coimmunoprecipitation experiments. We will also rescue the
existing gcJ70 chromosomal mutations by plasmid gap-repair and sequence the rescued alleles. These mutations
may identify a functional domain of GCDIO that is distinct from those involved in RNA binding and eIF-3
subunit interactions but which is critical for repression of GCN4 translation; eg. a domain that interacts with
eIF-2. If so, we will begin in vitro mutagenesis of GCDIO to isolate additional mutations affecting this domain
by mutagenizing a high-copy plasmid bearing GCDIO and selecting for growth on 5-FT plates (Gcd' phenotype)
in a strain containing wild-type chromosomal GCDIO. By selecting for dominant Gcd' mutations, we should
isolate mutations that alter GCDIO function rather than simply reducing its expression or ability to interact with
other components of eIF-3.

2. Genetic identification of proteins that interact with GCDIO.

We are in the process of localizing and sequencing the remaining high copy suppressor of gcd 10 mutations to
determine if it encodes an additional subunit of eIF-3. We also will begin to screen high copy plasmid genomic
libraries for suppressors of a prtl temperature-sensitive mutation, and a Pqali-cDNA expression libraty for
suppressors of both gcdlO and prtl mutations. This suppressor analysis may identify proteins that specifically
interact with these subunits of eIF-3 in addition to isolating the remaining subunits of the eIF-3 complex. If
we do not obtain any new suppressor genes in these approaches, we will turn to the isolation of synthetic lethal
mutations that render gcdlO ox prtl mutants inviable at a permissive or semi-permissive temperature.

John Hershey's lab at U.C. -Davis is obtaining peptide sequences from the 135, 33, 29 and 21 kDa subunits of
yeast eIF-3 and has agreed to share this information with us. By this approach, they have already identified
the 39 and 135 kDa subunits as sequenced ORFs of previously unknown function, encoded on Chromosome
XIII, whereas none of our suppressors maps to XIII. Presumably, the structural genes for the remaining
subunits of eIF-3 will be cloned either as one of our high copy suppressors or from the peptide sequences
obtained in Hershey's lab. It is important to verify that all of the proteins that co-purify with eIF-3 in Hershey's
purification scheme can be co-immunoprecipitated with the GCDIO and PRTl proteins. Thus, we will construct
epitope tagged alleles of the 39 and 135kDa subunits ( and any others that are identified in the yeast database

IBBB— BBBW

ZOl HD 01004-12 LMG

from the peptide sequences) and replace the wild-type chromosomal alleles with the tagged versions. These
strains will be used to determine whether the epitope-tagged proteins can be quantitatively immunoprecipitated
with GCDIO and PRTl. These cloned genes can be used in the future to study interactions between GCDIO
and other subunits of eIF-3 with the same techniques described above for studying GCD7-GCD2 interactions.

3. Analysis of GCD13 and GCD14.

To learn more about the function of GCD14, we will test it for RNA binding activity using the assays described
above for GCDIO. We will also analyze existing temperature-sensitive gcdl4 mutations for polysome profiles
indicative of general defects in translation initiation. We will construct an epitope-tagged functional allele of
GCD14, or raise antibodies against a trpE fusion protein expressed in bacteria. Using antibodies against
GCD14, we will investigate whether the protein is associated with polysomes, 40S or 60S subunits fractionated
by velocity sedimentation on sucrose gradients, and also test for interactions with eIF-2eIF-2B or eIF-3 by co-
imunoprecipitation experiments. The isolation of dosage suppressors of gcdl4 mutations will also be initiated.
We will screen the Pc^li"'^^^^ expression library for a GCDJ3 cDNA clone.

D. Regulation of protein kinase GCN2

1. Analysis of the N-terminal regulatory domain and potential autophosphorylation sites.

Use Barthelmess' group isolated a gene from A', crassa {cpc-3) encoding a GCN2 homologue (NcGCN2), with
40% and 32% sequence identity to the protein kinase and histidyl-tRNA synthetase-related domains of GCN2,
respectively. Deletion of cpc-3 impairs derepression of the GCN4 homologue in Neurospora, called cpc-1. In
collaboration with their lab, we are completing the sequence of an ca. 500-residue N-terminal domain of
NcGCN2 and find that it shows significant similarity to the corresponding region in GCN2 (25% identity). In
both proteins, this region contains a degenerate kinase catalytic domain of ca. 300 residues. In GCN2, there
are consensus copies of kinase subdomains VII-XII in the correct order, whereas the NcGCN2 pseudo kinase
domain (OPK) contains subdomains 11, and VIII-XI. N-terminal to the OPK, both GCN2 and NcGCN2 contain
an ca. 80-amino acid segment rich in positively and negatively charged residues. It seems likely that this
charged segment is located on the surface of the protein and, could perhaps be a binding site for an effector
protein.

Two instances of cI>PK domains have been reported in mammalian proteins. The JAK family of tyrosine protein
kinases (that interact with cytokine receptors) contain a OPK domain of unknown function in addition to a
conventional PK domain, whereas atrial natriuretic peptide (ANP) receptor contains a OPK domain N-terminal
to a guanylate cyclase catalytic domain. For the latter, the OPK segment appears to be a negative regulatory
domain that mediates stimulation of the cyclase activity by ATP and also provides a binding domain for a novel
protein phosphatase that may be involved in desensitizing the the ANP receptor by dephosphorylation. By
analogy with the ANP receptor, the OPK domain of GCN2 may play an important role in regulating kinase
activity in response to amino acid or purine limitation, or under unknown conditions of starvation or stress, by
binding another kinase or phophatase that modifies the authentic kinase domain, as there is evidence that GCN2
is phosphorylated by another kinase in vivo. The OPK domain contains all the conserved residues in the C-
terminal lobe located near the peptide substrate binding groove in PKA therefore, it might compete with the
authentic kinase domain for binding eIF-2a. Alternatively, it might physically interact with the authentic kinase
domain, or mediate dimer formation between two molecules of GCN2.

We will take several genetic approaches to address the importance of the OPK domain. In one approach, we
will use site-directed mutagenesis to substitute conserved amino acids in OPK subdomains VII-XII with
alanines. In addition, we will construct chimeras that replace the yeast OPK domain with that from the
Neurospora homologue. If this chimera is functional, it will imply that the exact set of invariant residues left
intact is not critically important for OPK function. We will also generate small in-frame deletions throughout

ZOl HD 01004-12 LMG

the region (using two-codon Sad site insertions made previously, and mutagenize the OPK domain at random
sites in the E. coli mutator strain XL 1 -Red (Stratagene), and screen the mutant plasmids for reduced GCN2
function (Gen" phenotype) or constitutive activation of GCN2 (Gcd' phenotype) using amino acid analogues as
indicators of GCN4 expression. This random mutagenesis should identify residues in the ct)PK that are not
conserved among protein kinases, but which are involved in the regulatory function of the domain, either as
binding sites for other regulatory proteins or as points of contact between the pseudo and authentic kinase
domains or between protomers in a GCN2 homodimer. We will also make deletions in the 80-residue N-
termina! highly charged segment and, depending on their phenofypes, introduce clustered alanine substitutions
throughout this segment of GCN2. In addition, we will replace it with the corresponding segment from
Neurospora NcGCN2 which, ahhough highly charged, shows little primary sequence similarity to the yeast
segment.

Another very interesting target for mutagenesis that could be involved in the function of the OPK domain is
a segment between subdomains VII and VIII of the authentic PK domain of GCN2. The activity of several
protein kinases, including PKA, MAPK, MAPKK and CDK requires phosphorylation or autophosphorylation
of one or two closely spaced Ser or Thr residues in this segment . In the case of CDK, this phosphorylation
is thought to alter the conformation of a loop that protrudes into the active site and prevents proper positioning
of the protein substrate. It can also promote proper interactions between the N-terminal and C-terminal lobes
of the kinase domain, as in the case of PKA and MAPK. GCN2 contains a larger than usual segment separating
subdomains VII and VIII, containing a total of 7 serine and threonine residues. One or more of these residues
could be critical sites of autophosphorylation or phosphorylation by another kinase, or sites of dephosphorylation
by a protein phosphatase. We have mutated one of these residues, Thr-813, to Gly and observed a leaky Gen'
phenotype. We will mutate additional Thr and Ser residues in this segment to Ala, singly and in combination
with an Ala substitution at Thr-813, and analyze the effects on GCN4 expression in vivo. If one or more of
these residues is found to be critical for GCN2 function in vivo, we will determine whether function can be
restored by substitutions with Asp or Glu, which may functionally substitute for phosphoserine or
phosphothreonine. In addition, we will determine whether the mutations reduce autophosphorylation of GCN2
in our standard immune complex assays, or its level of phosphorylation in vivo by labeling cells with
radioactive phosphate, immunoprecipitating GCN2 from whole cell extracts, and analyzing the labeled protein
by SDS-PAGE. If there are multiple phosphorylation sites, it will be necessary to analyze the tryptic
phosphopeptides produced in vitro and in vivo by mutant and wild-type proteins. If the critical site is for
autophosphorylation, it will be necessary to show that substitution of the site with Ser (if the wild-type residue
is Thr) or Thr (if the wild-type residue is Ser) leads to the production of a novel phosphopeptide. If
autophosphorylation occurs in trans following dimerization, we may also suceed in showing that wild-type
GCN2 can phosphorylate the critical residue using a K559V defective kinase as substrate.

2. Analysis of GCN1/GCN20 interactions.

The importance of the EF-3-related domain of GCNl for protein-protein interactions with GCN20 suggests that
the positive regulatory function of GCN1/GCN20 is related to the role of EF-3 in translation elongation. To
pursue this possibility further, we would like to determine whether residues in GCNl that are conserved with
EF-3 are of critical importance for GCNl function. We also wish to identify specific amino acids that are
required for protein-protein interactions with GCN20, as these may illuminate the role of GCN20 in the putative
EF-3-like function of GCNl. GCNl encodes a very large protein of 2,672 amino acids, and in addition to
mutational analysis of the EF-3-related domain (residues 1329-2140), we would like to determine whether other
segments of the protein are important for its regulatory function. We will begin a mutational analysis of GCNl
by testing internal deletions of ca. 300-amino acid segments, constructed using naturally occurring and
engineered restriction sites for: (1) loss of GCNl function, revealed by sensitivity to 3-AT; and (2) constitutively
activated GCNl function, leading to 5-FT resistance. This approach may allow us to eliminate relatively large

iiiiiimiiiyiiiiiiiiMiiii

ZOl HD 01004-12 LMG

segments of GCNl that may not be required for its regulatory function. In a second approach, we have used
the E. coli mutator strain XL-1 Red (Stratagene) to carry out random mutagenesis of GCNl, and we are
subcloning segments of 1-2 kb from this plasm id pool into an unmutagenized GCNl plasmid pool. Screening
of these pools will be restricted to those segments found to be important for GCNl function by the internal
deletion analysis, beginning with the segment corresponding to the EF-3-related region of GCNl. Mutant alleles
will be subjected to Western analysis and those which produce full-length GCNl at wild-type levels will be
sequenced to identify the mutation. The mutant proteins will be tested for protein-protein interactions with
GCN20 using the co-immunoprecipitation assays described above. Any gcnl mutations that reduce co-
immunoprecipitation of the two proteins will be examined for the ability to be suppressed by overexpression
of GCN20 from a high copy plasmid. If an in vitro assay for GCN2 kinase activity showing GCNl -dependence
is available at this time, the GCNl mutations will be analyzed for effects on GCN2 kinase function in vitro.
Any GCNl" alleles we obtain will be tested to determine whether they are dependent on GCN20 for their
derepressed phenotype.

3. Genetic screening for novel proteins that interact with GCN2.

At present, we have not detected a physical interaction between the GCN1/GCN20 complex and GCN2 and
little is known about how these proteins regulate GCN2 kinase activity by uncharged tRNA. The location of
GCN2 on the ribosome and possible roles of specific ribosomal proteins in this regulation are also unknown,
as is the mechanism for coupling GCN2 kinase activity to purine levels. It is unlikely that the gen class of
mutations has been saturated, and there may be additional proteins besides GCNl and GCN20 involved in
controlling GCN2 activity. For these reasons, we will pursue several genetic approaches aimed at the
identification of proteins that interact with GCN2. In one approach, we will use yeast two-hybrid constructs
containing different segments of GCN2 to screen the Elledge cDNA library and the Clontech genomic library.
We will begin with a bait construct containing the OPK domain of GCN2 and proceed as necessary to one
containing the HisRS-like domain, following the same plan outlined above for analyzing two-hybrid interactions
with GCNl bait plamids.

In a second approach, we will attempt to saturate the gen class of mutations by random insertional mutagenesis
of the yeast genome using the insertion library constructed in Mike Snyder's lab. This is a plasmid library of
genomic fragments containing Jn3:.!acZ::LEU2 transposons inserted at random sites. The mutagenized yeast
fragments are removed from the plasmid vector by digesting with NotI and then used to make chromosomal
insertions by transforming a yeast strain to Leu*. Using this library, we will mutagenize a haploid strain that
is wild-tj'pe for all known GCjVand GCD genes (except for GCN4) and contains an integrated HlS3-gus fusion
under the control of GCN4 and an integrated GCN4-lacZ fusion. The strain will carry one of our mutant alleles
of GCN4 which contains only the N-terminal activation domain. This GCN4 allele confers high-level activation
of GCN4-regulated genes under our standard starvation conditions but is defective relative to wild-type
GCN4 under more severe starvation conditions. By incorporating this attenuated allele in place of wild-type
GCN4, we hope to increase the sensitivity of our screen for gen mutations. (In addition, we may identify
mutations in nonessential transcriptional co-activator proteins, such as GCN5, that impair transcriptional
activation mediated by the N-terminal activation domain; see below). Transformants containing insertions in
different loci will be screened for increased sensitivity to 3-AT, sulfometuron (inhibits lie and Val synthesis),
and azaadenine (azA) (inhibits purine biosynthesis), indicating a Gen' phenotype. Transformants showing
sensitivity to 3-AT and SM or to azA will be subjected to quantitative assays of the two integrated fusions after
growth in the presence of 3-AT or aZA and ranked according to the magnitude of their derepression defects.
(Note that 1/6 of the chromosomal insertions will fuse lacZ in-frame to the disrupted gene. For this class of
insertion mutants, we will conduct Western analysis of GCN4 expression using antibodies against GCN4.) Any
mutants showing reduced GCN4 expression under starvation coditions will be analyzed genetically to determine

IIIIIIIIIIIIIIIHIIIIIIIIIII

ZOI HD 01004-12 LMG

whether this phenotype results from a single insertion. Mutants showing reduced HIS3-gus expression but
normal GCN4 expression will be set aside for studies on GCN4 transcriptional co-activators (see below). For
all others, we will isolate genomic sequences adjacent to the insertion site using the plasmid integration-excision
scheme of Bums et al. and determine the sequence of 100-200 nt. If the sequence does not correspond to a
known GCN gene or match an entry in the yeast database, we will determine the map location by hybridization
to the Riles-Olson lambda library and clone the wild-type gene from the appropriate lambda or cosmid clone
purchased from ATCC. The gene will be localized by complementation testing, completely sequenced, and a
chromosomal deletion will be constructed to determine the null phenotype with respect to GCN4 expression and
eIF-2a phosphorylation.

E. Molecular mechanism of transcriptional activation by GCN4

1. Biochemical analysis of potential interactions between GCN4 and transcriptional mediators.

Our mutational analysis of GCN4 has identified two functionally distinguishable activation domains that each
contain multiple clusters of 2-3 aromatic or bulky hydrophobic amino acids that are critically required for the
function of that domain. Five such clusters have been identified in the NTAD and 2 are located in the CAAD.
Some or all of these residues could mediate important interactions between GCN4 and its target proteins in the
transcriptional machinery. Genetic analysis has implicated the ADA2, ADA3 and GCN5 proteins as mediators
of transcriptional activation by GCN4 and several other acidic activators; however, at present, there is no
evidence that GCN4 physically interacts with these proteins or with any of the general transcription factors.
We plan to use the point mutations we generated in the activation domains of GCN4 to identify target proteins
in the transcription machinery and to identify the domains in these proteins that contact GCN4.

We are testing GCN4-GST fusion proteins for interactions with candidate proteins synthesized by in vitro
transcription/translation, focusing initially on the TATA-binding protein TBP. A panel of GCN4-GST fusion
proteins are being overexpressed in E. coli, including full-length GCN4, deletion derivatives lacking the NTAD,
CAAD, or both, and several full-length constructs containing mulitple point mutations in both the NTAD and
CAAD. Bacterial extracts containing these proteins are mixed with aliquots of in vitro translation extracts
containing "S-methionine-labeled yeast TBP, after which we determine the amount of radiolabeled TBP that
is retained with the GST-GCN4 proteins on glutathione-agarose beads (GST pull-down assay) . We have also
carried out reciprocal reactions in which "S-methionine-labeled GCN4 proteins synthesized in vitro are tested
for interactions with a GST-TBP fusion protein. In control experiments, we have reproduced the interaction
between VP16 and TBP that depends on the C-terminal activation domain of VP16. In addition, we observed
GCN4-GCN4 homodimer formation. Although we detected an interaction between TBP and GCN4, it appeared
to be relatively weak compared to the TBP-VP16 interaction. In addition, the DNA binding-dimerization
domain at the C-terminus of GCN4 seemed to be both necessary and sufficient for the in vitro interaction with
TBP. Consequently, we will not pursue this interaction for the time being. We also used the GST pull-down
assay to explore interactions between yeast TFllB and GCN4, but thus far have not observed any interaction
between them. In future experiments we will to examine possible interactions between GCN4 and the individual
ADA2, ADAS and GCN5 proteins synthesized by in vitro transcription/translation, as described above.

It is possible that GCN4 can interact with the ADA2, ADA3 and GCN5 proteins only when the latter are
present in a complex. To explore this possibility, we will add GST-tagged mutant and wild-type GCN4 proteins
expressed in E. coli to whole cell yeast extracts prepared from strains expressing either untagged ADA2 or a
functional HA-tagged form of ADA2 (provided by L. Guarente). We will then determine what fraction of the
GCN4 proteins can be co-immunoprecipitated with HA-ADA2 and what fraction of HA-ADA2 can be co-
purifed with GCN4 on a glutathione-agarose column. Similar experiments have been used to demonstrate that
GAL4-VP16 and native GAL4 can interact with ADA2 or with the mediator complex.

[gaMiHiwuiiiuiiiiJiiiiiiiMiiimimnmiwTwnwimmiiniiBBinniiiniiiiiM

mwBwmiHiiinmiiiiiimrmMiiiiiiiiiiwim^^

ZOl HD 01004-12 LMG

We will also examine whether components of the holoenzyme mediator complex, purified from yeast as
described by Kim et al, can be recovered with GCN4-GST proteins on glutathionine agarose. This last
experiment will be done in collaboration with Christoph Hengartner in Rick Young's who has provided us with
partially purified holoenzyme and antibodies against several of the SRB protein components of the mediator
complex. In addition, Tony Weil has supplied us with a partially purified preparation of a TFIID-like complex
from yeast, consisting of TBP and associated factors (A. Weil, personal communication). We will probe for
interactions between these proteins and GST-GCN4 proteins using glutathione agarose chromatography and
antibodies against some of the TFIID protein components provided by Weil's group. If we detect an interaction
between GCN4 and the mediator or TFIID complex that is abolished by one or more mutations in the GCN4
activation domain, we will follow up with GST-GCN4 pull-down experiments using individual subunits of the
appropriate complex. Young's and Weil's groups are in the process of purifying these proteins after over
expressing them in insect cells from bacculovirus vectors, and they have indicated their willingness to
collaborate with us in testing the purified proteins for specific binding to GST-GCN4 proteins in vitro.

2. Genetic screens for transcriptional factors that interact with GCN4.

a. Suppressors of a mutant GCN4 protein that is lethal when overexpressed.

We have set up several genetic screens aimed at identifying functionally important interactions between GCN4
and components of the transcriptional machinery. One approach already underway is to isolate chromosomal
mutations that overcome the dominant lethal effect of overe.xpressing a mutant GCN4 protein which lacks the
CAAD but retains the NTAD, and is produced at ca. 4-fold higher levels than wild-type GCN4. Because the
toxic effect of this construct is dependent on both a functional DNA binding domain and activation determinants
in the NTAD, we presume that the mutant GCN4 protein sequesters one or more essential transcription factors.
We reasoned that it might be possible to isolate mutations that eliminate a nonessential mediator that is required
for sequestering the essential transcription factor by the mutant GCN4 protein, as this approach was used
previously to isolate mutations in ADA2 that overcome the toxicity of an overexpressed GAL4-VP16 protein
in yeast. Since we found that the toxicity of overexpressing the mutant GCN4 protein is not suppressed by
deletion ofADA2, we hope to identify a distinct mediator that interacts with the NTAD of GCN4. It might also
be possible to isolate mutations in a general transcription factor that reduces its affinity for GCN4 without
eliminating essential interactions with other activators.

We have isolated spontaneous revertants of the lethal phenotype conferred by overexpressing the mutant GCN4
protein from an uORF-less construct carried on a single-copy URA3 plasmid. After removing the toxic GCN4
allele from the revertants by 5-FOA counter-selection, we identified strains which exhibited 3-AT-sensitivity,
indicating a possible defect in the ability of the resident wild-type GCN4 to activate transcription. Meiotic
analysis was used to identify those revertants in which 3-AT sensitivity (in the presence of wild-type GCN4)
and suppression of the lethal GCN4 mutant allele were conferred by a single mutation. Several of these mutants
also displayed a slow-growth phenotype on rich medium (Slg) that co-segregated with the other phenotypes.
For most of these Slg" revertants, this phenotype could be partially complemented by introducing a single-copy
plasmid bearing SPT13, encoding yeast TBP. Accordingly, the endogenous SPT13 gene was isolated from these
revertants by gap repair and was found to be indistinguishable from wild-type SPT13 in complementing the
Slg" phenotype of our revertants. Hence, none of these revertants appear to contain suppressor mutations in
SPTJ3. The fact that an extra copy of SPT13 partially complemented their Slg" phenotype does suggest,
however, that they contain mutations affecting a general transcription factor. We have verified that the
suppressor mutations do not reduce the steady-state level of the GCN4 protein, and we are attempting to clone
the wild-type alleles for three of the suppressor mutations with particularly strong Slg" phenotypes.

