CULAR

S^^^S^SB^E3

Marine Biological Laboratory Library y;

Woods Hole, Mass. E

Presented by

The Ronald Press Co,

New York
April 26, 1961

B3^^^^^^^^^^^^^SE

The Society of General Physiologists

Sixth Annual Symposium

University of Illinois
Urbana, Illinois, September, 1959

Macromo lee ular
Complexes

DAVID F. WAUGH • ALAN J. HODGE • FRANCIS 0. SCHMITT

MELVIN J. GLIMCHER • J. J. WOLKEN

H. FERNANDEZ-MORAN • HANS A. WENT

DANIEL MAZIA • J. A. BERGERON • R. C. FULLER

WALTER J. NICKERSON • G. FALCONE

GIAN KESSLER • R. D. PRESTON

Edited by
M. V. EDDS, JR.

BROWN UNIVERSITY

THE RONALD PRESS COMPANY • NEW YORK

Copyright © 1961 by
The Ronald Press Company

All Rights Reserved

the publisher.

Library of Congress Catalog Card Number: 61-6324

PWNTED IN THE UNITED STATES OF AMERICA

Preface

Efforts to analyze and interpret life processes, especially as they
relate to development, are ultimately confronted with such ques-
tions as: Do the orderly patterns constituting given levels of the
organizational hierarchy regularly arise solely as a consequence of
the specific interactive properties of the units comprising the next
lower levels? Given the proper units and the proper conditions, is
the spontaneous emergence of order at a higher level inevitable?

Considerable evidence attests that, at least in some cases, these
questions may be answered affirmatively. Examples include the in
vitro aggregation of certain macromolecules, among which collagen
is the prototype, into fibrils of near-crystalline regularity, and the
selective aggregation of previously dissociated cells into recogniz-
able tissues. The spontaneous emergence of orderly patterns could
also be illustrated by examples from levels higher than the cell or
lower than the macromolecule. Between these levels, however, are
others about which information is less certain. The time is clearl\'
ripe to begin inquiring into the nature and origin of heterogeneous
aggregates of macromolecules, with special emphasis on the inter-
active properties of their subunits. Actually, the subunits them-
selves have been identified only in a few cases, and concepts of their
structural and functional interrelations are still based largely on
inference. But, as F. O. Schmitt has written, we have already
learned enough to "glimpse, however imperfectly, the portentous
possibilities now opening up before us."

Some of these possibilities are explored in this book embodying
the proceedings of a symposium sponsored by the Society of Gen-
eral Physiologists. As the text reveals, exploration of the borderland
between the realms of the macromolecular chemist and the func-
tionally oriented electron microscopist has already yielded im-
portant information about the organization of living matter; it is
thus more than an article of faith that many biological problems
can be effectively attacked at this level. Not the least of these prob-

jy PREFACE

lems are those related to development. It is obvious, but often
neglected, that the genesis of form and function occurs, and must
be studied, at all levels of the organizational hierarchy. Most of the
attempts to relate chemistry and morphogenesis have stressed en-
tities on both sides of the borderland but have been silent on v^'hat
lies between.

The basic premise of the symposium was that further clarification
of the relation between chemistry and morphogenesis will depend
heavily on the analysis of aggregates of two or more species of
macromolecules. The speakers, who were selected because their
main competence lay neither in chemistry nor in morphogenesis,
but somewhere in between, were encouraged to discuss the thesis
that many aspects of morphogenesis can be attacked in relatively
"simple" systems b)' exploring in detail the properties of the macro-
molecular complexes which compose them. To this challenge the
speakers responded with solid information, both confirming the
thesis and raising a host of new morphogenetic problems. Excur-
sions into other areas were inevitable and welcome. Thus, much
was said that is related to development only in the sense that it
poses the question: How does it develop? But whatever the bias of
individual papers, the entire volume is a record of a combined effort
to place a point of view before all who may profit from its con-
templation.

This symposium was supported by a P.H.S. Research Grant
(RG-6228) from the Division of General Medical Sciences, Public
Health Service; the symposium was held at the University of Illinois,
September 7-9, 1959.