ZOl HD 01004-12 LMG

In a related approach, we will screen a high copy-number plasmid library for genes that when overexpressed
can overcome the lethality of overexpressing the mutant GCN4 protein described above. If this GCN4 construct
is lethal because a co-activator or general factor is being sequestered by the mutant GCN4 protein, then
overexpressing the sequestered factor should relieve the toxicity. Even if the sequestered factor is a multimeric
complex, this approach could succeed because overexpressing only one subunit of the complex might be enough
to sequester the toxic GCN4 protein and prevent it from interacting with the native co-activator complex. A
high copy-number URA3 plasmid library will be introduced into a wild-type GCN4 yeast strain and a pool of
ca. 10* transformants will be isolated and transformed with the toxic GCN4 allele on a single copy-number
LEU2 plasmid. Transformants that arise will be tested for the ability to lose the library plasmid and grow on
5-FOA. Failure to grow on 5-FOA except after loss of the LEU2 plasmid will indicate the presence of a dosage
suppressor of the lethal GCN4 construct. Transformants of this type will be analyzed to determine whether the
suppressor plasmid lowers the expression of the mutant GCN4 protein, and then tested for exacerbation or
suppression of the Gen' phenotype of a panel of leaky gcn4 alleles bearing different mutations in the activation
domain. Exacerbation of the Gen' phenotype could occur if the encoded protein is a component of a
heteromeric co-activator complex that can compete with native co-activator for interaction with defective GCN4
proteins. Suppression of the Gen' phenotype could occur if the overexpressed protein is a monomeric co-
activator whose interaction with GCN4 is impaired by one the gcn4 mutations being tested. We will also
determine the null phenotype of the suppressor gene. If a deletion strain is viable, we will examine it for Gen'
phenotypes (3-AT-sensitivity and defective derepression oi HIS-lacZ fusions) and for defects in activation by
GAL4 and HAP4 using the appropriate lacZ reporter constructs dependent on these activator proteins. We will
epitope-tag the gene with the myc epitope and use anti-myc antibodies to probe for co-immunoprecipitation of
the suppressor gene product with components of the ADA2/ADA3/GCN5 complex, or with the mediator or
TFIID complexes described above. The suppressor protein will also be tested for direct interactions with GCN4
protein in vitro using the GST pull-down assay described above.

b. Suppressors of a leaky GCN4 allele with point mutations in the activation domains.

In an independent genetic screen, we will isolate chromosomal suppressors of the 3-AT-sensitive phenotype of
the gcn4-I843 allele, which contains alanine substitutions in two of the activation modules in the CAAD (at
Met- 107, Tyr-110, Leu-113 and Trpl20, Leu- 123, and Phe-124) and in one activation module of the NTAD
(Phe-97 and Phe-98). Because it contains mutations in both the CAAD and NTAD, this allele confers very
weak activation in vivo despite an ca. 4-fold higher than wild-type level of protein. Assuming that these
mutations weaken interactions between GCN4 and one or more co-activator proteins or basic transcription
factors, we anticipate that chromosomal mutations can be isolated in these factors that will restore their
interaction with mutant GCN4. By testing these mutations for the ability to suppress other gcn4 alleles that
contain a different set of point mutations, we may identify suppressors that specifically restore interactions with
particular activation modules of GCN4. We may also isolate mutations in other components of the transcription
machinery which bypass the requirements for those activation modules which are lacking in the gcn4 allele.
It is conceivable that certain transcription factors that are normally recruited or activated by GCN4 could be
mutated to mimic the effect on them that wild-type GCN4 normally exerts. Even if they map in known
proteins, such bypass mutations would highlight the importance of these proteins in transcriptional activation
by GCN4.

We have constructed a pair of isogenic strains containing gcn4-1843 integrated on a URA3 plasmid at ura3-52,
plus a GCN4-lacZ fusion at TRPl and a H1S3-GUS construct at LEU2. We are collecting 3-AT-resistant
revertants in one of these strains and cross-screening them for resistance to sulfometuron-methyl (SM) and 5-
methyltryptophan (5-MT) (that inhibit isoleucine-valine and tryptophan biosynthesis, respectively), and for
resistance to cyclohexamide (to screen out multiple drug resistance). Revertants that are resistant to 3-AT, SM

ZOl HD 01004-12 LMG

and 5-MT, but sensitive to cyclohexamide, will be assayed for expression of the HIS3-GUS and GCN4-lacZ
reporters. Those showing increased HIS3-GUS expression but unchanged GCN4-lacZ expression will be
analyzed to determine whether they contain a back-mutation at GCN4 or an unlinked suppressor. This will be
done by evicting the gcn4-1843 allele on the integrating URA3 plasmid on 5-FOA medium, followed by
transformation with a single copy plasmid bearing gcn4-1843. Revertants containing trans-acting suppressors
will revert to a 3-AT-sensitive phenotype upon eviction ofihQgcn4-1843 allele and re-acquire 3-AT-resistance
upon reintroduction of gcn4-l 843 on a plasmid. Revertants that retain the S-AT phenotype after eviction of
the gcn4-J843 allele will be set aside for the present, because they contain GCN4-independent suppressors.
All revertants that contain unlinked GCN4-dependent suppressors will be crossed to the isogenic strain
containing the gcn4-1843 allele and subjected to meiotic analysis. Revertants found to contain single suppressor
mutations responsible for their 3-AT-resistant phenotypes will be tested for allele-specific suppression by
transforming the versions of the revertant strains which lack the integrated gcn4-1843 with plasmid-bome gcn4
alleles containing different point mutations in the activation domain. Suppressors that fail to suppress, or which
even exacerbate, the 3-AT-sensitive phenotype of other leaky gcn4 alleles will be particularly interesting. These
may contain a mutation in a co-activator protein that restores its interaction with an altered GCN4 activation
module but that weakens its interaction with the wild-type version of that module. Genetic analysis will be used
to assign suppressors to complementation and linkage groups and the wild-type genes will be cloned by
complementing either the suppressor phenotype, or a suitable secondary phenotype associated with the
suppressor mutations.

The same strain just described containing gcn4-1843 is also being used to screen high copy plasmid yeast
libraries for dosage suppressors. A mediator protein or basic transcription factor that functions as an individual
polypeptide might suppress a GCN4 activation defect when overproduced. In addition, if GCN4 is
phosphorylated as a means of stimulating its activation function, we might identify a protein kinase, an activator
of a kinase, or an inhibitor of a phosphatase as a dosage suppressor of a defective gcn4 allele. The virtue of
this approach is that a plasmid library can be screened rapidly by direct selection for 3-AT-resistant
transformants, and plasmid-associated suppressors can be readily distinguished from chromosomal suppressors
by determining whether the suppressor phenotype disappears when the plasmid is lost from the strain. The same
genetic criteria listed above for chromosomal suppressors will be applied to evaluate high copy-number plasmid
suppressors.

F. Regulation of purine nucleotide biosynthesis

Our interpretations of the mutational analysis of the ADE5, 7 promoter will be evaluated by conducting gel
retardation assays on various mutant and wild-type DNA fragments using BASl, BAS2 and ABFl proteins
expressed in E. coli, or produced by in vitro transcription/translation, following published procedures. From
these experiments we hope to learn whether the requirement for particular nucleotide positions for transcription
and adenine repression in vivo can be accounted for by their effects on binding BASl, BAS2, or ABFl in vitro.
They should also reveal the identify of the bindingsite(s) for BAS2 at ADE5, 7 and indicate whether the binding
of one protein is dependent on the other two. We will continue with our analysis of protein fusions to identify
the adenine-repressible activity of BASl or BAS2. In addition, we will proceed with our genetic selections for
trans-acting mutations that abolish adenine-mediated repression of transcription to identify additional factors
that might be involved in regulating BASl or BAS2 by adenine and to delineate the regulatory domains in
BASl or BAS2, all as described above in the progress report.

UUUJKUUIMHB^BBBBn^^BBI

ZOl HD 01004-12 LMG
Publications:

Drysdale CM, Duenas E, Jackson BM, Reusser U, Braus GH, Hinnebusch AG. Transcriptional activator GCN4
contains multiple activation domains that are critically dependent on hydrophobic amino acids. Mol Cell Biol,
1995;15:1220-33.

Flowers KM, Kimball SR, FeldhofT RC, Hinnebusch AG, Jefferson LS. Molecular cloning and characterization
of cDNA encoding the a-subunit of the rat protein synthesis initiation factor eIF-2B. Proc Natl Acad Sci USA,
1995;92:4274-78.

Garcia-Barrio MT, Naranda T, Vazquez de Aldana CR, Cuesta R, Hinnebusch AG, Hershey JWB, Tamame M.
GCDIO, a translational repressor of GCN4, is the RNA-binding subunit of eukaryotic translation initiation
factor-3. Genes Dev, 1995;9:1781-96.

Romano PR, Green SR, Barber GN, Mathews MB, Hinnebusch AG. Structural requirements for double-stranded
RNA binding, dimerization, and activation of the human eIF-2alpha kinase DAI in Saccharomyces cerevisiae.
Mol Cell Biol, 1995;15:365-78.

Vazquez de Aldana C, Wek RC, Segundo PS, Truesdell AG, Hinnebusch AG. Multicopy tRNA genes
functionally suppress mutations in yeast elF-2a kinase GCN2: Evidence for separate pathways coupling GCN4
expression to uncharged tRNA. Mol Cell Biol 1994;14:7920-32.

Vazquez de Aldana CR, Marton MJ, Hinnebusch AG. GCN20, a novel ABC protein, and GCNl reside in a
complex that mediates activation of the elF-2a kinase GCN2 in amino acid-starved cells. EMBO J,
1995;14:3184-99.

Runnn

mmmmmim

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBUC HEALTH SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD 01009-03 LMG

PERIOD COVERED

October 1, 1994 to September 30, 1995

TITLE OF PROJECT (80 characters or less. Title must fit on one line tietweert the borders!

Regulation and Function of Genetic Elements

PRINCIPAL INVESTIGATOR IList other professional personnel below the Principal Investigator! IName. title, laboratory, and institute affiliation)

PI: H.L. Levin Senior Staff Fellow LMG:NICHD

A. Atwood Bio Lab Tech LMG:NICHD

J. Lin IRTA LMG:NICHD

M. Williams Visiting Fellow LMG:NICHD

J. Choi Pre -IRTA LMG:NICHD

COOPERATING UNITS lif anyl

Dr. Jef Boeke, Johns Hopkins University, Baltimore, Maryland
Dr. Michael Benedik, University of Houston, TX

LAB/BRANCH

Laboratory of Molecular Genetics

SECTION

Unit on Yeast Molecular Genetics

INSTITUTE AND LOCATION

NICHD, NIH, Bethesda, Maryland 20892

TOTAL STAFF YEARS;
4.3

PROFESSIONAL;
4.0

OTHER:
0.3

CHECK APPROPRIATE BOXIES)

n (a) Human subjects
D (a1) Minors
D (a2) Interviews

D (b) Human tissues H (c) Neither

SUMMARY OF WORK lUse standard unreduced type. Do not exceed the space provided.)

Our research program focuses on the mechanisms of retroelement action. Our approach to understanding the complex
interactions between the retroelement and its host is to study retrotransposons, a family of elements that are closely
related to retroviruses. A significant advantage to studying retrotransposons is they exist in hosts such as yeast that
can readily be studied using sophisticated molecular genetic techniques. In the process of characterizing yeast
transposition, we have collected strong evidence that Tfl reverse transcriptase uses a novel self-priming mechanism
to initiate cDNA synthesis. This is in complete contrast to the tRNA mechanisms thought to be used by all other
LTR-containing elements. In this report we describe the characterization of the minus-strand strong-stop DNA that
provides additional support for the self-priming mechanism. Genetic and biochemical analysis of Tfl RT mutations
in the active site of the polymerase allowed us to observe priming intermediates consisting of transcripts that had the
first 1 1 bases removed. This data suggested a molecular model for priming that includes a cleavage of the first 1 1
bases of the transcript and the priming of reverse transcription from the 3 'OH of the 11th base.
The analysis of a large family of point mutations near the primer binding site (PBS) confirmed the presence of a new
39 base pair RNA structure that is essential for transposition. As this large structure includes the 1 1 base pairs
shown to be important for priming, we speculate that the newly detected structure may also participate in the self-
priming mechanism. The assembly of functional Tfl particles has been a paradox since other retroelement particles
assemble with a molar excess of capsid protein that accumulates because the levels of the Pol proteins are restricted
by reading frameshifts or stop codons. Tfl however, expresses all its protein from within a single open reading
frame as a primary translation product and we found that Tfl particles contain a 26-fold excess of Gag compared to
IN protein. By looking at cultures in different stages of growth, we have been able to observe an IN degradation
process that leads to this excess of Gag. In addition, we found that most of the IN degradation occurred before Tfl
cDNA is synthesized indicating that the particles with a 26-fold excess of Gag are intermediates in transposition. We
have used immunoblot analysis to reveal that IN degradation occured in cells starved for glucose and not in those
cell starved for nitrogen, suggesting that the loss of IN is not a time-dependent process but occurred only after certain
growth conditions are met. To determine the factors required for transposition, we have created a large set of mutant
strains that are defective for transposition. Thus far, we have identified six host genes that are required for either
protein accumulation, particle stability, or integration.

PHS 6040 (Rev. 5/92)

mmsstmsmm^BK

nmmamsm

ZOl HD 01009-03 LMG
Project Description

Objectives:

Our efforts are designed to elucidate the molecular basis of each step in the process of retroeiement
proliferation. The most medically important class of these elements are the retroviruses, members of
which are responsible for AIDS as well as several types of cancer caused by the proviral induction of
oncogene expression. However, many aspects of particle assembly, reverse transcription, nuclear entry
and chromosome integration are difficult to study because retrovirus hosts possess a high level of
genetic complexity. Our approach to understanding these events is to study retrotransposons, a family
of elements that are closely related to retroviruses. A significant advantage of studying retrotransposons
is they exist in yeast, a host that can be studied readily by using sophisticated molecular genetic
techniques. The similarities of retrotransposons to retroviruses includes the presence of two long
terminal repeats (LTRs) and open reading frames (ORFs) with coding sequences homologous to
retroviral protease (PR), reverse transcriptase (RT), and integrase (IN). The first step in the
transposition pathway is synthesis of a full-length mRNA with sequence that begins in the 5' LTR and
terminates in the 3' LTR. This is directly analogous to the initial step in retrovirus particle formation.
Retroviral and retrotransposon mRNAs are translated into proteins that assemble along with the mRNA
into large particle structures. Both retroviral and retrotransposon particles undergo a maturation process
that includes the proteolytic processing of precursor proteins and the reverse transcription of the mRNA.
Retroviral particles are able to escape the host cell and integrate their DNA into the genome of
neighboring cells, while retrotransposon particles complete transposition by simply inserting their DNA
into the genome of the original host cell. Because each step in retrotransposition is directly related to
a retrovirus process, results from the investigation of yeast retrotransposition will be relevant to aspects
of retrovirus behavior.

The retrotransposon we study is the Tfl element of the fission yeast, Schizosaccharomyces pombe.
The transposon is 5 kb and contains a single ORF of 1340 amino acids with coding sequences
homologous to PR, RT, and IN. We have previously demonstrated, using an //; vivo assay, that at least
one of our cloned copies of Tfl is active and can transpose at a significant frequency. The system we
have developed allows us to use a battery of powerful molecular and genetic techniques to identify and
characterize factors that contribute to the transposition process. We are initially focused on the
interactions between the host cell proteins and the element because it is these that are the least
understood in the retrovirus systems. Genetic experiments on retroviruses have defined the principal
role of each of the major virus proteins by altering the virus encoded factors and assaying for an effect
on virus functions. Studying retroelements in yeast allows us to explore the contributions of the host
proteins to the entire process of transposition.

The genetic approach we are using to identify proteins that function in transposition is to use a new
form of the transposition assay to screen individual colonies for mutations that alter the frequency. We
developed a genetic screen at NIH for transposition that depends on the overexpression of Tfl
transcripts containing a selectable marker that is reverse transcribed during transposition and inserted
into the host genome. We experimented with various combinations of growth media and expression
systems with the result that we increased our previously reported transposition frequency of 0.5% per
10 cell generations to 18%. This improvement allowed us to develop this new transposition assay that
is now being used to screen large numbers of cells for mutations that alter the transposition frequency.
Mutations in genes that contribute to transposition will be used to clone the genes required for
transposition from an S. pombe genomic library.

ZOl HD 01009-03 LMG

Another goal associated with the mechanisms of transposition is to learn which parts of the Tfl proteins
are required for mobility and what are their roles in the process. Towards this goal, we have started
to assemble a large set of random mutations in Tfl itself that block transposition. By characterizing
the positions and properities of these mutations, we hope to associate each protein with a part of the
transposition pathway and perhaps attribute new functions to some of these domains.
Further characterization of Tfl protein expression is also a goal because of its unique aspects among
LTR-containing retroelements. Retroviruses and LTR-containing retrotransposons encode their Gag
(capsid) and Pol (PR, RT, and IN) proteins in separate ORFs that are interrupted by a frameshift or
a stop codon. This organization causes a molar excess of Gag to Pol protein to be expressed which is
important for the assembly of particles that normally contain as much as 20-200 times more Gag than
Pol. Tfl coding sequence for Gag and Pol is contained within a single ORF and we have used three
separate experiments to show that all Tfl protein is expressed within a single primary translation
product. Last year we determined that the Tfl proteins do form virus-like particles similar to those
made by the other retrotransposons studied to date. One of our present objectives is to determine the
ratio of Gag to Pol proteins within these particles and develop a model for how these ratios are achieved
as expressed from a single ORF.

A unique aspect of Tfl is that its sequence contains no homologies to tRNA molecules at the site where
reverse transcription is initiated, the primer binding site (PBS). This is in direct contrast to all other
LTR-retroelements which have 15-20 bases of perfect homology to specific tRNA molecules that serve
as primers for reverse transcriptase during the first step of cDNA syntheis. Although we have detected
the presence of an RNA primer at the Tfl PBS, it appears not to be a tRNA. Our goal is to determine
the identity of the reverse transcriptase primer as well as the mechanism used. Recent results indicate
that Tfl uses a novel selfpriming mechanism of reverse transcription that defines a new family of
retroelements.

Methods Employed

We have used a wide variety of molecular, genetic, and biochemical techniques in experiments
described in this report. Our genetic approaches include several standard yeast techniques such as
transformation, gene replacement, mating, tetrad dissection, diploid formation and mutagenesis. We are
also using the two-hybrid system as a general method for identifying any S. ponibe proteins that interact
with Tfl proteins.

Our standard transposition assays take advantage of two drug marker selections. We use 5-fluoroorotic
acid to select for cells that lose plasmid copies of the URA3 gene. Our transposons are marked with
the bacterial neo gene that provides S. pombe with resistance to high concentrations of G418. We have
also started to use an artificial S. pombe intron placed within marker genes in the transposon to assay
for reverse transcription events.

Biochemical methods include the use of various yeast expression systems to improve production of the
Tfl proteins. We now use the nmtl promoter of 5. pombe to regulate high levels of protein synthesis.
The Tfl ORF is fused to the nmtl promoter in a multicopy plasmid. The presence of 10 ^M vitamin
Bl reduces expression 50-fold, nmtl is the strongest powie promoter thus far characterized. Extracts
from these expression strains are then subjected to sucrose gradient sedimentation as well as other
purification protocols. Fractions from these gradients are immunoprecipitated by using antibodies raised
against Tfl proteins.

"-" — —.-«»«

ZOl HD 01009-03 LMG

A full range of molecular biology techniques are utilized to make reagents and characterize them.
Constructs are made by using restriction enzymes, phosphatases, DNA polymerases, PCR and ligase.
We routinely sequence the DNA of complex constructs. Immunoblots, DNA blots and RNA blots are
all used to characterize various strains.

Major Findings:

A. Priming of Tfl reverse transcription.

1. Evidence that Tfl uses an unusual mechanism of self-primed reverse transcription.

The role of tRNA in the priming of reverse transcription is thought to be common to all LTR-containing
elements since the retroviruses and LTR-retrotransposons that have been sequenced have a PBS that
is complementary to known tRNA species (Figure lA). In light of the conservation of reverse
transcription priming, it is surprising that no tRNA primer has been identified for Tfl. In addition to
the absence of tRNA homology, Tfl also lacks the highly conserved UGG in the PBS of retroelements
that hybridizes to the last 3 bases of the tRNA molecule. Evidence for the presence of an RNA primer
that functions precisely at the conventional PBS location of Tfl came from previous analyses of DNA
isolated from virus-like particles. This DNA was used as template in primer extension analyses and
treatment with RNase demonstrated that there are at least 9 bases of RNA at the 5' end of the minus
strand strong-stop DNA. The position of this RNA is immediately downstream of the 5' LTR and
defines the PBS for Tfl. Instead of postulating that an uncharacterized tRNA serves as the primer, we
proposed in our last annual report that Tfl mRNA undergoes a novel self-priming mechanism of reverse
transcription. The first 1 1 bases of the Tfl transcript are exactly complementary to the PBS location
identified by the primer extension experiments. This observation raised the possibility that the first 1 1
bases of the transcript folds back and anneals to the PBS. An RNA endonuclease would be able to cut
the first 1 1 bases from the transcript so that the 3' OH required to prime DNA synthesis would become
available (Figure IB). Alternatively, RT might prime DNA synthesis from a 2'OH of an internal
nucleootide as has been reported for the bacterial retron reverse transcriptase [Hsu, 1992 #1424;
Lampson, 1989 #836].

Experiments decribed in our last annual report showed that the mutations in Figure IC at the PBS or
the 5' end of the transcript reduced transposition more than 30-fold. When two mutations were
combined to reestablish complementarity, transposition frequencies were restored to 80% of the wild
type level [Levin, 1995 #1575]. Consistent with the hypothesis that the mutations block priming, we
fould that the individual mutations resulted in a drop in the levels of minus-strand strong-stop DNA as
isolated from Tfl particles. Strains with the compensating double mutations produced wild type levels
of strong-stop DNA.

2. Analysis of minus-strand strong-stop DNA.

In all cases where the strong-stop species was observed on DNA blots, the signal was a doublet
composed of two closely migrating bands. In hopes of learning more about the priming mechanism,
the doublet nature of the minus-strand strong-stop species was investigated. We used DNA blot analysis
of sequence gels in combination with radiolabeled probes consisting of several oligonucleotides to map
the 5' and 3' termini of the two strong-stop species. The sequence ladders on these gels indicated that
the size of the two strong-stop species differed by about 11 bases (+/-1 base). The results indicated
that the larger band is the full-sized strong stop species with a 3' end that corresponded with the first

ZOl HD 01009-03 LMG

base of the Tfl transcript and a 5' end that has the RNA primer with the PBS sequence. The smaller
species has the same RNA primer at the 5' end but the 3' end is missing the 1 1 bases that are templated
by the beginning of the Tfl transcript. The shortened strong-stop molecule is likely due to a premature
termination of reverse transcription resulting from either a cleavage of the mRNA or an attachment of
a 2'-5' linkage at the 11th base of the Tfl transcript. Regardless of which of the two events caused
the termination of reverse transcription, the presence of the shortened strong-stop product provides
support for the self-priming model.

3. In vivo experiments to distinguish between two types of self-priming mechanismOespite the genetic
and biochemical evidence that the folded RNA structure is important for priming, several questions
about the mechanism are left unanswered. One very significant question is how does the RNA structure
allow for priming given that no 3' OH is made available by the formation of the fold-back loop. A
likely possibility is that an endonuclease is responsible for cleaving off the 1st 11 bases of the Tfl
transcript so that the 3'OH of the 11th base can serve as the site of priming. Another explanation for
the lack of a 3' OH is that the Tfl reverse transcriptase may actually prime DNA synthesis from a 2'
OH forming a branched intermediate similar to that produced by the bacterial retron reverse
transcriptases [Hsu, 1992 #1424; Hsu, 1990 #998]. Our observation that minus strand strong-stop DNA
synthesis is blocked near the 11th base of the Tfl transcript supports the idea that either a cleavage
event or a 2' OH linkage occurs at the 1 1th base of the transcript.

To explore the nature of the priming mechanism, we mapped the 5' ends of Tfl transcripts extracted
from particles to test for the presence of Tfl mRNA missing its first 1 1 bases. The result was that large
amounts of Tfl transcript were detected, but all of it had 5' ends at the previously mapped 5' start site
[Levin, 1992 #1196]. One complication of this result is that the putative cleavage event is likely
coupled to reverse transcription so that if transcripts are cleaved, they would immediately be degraded
by RNaseH. In addition, our inability to detect evidence for Tfl transcripts connected to strong stop
DNA by a branch point might have been due to RNaseH degradation of the mRNA.

To preserve priming intermediates from degradation, we used site-directed mutagenesis to generate five
Tfl plasm ids each with a mutation in an active site residue of RNaseH. If these mutations greatly
reduce RNaseH activity, any cleaved transcripts would be protected from degradation and thus could
be observed by SI nuclease mapping or primer extension. In addition, mutations that leave RT
polymerase activity intact would also allow us to observe Tfl transcripts attached by a branch point to
strong-stop DNA if a 2' OH group is used for priming. Several groups have identified 3 amino acid
positions within RNaseH of HIV and moloney leukemia virus that can be mutated to produce an RT
with normal polymerase activity and no RNaseH function [Repaske, 1989 #1479; Mizrahi, 1994 #1478].
Using these studies as a guide, as well as the identification of active site residues from X-ray crystal
structures of other RNaseH molecules [Davies, 1991 #1598; Katayanagi, 1990 #1599; Yang, 1990
#1600; Kohlstaedt, 1992 #1601], we individually changed two essential aspartic acids (443 and 498,
HIV numbers) to asparagines and alanines while a glutamic acid (478) was changed to a glutamine.
These mutations were incorporated into the Tfl transposition plasmid and their proteins expressed in
our S. ponibe strains. Transposition assays of the mutant Tfl elements were used to verify a lack of
mobility, and in vitro RT assays of the Tfl particles were used to identify the mutations that preserve
DNA polymerase activity. All 5 mutations caused a severe reduction in transposition frequency.
Replacement of either of the aspartic acids with glutamine resulted in particles with wild type levels
of RT activity as measured by in vitro assays that include synthetic primer and template. We found that
the alanine replacements of the aspartic acids retained approximently half of the RT activity and the
glutamine replacement of the glutamic acid had no detectable activity. Extracts from partially purified

iSm^HUBBHBH^iHUHin^

wBxamsmim

ZOl HD 01009-03 LMG

particles were used to isolate Tfl mRNA to be examined for cleavage at the 5' end as well as for
covalent linkage to strong-stop DNA. None of the RNA isolated from any of the mutant or wild type
strains exhibited evidence of a 5' end cleavage as indicated by primer extension studies. These results
ostensibly suggested that cleavage of the first 1 1 bases of the transcript was not required for self-
priming. However, examination of the particles for DNA indicated that none of the mutant RT
polymerases were able to produce minus-strand strong-stop material. Therefore, the lack of any
observable cleaved mRNA could have been due to a complete lack of priming events caused by the
RNaseH mutations. In addition, the absence of minus-strand strong-stop DNA in the mutants suggested
that the priming event may actually require RNaseH activity.

An additional in vivo experiment designed to detect the generation of a primer via a cleavage event of
the Tfl transcript avoided mutations in RNaseH in case it itself possesses the putative cleavage avtivity.
We used versions of Tfl RT that have defective polymerase activity with the objective of trapping
priming intermediates that have undergone the putative cleavage event but not extension. We expected
that mutations in the polymerase domain of RT would allow the accumulation of cleaved Tfl transcripts
because RNaseH would be unable to degrade the cleaved mRNA due to the lack of DNA:RNA duplex.
3 Tfl assay plasmids with mutations in the conserved YXDD box of the RT polymerase were created.
RNA from mutant and wild type particles was examined by primer extension and the results indicated
that 20%-50% of the RNA from the mutant particles had suffered a cleavage event that removed just
the first 1 1 bases. Wild tj'pe particles showed no cleavages indicating that RNaseH was degrading the
RNA as cleavage events occurred. SI nuclease mapping experiments are in progress to independently
examine the 5' end of the mutant RNA for cleavage sites. Should these confirm the presence of
cleaved RNA in particles, we will combine the RT polymerase mutations with the RNaseH mutations
to test the role of RNaseH activity in the cleavage event. A role of RNaseH in the cleavage of the Tfl
self-priming RNA would be particularly interesting given recent reports that RNaseH from retroviruses
possess an activity designated RNaseH* that results in cleavage of RNA in an RNA:RNA duplex
[Hostomsky, 1994 #1385].

4. The 1 1 base pair RNA duplex required for self-priming comprises one third of a much larger
structure of continuous RNA base pairs.

An important question left unanswered by our current self-priming model resulted from the close
inspection of the RNA sequence in this leader region. Figure 2 shows a large potential RNA structure
containing the II base pairs of the PBS as well as 28 other continuous base pairs. We are interested
in defining the function of this large potential structure in priming as well as other processes such as
packaging of Tfl mRNA into the virus like particles. To this end, we have randomly mutagenized the
entire region of secondary structure by PCR and isolated a total of 78 mutations, 40 of which represent
single base mutations that reduce transposition significantly. These strains have been subjected to our
quantitative transposition assay and the results indicated that approximently half exhibited at least 50-
fold less transposition activity while the others showed 5 to 50-fold lower transpositon. Mutations of
both classes were distributed throughout the PBS region, the adjacent 1 1 base pair region, and the 5
base pair section just upstream of the PBS. To address the effect of the mutations on expression, each
of these Tfl elements was subjected to immunoblot analysis and found to synthesize wild type levels
of Gag and IN protein.

We have recently developed a DNA blot method for the detection of Tfl cDNA that enables us to
measure amounts of mature double-stranded reverse transcript. S. pombe cell extracts are phenol
extracted and ethanol precipitated. The resulting nucleic acid is restriction digested with enzymes that

mmamm

ZOl HD 01009-03 LMG

produce cDNA fragments with very different sizes from the plasm id Tfl sequences. This method was
used to screen each of 40 different Tfl elements with mutations in the RNA leader structure. The
results showed that each of the mutants produced lower levels of cDNA than wild type. When the
amounts of cDNA were quantitated, the reduction in reverse transcription caused by each mutation
correlated well with the magnitude of their transposition defect. Although this data suggests each
mutation only effects priming levels, we are also testing the mutations for defects in the packaging of
RNA into particles, another activitiy that would result in lower levels of cDNA synthesis.

B. The ratio of Gag compared to IN in Tfl particles.

1. The Tfl Gag protein accumulates to significantly higher levels than IN.

We have previously shown that the Tfl primary translation product is a single 140 kDa polypeptide that
is cleaved by PR to form the mature Gag, PR, RT and fN proteins. Further characterization of the Tfl
proteins is important because their expression is unique among LTR containing retroelements.
Retroviruses and LTR-containing retrotransposons encode their Gag (capsid) and Pol (PR, RT, and IN)
proteins in different ORFs that are separated by a frameshift or a stop codon. This organization causes
a molar excess of Gag with respect to Pol protein to be expressed because a small number of ribosomes
are able to translate past the stop codon of Gag. Excess Gag is important for assembly of particles
because Gag serves as a structural component of the particle. Since the Tfl coding sequence for Gag
and Pol is contained within a single ORF, there is no obvious mechanism for overproducing Gag
protein. The high levels of transposition exhibited by Tfl and the particle nature of its proteins
compelled us to develop methods for determining the ratio of Gag to Pol proteins in whole yeast
extracts as well as in partially purified virus-like particles.

As reported last year, two different methods were developed to calibrate the immunoblot signals
produced by anti-Gag and anti-IN antisera so that quantitative blotting could be used to evaluate the
relative levels of these two proteins. Extracts from a PR' S. pombe strain that contained only the 140
kD primary translation product were used to calibrate anti-Gag and anti-fN antisera since the Tfl
product contained equal amounts of Gag and FN epitopes. Our second method for antisera calibration
was based on Gag and FN protein expressed in bacteria and purified using Ni columns. Both methods
for comparing the strength of the antisera generated similar results that indicated the anti-fN antiserum
produced 1.5 to 3 times more signal than the anti-Gag antiserum. Using the calibrated antisera, we
found immunoblots of 5. pombe proteins showed that whole cell extracts as well as partially purified
particles contained 26 times more Gag than fN. This significant excess of Gag in the Tfl particles
represented a composition that resembled that found in retrovirus and retrotransposon particles
[Stromberg, 1974 #1451; Panet, 1975 #1450; Farabaugh, 1993 #1389].

Despite the observation that Tfl particles appeared to assemble with the high levels of Gag expected
for typical retroelement particles, the mechanism for adjusting the ratios of Gag to fN was very different
than the frameshifting mechanism. We surmised that since Tfl proteins are all expressed at the same
levels, a selective degredation mechanism must result in the reduction of FN levels. In support of this
reasoning, we found that extracts made from log-phase yeast cultures contained equal amounts of Gag
and fN while cultures of higher cell density contained the lower amounts of IN relative to Gag that are
usually observed.

2. The analysis of Tfl particle development in S. pombe cultures.

,^1

unjuuivujym

ZOl HD 01009-03 LMG

This year we initiated detailed time-course experiments to careftiily monitor the changes in IN levels
as cells reached stationary phase. In agreement with observations reported last year, log phase cells in
these cultures possessed approximently 1:1 ratios of Gag to IN and the stationary phase cells showed
much less IN than Gag. Most of the IN degradation occurred within a 6 hour period between the OD^oo
levels of 9.2 and 14.0. The final Gag to IN ratio observed in this experiment was 40:1. To determine
if the Tfl particles that contained excess Gag played a functional role in transposition, we examined
the same cells from the time course for reverse transcription products. A blot of DNA extracted from
whole cells, cut with BstXl, and hybridized with a neo probe was used to measure the synthesis of Tfl -
neo cDNA. The restriction cut allowed us to distinguish between neo sequences derived from reverse
transcripts verses plasmids because Tfl has a single BstXl site 2 kb from the 3' end of the element.
Additional evidence that the 2 kb band was derived from the Tfl cDNA came from examination of an
isogenic strain with a Tfl plasmid that had a frameshift mutation in PR and therefore produced no RT.
This mutant strain contained the plasmid band but not the 2 kb cDNA band. The 2 kb cDNA band
derived from the double-stranded Tfl cDNA was observed to increase greatly in the cells that contained
reduced IN protein. The levels of the 2 kb band were quantitated and normalized to the amount of
plasmid derived signal in each lane. The increase in Tfl cDNA content observed during the entire time
course was 8-fold and the amounts were linearly increasing even in cells from the last time point. The
bulk of this increase occurred after the Gag to IN ratios increased to the levels observed in stationary
phase. The fact that most of the cDNA is produced following degradation of the majority of IN protein
was consistant with the hypothesis that particles containing a large molar excess of Gag compared to
IN are functional in producing double-stranded cDNA. We cannot however determine whether the
small amounts of cDNA produced in log phase cells are synthesized by particles of this type or by
particles containing equal molar amounts of Gag and IN.

3. Examination of stationary phase cultures for conditions that trigger fN degradation.

The degradation of Tfl IN in cells from stationary phase cultures is a regulated process that results in
the high ratios of Gag to IN typical of retroelements. Several aspects of stationary phase conditions are
candidates for initiating the degradation of IN. Nitrogen, sulfate, and glucose are all nutrients that may
be limiting in stationary phase cultures and thus initiate IN degradation. The low pH of these cultures
or the accumulation of ethanol are also properities of stationary phase that might result in IN
degradation. To determine which if any of these conditions might effect the ratio of Gag to IN, we
grew cells expressing Tfl to midlog densities and placed them in media that contained different levels
of the agents being tested. The results form this survey indicated that cells starved for nitrogen or
sulfate exibited the normal degradation of IN seen in stationary phase cells. However, we found that
cells starved for glucose, were unable to degrade IN and therefore resulted in stationary cells that
contained equal amounts of Gag and IN long after cells in complete medium exhibited the drop in IN
levels. These results suggested that the turnover of IN may be a process that only occurs in cells
arrested in G,, since nitrogen and sulfate starvation ofS. pombe causes G^, arrest while glucose starvation
causes a nonspecific arrest [Moreno, 1991 #1429]. Never the less, we plan to use plates with reduced
glucose to assay the transposition frequency of cells that possess equal levels of Gag and IN. These
experiments will indicate whether IN degradation is an esential feature of transposition.

C. The contribution of Tfl RT to transposition.

To identify essential functions of RT for transposition, we have characterized a set of point mutations
created either by random or site-directed mutagenesis.

wmmBm

ZOl HD 01009-03 LMG

1. Mutations in RNaseH tiiat block transposition but not reverse transcription.

We have previously reported the isolation of a large set of random point mutations in Tfl that reduced
transposition activity. These were screened using our homologous recombination assay to identify
mutations that, never the less, produced normal levels of reverse transcript. We identified 7 Tfl
mutations that exhibited these properities and as expected, most of the base changes were found to be
located within IN. Surprisingly, 2 of the mutations were within the C terminus of the RNaseH domain
of RT. This was an unexpected result because studies of RNaseH mutants in retroviruses have
indicated that the synthesis of mature double-stranded cDNA requires RNaseH activity to release single
stranded DNA intermediates from their RNA templates [Champoux, 1993 #1442; Blain, 1995 #1602].
To establish if the level of reverse transcription is effected in the two mutant strains, we extracted total
DNA from each and subjected it to DNA blot analysis using conditions that allowed us to visualize
restriction products from both ends of the cDNA. Consistant with the homologous recombination
assays, the blot results showed that one of the strains had fully normal levels of reverse transcript while
the other exhibited about 2-fold less cDNA than wild type Tfl. One possible explanation of our results
is that the mutants specifically affect the ability of RNaseH to cleave the residule RNA primers off the
5' ends of the cDNA without reducing the enzyme's general ability to degrade RNA annealed to DNA.
The result of this defect would be to allow RT to product completed cDNA but the presence of the
RNA primers would inhibit the integration reaction. Whether or not this is the cause for the low
transposition frequencies of the 2 RNaseH mutations, we appear to have identified a role of RNaseH
in the integration process.

2. Site-directed mutations of Tfl RT define essential functions.

Despite the presence of highly conserved amino acid residues in the sequence of Tfl that indicate the
presence of an RT, we had no direct evidence that the domain actually encodes a functional reverse
transcriptase. As discussed in section A3 (/a/ vivo experiments to distinguish between two types of self-
priming mechanisms), we generated 8 site-directed mutations within conserved residues of Tfl RT.
Within the RNaseH domain, we changed two conserved aspartic acids (443 and 498, HIV numbers) to
asparagines and alanines while glu478 was changed to a glutamine. The result of each of these 5
RNaseH mutations was a Tfl element that exhibited no detectable transposition activity. In addition,
the mutations resulted in undectable levels of homologous recombination between the Tfl plasmid and
cDNA, indicating the levels of reverse transcripts was low. To gauge the effect of these RNaseH
mutations on the RT polymerase activity, we partially purified particles from the mutant transposons
and assayed them for RT activity by adding oligo dT as primer and poly rA as template. The results
indicated that D443N, D498N, D443A, and D498A all showed high levels of activity that were either
equilivant to wild type or within 2-fold of wild type RT. The high level of RT activity possessed by
these mutant RTs indicate that the defects in transposition were not due to lower stability of RT but
probably the lack of RNaseH activity. This conclusion was supported by the absence of any products
of reverse transcription as measured by DNA blotting. As a result, these mutations provided the first
physical data that the Tfl RT possesses RNaseH activity essential for transposition.

To address the role of the polymerase domain of RT, three additional mutations were produced in the
conserved "YXDD" box that constitutes a portion of the polymerase active site in other retroelements.
The mutations created were D185N, DI85L, and D186N (HIV number system). The transposition
frequencies of all three strains were undectable as were their frequencies of homologous recombination.
In vitro assays of particles produced by these mutant strains indicated the levels of RT activity was

msm

ZOl HD 01009-03 LMG

reduced by at least 60-fold as a result of each mutation. The results from the mutations in polymerase
and RNaseH strongly indicate Tfl does encode an RT that possess functions essential for transposition.

D. The characterization of host mutations that inhibit transposition.

In order to improve our understanding of the retrotransposition process and by analogy, retrovirus
infection, we initiated a large scale genetic screen for S. pombe strains that were defective for
transposition. As reported last year, cultures of S. pombe that contained our Tfl assay plasmid were
mutagenized with EMS under conditions that caused 80% lethality on minimal medium. After these
strains were colony purified, 5,000 were patched onto our transposition assay media and screened for
reduced frequencies of transposition. Approximently 176 strains from two mating types reproducibly
transposed at significantly lower levels than the parent strains. All these candidates were tested for the
presence of single genetic lesions affecting transposition and for the magnatude of the deficiency caused
by each mutation. As a result, we chose to focus our experiments on 6 mutant strains that had
transposition frequencies 5 to 25-fold lower than that of wild type cells.

1. Recombination analysis of 6 mutant strains to identify the minimum number of genes that effect
transposition.

Each of the original mutant strains were crossed with others from the same set and the resulting spores
were tested for transposition to determine if any two strains carried mutations within the same gene.
If two strains are mated that each possess mutations in the same gene, none of the resulting spores will
transpose at wild type frequencies. If the mutations exist in different genes that segregate
independently, one quarter of the spores will show normal transposition activity. The results indicated
that each of the 6 mutant strains contained genetic lesions in different genes that we have termed hopl
through hop6.

2. The evaluation of each transposition mutation for effects on specific steps in the transposition
pathway.

I. Hopl

The strains with the hopl-1 mutation grow at wild type rates and possess a mutation that reduced its
transposition frequency by 25-fold when measured with our quantitative assay. The frequency of
homologous recombination between Tfl cDNA and the Tfl plasmid was also measured quantitatively
and found to be 50-fold lower in hopl-1 cells compared to wild type levels. The low frequencies of
homologous recombination suggested that the hopl-1 mutation caused a defect early in the transposition
process that resulted in lower levels of reverse transcript. Results from DNA blots of hopl-1 strains
were consistent with the recombination data in that no reverse transcripts were detected. Immunoblots
of hopl-1 strains demonstrated a dramatic lack of Gag and FN protein accumulation that very likely
caused the low levels of cDNA due to a simple lack of particle formation. We have measured mRNA
levels of Tfl-neo in hopl-1 strains and find at most a reduction of 2-fold that may not be the main
cause of the transposition defect but merely the result of transcript destabilazition due to low rates of
Tfl translation. We are currently pursuing the possibility that the lack of Tfl protein is due to either
reduced translation of the Tfl transcript or increased degradation of the Tfl proteins. We have made
lacZ fusions to the first codon of the Tfl ORF as well as to internal locations to identify which Tfl
sequences are effected by the hopl-1 defect. The isogenic S. pombe strains with and without the hopl-1

.'^4

woKsmm

ZOl HD 01009-03 LMG

mutation are now being constructed to test the lacZ plasmids. In addition, we have fused the complete
Tfl ORF to the nmtl promoter after the nmtJ 5' untranslated region to determine if the hopl-1 mutation
caused a lack of Tfl protein accumulation due strictly to protein degradation and not lack of translation.
Preliminary results indicate that when the Tfl proteins are expressed from the nmtl transcriptional
leader, the hopl-1 mutation has no effect on Gag or IN acumulation suggesting that the lack of Tfl
protein accumulation caused by hopl-1 was due to a defect in translation linked to the 5' untranslated
region of the Tfl mRNA.

Because the lack of homologous recombination is so dramatic, we were able to identify plasmids from
an S. pombe genomic library that suppressed the hopl-1 defects. We are now characterizing these
plasmid candidates to determine if they contain the hopl gene.

2. hop2

Strains with the hop2-l mutation have an 8-fold transposition defect and a 4-fold drop in their
homologous recombination frequency. The phenotype of the hop2 strains is similar to hopl cells in that
the levels of Tfl protein accumulation was low although the amounts of cDNA detected by DNA blot
were not greatly reduced. We are testing these strains for defects in translation and protein stability
using the same lacZ fusions developed for characterizing hopl.

3. hop3

The hop3-l mutation caused a 10-foId drop in both the transposition and homologous recombination
assay frequencies. Unlike the hopl and hop2 mutations, hop3 cells were able to produce wild type
levels of mature Tfl proteins in log phase cultures. However, 1 day after hop3-l cells reached
stationary phase, they exhibited a sharp drop in Gag levels suggesting a loss of protein stability.
Although we observed the hop3-l cells contained normal levels of reverse transcript, the loss of Gag
and possibly IN could result in lower IN activity or less efficient cDNA presentation to the nucleus.
To reveal the cause of low Tfl protein accumulation in stationary phase cells we are testing these
strains for defects in translation or protein stability using the same lacZ fusions developed for
characterizing hopl. Preliminary results indicate that the Gag protein in stationary phase hop3-l cells
is absent even when expressed from a plasmid with the nmtl 5' untranslated region fused to the Tfl
ORF. This suggests that the lack of Gag is due to degradation, not a defect in translation.

4. hop5

Significance to Biomedical Research and the Program of the Institute.

The medical significance of our research stems from the close relationship between retroviruses and
retrotransposons. The high level of similarity in structure between these two types of retroelements

.'^5

ZOl HD 01009-03 LMG

results in the large number of mechanisms that the elements share. The mechanism of IN mediated
insertion for retrotransposons and retroviruses has been shown to require the same reaction
intermediatesb [Eichinger, 1988 #419]. The early steps of reverse transcription are primed from the
same position of the elements and both types of retroeiements use polypurine sequences for priming
plus strand reverse transcription. The structure of transposon virus-like particles is analogous to
retrovirus particles in their composition of capsid and Pol proteins. The medical importance of
retroviruses is dominated by the role of HIV in the AIDS epidemic. HIV is a retrovirus that appears
to undergo the same types of reverse transcription, proteolysis, integration and particle formation as do
retrotransposons. Because many aspects of retrovirus function, including particle assembly, reverse
transcription, nuclear entry and chromosome integration, are difficult to study in hosts that possess a
high level of genetic complexity, the molecular genetic analysis of retrotransposons in yeast provides
a promising approach to answering many of the important questions that are at the center of the AIDS
crisis.

The study of transposons in general and retrotransposons in particular is relevant to the understanding
of neoplastic disease mechanisms. Transposons are endogenous mutagens of cells; chemical
mutagenesis is strongly correlated with carcinogenesis; therefore studies of these endogenous biological
mutagens is highly pertinent to an understanding of cancer. Moreover, the growth of a wide variety
of tumors is associated with activation of endogenous retroviral expression and rearrangements of the
genomic DNA in tumor cells. Many studies clearly indicate that the activation of cellular oncogenes
is directly responsible for mammalian neoplasms. In many cases, such gene activation is caused by
retrovirus or retrotransposon transpositional insertion into a proto-oncogene locus. For this reason, the
thorough understanding to be gained from studies of retrotransposons inhabiting the genomes of
genetically tractable organisms, particularly yeast, is directly relevant to the understanding of both
oncogene activation mechanisms and the DNA rearrangements that occur in cancer cells.

Proposed Course

A. Self-primed reverse transcription.

1. The mechanism of mRNA cleavage.

We are pursuing several approaches to identify the chemical nature of the self-priming mechanism of
Tfl reverse transcription. We have previously shown that the interaction between the PBS and the
5'end of the Tfl transcript is required for synthesis of the minus-strand strong-stop DNA. The chemical
nature of the priming event is still unknown. One possibility is that a 3'OH is provided by a site-
specific nicking enzyme that creates a primer by cutting off the first 1 1 bases of the Tfl transcript. Our
data from the RT polymerase mutations has provided the first evidence for a cleavage of the Tfl
transcript after the 11th base. Tfl mRNA isolated from particles with the mutant reverse transcriptases
was analyzed by primer extension and found to have been cleaved between the 11th and 12th bases.
We will also subject the particle UNA preparations to SI nuclease analysis to seek independent evidence
for the absence of the first 1 1 bases. Because the mutations in the conserved residues of RNaseH
surprisingly resulted in much lower amounts of minus-strand strong-stop DNA, we will test the
hypothesis that RNaseH is the source of the cleavage activity. We will compare Tfl elements with
mutations in the polymerase domain to Tfl elements with mutations in both the polymerase and
RNaseH domains to determine if the addition of the RNaseH mutation reduces the level of cleaved
mRNA.

Bssn

ZOl HD 01009-03 LMG

2. In vitro analysis of the self-priming mechanism.

As an additional means of studying the chemistry of priming, we will develop an in vitro priming assay
to generate large quantities of priming intermediates. Initially, we will attempt to produce a priming
reaction from purified RT and Tfl transcript. RT expression vectors will be produced and tested for
use in bacteria as well as baculovirus and Pichia pastoris. Once RT is successfully purified from an
expression system, we will combine RT, in vitro transcribed Tfl mRNA, and ^-P labeled dNTP's under
a varity of conditions. These reactions will be tested for the synthesis of minus-strand strong-stop
DNA. Mutations in RNaseH and/or the polymerase domain of RT will be tested in the assay to
characterize the source of the cleavge activity. We will also test reactions with mRNA alone to
determine if the cleavage is autocatylitic. The products of the in vitro reactions will also be examined
for evidence of a 2'OH priming mechanism. The products of priming reactions containing radiolabeled
Tfl mRNA will be digested with the nucleases PI and T2 to determine if the bond between the RNA
primer and the synthesized DNA is a 2'-5' or a 3'-5' linkage [Ruskin, 1990 #917; Ruskin, 1985 #747].

3. The function of the 39 base pair RNA structure.

In conjunction with the study of the mRNA cleavage mechanism, we will continue to explore the role
of the 39 base pair RNA structure that contains the PBS. The data from our mutagenesis of this
structure indicated that sequence along its entire length are critical for transposition. We now plan to
evaluate if these mutations can be rescued by the addition of compensitory mutations, suggesting that
the RNA structure is important and not the individual sequences. Mutations from each of the 4 sections
of the RNA structure will be further studied to identify their individual contributions to the transposition
process. These mutations will be tested for their effects on translation, mRNA packaging into particles,
self-priming, reverse transcription, and integration.

B. The mechanism of Tfl VLP assembly.

1. Identification of growth conditions that induce IN degradation

We have shown that cells starved for nitrogen or sulfate undergo IN degradation while cultures starved
for glucose maintain equal molar amounts of Gag and IN. This observation will allow us to test the
transposition efficiency of Tfl in cells that starve for glucose to evaluate the importance of FN
degradation to transposition. If functional particle development requires IN degradation then the cells
starved for glucose will be unable to transpose even several days after stationary phase is reached. If
however, the degradation of IN is not required for transposition, the glucose starved cells should
undergo normal levels of transposition. This question of IN degradation will be addressed below in
separate experiments.

2. The role of IN degradation in transposition.

Although we propose that the reduction in mature IN protein observed in stationrary phase cells is a
regulated step of Tfl particle assembly, the possibility exists that assembly occurs before IN degradation
and excess FN not incorporated into the VLP's is degraded simply due to greater exposure to cellular
proteases. Our observation that much of Tfl reverse transcription occurred after IN degradation
indicated that the particles with excess Gag are functional and represent intermediates in the
transposition process. The idea that IN degradation is a programmed aspect of particle development
is consistent with our observation that the majority of the RT is also degraded in stationary phase cells

IIIIIIIIIHIIIIIIIIIIHI

ZOl HD 01009-03 LMG

(J. Lin, A. Atwood, and H. Levin, unpublished). A resolution to the question of how important is the
reduction of Pol protein to the assembly process requires that we develop direct genetic means of
manipulating the Gag to Pol ratios in stationary phase ceils.

We plan to test the importance of high Gag to FN ratios in transposition by overexpressing FN from a
separate plasmid with a promoter that is active in stationary phase cells. If particle development
requires IN levels be reduced much below the amount of Gag protein, the increase in IN provided by
the overexpression plasmid should cause a defect in transposition. The fbpl promoter is active in
stationary phase and will be tested for use in this experiment [Hoffman, 1989 #858].

Another approach we will take to define the importance of Gag to IN ratios is to reduce Pol protein
levels throughout particle formation by inserting "leaky" frameshift mutations between Tfl Gag and PR,
PR and RT, and between RT and IN. The frameshift sequences from Ty3, Tyl and Ml have been
characterized in S. cerevisiae and will be tested in S. pombe for the purposes of this experiment
[Dinman, 1992 #1361; Farabaugh, 1993 #1389; Belcourt, 1990 #824]. If the excess Pol proteins are
simply an impediment to assembly, the reduction in expression caused by the frameshift mutations will
have no detrimental effect on transposition and may actually cause an improvement in activity.

3. Characterization of the activity that degrades IN.

An important aspect of Tfl particle assembly that we will address here is the source of the IN
degradation activity. The protease or proteases that degrade the IN and RT in stationary phase cells
could either be a transposon encoded protein such as PR, or it could be a host protein. We will study
the stability of mature-sized IN expressed on a plasmid in the absence of all the other Tfl proteins to
determine if host factors alone are sufficient to regulate the IN degradation in stationary phase cells
without the association of other Tfl proteins. If no IN degradation is observed in these stationary phase
cells, or if IN is unstable even in log-phase cells, the possibility will be tested that regulated degradation
of IN requires the presence of Gag, RT, and IN assembled into particles. We plan to construct Gag,
PR, and RT expression plasmids so that mature-sized Gag, PR, RT, and IN can be independently
expressed within the same cells to evaluate the requirements for fN degradation. IN and PR will be
coexpressed in one experiment as a direct test of PR catalyzed degradation of IN. In additional
experiments. Gag, PR, RT and IN will be coexpressed in several combinations to establish the minimal
requirements to reproduce the regulated degradation of IN. If IN regulation can be observed in one of
these combinations, the requirement for PR will be evaluated.

The independent expression of all of the mature Tfl proteins will present an opportunity to determine
if particle assembly requires intermediates composed of primary translation products. One model for
particle assembly predicts that the attachment of Gag to the Pol proteins in the primary translation
product may be required to localize RT and IN to the sites of particle formation. Alternatively, the
result that Tfl proteins are completely processed into mature products that appear to form particles
before RT and IN are degraded suggests that the independent expression of Gag, RT and IN may result
in functional particles. The ability of independently expressed mature-sized proteins to form functional
particles will be tested by expressing Gag from a TfI-«eo transcript that has a frame shift placed before
PR.

4. The study of particle stability to measure the structural integrity of log phase particles compared to
stationary phase particles.

ZOl HD 01009-03 LMG

The Tfl protein isolated from stationary phase cells was found to sediment in sucrose gradients as
particles that contained 50-fold more Gag than FN [Atwood, 1995 #1576]. Particles isolated from log
phase cells have very different properities including the association of a much greater amount of IN and
RT [Atwood, 1995 #1576]. We wish to probe these two forms of Tfl particles for evidence of a
maturation process that could result in greater structural integrity. The stability of the log and stationary
phase particles will be evaluated by their resistance to perturbations such as treatment with denaturing
agents and proteases. Particles from log and stationary phase cells will be isolated in sucrose gradients.
The strength of the Gag and IN association in these particles will be tested in various concentrations
of denaturing agents (e.g. tween, urea and guanidinium hydrochloride) to evaluate the stability of log-
phase particles compared to stationary phase particles. After the denaturing treatments, the particles will
be refractionated on sucrose gradients and the fractions analyzed on immunoblots to determine the effect
of the treatments on particle integrity. The structure of log phase and stationary phase particles will also
be probed by partial protease treatments. The sensitivity of Gag and IN to proteases will also be
monitored by immunoblot analysis.

C. The primer cleavage activitiy of RNaseH.

Two different mutations in the RNaseH domain of RT were identified that caused a sharp drop in
transposition activity despite the results of DNA blot analysis that showed these strains produced normal
levels of full-length reverse transcript. This was an unexpected result because previous studies of
RNaseH mutants have indicated that the synthesis of mature double-stranded cDNA requires RNaseH
activity to release single stranded DNA intermediates from their RNA templates [Champoux, 1993
#1442; Blain, 1995 #1602]. One likely explanation of these results is that the mutants specifically affect
the ability of RNaseH to cleave the residule RNA primers off the 5' ends of the cDNA without reducing
the enzyme's general ability to degrade RNA annealed to DNA. The result of this defect would be to
allow RT to product completed cDNA but the presence of the RNA primers would inhibit the
integration reaction. Although this primer cleavage activity has been observed in vitro [Champoux,
1984 #1603; Rattray, 1987 #1606; Furfme, 1991 #1604; Pullen, 1992 #1605], no evidence has yet been
reported indicating the two RNaseH activities have different active site requirements. We will test this
hypothesis using a primer extension method and a DNA blot technique to detect the cDNA primers.
If the mutants exhibit normal primer removal, we will consider alternative causes of the reduced
transposition such as participation of RNaseH in the integration reaction. This issue could be addressed
using the type of m vitro integration assays developed for other retrotransposons [Kirchner, 1995 #1474;
Braiterman, 1994 #1492; Braiterman, 1994 #1493].

D. The characterization of host gene function in Tfl retrotransposition.
1. hopJ.

The observation that hopl-1 mutant cells do not accumulate Tfl protein could be due to a defect in
translation or an increase in degradation. We have now assembled a set of Tfl expression plasmids that
contain lacZ fused to several different regions of Tfl. These plasmids will be expressed in S. pombe
to determine whether the Tfl 5' untranslated region or the Tfl coding sequence must be fused to lacZ
for beta-galactosidase activity to be reduced in hopl-1 cells. We have also constructed a plasmid
containing the Tfl ORF fused to the nmtl promoter with the nmtl 5' untranslated region. This plasmid
will be transformed into wild type and hopl-1 cells to test the stability of normal Tfl proteins expressed
from this heterologous transcript. Any effect of the hopl-1 mutant on Gag levels expressed from this

HHm

ZOl HD 01009-03 LMG

plasmid would suggest that Tfl protein levels drop due to increased degradation. The results produced
using all the fusion plasmids will allow additional plasmids to be produced that will help to further
define which sequences of Tfl are most effected by hopl-1.

We will seek additional information about the function of hopl by continuing our effort to clone the
gene. We now have one 6 kb genomic clone that suppresses the hopl-1 phenotype and is geneticlly
linked to the site of the original hopl mutation. Should this 6 kb insert contain the hopl gene, we will
analyze its sequence for information related to its function.

2. hop2.

Strains with the hop2-l mutation have an 8-fold transposition defect and a 4-foId drop in their
homologous recombination frequency. Because the phenotype of the hop2 strains is similar to hopl
cells in that the levels of Tfl protein accumulation was low, we are testing these strains for defects in
translation or protein stability using the same lacZ fusions developed for characterizing hopl. We will
also attempt to isolate the hop2 gene using an S. pombe genomic library to identify complementing
clones.

3. hopS.

The hop3-l mutation caused a 10-fold drop in both the transposition and homologous recombination
assay frequencies. Unlike the hopl and hop2 mutations, hop3 cells were able to produce wild type
levels of mature Tfl proteins in log phase cultures. However, 1 day after hop3-l cells reached
stationary phase, they exhibited a sharp drop in Gag levels suggesting a loss of protein stability.
Although we observed the hop3-l cells contained normal levels of reverse transcript, the loss of Gag
and possibly FN could result in lower IN activity or less efficient cDNA presentation to the nucleus.
To reveal the cause of low Tfl protein accumulation in stationary phase cells we are testing these
strains for defects in translation or protein stability using the same lacZ fusions developed for
characterizing hopl. Preliminary results indicate that the Gag protein in stationary phase hop3-l cells
is absent even when expressed from a plasmid with the ntntl 5' untranslated region fused to the Tfl
ORF. This suggests that the lack of Gag is due to degradation, not a defect in translation. We will
therefore attempt to identify which Tfl proteins are specificly destablized in hop3-l cells. We also plan
to isolate the hop3 gene using the genomic library to complement the lack of transposition exhibited
by hop3-l cells.

4. hop 5

Cells with the hop5-l mutation transpose with frequencies 12-fold lower than wild type strains. Results
from the homologous recombination assay also indicated a 12-fold defect. As observed for hop3-l
cells, DNA blots made from hop5-l strains showed normal levels of mature reverse transcript were
produced. Immunoblots indicated that this mutation had no effect on the levels of Tfl protein
accumulation. We speculate that the defect caused by hop5-l may occur in the presentation of the
cDNAifN complex to the nucleus. Alternatively, the transposition and recombination defect could be
caused by a mutation in a general recombination factor that contributes to both processes. To test this
possibility, we will measure the efficency of homologous recombination between plasmid and genomic
copies of ade 6 in both wild type and hop5-l cells. We will also measure the frequency of switching
at the mating type cassette as an independent measure of a homologous recombination process.

^am

MiiiyiiraiiM

ZOl HD 01009-03 LMG

Additional information about the function of hop5 will be sought by the isolation and characterization
of its sequence.

Publications

Atwood A, Lin J, Levin H. The retrotransposon Tfl assembles virus-like particles with excess Gag due
to specific degradation of Pol protein, Mol Cell Biol 1995; in press.

Levin HL. A novel mechanism of self-primed reverse transcription defines a new family of
retroelements, Mol Cell Biol 1995;15:3310-17.

References

Atwood, A., J. Lin, et al. (1995). "The retrotransposon Tfl assembles virus-like particles with excess

Gag due to specific degradation of Pol protein." Mol. Cell. Biol.: in press.
Belcourt, M. F. and P. J. Farabaugh (1990). "Ribosomal frameshifting in the yeast retrotransposon Ty:

tRNAs induce slippage on a 7 nucleotide minimal site." Cell 62: 339-352.
Blain, S. and S. Goff (1995). "Effects on DNA synthesis and translocation caused by mutations in the

RNase H domain of Moloney Murine Leukemia virus reverse transcriptase." J. Virol 69: 4440-

4452.
Braiterman, L. T. and J. D. Boeke (1994). "In vitro integration of retrotransposon Tyl: a direct physical

assay." Mol Cell Biol 14(9): 5719-30.
Braiterman, L. T. and J. D. Boeke (1994). "Tyl in vitro integration: effects of mutations in cis and in

trans." Mol Cell Biol 14(9): 5731-40.
Champoux, J. (1993). Roles of ribonuclease H in reverse transcription. Reverse Transcriptase. A. Skulka

and S. Goff Cold Spring Harbor, NY, Cold Spring Harbor Laboratory Press: 103-117.
Champoux, J., E. Gilboa, et al. (1984). "Mechanism of RNA primer removal by the RNase H activity

of avian myeloblastosis virus reverse transcriptase." J. Virol 49: 686-691.
Davies, J., Z. Hostomska, et al. (1991). "Crystal structure of the ribonuclease H domain of HIV- 1

reverse transcriptase." Science 252: 88-95.
Dinman, J. D. and R. B. Wickner (1992). "Ribosomal frameshifting efficiency and gag/gag-pol ratio

are critical for yeast Ml double-stranded RNA virus propagation." Journal of Virology 66:

3669-3676.
Eichinger, D. J. and J. D. Boeke (1988). "The DNA intermediate in yeast Tyl element transposition

copurifies with virus-like particles: cell-free Tyl transposition." Cell 54(955): 955-966.
Farabaugh, P. J., H. Zhao, et al. (1993). "A novel programed frameshift expresses the P0L3 gene of

retrotransposon Ty3 of yeast: frameshifting without tRNA slippage [published erratum appears

in Cell 1993 Nov 19;75(4):826]." CeU 74(1): 93-103.
Furfine, E. and J. Reardon (1991). "Reverse transcriptase-RNase H from the human immunodeficiency

virus. Relationship of the DNA polymerase and RNA hydrolysis activitiies." J Biol Chem 266:

406-412.
Hoffman, C. S. and F. Winston (1989). "A transcriptionally regulated expression vector for the fission

yeast Schizosaccharomyces ponibe." Gene 84: 473-479.
Hostomsky, Z., S. H. Hughes, et al. (1994). "Redesignation of the RNase D activity associated with

retroviral reverse transcriptase as RNase H." J Virol 68(3): 1970-1.
Hsu, M., S. Eagle, et al. (1992). "Cell-free synthesis of the branched RNA-linked msDNA from retron-

Ec67 of Escherichia coli." J. Biol. Chem. 267: 13823-13829.

fomm

mm^^^^^tmm^soBmmmmmma^K^^^^^^^mmBmm^sm&i^

ZOl HD 01009-03 LMG

Hsu, M. Y., M. Inouye, et al. (1990). "The msDNA-Ec67 retron: a potential E. coli transposable

element encoding both reverse transcriptase and Dam methylase proteins." Proc. Natl. Acad.

Sci. USA 87: 9454-9458.
Katayanagi, K., M. Miyagawa, et al. (1990). "Three dimensional structure of ribonuciease H from E.

coli." Nature 347; 306-309.
Kirchner, J., C. Connolly, et al. (1995). "Requirement of RNA polymerase III transcription factors for

in vitro position-specific integration of a retroviruslike element." Science 267: 1488-1491.
Kohlstaedt, L., J. Wang, et al. (1992). "Crystal structure at 3.5 A resolution of HIV-1 reverse

transcriptase complexed with an inhibitor." Science 256: 1783-1790.
Lampson, B. C, M. Inouye, et al. (1989). "Reverse transcriptase with concomitant ribonuciease H

activity in the cell-free synthesis of branched RNA-linked msDNA of Myxococcus xanlfms."

Cell 56: 701-707.
Levin, H. L. (1995). "A Novel mechanism of self-primed reverse transcription defines a new family of

retroelements." Mol Cell Biol 15: 3310-3317.
Levin, H. L. and J. D. Boeke (1992). "Demonstration of retrotransposition of the Tfl element in fission

veast." EMBO J. 11: 1145-1153.
Mizrahi, V., R. Brooksbank, et al. (1994). "Mutagenesis of the conserved aspartic acid 443, glutamic

acid 478, and aspartic acid 498 residues in the ribonuciease H domain of p66/p51 human

immunodeficiency virus type I reverse transcriptase." Journal of Biological Chemistry 269:

19245-19249.
Moreno, S., A. Klar, et al. (1991). "Molecular genetic analysis of fission yeast Schizosaccharomyces

pombe." Methods in Enzymologv 194: 795-823.
Panet, A., D. Baltimore, et al. (1975). "Quantitation of avian RNA tumor virus reverse transcriptase by

radioimmunoassay." Journal of Virology 16: 146-152.
Pullen, K., L. Ishimoto, et al. (1992). "Incomplete removal of the RNA primer for minus-strand DNA

synthesis by human immunodeficiency virus type 1 reverse transcriptase." J Virol 66: 367-373.
Rattray, A. and J. Champoux (1987). "The role of Moloney murine leukemia virus RNase H activity

in the formation of plus-strand primers." J Virol 61: 2843-2851.
Repaske, R., J. Hartley, et al. (1989). "Inhibition of RNase H activity and viral replication by single

mutations in the 3' region of moloney murine leukemia virus reverse transcriptase." Journal of

Virology: 1460-1464.
Ruskin, B. and M. R. Green (1985). "An RNA processing activity that debranches RNA lariats."

Science 229: 135-140.
Ruskin, B. and M. R. Green (1990). "RNA lariat debranching enzyme as tool for analyzing RNA

structure." Meth. Enzvmol. 181: 180-188.
Stromberg, K., N. Hurley, et al. (1974). "Structural studies of avian myeloblastosis virus: Comparison

of polypeptides in virion and core components by dodecyl sulfate-polyacrylamide gel

electrophoresis." Journal of Virology 13: 513-528.
Yang, W., W. Hendrickson, et al. (1990). "Structure of ribonuciease H phased at 2 A resolution by

MAD snalysis of the selenomethionine protein." Science 249: 1398-1405.

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBUC HEALTH SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

ZOl-HDOlOlO-01

PERIOD COVERED

October 1, 1994 through September 30, 1995

TU£ OF PROJECT ISO^ttaraaters or less, fitle must fil-pn ona line -be

Regulation or huKaryotic Protein byntnesis

•between the borders.)

PRINCIPAL INVESTIGATOR lU'st other professional personnel below the Principal Investigator.! (Name, title, laboratory, and institute affiliation)

PI: Thomas E. Dever Unit Head LMG:NICHD

Others: Kobayashi, Makiko Visiting Fellow LMG:NICHD

Locke, Emily Biologist LMG:N1CHD

Ung, Tekly Volunteer LMG:NICHD

COOPERATING UNITS lif any)

Laboratory of Molecular Genetics

SECTION

Unit on Translational Regulation

INSTITUTE AND LOCATION

NICHD, NIH. Bethesda, Maryland 20892

TOTAL STAFF YEARS:

3.4

PROFESSIONAL:

OTHER:

1.5

CHECK APPROPRIATE BOX(ES)

n (a) Human subjects D (b) Human tissues
n (a1) Minors
n (a2) Interviews

SUMMARY OF WORK lUse standard unreduced type. Do not exceed the space provided.)

The regulation of protein synthesis in mammalian cells under stress conditions as well as the
regulation of GCN4-specific translation in the yeast Saccharomyces cerevisiae under amino acid
starvation conditions is mediated by phosphorylation of the eukaryotic translation initiation factor
(eIF)-2. Phosphorylation of serine-51 on the a subunit of eIF-2 by the mammalian PKR kinase
inhibits general translation, while phosphorylation of serine-51 by the yeast GCN2 kinase is required
to increase expression of GCN4. We have been studying how these protein kinases specifically
recognize and phosphorylate eIF-2a on serine-51. Over 100 mutations throughout the eIF-2a
protein have been identified that prevent induction of GCN4 expression when GCN2 is activated.
Among the residues immediately flanking serine-5 1 , the positions -1 and -2 appear most important
for regulation. Examination of eIF-2a phosphorylation in vivo in strains carrying various mutant
alleles reveals that some of the substitutions markedly affect the ability to phosphorylate serine-51.
We have also been studying the vaccinia virus K3L protein, a pseudosubstrate inhibitor of PKR. We
have established a system to suppress the toxicity of PKR expression in yeast by co-expressing K3L.
Using this system both loss-of-function and increased activity K3L mutants have been isolated.
Residues near the carboxyl-terminus of K3L, that are conserved in eIF-2a (residues 73-83), are
critical for K3L activity. In addition, the corresponding residues in eIF-2a are required for proper
regulation of GCN4 expression. Finally, we have begun characterization of a novel eIF-2a kinase
inhibitor from the baculovirus Autographa califomica.

PHS 6040 (Rev. 5/92)

ZOl HDOlOlO-01-LMG
Project Description:

Objective: To understand tiie mechanism and regulation of protein biosyntiiesis in eukaryotic organisms,
especially focusing on the phosphorylation of eIF-2a by the yeast Saccharomyces cerevisiae GCN2 kinase
and by the mammalian PKR kinase, and to understand how viruses circumvent this regulatory network.

Major Findings:

1. Identification of eIF-2a mutants defective for GCN4 translational regulation.

The yeast GCN2 kinase phosphorylates eIF-2a on serine-51 in response to an amino acid starvation.
Phosphorylated eIF-2 acts as a competitive inhibitor of eIF-2B, the guanine nucleotide exchange factor for
eIF-2. Thus, phosphorylation of eIF-2 results in an inhibition of general translation. In yeast, this
phosphorylation of eIF-2 and subsequent inhibition of eIF-2B activity not only limits general translation, but
also specifically stimulates GCN4 expression. Increased expression of GCN4 is required for yeast strains
to grow under amino acid starvation conditions. Therefore, yeast mutants that are unable to phosphorylate
eIF-2a or which fail to inhibit eIF-2B cannot induce GCN4 expression and are thus sensitive to amino acid
starvation conditions. To identify how the eIF-2a kinases, such as GCN2, specifically recognize serine-51
on the a subunit of eIF-2 we have begun a mutagenic analysis of eIF-2a. The residues immediately flanking
serine-5 1 were randomly mutated and the mutant alleles were introduced into yeast in place of the wild-type
eIF-2a. Pools of mutants were screened to identify eIF-2a mutations that block the growth of the yeast cells
under amino acid starvation conditions. Following the identification of the mutants sensitive to starvation
conditions, the plasmids carrying the mutant eIF-2a alleles were isolated and sequenced to identify the amino
acid substitution in eIF-2a that blocks the translational regulation of GCN4 expression. The sequence
flanking serine-51 from residue 48 (-3) to residue 54 (+3) is: SELSjiRRR. No loss-of-regulation mutations
were identified at residues S48, R52 or R53. Only a single mutation was identified at the +3 position: R54G.
However, 10 mutations at position -1 (L50) and all 19 possible substitutions at position -2 (E49) blocked the
regulation. There are at least two possible mechanisms by which a mutation in eIF-2a could block regulation
of GCN4 expression: (1) the mutation could block the ability of GCN2 to phosphorylate eIF-2a or (2) the
mutation could prevent phosphorylated eIF-2 from regulating eIF-2B. One-dimensional polyacrylamide
isoelectric focusing slab gels can be used to examine eIF-2a phosphorylation in yeast cells. To date, only
one of the loss-of-regulation mutations, a leucine to proline change at position 50 (L50P), has been found
to severely inhibit phosphorylation of eIF-2a by GCN2.

In addition to examining the residues immediately flanking serine-51, we have also mutated other
residues based on their conservation in the vaccinia virus K3L protein. The vaccinia virus K3L protein is
an 88 amino acid pseudosubstrate inhibitor of the mammalian anti-viral PKR kinase. K3L shares 28% amino
acid sequence identity with the amino-terminal 90 residues of eIF-2a. This homology is most striking
between residues 72 and 83 of eIF-2a where 10 of 12 residues are conserved between K3L and human elF-
2a. Based on the strong sequence conservation and the fact that both K3L and eIF-2a bind to PKR we
hypothesized that the conserved residues may be important for the kinases to recognize and phosphorylate
eIF-2a. To test this prediction the residues in yeast eIF-2a that are conserved in K3L were randomly
mutated and the mutant pools screened to identify eIF-2a alleles that block regulation. To date, 14 residues
have been examined and 72 loss-of-regulation mutations have been identified. It appears that for several of
the residues in the conserved block between positions 72 and 83 any mutation will obstruct regulation.
Preliminary analysis of several of these mutations using the isoelectric focusing gels suggests that they do
not impede the ability of GCN2 to phosphorylate eIF-2a. Therefore, these mutations are probably affecting
the inhibition of eIF-2B by phosphorylated eIF-2. Since both K3L and eIF-2a bind to the eIF-2a kinases

^^^■n

ZOl HDOl 010-0 1-LMG

it seems reasonable to suggest that the residues conserved between K3L and eIF-2a are probably important
for this interaction. However, the mutational and isoelectric focusing analysis of some of these conserved
residues suggests that they are affecting the interaction between eIF-2 and eIF-2B. Taken together these
results suggest a model in which the kinase recognition and eIF-2B recognition surfaces on eIF-2a overlap.

2. Mutational analysis of the vaccinia virus K3L protein.

As introduced above, the vaccinia virus K3L protein is a pseudosubstrate inhibitor of the elF-2a
kinase PKR, a component in the mammalian antiviral defense mechanism. To increase our understanding
of how the elF-2a kinases recognize their substrate we have carried out an analysis of the K3L protein. First,
we obtained from Dr. Patrick Romano a yeast strain in which the PKR kinase, expressed under the control
of a GAL promoter, was integrated into the yeast genome. This strain grows well on glucose medium,
however the strain is dead on galactose medium due to severe inhibition of translation caused by
phosphorylation of eIF-2a. The vaccinia virus K3L protein gene was cloned into the vector pEMBLyex4
such that K3L expression was also under GAL control. When the K3L expression plasmid was introduced
into the yeast strain expressing PKR we found that K3L could suppress the lethality caused by high level
expression of PKR in yeast. These results established the yeast system as a tool to study K3L.

Two types of mutational analyses have been conducted on K3L: a random mutational analysis to
identify superactive K3L alleles and a site-directed mutational analysis to assess the importance of residues
conserved between K3L and eIF-2a. For the random mutational analysis the K3L gene was amplified under
low-fidelity conditions using the PCR. The mutant pool of K3L genes was subcloned into pEMBLyex4 and
the plasmids were introduced into the yeast strain containing the GAL-PKR construct. The colonies were
screened to identify K3L mutants that were better suppressors of PKR. Five such mutants have been
identified and one, which is significantly better than the others, has been more extensively characterized.
This latter K3L upmutant gene was sequenced and found to contain two nucleotide substitutions. One was
a silent mutation that did not alter the amino acid sequence, and the second caused a histidine to arginine
change at residue 47 (K3L-H47R). While wild-type K3L yielded a partial suppression of the slow-growth
phenotype due to high level expression of PKR in yeast, the K3L-H47R mutant afforded almost complete
suppression of the toxic effects associated with PKR. Immunoblot analysis reveals that the K3L-H47R
protein is not expressed to higher levels than the wild-type protein nor is PKR expression lower in the K3L-
H47R strain than in the wild-type K3L strain. Thus the increased suppression is not due to trivial reasons
like increased K3L expression or decreased PKR levels. One hypothesis we are currently pursuing is that
K3L-H47R will bind tighter to PKR than wild-type K3L. We have been able to co-immunoprecipitate K3L
and K3L-H47R with PKR. Currently, we are altering the wash conditions to see if we can detect a difference
in the strength of the association between PKR and K3L-H47R versus wild-type K3L. A point of interest
relating back to the first project is that the K3L-H47R mutation makes K3L more similar to eIF-2a. In fact,
the H47 residue in K3L aligns with R52 in eIF-2a immediately adjacent to the phosphorylation site at serine-
51. A possible interpretation of these results is that by increasing the similarity between K3L and eIF-2a
around the serine-51 region, K3L will interact more avidly with PKR.

The site-directed mutational analysis of K3L has focused on the carboxyl-terminal region. As stated
above, this region of K3L contains a stretch where 10 out of 12 residues are perfectly identical to mammalian
eIF-2a. The sequence KGYID found between residues 74 and 78 in K3L is perfectly conserved in all eIF-2a
and K3L proteins that have been identified. As mentioned previously mutations in this region of yeast elF-
2a block the regulation of GCN4 expression. Mutation of K74, Y76 or D78 of K3L to alanine blocks the
ability of either wild-type K3L or the upmutant K3L-H47R to suppress PKR. In addition, immunoblot
analysis reveals that the Y76A mutation does not alter K3L protein levels ruling out the trivial possibility

wmm

■m

ZOl HDOlOlO-01-LMG

that the loss of suppression is due to lower K3L protein levels. A second set of site-directed mutations were
introduced to determine the carboxyl-terminal boundary of K3L required for its PKR inhibitory activity.
Truncation of the 88-residue K3L protein at residue 73, eliminated the KGYID sequence and the ability to
suppress PKR. Truncation of only the last five residues had no effect on K3L activity. While after
truncation of three additional residues (8 total), the ability of K3L or K3L-H47R to inhibit PKR was lost.
Recent results reveal that after removal of the carboxyl-terminal 6 amino acids K3L-H47R retains its
suppressing activity suggesting that the carboxyl-terminal boundary for K3L activity is either 6 or 7 residues
from the end of the protein.

3. Identification of a baculovirus inhibitor of eIF-2a kinases.

Many viruses have evolved ways to counteract the antiviral defense mechanisms of eukaryotic cells.
One of the defense mechanisms cells employ is phosphorylation of eIF-2a by the PKR kinase. As described
above, vaccinia virus expresses the K3L protein, a pseudosubstrate inhibitor of PKR. Other viruses express
different protein or RNA inhibitors of PKR and still others activate latent cellular inhibitors of PKR or even
degrade PKR. Since our studies of the K3L protein have provided insights into how PKR recognizes eIF-2a,
we have examined the literature closely to see if we could identify any new viral inhibitors of PKR that
might offer new insights into this mechanism of translational regulation. When the sequence of the
baculovirus Autographa californica was reported late in 1994 an open reading frame designated ORF123
encoding a truncated protein kinase termed pk2 was identified. The amino acid sequence of pk2, though
truncated, was most similar to the eIF-2a kinase family. Since many viruses encode inhibitors of the eIF-2a
kinases, we reasoned that pk2 might be an inhibitor of PKR. To test this hypothesis the pk2 gene was
isolated from baculovirus using the PCR and the gene was inserted into the vector pEMBLyex4.
Introduction of the pk2 expression plasmid into the yeast strain that expresses PKR under the control of the
GAL promoter revealed that pk2 could suppress the toxicity of PKR in yeast. Not only could pk2 suppress
PKR toxicity, but pk2 was also able to suppress both wild type and hyperactive alleles of the yeast elF-2a
kinase GCN2. These results demonstrate that pk2 is an eIF-2a kinase inhibitor, and since baculovirus is an
insect virus this is the first indication that insects may also use phosphorylation of eIF-2a as a means to block
viral infection.

The pk2 protein is similar to the carboxyl-terminal 1/2 of an eIF-2a kinase domain. This raised the
possibility that the corresponding fragment from an authentic eIF-2a kinase domain may also act in a
dominant negative manner. The amino acid sequences from GCN2 and pk2 were aligned and
oligonucleotides were identified to amplify the corresponding region of the GCN2 kinase using the PCR.
Following cloning into pEMBLyex4, the truncated gcn2 kinase domain was introduced into various strains.
High level expression of the truncated gcn2 kinase domain was found to inhibit the toxicity associated with
the hyperactive GCN2' kinases and thus reveal that this region of an eIF-2a kinase can act in a dominant
negative manner.

4. Examination of substrate specificify of the elF-2a kinases.

Most protein kinases can be classified into one of two classes either serine/threonine specificify or
fyrosine specificify. However, a few kinases have been identified with dual specificify that will
phosphorylate both fyrosine and serine/threonine residues. To date the only known physiologically relevant
substrate of the eIF-2a kinases is eIF-2a, however recent reports suggest that PKR may have alternative
substrates. While the eIF-2a kinases have only been reported to phosphorylate serine residues the alternative

ZOl HDOlOlO-01-LMG

phospho-acceptors threonine and tyrosine have not been examined. Intriguingly, the mouse PKR kinase was
originally identified as TIK in a screen for tyrosine tcinases. However, in this study there was no evidence
that PKR was phosphorylated on or that PKR could phosphorylate other proteins on tyrosine residues. To
further investigate the substrate specificities of the eIF-2a kinases and as a possible means to examine the
differences between the serine/threonine and tyrosine kinases we mutated the serine-51 residue in eIF-2a
to threonine and tyrosine. Yeast cells expressing the eIF-2a-S5 IT allele were practically indistinguishable
from wild-type cells with a good growth rate and the ability to grow under amino acid starvation conditions,
thus indicating that GCN2 can phosphorylate the threonine at residue-51 on the eIF-2a-S51T protein. In
addition, high level expression of PKR in the eIF-2a-S5 IT strain was lethal again demonstrating that the elF-
2a kinases can phosphorylate a threonine residue, and that the phosphothreonine residue maintains the
normal regulation of eIF-2B.

When serine-5 1 was replaced with tyrosine initial results suggested that tyrosine was not a substrate.
Yeast strains expressing GCN2 and carrying the eIF-2a-S51Y allele are sensitive to amino acid starvation
conditions suggesting that GCN2 cannot phosphorylate eIF-2a-S51Y. In addition high level expression of
PKR in the eIF-2a-S51Y strain is not lethal. Both of these results are consistent with the eIF-2a kinases
being unable to phosphorylate tyrosine residues. However, upon closer inspection a very modest slow-
growth phenotype was observed in eIF-2a-S51Y strains expressing high levels of PKR. In addition, these
strains expressing PKR to high levels could grow under amino acid starvation conditions suggesting that
PKR may be phosphorylating the tyrosine residue and inducing GCN4 expression. To provide convincing
evidence that PKR can in fact phosphorylate eIF-2a-S5 1 Y, isoelectric focusing analysis was performed. A
signal consistent with phosphorylation of eIF-2a was observed in strains expressing PKR and either wild-
type eIF-2a or eIF-2a-S51Y, but not eIF-2a-S51A. In addition, the phosphorylation of eIF-2a-S51Y was
dependent on wild-type PKR and was not observed with an inactive mutant of PKR. To confirm that the elF-
2a-S51 Y protein was indeed phosphorylated on tyrosine the immunoblot from the isoelectric focusing gel
was probed with anti-phosphotyrosine antibodies. A signal was specifically detected in the strains
expressing both PKR and elF-2a-S51Y, but not wild-type eIF-2a. Thus PKR can phosphorylate a tyrosine
residue in vivo. Surprisingly, GCN2' kinases were also found to phosphorylate eIF-2a-S5IY when
examined on the isoelectric focusing gels despite the lack of any noticeable phenotypes. The finding that
PKR can phosphorylate a tyrosine residue in vivo raises the possibility that alternate substrates for PKR or
any of the eIF-2a kinases may be phosphorylated on tyrosine residues rather than exclusively on serine
residues.

Proposed Course of Project:

1 . eIF-2tt mutants defective for translational regulation. We plan to continue our analysis of eIF-2a

mutants identifying alleles that fail to properly induce GCN4 expression under amino acid starvation
conditions. More importantly, we will examine the phosphorylation of serine-5 1 in the mutant proteins using
the isoelectric focusing gels. While only a single mutation has been identified that severely impairs
phosphorylation, a few mutations have been identified that appear to reduce phosphorylation. By
constructing chimeric eIF-2a proteins containing combinations of mutations that alone modestly impair
phosphorylation, we hope to obtain additional alleles that are severely impaired for phosphorylation by
GCN2 in vivo. The eIF-2a alleles that are defective for phosphorylation by GCN2 in vivo will also be
examined both in vivo and in vitro with PKR. The ability of these alleles to suppress the toxicity of high
level PKR expression in yeast will be monitored, and the ability of PKR to phosphorylate these proteins will
be examined both in vivo (using the isoelectric focusing gels) and in vitro. The in vitro kinase assays will
employ either recombinant PKR or PKR immunoprecipitated from yeast and the substrate will be

ZOl HDOlOlO-01-LMG

recombinant eIF-2a alleles expressed using bacterial expression systems. Alleles of eIF-2a that are
defective for phosphorylation by PKR will be used in a reversion analysis to identify PKR mutants with
altered substrate specificity. Pools of mutant PKR genes will be introduced into yeast cells expressing the
mutant eIF-2a alleles, and PKR mutants that restore phosphorylation will be identified as causing a slow-
growth phenotype or increased resistance to amino acid starvation conditions.

2. Analysis of the vaccinia virus K3L protein. The first priority is to determine if the superactive K3L
mutants bind to PKR with greater avidity. As discussed above, K3L can be co-immunoprecipitated with
PKR and we are now varying conditions to see if K3L-H47R binds tighter than wild-type K3L. Similarly,
the loss-of-function mutants in K3L will also be examined using the co-immunoprecipitation assay to
determine if these mutations alter the binding affinity of K3L for PKR. A new screen for additional K3L
upmutants will be performed. Through the identification of key residues in the K3L protein we hope to
uncover the recognition elements the kinases use when modifying eIF-2a. In a second screen we will
identify PKR mutants that are resistant to K3L. The K3L upmutant K3L-H47R will be introduced into a
yeast strain under the control of a GAL promoter. Wild-type PKR will be non-toxic in this strain due to
inhibition by K3L. The PKR gene also under GAL control will be randomly mutated and a pool of mutant
PKR genes will be introduced into the yeast strain expressing K3L-H47R. PKR mutants resistant to K3L-
H47R will be identified as causing a slow-growth or no-growth phenotype on galactose medium. These PKR
mutants will also be examined to determine their resistance to the various K3L upmutants uncovered in the
other screens. The identification of PKR mutants resistant to K3L will help identify the substrate binding
region in PKR; and through the examination of the various K3L and PKR mutants we may be able to map
the interacting surfaces in PKR and K3L. These results should help identify how PKR and the other eIF-2a
kinases interact with the authentic substrate eIF-2a.

3. Analysis of the baculovirus pk2 protein. A primary goal for this study is to determine how pk2
inhibits the eIF-2a kinases. First, we will examine eIF-2a phosphorylation in yeast cells expressing pk2 and
various eIF-2a kinases. Based on preliminary results we expect pk2 to lower eIF-2a phosphorylation levels.
Two models can be proposed for the suppressing activity of pk2: (1) pk2 binds to eIF-2 and prevents the elF-
2a kinases from gaining access to their substrate, or (2) pk2 heterodimerizes with the eIF-2a kinases and the
heterodimers are less active. According to the first model it would be expected that overexpressing eIF-2
or eIF-2a alone would reverse the effects of pk2. While overexpression of GCN2 or another eIF-2a kinase
would be expected to counteract pk2 according to the second model. High level expression of pk2 causes
a slow-growth phenotype in yeast cells independent of any eIF-2a kinases. The cause of this slow-growth
phenotype is unknown, however suppressors of the slow-growth may provide insights into proteins which
interact with pk2. To identify dosage-dependent suppressors the pk2 gene under the control of the GAL
promoter will be introduced into a yeast cell. This strain will subsequently be transformed with a high copy-
number library of yeast genes and fast growing transformants will be identified. Prior to this random screen
for dosage-dependent suppressors, various candidate genes (for example, eIF-2a and GCN2) on high copy
number vectors will be tested. While it is not known if the slow-growth phenotype caused by high-level
expression of pk2 is related to the ability of pk2 to suppress the eIF-2a kinases, the identification of proteins
that interact with pk2 is of interest since these proteins may also interact with the eIF-2a kinases. Another
means to identify proteins that interact with pk2 is to perform co-immunoprecipitation reactions and examine
if candidate proteins such as eIF-2 can be co-precipitated with pk2. To facilitate this analysis we have
engineered a c-myc epitope tag on the carboxyl-terminus of pk2.

In addition to examining authentic pk2, we will continue our examination of the related domain from
the eIF-2a kinases. As reported above, the corresponding region of the GCN2 kinase will act to suppress
at least two different GCN2' kinases. Similar regions of PKR and HRI have also been examined, however

ji^illlllWU

BmawnnmiiBiiwrmmrmmrinmm

ZOl HDO 101 0-01 -LMG

these proteins fail to suppress either PKR or GCN2. A caveat in these results is that the amino-temiinal
boundary is not exactly the same for the various constructs, therefore we will express several different PKR
and HRI constructs varying the amino-terminus of the protein to see if any of these truncated kinase domains
can inhibit the activity of a wild-type eIF-2a kinase (GCN2 or PKR).

4. Substrate specificity of the eIF-2a kinases. This project is close to completion. A quantitative
measure of the ability of the various combinations of kinases and substrates to activate GCN4 expression
will be obtained by measuring the expression from a GCN4-lacZ allele. In addition, to even more
convincingly show that the elF-2a-S51Y protein is phosphorylated on a tyrosine residue, extracts from
various cells will be treated with phosphotyrosine-specific phosphatases prior to analysis using the isoelectric
focusing gels. The phosphotyrosine phosphatase should specifically affect eIF-2a-S5 1 Y but not the wild-
type protein. Mutant alleles of the GCN2 or PKR kinase that more efficiently phosphorylate eIF-2a-S5 1 Y
could be identified to increase our understanding of the structural differences that distinguish the tyrosine-
specific from the serine/threonine class of protein kinases, however, these projects are not planned for the
immediate future.

5. Molecular analysis of yeast FUN12. The recent report of the DN A sequence of the entire yeast
chromosome I identified a novel gene tenned FUN 12 that shares amino acid sequence similarity to bacterial
IF-2 proteins. This protein is not the yeast mitochondrial IF-2 because this latter gene has already been
identified elsewhere in the genome. IF-2 in bacteria functions to deliver the Met-tRNA,'''" to the ribosome,
the same role performed by eIF-2 in eukaryotes. Since yeast already contains eIF-2, the function of FUN 12
is unknown. The translational regulation of GCN4 is exquisitely sensitive to translation initiation levels and
especially the levels of Met-tRNAj"^', so the regulation of GCN4 expression is an ideal system to study
FUN 12.

Preliminary results have revealed that overexpression of FUN 12 will suppress the slow-growth
phenotype of at least one of the GCN2'^ kinases. This early result suggests that FUN 12 functions in
cytoplasmic translation initiation in yeast. To further explore the role of FUN 12 in translation initiation, the
FUN 12 gene will be randomly mutated and the mutants will be screened to identify alleles that alter GCN4
translational regulation. We will screen for both alleles that block induction of GCN4 expression under
amino acid starvation conditions as well as alleles that lead to high unregulated levels of GCN4 even in non-
starvation conditions. The identification of either class of mutation will confirm that FUN12 functions in
translation initiation. During this screen we will also identify any temperature sensitive (ts) alleles of
FUN 12. The biochemical analysis of FUN 12 ts' mutants should help define the step of translation at which
FUN 12 acts. If the mutant screens fail to identify any interesting FUN 12 alleles, the suppression of the
GCN2'' kinases can be used in an analysis of the structural domains of the FUN 12 protein. In addition to
random mutants, site-directed mutations will be introduced into the FUN 12 GTP-binding motif Mutations
in this region of bacterial IF-2 have been identified that lead to a slow-growth phenotype, we will examine
the effects of the corresponding mutations in yeast FUN12. The identification of a protein in yeast showing
homology to bacterial IF-2 proteins is quite surprising, through molecular genetic analysis of this gene in
yeast we hope to identify the role for this protein in eukaryotic translation. These results take on added
significance with our recent identification of an expressed sequence tag (EST) from mouse that bears striking
homology to FUN 12, raising the possibility that this protein is conserved in humans as well.

^■BB

ZOl HDOlOlO-01-LMG

Significance to Biomedical Research and the Program of the Institute.

The eIF-2a kinases play a key role in the antiviral and stress responses of mammalian as well as yeast cells.
The ability of cells to alter gene expression under stress conditions is essential to the survival of the
organism. In addition, the genetic characterization of substrate recognition by the eIF-2a kinases should
provide insights into the mechanism of substrate recognition by other protein kinases. Finally, increasing
our understanding of viral inhibitors of the eIF-2a kinases will not only provide insights into the molecular
mechanisms regulating these kinases, but may also identify ways to defeat these viral defense mechanisms.

Publications:

Dever TE, Yang W, Astrom S, Bystrom AS, Hinnebusch AG. Modulation of tRNA>'^ eIF-2 and eIF-2B
expression shows that GCN4 translation is inversely coupled to the level of the eIF-2'GTP'Met-tRNAi'^''"
ternary complexes, Mol. Cell. Biol. (In Press).

^BBmmem^mKm

uMB-f^96

MBWq

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBUC HEALTH SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD01900-05 LME

PERIOD COVERED

October 1, 1994 to September 30, 1995

TITLE OF PROJECT (80 chamcters or less. Title must fit on one line between the borders.)

Developmental Regulation of Differential Gene Expression

PRINCIPAL INVESTIGATOR (List other professional personnel be/ow the Principal Investigator.) (NarDe, title, laboratory, and institute affiliation)

PI: Alan Wolffe, Visiting Scientist LME/NICHD

OTHERS: Stefan Dimitrov, Visiting Scientist LME/NICHD

James Godde, Adjunct Scientist LME/NICHD

Michael Harrier, Adjunct Scientist LME/NICHD

Cathy Jozwik, IRTA Fellow LME/NICHD

Stefan Kass, Adjunct Scientist LME/NICHD

Hitoshi Kurumizaka, Visiting Fellow LME/NICHD

Ninnlfitta 1 anrisherger. Visiting FRilnw LME/NICHD

Funds Meric, Adjunct Scientist LME/NICHD
Karl Nightingale, Visiting Fellow LME/NICHD

Ken Matsumoto, Visiting Fellow LME/NICHD
Danielle Patterton, Adjunct Technician LME/NICHD

Dmitry Pruss, Visiting Fellow LME/NICHD
John Strouboulis, Adjunct Scientist LME/NICHD
KiyoR iJra. Adjunct Sf^ientist LME/NICHD

COOPERATING UNITS (if any)

Laboratory of Molecular Embryology, NICHD (M. Dasso, Y.-B. Shi); Lab. of Mol.
Carcinogenesis, NCI (M. Bustin) ; Universite de Lausanne (W. Wahli) ; Lab., of Chem.
Physics. NIDDK fM. Clore. A.M. Gronenborn^

LAB/BRANCH

Laboratory of Molecular Embryology

SECTION

Section on Molecular Biology

INSTITUTE AND LOCATION

NICHD, NIH, Bethesda, MD 20892-2710

TOTAL STAFF YEARS:
15

PROFESSIONAL:
15

OTHER:
0

CHECK APPROPRIATE B0X(ES1

D (a) Human subjects D (b) Human tissues
D (a1) Minors
D (a2) Interviews

SUMMARY OF WORK (Use standard unreduced type. Do not exceed the space provided.)

Our work is focussed on the molecular mechanisms responsible for establishing and
maintaining stable states of gene expression during vertebrate embryogeneis.
Progress has been achieved in the following key areas:

1. We have determined that nucleosome mobility has a key role in facilitating
transcription of chromatin templates and that linker histones repress transcription
by restricting this mobility.

2. We have determined that chromatin structure has an essential role together with
Xenopus heat shock transcription factor in regulating Xenopus hsp70 gene
transcription in Xenopus oocytes.

3. We have determined that DNA methylation does not influence the association of
linker histones with DNA, and that incorporation of HMG14 and 17 into nucleosomea
during chromatin assembly does influence chromatin compaction.

4. We have found that nucleosome dilution in vitro can facilitate transcription
factor access to nucleosomal DNA in vitro.

5. We have examined the nucleic acid binding specificities of two proteins capable
of interacting with either DNA or RNA. TFIIIA which binds specifically to 53 rRNA
and 5S DNA, does not interact specifically with RNA-DNA heteroduplexes countaining
5S RNA and DNA sequences. FRGY2 is found to recognize specific RNA sequences vie.
the highly conserved cold shock domain.

PHS 6040 (Rev. 5/92)

sssssmwaaam

Project Number: ZOl HD 01900-05 LME

Objectives:

The object of this project is to determine the molecular processes responsible for establishing and
maintaining stable states of gene activity during vertebrate development.

Methods Employed:

High resolution gel electrophoresis is used to resolve DNA sequence at the nucleotide level, various
chemical and enzymatic probes determine the nature and pattern of transcription factor binding to DNA.
Transcription factors and RNA polymerase are fractionated and purified by standard chromatographic
procedures. High efficiency in vitro transcription assays are used to assess the function of transcription
factor binding to specific DNA sequences and the function of transcription factor interactions with other
transcription factors. These interactions are quantitated further by competitive footprinting and gel
retardation assays. Chromatin structure of active and inactive genes is assessed using enzymatic and
chemical probes, followed by gel electrophoresis and sucrose gradient sedimentation. Histones and
transcription factors associated with chromatin are monitored by electrophoresis and immunoblotting
using polyclonal antibodies. Sub-cellular localization is assessed using immofluorescence and
microscopy. Chromatin is assembled in vitro using cell-free preparations of Xenopus eggs and embryos.
Microinjection of oocytes and embryos is used to assess the functional role of DNA sequence and
protein-nucleic acid interactions in vivo. mRNA is isolated, cDNA synthesized and fusion proteins
expressed in bacteria. Mutant proteins are constructed by reverse genetic techniques and their functions
assayed in vitro as described above.

Major Findings:

GENE REGULATION DURING EARLY EMBRYOGENESIS

Constraints on transcriptional activator function contribute to transcriptional quiescence during early
Xenopus embryogenesis

We have examined the cause of transcriptional quiescence prior to the mid-blastula transition (MBT)
in Xenopus laevis. We have found distinct requirements for transcription of class II and class III genes.
An artificial increase of the amount of DNA present within the embryo over the amount of DNA present
within the embryo over that found at the MBT allows precocious transcription of tRNA genes, but not
of the adenovirus E4 or human cytomegalovirus (CMV) promoters. Thus titration of an inhibitor by
exogenous DNA determines class III but not class II gene activation. We have demonstrated that the
action of the inhibitor depends on the association of core histones with DNA. The addition of exogenous
TBP, together with an increase in the amount of DNA within the embryo, allows significant basal
transcription of class II genes prior to the MBT, whereas it does not increase transcription of tRNA
genes. To examine the activation of transcription above basal levels, we used a defined minimal
promoter containing five Gal4 binding sites and the activator Gal4-VP16. Precocious transcriptional
activation is directed by Gal4-VP16 prior to the MBT, demonstrating that a functional transcriptional
machinery exists at this early developmental stage. Furthermore, since this activation can occur in the
absence of exogenous TBP or chromatin titration, a transcription factor that can penetrate chromatin is

p»

ffwiimPMiinirnniiniiwiniiiirniiwiTi

Project Number: ZOl HD 01900-05 LME

sufficient for recruitment of this machinery to a promoter. Our results support the hypothesis that the
temporal regulation of transcription during early embrogenesis in Xenopus reflects not only a titration
of inhibitors by DNA, but also a deficiency in the activity of transcriptional activators prior to the MBT.

The heat shock response in Xenopus oocytes, embryos and somatic cells: an essential regulatory role for
chromatin

The heat shock response in Xenopus laevis has been reported to be developmentally regulated at the
transcriptional level. We find that the heat shock response of an exogenous Xenopus hsp70 gene
introduced into Xenopus oocytes, embryos, and somatic cells is dependent on the transcriptional assay
conditions employed. Under conditions of efficient chromatin assembly, transcription from the Xenopus
hsp70 gene promoter is repressed in oocytes and embryos, yet the promoter responds to heat shock by
activating transcription. Under conditions of inefficient chromatin assembly, the Xenopus hsp70 gene
is constitutively active in oocytes and somatic cells. Our results resolve previous controversy concerning
the existence of a heat shock response for the hsp70 promoter in oocytes and illustrate the importance
of considering chromatin assembly as a contributory factor in reconstructing the developmental control
of gene expression.

The role of chromatin and Xenopus heat shock transcription factor (XHSFl) in the regulation of
transcription from the Xenopus hsp70 promoter in vivo

Xenopus oocytes activate transcription from the Xenopus hsp70 promoter within a chromatin
template in response to heat shock. We have isolated and characterized a cDNA encoding Xenopus heat
shock transcription factor (XHSFl) causes the activation of the wild type hsp70 promoter within
chromatin. XHSFl activates transcription at normal growth temperatures (18°C), but heat shock (34°C)
facilitates transcriptional activation. Titration of chromatin in vivo leads to constitutive transcripaon
from the wild type hsp70 promoter. The Y-box elements within the hsp70 promoter facilitate
transcription in the presence or absence of chromatin. The presence of the Y-box elements prevents the
assembly of canonical nucleosomal arrays over the promoter and facilitates transcription. In a mutant
hsp70 promoter lacking Y-boxes, exogenous XHSFl activates transcription from a chromatin template
much more efficiently under heat shock conditions. Activation of transcription from the mutant promoter
by exogenous XHSFl correlates with the disappearance of a canonical nucleosomal array over the
promoter. Chromatin structure on a mutant hsp70 promoter lacking Y-boxes can restrict XHSFl access,
however on both mutant and wild type promoters chromatin assembly can also restrict the function of
the basal transcriptional machinery. We suggest that chromatin assembly has a physiological role in
establishing a transcriptionally repressed state on the Xenopus hsp70 promoter in vivo.

Nuclear localization and transcriptional activation by Xenopus heat shock transcription factor OCHSFl):
the definition of dominant negative mutants that inhibit the heat shock response in Xenopus oocytes

We have defined regions of the Xenopus heat shock transcription factor XHSFl that are essential
for nuclear accumulation of HSF and transcriptional activation of the Xenopus hsp70 promoter within
Xenopus oocyte nuclei. In addition we have made use of mutant XHSFl proteins to establish that
distinct molecular mechanisms exist capable of inhibiting heat inducible transcription from the Xenopus

UWUUHliUMl

Project Number: ZOl HD 01900-05 LME

hsp70 promoter in Xenopus oocytes. Deletion of the carboxyl (C)-terminal 38 amino acids of XHSFl
prevents nuclear accumulation and transcriptional activation, however expression of this truncated protein
does not impede the endogenous heat shock response. Deletion of a further segment of HSF including
the C-terminal hydrophobic domain, but retaining the amino (N)-terminal DNA-binding and trimerization
domains creates proteins that exert a dominant negative influence on the heat shock response. Inhibition
of heat inducible transcription by these truncated proteins depends on their capacity to retain wild type
HSF in the cytoplasm. In contrast, deletion of the N-terminal 23 amino acids disrupts DNA binding and
transcriptional activation but does not influence the capacity of HSF to localize to the oocyte nucleus.
This N-terminal deletion mutant also functions as a dominant negative inhibitor of the heat shock
response. Thus mutant HSF proteins can exert dominant negative effects through two independent
pathways that operate either in the nucleus or cytoplasm.

CHROMATIN STRUCTURE AND FUNCTION

A positive role for nucleosome mobility in the transcriptional activity of chromatin templates: restriction
by linker histones

We have established a model system for investigation of the selective repression of transcription from
nucleosomal templates by linker histones. We make use of physiologically spaced dinucleosomal
templates in which it is possible to assay both chromatin structural and transcriptional characteristics.
We find that nucleosome mobility is a characteristic of transcriptionally competent chormatin templates.
Reconstitution of chromatin with linker histones restricts nucleosome mobility and locks the nucleosome
into a unique position. This fixation of histone-DNA contacts is concomitant with transcriptional
repression. Thus, stable states of gene expression can be established at the nucleosomal level.

Disruption of reconstituted nucleosomes: the effect of particle concentration. MgCU and KCl
concentration, the histone tails and temperature

We have found that reconstituted nucleosome cores containing specific DNA sequences dissociate
on dilution. This disruption of histone-DNA contacts leading to the release of free DNA is facilitated
by the presence of the core histone tails, MgCl2 (5mM), KCl (60mM), and temperatures above 0°C.
Under reaction conditions that are commonly used to assess trans-acting factor access to nucleosomal
DNA, histone-DNA contacts are on the threshold of instability. We have found that dilution of
reconstituted nucleosomes containing a TATA box can facilitate TBP access to DNA.

Methylation at CpG sequences does not influence histone HI binding to a nucleosome including a
Xenopus borealis 5S rRNA gene

We have found that methylation of the 12 dinucleotide CpGs within a GC-rich DNA fragment
containing a Xenopus borealis 5S rRNA gene does not influence histone HI binding to naked or
nucleosomal 5S DNA. Thus a simple mechanism in which histone HI selectively associates with
nucleosomes containing methylated CpG cannot explain the repressive effects of methylation on gene
activity.

Project Number: ZOl HD 01900-05 LME

Incorporation of chromosomal proteins HMG-14/HMG-17 into nascent nucleosomes induces an extended
chromatin conformation and enhancers the utilization of active transcription complexes

In collaboration with M. Bustin (NCI) we have examined the role of chromosomal proteins HMG-14
and HMG-17 in the generation of transcriptionally active chromatin in a Xenopus laevis egg extract
which supports complementary DNA strand synthesis and chromatin assembly. Chromosomal proteins
HMG-14/HMG-17 enhanced transcription from a chromatin template carrying a 5S rRNA gene, but not
from a DNA template. The transcriptional potential of chromatin was enhanced only when these proteins
were incorporated into the template during, but not after, chromatin assembly. HMG-14 and HMG-17
stimulate transcription by increasing the activity, and not the number, of transcribed templates. They
unfold the chromatin template without affecting the nucleosomal repeat or decreasing the content of
histone B4. We suggest that HMG-14/HMG-17 enhance transcription by inducing an extended
conformation in the chromatin fiber, perhaps due to interactions with histone tails in nucleosomes. By
disrupting the higher order chromatin structure HMG-14/HMG-17 increase the accessibility of target
sequences to components of the transcriptional apparatus.

NUCLEIC ACID BINDING SPECIFICITIES OF THE MULTIFUNCTIONAL XENOPUS OOCYTE
PROTEINS TFIIIA AND FRGY2

The interaction of TFIIIA with specific RNA-DNA heteroduplexes

We have examined the association of transcription factor TFIIIA with RNA-DNA heteroduplexes
containing sequences from the Xenopus borealis 5S rRNA gene. Under conditions where TFIIIA
selectively binds to 5S rRNA or to the internal control region of the 5S rRNA gene, no specific
association of TFIIIA with RNA-DNA heteroduplexes containing either strand of 5S DNA could be
detected. Our results exclude specific models of TFIIIA recognition of the internal control region in an
A-type DNA configuration or of DNA-RNA hybrids during the process of transcribing the 5S rRNA
gene.

Sequence specific RNA recognition by the Xenopus Y-box proteins: an essential role for the cold shock
domain

The Xenopus Y-box protein FRGY2 has a role in the translational silencing of masked maternal
mRNA. We have determined that FRGY2 will recognize specific RNA sequences. The evolutionarily
conserved nucleic acid-binding cold shock domain is required for sequence-specific interactions with
RNA. However RNA-binding by FRGY2 is facilitated by amino and carboxyl terminal regions flanl:ing
the cold shock domain. The hydrophilic carboxyl-terminal tail domain of FRGY2 interacts with RNA
independent of the cold shock domain, but does not determine sequence-specificity. Thus, both
sequence-specific and non-specific RNA recognition domains are contained within the FRGY2 protein.

Publications:

Almouzni G, Wolffe AP. Constraints on transcriptional activator function contribute to transcriptional
quiescence during early Xenopus embryogenesis, EMBO J 1995;14:1752-1765.

mniniFiwimmi»iyii|ii>iiiiifii[iirnFimMMinTinniinTnFirt]

Project Number: ZOl HD 01900-05 LME

Bouvet P, Matsumoto K, Wolffe AP. Sequence specific RNA recognition by the Xenopus Y-box
proteins: an essential role for the cold shock domain, J Biol Chem 1995;(in press).

Dasso M, Dimitrov S, Wolffe AP. Nuclear assembly is independent of linker histones, Proc Natl Acad
Sci USA 1994;91:12477-12481.

Dimitrov S, Wolffe AP. Chromatin and nuclear assembly: experimental approaches towards the
reconstitution of transcriptionally active and silent states, Biochem Biophys Acta 1995;1260:1-13.

Godde JS, Wolffe AP. Disruption of reconstituted nucleosomes: the effect of particle concentration,
MgCl2 and KCl concentration, the histone tails and temperature, J Biol Chem 1995;(in press).

Hayes JJ, Wolffe AP. Chromatin and transcription. Nucleic Acids and Molecular Biology 1995;9:22-41.

Hayes JJ, Dimitrov S, Wolffe AP. Physical and chemical analysis of the dynamics of nucleosome and
chromatin structure, Chemtracts 1995; (in press).

Landsberger N, Wolffe AP. Chromatin and transcriptional activity in early Xenopus development,
Seminars in Cell Biology 1995;6:191-199.

Landsberger N, Wolffe AP. The role of chromatin and Xenopus heat shock transcription factor (XHSFl)
in the regulation of the Xenopus hsp70 promoter in vivo. Mol Cell Biol 1995; (in press).

Landsberger N, Ranjan M, Almouzni G, Stump D, Wolffe AP. The heat shock response in Xenopus
oocytes, embryos and somatic cells: a regulatory role for chromatin. Develop Biol 1995;160:62-72.

Landsman D, Wolffe AP. Common sequence and structural features in the heat shock factor and ets
families of DNA binding domains. Trends Biochem Sci 1995;20:225-226.

Nightingale K, Wolffe AP. The interaction of TFIIIA with specific RNA-DNA heteroduplexes, J Biol
Chem 1995; (in press).

Pruss D, Hayes JJ, Wolffe AP. Histone and DNA contributions to nucleosome structure. The Nucleus
I, The Nucleosome 1995;3-29.

Pruss D, Hayes JJ, Wolffe AP. Nucleosomal anatomy - where are the histones? BioEssays 1995; 17: 161-
170.

Strzelecka TE, Hayes JJ, Clore GM, Gronenbom AM. DNA binding specificity of the Mu Ner protein,
Biochemistry 1995;34:2946-2955.

Stump DG, Landsberger N, Wolffe AP. The cDNA encoding Xenopus laevis heat-shock factor 1
(XHSFl): nucleotide and deduced amino acid sequences, and properties of the encoded protein, Gene
1995;160:207-211.

fgfffjffffff^m,aimnm«>t,»>mmn i^„j^i,,^w^j^r,^nnrwmm,mmmmm„,mnn,wnnrf^

Project Number: ZOl HD 01900-05 LME

Trieschmann L, Alfonso PJ, Crippa MP, Wolffe AP, Bustin M. Incorporation of chromosomal proteins
HMG-14/-17 into nascent nucleosomes induces an extended chromatin conformation and enhances the
utilization of active transcription complexes, EMBO J 1995;14:1478-1489.

Ura K, Hayes JJ, Wolffe AP. A positive role for nucleosome mobility in the transcriptional activity of
chromatin templates: resitriction by linker histones, EMBO J 1995;14:3752-3765.

Ura K, Wolffe AP. Reconstruction of transcriptionally active and silent chromatin, Methods Enzymol
1995; (in press).

Wolffe AP. Chromatin structure and DNA replication: implications for transcriptional activity. DNA
replication in eukaryotic cells, Cold Spring Harbor Press, CSH 1995;(in press).

Wolffe AP. Chromatin Structure and Function. 2nd ed. London: Academic Press, 1995.

Wolffe AP. DNA deliberations. Trends Biochemical Sciences 1995;20:330-331.

Wolffe AP, Drew HR. DNA structure in chromatin. Frontiers in Molecular Biology, IRL Press, Oxford
1995;27-48.

Wolffe AP. Epigenetic inheritance: the chromatin connection. Paternal Imprinting: Causes and
Consequences, Cambridge University Press, Cambridge 1995;49-70.

Wolffe AP. First class transcription, BioEssays 1995;17:272-273.

Wolffe AP. Genes, simple; Transcription complex. Trends in Genetics 1994; 11:32,

Wolffe AP. Genetic effects of DNA packaging, Scientific American: Science and Medicine
November/December 1995;32-41.

Wolffe AP. Histone deviants, Curr Biol 1995;5:452-454.

Wolffe AP. Histones and transcriptional control, Eukaryotic Gene Transcription, Oxford University
Press, Oxford 1995;34-58.

Wolffe AP. The cold stress response in bacteria. Science Progress 1995;(in press).

Wolffe AP. The inheritance of chromatin states. Developmental Genetics 1994;15:463-470.

Wolffe AP, Meric F. Coupling transcription to translation: a novel site for the regulation of eukaryotic
gene expression. International Journal of Biochemistry and Cell Biology 1995; (in press).

IIIIMIHIIIllllllllllliHIIIimi

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBUC HEALTH SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD01901-01 LME

PERIOD COVERED

October l3t, 1994 to September 30, 1995

TITLE OF PROJECT 180 cftaracters or /ess. Title must fit on one line between the borders.)

Gene Regulation by Thyroid Hormone during Tissue Remodeling

PRINCIPAL INVESTIGATOR {List other professional personnel below the Principal Investigator.) (Name, title, laboratory, and institute affiliation)

PI: Yun-Bo Shi, Senior Staff Fellow LME/NICHD

OTHERS :

Monika Puzianowska-Kuznicki, Visiting Fellow LME/NICHD

Melissa Stolow, Adjunct Scientist LME/NICHD

Jiemin Wong, Adjunct Scientist LME/NICHD

Yuan Su, Adjunct Scientist LME/NICHD

COOPERATING UNITS lif any)

Dept. of Anatomy, Dokkyo University, School of Medicine, Japan (A.
Section of Mol. Biol., LME, NICHD (A. Wolffe)

Ishizuya-Oka) ;

LAB/BRANCH

Laboratory of Molecular Embryology

SECTION

Unit on Molecular Morphogenesis

INSTITUTE AND LOCATION

NICHD, NIH, Bethesda, MD 20892-5455

TOTAL STAFF YEARS:
4.8

PROFESSIONAL:
4.8

OTHER;
0

CHECK APPROPRIATE BOXIES]

D (a) Human subjects
D (a1) Minors
D (a2) Interviews

D (b) Human tissues M (c) Neither

SUMMARY OF WORK (Use standard unreduced type. Do not exceed the space provided.)

The research in the Unit of Molecular Morphogenesis is focussed on the
understanding of the molecular mechanism of amphibian metamorphosis. The control
of this developmental process by thyroid hormone (TH) offers a unique paradigm in
which to study genes that are important for post-embryonic organ development. We
began to study metamorphosis by choosing the remodeling of the tadpole intestine in
Xenopus laevis as a model system. The tadpole intestine is a simple tubular
structure consisting of primarily a single layer of primary epithelial cells.
During metamorphosis, it is transformed into a multiply folded adult epithelium
with elaborate connective tussue and muscles through specific cell death and
selective all proliferation and differentiation. We have isolated and identified
many TH-response genes in the intestine during this transition. Among them are the
TH receptor (TR) P genes. Analysis of the receptor gene expression in different
tissues shows that not only TRP but also TRa and RXR genes are regulated in an
organ-dependent manner during metamorphosis. RXRs are known to be able to form
heterodimers with TRs. Indeed, Xenopus TR/RXR heterodimers can bind to their
binding site even in a chromatinized template and activate or repress transcription
in the presence or absence of TH, respectively. Thus, our results strongly suggest
TR/RXR heterodimer are the functional complexes mediating the effect of TH during
metamorphosis. In addition, we have also obtained evidence that two NFI genes,
which are also regulated by TH, are involved in the adult organ development.
Finally, a putative morphogen, hedgehog, and the matrix metalloproteinase,
stromelysin-3, appear to participate in the regulation of cell-cell and cell-ECM
interactions during intestinal remodeling.

PHS 6040 (Rev. 5/921

igTiinniii¥innniiiiiHMinniiiinwimBiHwm«i9WMMHB»»itfii>iii^

Project Number: Z01 HD 01901-01 LME

Project Description:

Objectives:

To understand the molecular mechanism of gene regulation by thyroid hormone and the roles of
thyroid hormone response genes, especially those influence cell-cell and cell-ECM interactions, during
frog organ development.

Methods Employed:

In vitro biochemical studies of protein function and nucleic acid manipulation; functional
investigation of transcription factors in oocytes, in situ hybridization and immunohisto-chemical
analyses of gene expression.

Major findings:

Regulation and function of TH receptors (in collaboration with Dr. Alan Wolffe)

TH is known to regulate gene expression by controlling the transcription of target genes directly
through thyroid hormone receptors (TRs). Work from other laboratories on TRs in birds and mammals
has shown that TRs belong to the superfamily of the steroid hormone receptors. These receptors are
sequence specific DNA binding proteins that regulate the transcription of genes containing their
binding sites (TREs or thyroid hormone response elements for TRs) in a ligand dependent manner.
While TRs can bind to TREs as monomers and homodimers, they most likely function as heterodimers
formed with RXRs (9-cis retinoic acid receptor).

Toward the first step to investigate the roles of TR/RXR during metamorphosis, we have determine
the mRNA levels of TR (TRa and TR;?) and RXR (RXRa and RXRk) genes during metamorphosis in
different organs, the limb, tail, and intestine, which undergo very contrasting changes, i.e. total
resorption, de novo development, and remodeling, respectively. In general, TR and RXR genes are
coordinately regulated in all organs. Thus, high levels of their mRNA are present during early stages
of limb development when morphogenesis occurs and in the tail toward the end of metamorphosis
when it is being resorbed. In the intestine, moderate levels are present throughout the remodeling
period. Furthermore, by introducing TRs and/or RXRs through microinjection of their mRNAs into
Xenopus oocytes, we have established an in vivo TH-response system. By using as a reporter the
promoter of Xenopus TRyffA genes, which we have shown previously to be regulated by TH, we have
demonstrated that TRs or RXRs alone have little or no effect on the activity of the promoter. On the
other hand, coinjection of TR and RXR mRNAs lead to transcriptional repression of the basal activity
in the absence of TH and strong activation of the promoter when TH is present. In vitro DNA binding
experiments indicate that only TR/RXR heterodimers bind strongly to the TRE in the TR^ promoter.
Thus, while we have not analyzed the levels of TR and RXR proteins during development, our anal' sis
of mRNA levels and functional studies of TRs and RXRs provide strong evidence that TRs function
as heterodimers with RXRs during metamorphosis.

To investigate the mechanism of transcription regulation by TR/RXR heterodimers, we again made
use of the oocyte system. In collaboration with Dr. A. Wolffe, we have reconstituted in vivo the TH-
regulated TRy?A promoter in chromatin and studied the effect TR/RXR on the promoter activity. Dr
Wolffe's group has shown previously that when a double stranded (ds) promoter plasmid is injected

iiiiniiiiiiiiiiiiiMmiiiiiiiiiiiiiiMmwMniiimnii""™'"!"'^ ■iiiitiiniiiMiUBimMKiiiiiniKiiiM'imiminiinMiimniiiiiiiiiiiiiiiiin

Project Number: Z01 HD 01901-01 LME

into oocytes, it are chromatinized such that the promoter is highly active. In contrast, when a single
stranded (ss) promoter plasmid is injected into oocytes, it is rapidly replicated and undergoes
replication-coupled chromatin assembly to produce a template of a very low basal promoter activity.
We have found that heterodimers of TR and RXR within the chromatin templates, generated by
injection of either ds or ss DNA containing the TRyffA promoter can alteratively repress or active
transcription dependent upon the absence or presence of TH. Efficient transcriptional repression
requires the presence of unliganded TR/RXR heterodimers during replication-coupled chromatin
assembly. In vivo DNasel footprinting directly demonstrate that TR/RXR heterodimers can bind to the
TRE in chromatinized templates independently of TH. Such binding also occurs in vitro on a
reconstitate nucleosome containing the TRE. While TR/RXR binding alone does not affect the
nucleosome in vitro or chromatin structure in vivo, the addition of TH, which results in active
transcription in vivo, leads to strong chromatin disruption as revealed by micrococcal nuclease
digestion of the chromatin templates. Interestingly, this disruption is dependent on TH but can occur
even when transcription is blocked by a-amanitin. While the nature of this disruption is unclear, these
results indicate that the additon of TH leads to a conformational change in the TR/RXR heterodimer
to allow either the recruitment of other factors and/or conformational changes in transcriptional
machinery, which in turn causes chromatin structural alterations and activation of the promoter.

Regulation of NFI genes during organ development

We have identified two of the TH-response genes in the intestine as members of the nuclear factor
I (NFI) family. Mammalian NFIs are transcription factors that are believed to also participate in DNA
replication. The two Xenopus NFI genes are highly homologous to mammalian and avian NFI genes.
Their deduced amino acid sequences are over 84-95% identical to their respective counterparts in
birds and mammals. In contrast, the two Xenopus NFIs are much less homologous to each other,
sharing only 58% homology which largely resides in the DNA binding domain at the amino terminus.
However, both NFIs can bind to a consensus NFI binding site and activate the transcription of a
promoter bearing the binding site. Northern blot analysis reveals that both NFI genes are regulated
in a tissue- and developmental stage-dependent manner. They are first activated, independently of
thyroid hormone, to low levels at stages 23/24, around the onset of larval organogenesis. After stage
54, their mRNA levels are dramatically up-regulated by endogenous TH. High levels of their mRNAs
are present during tail resorption, limb morphogenesis, and intestinal remodeling, even though each
of which occurs at very developmental different stages. Furthermore, gel mobility shift assay shows
that proteins recognizing the NFI binding site are present in different organs and their levels are
regulated similarly to NFI mRNA levels. These results thus strongly suggest that NFIs play important
roles during post-embryonic organ development, in contrast to the general belief that NFIs are
ubiquitous transcription factors.

Cell-cell and cell-extracellular matrix interactions during tissue remodeling (in collaboration with Dr.
Atsuko Ishizuya-Oka)

Each metamorphosing organ consists of many different cell types. Some of them are in direct
contact with each other. Others are separated by extracellular matrices (ECM). Proper interactions
among these cells are likely to be important for tissue remodeling. Thus, it is not supprising that
among the TH-response genes are those encoding a matrix metalloproteinase and a putative
morphogen, hedgehog.

Project Number: Z01 HD 01901-01 LME

The hedgehog gene was first cloned as a segment polarity gene in Drosophila. Its homologs in
mammals and birds are capable of functioning as morphoges in processes such as neural induction
and limb morphogenesis. We have found that the Xenopus hedgehog gene is activated by neurula
stage during embryogenesis. Whole mount in situ hybridization analysis showed that at these early
stages, the hedgehog gene is expressed in the notochord and floor plate similar to that observed in
other animal species. During metamorphosis, hedgehog mRNA was found to be high in the intestine
during the period when adult epithelial morphogenesis takes place. Thus, Xenopus hedgehog seems
to play a role in organ development during both embryogenesis and metamorphosis.

Cell-cell interactions can also be regulated through the modification ECM. We have previously
reported the characterization of Xenopus stromelysin-3 (ST-3) gene, which encode a putative matrix
metalloproteinase (MMP). ST-3 is highly expressed in both the tail and intestine when extensive cell
death takes place. In the limb, it is expressed at low levels during limb morphogenesis when cell
death occurs in the interdigital region. However ST-3 is activated prior to any noticeable cell death.
In contrast to several other MMP genes. Its mRNA is present only in the fibroblastic cells but not in
the epithelial cells that will undergo apoptosis. Thus, if ST-3 plays a role in epithelial degenration, it
is likely through alteration of cell-cell and/or cell-ECM interactions. In support of this, in collaboration
with Dr. Ishizuya-Oka, we have found by in situ hybridization that ST-3 is highly expressed in the
fibroblastic cells surrounding the epithelium at the time when basal lamina, which separates the
epithelium and the connective tissue, becomes thick. Although the basal lamina is thick at this time,
it allows extensive cell-cell contact and cell-migration through it, in contrast to the thin but apparently
less permaeble pre- or post-metamorphic basal lamina. While the role of ST-3 in the ECM
modifciations remains to be established, such ECM changes are likely to be important of larval
epitheial cell death and adult epitheial proliferation and differentiation.

Proposed course of projects:

Gene regulation by thyroid hormone receptors

Our studies so far have shown that TR/RXR heterodimer can regulate transcription from a
chromatinized template. The activation in the presence of TH leads to chromatin disruption. We plan
to identify regions of TR and RXR that are important for chromatin disruption and transcription
regulation. We would also like to determine the mechanism of this disruption. Do TR and RXR
interact with any factors to bring about chromatin structural changes? What are these factors, if they
do? Are these also the same factors that are involved in promoter activation?

In our analysis of the receptor function in vivo, we found that TRoB, encoded by one of the two
TRo genes in Xenopus, failed to bind to a TRE or regulate transcription. Sequence comparison
showed several amino acid differences between TRoA and TRaB. We would like to determine if any
of these changes are responsible for the inactivity of TRoB and whether it is due to the failure to bind
a TRE or inability of the receptor to interact with transcriptional machinery.

Finally, we are also interested to study TR/RXR function in developing embryos. We will approach
this by introducing TR/RXR into fertilized eggs together with a reporter promoter. Alternatively, we
can analyzed the expression of endogenous genes in these embryos with over-expressed TR/RXR or
their mutants.

1..

injMiMi»wiiiiniinaiv»nwiiirniin»i

Project Number: Z01 HD 01901-01 LME
Cell-cell and cell-ECM interactions

We plan to investigate further how hedgehog and stromelysin-3 participate in intestinal remodeling.
We intend to obtain antibodies against both proteins and analyzed the temporal and spatial distribution
of the proteins by Western blot and immunohistochemistry. To study the biological functions we
intent to establish primary intestinal epithelial cell and/or organ cultures under conditions such that
they will respond to TH just like in intact tadpoles. At same time, we will try to generate functional
hedgehog and stromelysin-3 proteins by stable cell transfection. Once these are achieved, we can
directly investigate the function of hedgehog and stromelysin-3 by adding over-produced proteins to
organ or cell cultures or using antibodies to block the function of these proteins and examining the
response of the cells or organs to TH.

ECM is a complex structure made of many components. It is very likely that MMPs other than
stromelysin-3 are involved. We plan to investigate this by analyzing the mRNA or protein levels of
several other MMPs by Northern and Western blot analysis. Their regulation and function will be
studied in comparison to these of stromelysin-3.

Significance to biomedical research and the program of the Institute

While metamorphosis is unique to lower organisms, the basic developmental process is similar to
that in mammals. This is especially true for post-embryonic organ development. The regulation of
the process by TH offers an unique opportunity to identify the genes which are important for organ
development. The similarity in function of organs and in their developmental process between frogs
and mammals suggest that the same genes which are involved in metamorphosis will likely be
important for mammalian organogenesis. In addition, gene regulation by hormones is present in all
vertebrate organisms. TH regulation is an extreme example of hormonal regulation during
development. Understanding the mechanism of how TH controls the transcriptional state of target
genes, especially in a chromatin context, will be of general interest research to fields on hormones,
transcriptional regulation and the effects of chormatin on gene expression.

Publications:

Ishizuya-Oka A, Shimozawa A, Takeda H, Shi Y-B. Cell-specific and spatio-temporal expression of
intestinal fatty acid-binding protein gene in the small intestine during amphibian metamorphosis.
Roux's Archives of Dev Biol 1994;204:150-155.

Patterton D, Hayes WP, Shi Y-B. Transcriptional activation of the matrix metalloproteinase gene
stromelysin-3 coincides with thyroid hormone-induced cell death during frog metamorphosis. Dev Biol
1995;167:252-262.

Shi Y-B, Liang VC-T, Parkison C, Cheng S-Y. Tissue-dependent developmental expression of a
cytosolic thyroid hormone binding protein gene in Xenopus: Its role in the regulation of amphibian
metamorphosis. FEBS Letters 1994;355:61-64.

Shi Y-B. Thyroid hormone-regulated early and late genes during amphibian metamorphosis. In: Gilbert
LI, Atkinson BG, Tata JR, eds. Metamorphosis: Post-embryonic reprogramming of gene expression
in amphibian and insect cells. New York: Academic Press, (in press).

iiHm?iTnBiiiiiinniwwimimw^OTiwgifflffiff™»««^^ ■ winiiiiiiiyTimmFinwwnnniniinmmiri]

Project Number: Z01 HD 01901-01 LME

Shi Y-B, Ishizuya-Oka A. Biphasic intestinal development in amphibians: Embryogenesis and
remodeling during metamorphosis. Current Topics in Developmental Biology 1995;(in press).

Shi Y-B, Stolow MA, Puzianowska-Kuznicka M, Wong J. The new frontier of amphibian
metamorphosis. La Recherche 1995;(in press).

Stolow MA, Shi Y-B. Xenopus sonic hedgehog as a potential morphogen during embryogenesis and
thyroid hormone-dependent metamorphosis. NucI Acids Res 1995;(in press).

Wong J, Shi Y-B Coordinated regulation of and transcriptional activation by Xenopus thyroid hormone
and retinoid X receptors. J Biol Chem 1995;(in press).

iiii Willi iMMiMihiMW M > iiii imnnnnnnniwi

DEPARTMENT OF HEALTH AND HUMAN SERVICES - PUBUC HEALTH SERVICE

NOTICE OF INTRAMURAL RESEARCH PROJECT

PROJECT NUMBER

ZOl HD01902-01 LME

PERIOD COVERED

October 1, 1994 to September 30, 1995

TITLE OF PROJECT (80 characters or less. Title must fit on one line between the borders!

Analysis of the S phase checkpoint in higher eukaryotes

PRINCIPAL INVESTIGATOR (List other professional personnel below the Principal Investigator.) (Name, title, laboratory, and institute affiliation)

PI: Mary Dasso, Senior Staff Fellow LME/NICHD

OTHERS :

Hisato Saitoh, Visiting Associate LME/NICHD
Kathy Steinmann, Adjunct Scientist LME/NICHD
Robert Pu, IRTA Fellow LME/NICHD

COOPERATING UNITS (if any)

Dept. of Cell Biology, Johns Hopkins Medical School, Baltimore, MD

(W. Earnshaw, Ph.D.); Dept. of Molecular Biology, Holland Laboratories of the

American Red Cross. Rockville. MD (W.H. Burgess. Ph.D.W Dept. of Mol. Biol.. Grad,

LAB/BRANCH

Laboratory of Molecular Embryology

SECTION

Unit on Cell Cycle Regulation

INSTITUTE AND LOCATION

NICHD, NIH, Bethesda, MD 20892-5430

TOTAL STAFF YEARS:
3.5

PROFESSIONAL:
3.5

OTHER:
0

CHECK APPROPRIATE BOXIESI

D (a) Human subjects D (b) Human tissues
D (a1) Minors
D (a2) Interviews

SUMMARY OF WORK (Use standard unreduced type. Do not exceed the space provided. I

The eukaryotic nucleus is a highly ordered structure that carries out an assortment
of complex functions. In order for the nuclear tasks to be carried out properly,
they must be temporally and spatially coordinated amongst themselves and with
respect to the other functions of the cell. We are investigating how cells send
and interpret signals that coordinate these activities with each other and with the
cell cycle. Among the proteins that are thought to be important for maintaining
nuclear integrity, two proteins that appear to be essential for the spatial and
temporal order of the nucleus are Ran and RCCl. These two proteins interact
enzymatically with each other, and they are required for almost every nuclear
process including RNA transcription and processing, nuclear transport, DNA
replication, and cell cycle control. Ran is a small, very abundant Ras-like GTPase
that is mainly nuclear, while RCCl is a chromatin-bound protein that acts as a
guanine nucleotide exchange factor (GEF) for Ran. Our goal is to understand RCCi
and Ran at a molecular level and to discover how they interact with other cellular
components that regulate cell cycle progression and interphase nuclear activities.
We have studied RCCl and Ran in biochemical assays using purified components and in
Xenopus laevis egg extracts. We have also begun to purify other proteins that
interact with RCCl and Ran from Xenopus egg extracts. We have thereby discovered a
number of novel proteins that may be important for the activity of this GTPase
pathway. This work is complemented by an analysis of chemical agents that bypass S
phase cell cycle control in mammalian cells and by a search for the molecular
target(s) of these agents. In particular, we are examining the regulation of mRtlA
stability by checkpoint regulatory systems. We have found a number of candidate
mRNAs whose stability appears to be controlled by the replication state of the
nuclear DNA and we are currently in the process of characterizing these messages
further.

PHS 6040 (Rev. 5/921

Project Number: ZOl HD 01902-01 LMfi

Objectives:

The objective of this unit is to discover the mechanism by which cells transmit and interpret signals
that coordinate nuclear activities with each other and with the cell cycle. In particular, we nave
focussed on understanding the control of cell division with respect to the completion of DNA replication,
since mitosis is inhibited by signals from the nucleus until DNA synthesis is complete. We are pursuing
two molecular approaches to this problem: First, we are investigating the functions of RCCl and Ivan,
two nuclear proteins that are essential for coordination of nuclear activities to the cell cycle. Second,
we are characterizing cellular mRNAs whose stability is regulated by this checkpoint mechanism, and
whose function is essential for the initiation of mitosis.

Methods Employed:

In vitro biochemical studies of protein function and nucleic acid manipulation; functional
investigations of Ran GTPase pathway components in Xenopus laevis egg extracts; Molecular biological
analysis of mRNAs in mammalian tissue culture cell lines.

Major Findings:

The RCCl protein in interphase nuclear function (Collaboration with T. Nishimoto)

The Ran protein is a small GTPase that has been implicated in a large number of nuclear processes,
including transport, RNA processing and cell cycle checkpoint control. A similar spectrum of nuclear
activities has been shown to require RCCl, the guanine nucleotide exchange factor (GEF) for Ran. We
have used the Xenopus laevis egg extract system and in vitro assays of purified proteins to examine how
Ran or RCCl could be involved in these numerous processes. We employed mutant Ran proteins to
perturb nuclear assembly and function. The addition of a bacterially-expressed mutant form of Ran
(T24N-Ran), which was predicted to be primarily in the GDP-bound state, profoundly disrupted nuclear
assembly and DNA replication in extracts. We further examined the molecular mechanism by which
T24N-Ran disrupts normal nuclear activity. We found that T24N-Ran binds tightly to the RCCl protein
within the extract, resulting in its inactivation as a GEF. The capacity of T24N-Ran blocked interphase
extracts to assemble nuclei from demembranated sperm chromatin and to replicate their DNA could be
restored by supplementing the extract with excess RCCl and thereby providing excess GEF activity.
Conversely, nuclear assembly and DNA replication were both rescued in extracts lacking RCCl by the
addition of high levels of wild-type GTP-bound Ran protein, demonstrating that RCCl is not essential
for any function beyond its role as a GEF. Taken together, these results indicate that GTP-Ran is
essential for nuclear assembly and DNA replication in Xenopus extracts, and that these processes are
blocked when nucleotide exchange is either inhibited by T24N-Ran mutant protein or absent because of
RCCl -depletion.

RCCl- and Ran-associated protein in Xenopus extracts

It is currently unknown how the activity of RCCl is regulated or whether interactions with other
nuclear components facilitate its proper localization or control. It was originally thought that RCCl

HiwwiwwniiimiMiiiwiiiiiiiiiii

Project Number: ZOl HD 01902-01 LME

binds to DNA directly, since it both binds to chromatin in vivo and to DNA cellulose in vitro.
However, more recent evidence indicates that RCCl associates to chromatin through interactions with
other proteins. We have therefore been interested in discovering and characterizing proteins that interact
with RCCl and that may serve to regulate its distribution or activity. We sought to identify proteins
that interact with RCCl via their association with a glutathione-S-transferase-RCCl fusion protein
(GST-RCCl) in Xenopus laevis egg extracts. We found that at least four proteins bound specifically to
GST-RCCl in extracts. Consistent with the behavior of endogenous RCCl -containing complexes,
magnesium and guanine nucleotides released these proteins from their association to GST-RCCl protein.
Three of these proteins have previously been identified: Ran, RanBPl, hscVO. The fourth protein has
an apparent molecular weight of 340 kDa, and had not been previously described in association with
RCCl.

We have pursued a similar strategy to examine Ran-interacting proteins (RIPs), looking especially
for polypeptides that interact with both RCCl and Ran. To do this, we incubated a glutathione S
transferase-Ran fusion protein in Xenopus egg cytosol, as described for GST-RCCl above. We found
that a number of proteins were associated with the GST-Ran protein, some of which common to both
the GST-Ran and GST-RCCl associated fractions. The common bands had apparent molecular weights
of 340 kDa, 72 kDa and 30 kDa. The 72 kDa and 30 kDa bands in the GST-RCCl fractions had
previously been identified as the heat shock cognate protein hsc70, and RanBPl, respectively. There
were also bands that did not correspond to proteins associated with RCC 1 . These bands had apparent
molecular weights of 180 kDa, 120 kDa, 92 kDa, 88 kDa, 18 kDa and 10 kDa. We are currently in
the process of identifying these peptides and beginning a more complete biochemical analysis of their
interactions with Ran.

We have gone on to perform a more extensive characterization of the p340 protein. We raised
polyclonal antibodies against p340 and used immunogold electron microscopy (in collaboration with W.
Eamshaw and C. Cooke) to localize this protein within Xenopus tissue culture cells. We found that this
protein is associated with the cytosolic face of the nuclear pore, as well as with unidentified foci within
the cytosol. Our results suggest that it is likely that p340 is a component of a series of fibers that
electron microscopy has shown to emanate from the pore into the cytosol. These fibers may serve as
docking sites for proteins that will be transported into the nucleus. Thus, it will be important to
understand p340's activity in order to understand the mechanism of nuclear transport and how transport
is regulated. Toward this end, we purified p340 and subjected it to peptide sequencing (in collaboration
with W. Burgess). The peptides revealed that this protein has homology to nuclear pore proteins.
Comparison of our sequences with the cDNA of a hamster protein (RanBP2) that was discovered by T.
Nishimoto and colleagues suggests that these proteins are homologous. RanBP2 was found in a
two-hybrid screen for proteins that interact directly with Ran, and antibodies directed against RanBP2
are capable of blocking nuclear import. The p340/RanBP2 cDNA contains sequence motifs that may
provide clues regarding its function: The cDNA sequence contains RanBPl -like sequences (Ran-binuing
domains), a zinc finger domain and a cyclophilin-like region. Since p340 is a nuclear pore protein that
is required for nuclear transport and that interacts with Ran and other nuclear pore components, an
understanding of its activity would be a significant step toward understanding the nuclear pore as a
whole.

Project Number: ZOl HD 01902-01 LME

Cloning and characterization of the Xenopus RNAl and RanBPl homologues

Like other small GTPases of the ras superfamily, Ran has a low intrinsic GTPase activity that can
be activated by interacting with a GTPase activating protein, RanGAPl. RanGAPl was isolated as a
Ran GTPase activator that can increase the hydrolysis of GTP by three orders of magnitude. RanBPl
was purified as a Ran-binding protein and was shown to be able to increase RanGAPl induced GTP
hydrolysis of Ran-GTP. We have begun to examine the cellular roles of RanGAPl and RanBPl using
the Xenopus egg extract system and a variety of molecular biological techniques. Our goal is to define
how these proteins function biochemically, whether and how they are regulated, and how their activity
impacts on the control of nuclear functions and cell cycle control. As a first step toward characterizing
RanGAPl and RanBPl 's functions, we have cloned the Xenopus RanGAPl and RanBPl. This was
done using degenerate primers corresponding to the amino acid sequences that are conserved among the
known RanGAPl and RanBPl homologues. These primers were used to perform polymerase chain
reaction (PCR) using Xenopus cDNA library as the template to amplify DNA a fragment of putative
RanGAPl or RanBPl sequence. After confirming the amplified DNA fragments encode RanGAPl or
RanBPl by sequencing, these fragments were used as the probes to screen the same Xenopus cDNA
library to identify clones. The cDNAs of Xenopus RanBPl and RanGAPl show that these proteins are
well conserved between Xenopus and other species.

These cDNAs have been cloned in a bacterial expression vectors to allow production and purification
of recombinant RanGAPl and RanBPl. An epitope-tagged version of RanGAPl and RanBPl have also
been produced, so that soluble proteins can be purified in a one-step column chromatography procedure.
Such tagged proteins will allow the purification of RanGAPl and RanBPl interacting factors, in a
manner similar to the purification of RCCl- and Ran-interacting proteins discussed above. The purified
recombinant RanGAPl and RanBPl proteins that we have produced are currently being used to raise
polyclonal antibodies in rabbits. We are making a series of mutants of the RanBPl protein that will
alter sequences that are suspected to be important for its function. We are now expressing these proteins
for analysis in biochemical assays and in the egg extract. Another series of mutants is currently being
made to allow the expression of mutant RanGAPl proteins. These proteins and antibodies will be used
in the functional studies described below.

The regulation of mRNA stability in the cell cycle

Despite the importance of the DNA replication checkpoint for maintaining genetic stability in
eukaryotic cells, many of the intermediary components in the checkpoint pathway coupling the onset of
mitosis to the completion of DNA replication remain to be identified. One mechanism by which mitotic
onset is controlled is through changes in mRNA stability. At least some of the mRNA required for
induction of mitosis that accumulates during S phase arrest in baby hamster kidney (BHK) cell", is
unstable as long as DNA replication is ongoing. This mRNA is stabilized by drugs that inhibit DNA
replication (e.g., hydroxyurea), but becomes unstable again once the drugs are removed and replication
resumes. These properties will allow the identification of these mRNAs by the technique of differential
display. Using this strategy is important because it has the potential to identify new proteins th:U are
required for mitosis and that are regulated by the DNA replication checkpoint.

IBBB^

aamaaBim^asBSBSiises

Project Number: ZOl HD 01902-01 LME

The immediate goals of this work are to: a) Use differential display to identify mRNA sequences
whose levels decrease when DNA replication resumes in S phase arrested BHK cells, b) Use Northern
blot analysis first to confirm that mRNAs identified by differentially displayed sequences are present at
higher levels in S phase arrested BHK cells than in cells undergoing DNA replication and then to identify
the mRNAs whose differing levels in these two cell populations are the result of differences in mRNA
stability. c) Sequence the differential display products to identify conserved sequences as well as
sequences found in databases. We have been successful in identifying populations of candidate mRNAs
that are regulated by ongoing DNA replication in BHK cells through differential display. We are
presently in the process of confirming these data by Northern blot analysis, and we will then characterize
the mRNAs by sequencing their cDNAs. Within the coming year, we hope to find mRNA sequences
that serve as targets for destabilization by signals indicative of incomplete nuclear DNA replication.

Proposed course of projects:

Characterization of Ran-interacting proteins (RIPs) in Xenopus

We will continue the examination of Ran-interacting proteins (RIPs) by both characterizing
unidentified proteins that interact with Ran in Xenopus extracts and by a more in-depth examination of
the p340 protein. Unidentified proteins that associate with GST-Ran in Xenopus extracts include proteins
of apparent molecular weights 180 kDa, 120 kDa, 92 kDa, 88 kDa, 18 kDa and 10 kDa. We have
already examined whether these proteins associate with Ran in a nucleotide-dependent manner and
whether they can bind to 32P-GTP bound Ran in an overlay blot assay. During the coming year, we
intend to purify the proteins in sufficient quantity to obtain protein sequence data and to attempt to clone
the proteins. We then intend to pursue an analysis of each protein that is analogous to the analysis
underway for RanBPl and RanGAPl. We believe that characterizing the proteins that interact with Ran
is the best way to achieve a complete molecular picture of the Ran GTPase pathway.

We wish to examine the cellular role of p340/RanBP2 in three stages: First, in order to examine
how p340 interacts with other extract proteins, including Ran, RCCl and the other RIPs, we will express
domains of the mammalian RanBP2 protein in E. coli as fusion proteins and purify these domains by
affinity chromatography using the fusion peptide. It will be possible to map the interactions between
p340 and the other proteins by examining the interactions between the fusion proteins and the
p340-associated extract proteins. Using the isolated domains, it will also be possible to test for
biochemical activities: For instance, it will be of interest to determine whether the cyclophilin homology
domain of p340 has activity as a cis-trans-peptidylprolyl isomerase (PPIase). If it does, we would then
seek to determine whether p340's activity can be blocked by cyclosporin A. In this case, a number of
experiments will be performed to relate the pharmacological effects of this inhibitor to the activity of
p340 in nuclear assembly and transport. Similarly, it will be of interest to determine whether the
zinc-finger region of p340 can bind nucleic acids and whether this binding promotes RNA export.

The second stage of our studies on p340 is to undertake a direct analysis of p340's function using
Xenopus extracts. This will be accomplished by immunodepletion of the endogenous p340 and
examination of how nuclear assembly and transport are affected. We will further examine whether the
fusion proteins containing isolated domains of p340 are able to restore activity for nuclear formation.

lii

IIIIIMllllllllftlllMWIIimillHllllliWJI

Project Number: ZOl HD 01902-01 LME

nuclear transport and DNA replication in p340-depleted extracts. We will also make use of the fusion
proteins to determine whether they can act as dominant negative inhibitors of any of these processes. To
do this, we will add the fusion proteins in excess prior to the beginning of the nuclear assembly reaction
and then determine the effect on nuclear morphology, nuclear transport and DNA replication. Mutat:on:>
that block biochemical activities of the individual domains will be used to extend this analysis.

Third, we will perform an ultrastructural analysis of the p340 protein. There is currently good
evidence that at least part of p340 resides on the outer face of the nuclear pore. However, some
antibodies against p340 also recognize the inner face of the pore. It is therefore of interest to determine
how p340 is oriented within the pore. To do this, we will localize tagged fusion proteins by
immunogold electron microscopy. We would simultaneously raise monoclonal antibodies against
domains of the p340 protein and determine whether these antibodies give results that are consistent with
the observations from tagged proteins. Not only would these experiments be important for considenng
structural and functional questions related to p340, but they could also be extended to examine how the
distribution of p340 changes in response to nucleotide analogs and to other agents that disrupt nuclear
transport. Such experiments could be expected to reveal whether the localization of p340 is dynamic
during nuclear transport.

The roles of RNAl and RanBPl in cell cycle control and interphase nuclear function

During the coming year, we will investigate the cellular roles of RanBPl and RanGAPl in extract
depletion studies and mutational studies. In the depletion studies, RanGAPl or RanBPl will be removed
from extracts using the antibodies that we have recently generated. The depleted extracts will be used
in the nuclear assembly, DNA replication and nuclear import assays in order to determine how these
proteins are required for interphase nuclear functions. If depletion has a clear effect in one of these
assays, purified RanGAPl and RanBPl will be used to restore the defects caused by the depletion
treatment. [If the defect cannot be rescued, we would examine whether the depletion process removed
some other component(s) that is required for nuclear function.] While we expect that such depletion v/ill
result in non-functional nuclei, it will be of interest to determine whether defects are solely in one of the
nuclear functions, such as envelope formation or nuclear transport. If these depletion experiments are
successful, they will also establish an assay system in which to analyze RanGAPl and RanBPl mutants.

Mutational studies on RanGAPl and RanBPl will be focused on defining the domains of these
proteins. As mentioned above, constructs have been or will be made which will allow the expression
of mutant proteins or of protein fragments. The bacterially expressed proteins will be used for two tjpes
of experiments. First, they will be used in depleted extracts to define the critical domains of each of
these proteins. Second, they will be used as affinity reagents in experiments similar to those already
discussed for RCCl and Ran. The combination of these two approaches may simultaneously provide
evidence of which domains of these proteins control their localization and function and of what other
extract proteins interact with these domains. Finally, dominant negative mutants of both RanGAPl and
RanBPl will be of great interest if they can be generated, since they would be useful for bJocldng
individual steps in nuclear transport. Such mutant proteins could allow dissection the nuclear transport
pathway into sequential steps and the identification proteins that are involved in each of the steps.

ne^KBoamm

miiimwminrmnniimffifflmmwmnmff iiiiiiiiiiiiiii iiiiiiii inii iiii iiiiiiimi imnn—TH^

Project Number: ZOl HD 01902-01 LME

Systems regulating mRNA stability in the cell cycle

After confirming that the candidate mRNAs found in our differential screen are regulated in their
stability by the replication checkpoint, we will determine the sequences of these mRNAs. We then wish
to find regions of the mRNAs that confer instability. To do this, the cDNAs corresponding to these
mRNAs will be expressed from expression vectors in BHK cells. We will confirm that the mRNAs
expressed from these vectors are regulated in a manner similar to the endogenous mRNA, then we will
map the instability region by deletion analysis.

After the sequences responsible for mRNA stability are determined, we will pursue two avenues of
experimental investigation: First, we will seek to demonstrate what role the proteins encoded by these
mRNAs play in the regulation of mitosis. Destabilization of these mRNAs has been shown to block
caffeine-induced premature mitosis. A direct role in caffeine-induced premature mitosis could be
demonstrated if premature mitosis does not occur when the expression of a particular protein is blocked
through antisense RNA strategies or if premature mitosis is potentiated when we express a message
lacking the mRNA destabilizing element. Proteins that override the DNA replication checkpoint control
when inappropriately expressed would be expected to interact with or be key components of the
checkpoint pathway. Second, we would seek to identify the mechanisms through which the destabilizing
sequences confer regulation on these mRNAs. For instance, mRNA stability could be controlled by the
regulated binding of a protein to the stability sequence. We would test whether this is also the case for
checkpoint regulation of mRNA stability, and if so, we would seek to discover the identity of the binding
protein and how it is regulated by the checkpoint pathway.

Significance to biomedical research and the program of the Institute

Orderly progression through the cell cycle is required for the fidelity of transmission of genetic
material. Checkpoint controls insure that initiation of later cell cycle events depends on the completion
of earlier events. In eukaryotic cells, these controls are not only essential to the timing of events in
early development, but also critical for the regulation of cell cycle progression in somatic cells. Our
work concerns the molecular mechanism of these controls. For reasons that are discussed above, we
have chosen two points at which to investigate the S phase checkpoint: The Ran GTPase pathway and
the regulation of mRNA stability. Investigations of the Ran GTPase pathway will contribute to our
knowledge of the fundamental cell biology of the nucleus and nuclear transport. Eventually, we
anticipate that we will find how these essential nuclear activities are related to the signals that are
communicated between the nucleus and the cytoplasm in order to control the cell cycle. In this regard,
it is worth noting that an increasing number of oncogenes have been found to be nuclear pore proteins,
re-enforcing the notion that nuclear transport control may be critical to the regulation of cellular activity.
In a similar manner, we anticipate that a fundamental understanding of the cytosolic mRNA targets of
checkpoint control will contribute to our understanding of how the DNA replication checkpoint signaling
pathway functions. Once these mRNAs are identified, we will not only be able to investigate
downstream mitotic processes for which their protein products are required, but also upstream signal
transduction pathways through which information from the nucleus is relayed to cellular components in
the cytosol.

;:o

UUJMUll

Project Number: ZOl HD 01902-01 LME

Increasing our knowledge of how this checkpoint operates will ultimately help determine whether
DNA replication checkpoint failure contributes to disease states such as cancer. If so, this understanding
of the cell cycle may contribute not only toward basic biological knowledge but also toward the
development of diagnostic tests and treatments for cancer.

Publications:

Dasso M. The role of the Ran GTPase Pathway in Cell Cycle Control and Interphase Nuclear Functions,
Prog Cell Cycle Res 1995; (in press).

Dasso M, Seki T, Azuma Y, Nishimoto T. A Mutant for of the Ran/TC4 Protein Disrupts Nuclear
Function in Xenopus laevis Egg Extracts by Inhibiting the RCCl protein, a Regulator of Chromosome
Condensation, EMBO J 1994;23:5732-5744.

Dasso M, Dimitrov S, Wolffe AP. Nuclear Assembly and Replication is Independent of Linker Histone.
Proc Natl Acad Sci USA 1994;91:12477-12481.

Saitoh H, Dasso M. The RCCl protein interacts with Ran, RanBPl, hsc70 and a 340 kDa protein in
Xenopus extracts, J Biol Chem 1995;270:10658-10663.

famem

IBM

gs^BBn^RH

IHmilllWIMHHIg'WWlK

'•Co

ySH

JTI

^ vU J

^1 tiwj

^^.i^'

V <^^'
.^^

^^ / x/e/ / \^^^^ / \^ / \^^%/ \;e// \^^\.

^^^

%^i

^' ..^^ -\

* \ /V,:

ci:^'

y \

\
<*>.

\ v^^ .^j!:\ \ / «S. \ v^^ .^

^^*

■*,

.^^

oS^
^

*o.

«>.

^*' 1^ ^ **^ '

*o^*

#^ X

\ \ 4^

^^k^

.^•' X

E/ .^-^^ V ^

\ ■

.<^^

I ri_ X '^^" .^^^ X ^^^* .^^^ X '^^' .^^^ X ^^^* .^^^ X ^^^

^>\W Amazing Research. -V X<^ -^ /<^ 'A, yV<^ 'A^ '>^ V,

http://nihlibrary.nih.gov ^| |^§ |(^| |<^| |^|

10 Center Drive

Bethesda, MD 20892-1150

301-496-1080

V ^* ..^ X w j^ X ^^*' ..^- X ^fe^ ..^ X ^

t4^

'^^ ../ \

<&x.

# #

t«

>;. <^

HV^

?^ %

.f*

% I f §

\ ^^ Z^^"^' /' \ ""^ /' X "^^ #^ \ ''^ ^-"^ X '^^ J^ X *

o./ ^>.

-n^

<S>V/T<'

% <S^ %

% /

^-^^

'^^ .^^^H^ X ^v

X "^^

c^)::c«

'^S;*^'^ ..^"^ V '^'■'VN

i^'^ vp'if^'Q

X /"

,^^

#• i:

Internet Archive Audio

Featured

Top

Images

Featured

Top

Software

Featured

Top

Books

Featured

Top

Video

Featured

Top

Mobile Apps

Browser Extensions

Archive-It Subscription

Save Page Now

Full text of "Annual report : National Institute of Child Health and Human Development"

See other formats