M. V. Edds, Jr.

Contents

Molecular Interactions and Structure Formation in Bio-
logical Systems

David F. Waugh

The Tropocollagen Macromolecule and Its Properties of

Ordered Interaction ....... 19

Alan J. Hodge and Francis O. Schmitt

The Role of the Macromolecular Aggregation State and

Reactivity of Collagen in Calcification ... 53

Melvin J. Glimcher

The Chloroplast 85

J. J. Wolken

Lamellar Systems in Myelin and Photoreceptors as

Revealed by High Resolution Electron Microscopy . 113

H. Fernandez-Moran

Fibrillar Systems in the Mitotic Apparatus .... 161
Hans A. Went and Daniel Mazia

The Submicroscopic Basis of Bacterial Photosynthesis: The
i Chromatophore 179

J. A. Bergeron and R. C. Fuller

CONTENTS

PAGE

Polysaccharide-Protein- Complexes of Yeast Cell Walls . 205
Walter J. Nickerson, G. Falcone, and Gian Kessler

Cellulose-Protein Complexes in Plant Cell Walls .
R. D. Preston

Index

229

255

Macromolecular
Complexes

Molecular Interactions and Structure
Formation in Biological Systems

David F. Waugh ^

The three-dimensional display of structural complexity in cells
and tissues, now so convincingly visible in the evidence from elec-
tron microscopy, is at the colloidal level, thus the level where one or
two dimensions of the particles (rodlets, lamellae) which result
from molecular interaction are near the dimensions of the interact-
ing molecules themselves. We accept the fact that the functional
aspects of cells, tissues, organs, etc., must be a natural (thermo-
dynamically acceptable) result of the organization at the level
mentioned, and, although we are not in a position to state the type
or extent of knowledge which will be necessary to understand the
way in which subcellular colloidal aggregates modify each other's
activities to produce the net result of "life," several aspects of in-
teraction have been examined in a way which must excite all in-
terested in biological processes. These I have been asked to discuss
briefly.

Manipulation and interaction at the molecular level involve
selections of the interacting molecules; thus, the interactions are
specific. We are familiar with the high degree of specificity ex-
hibited by enzyme-substrate interactions ( Wilson, 1959 ) , in energy
transfer (Lehninger, 1959), in immunological interactions (Land-
steiner, 1945), in the formation of tissues such as collagen and
elastin (Schmitt, 1959), in the formation of organs, and during the
course of embryonic development, to cite a spectrum of examples.
As we shall see, specificity of interaction between molecules coming
into close contact may be understood on the basis of short-range
interactions between appropriately placed submolecular groups of
atonijS having different interaction characteristics. Catalyzed chem-

^ Massachusetts Institute of Technology, Cambridge, Massachusetts.

3

4 MACROMOLECULAR COMPLEXES

ical reactions and the development of the more condensed regions
of structure (lamellae, fibrils, etc.) may soon be quantitatively ac-
counted for on this basis.

The biological system is clearly one in which most reactions are
carried out in the presence of water. There is, however, apparently
an abundance of water, and control of the amount and disposition
of water is of paramount importance. The more condensed regions,
on whose surfaces or within which much of the molecular traffic
will take place, must be prevented from increasing in size (or pre-
cipitating out ) to the point where such interferes with the execution
of ordered series of linked reactions by the condensed structures and
with a reworking of the molecules of the condensed structures them-
selves. We suppose that the incorporation or holding of too much
water, however, might well interfere with molecular traffic, would
certainly reduce mechanical rigidity, and might allow, for example,
deformation of the organ or tissue as a whole to produce local ( sub-
cellular) deformations in colloidal structure from which the system
would not recover. Control of the amount and disposition of water
in localities where the condensed structure is organized in the fomi
of a continuous network (gel) is most readily understood. How-
ever, deformation of a cross-linked network must necessarily involve
both flow of the entrapped fluid and a rearrangement of the relative
positions of the elements of the network. A network arrangement
is most effective where the network supports some biological func-
tion such as the transmission of tension, as with collagen and elastin.
In the examples chosen, the fraction of the tissue or organ occupied
by water is low, and the tissues are relatively inert.

Many portions of the biological system, however, contain large
amounts of water, and interactions of the more condensed structures
may be taking place over intervening aqueous regions of consider-
able size. Here a distortion of the system need not change the
relative positions of the elements of the condensed structures; for
example, lamellae might slide past one another without altering the
interlamellar distance, thus without altering the local distribution of
water and solutes.

In considering molecular interaction and the formation of con-
densed structures, we must first note that the word "molecule" can
be given a restricted meaning with respect to small molecules:
namely, a group of atoms joined by covalent (infrequently coordi-
nation) bonds which are stable with respect to kinetic energy (kT).

MOLECULES AND STRUCTURE FORMATION 5

The fact that covalent bonds have associated with them preferred
or hmited distances, directions, and rotations in conjunction with
the sizes of the participating atoms (Pauhng, 1940) means that the
possible conformations of the molecule may be specified. As the
number of atoms increases, possibilities arise for various types of
interaction between groups of atoms within the same covalently
linked assembly. In molecules of large size, such as the proteins,
carbohydrates, and nucleic acids, these interactions are numerous
and involve all known types of forces. For example, these inter-
actions are involved in determining the structures of the poly-
peptide chain helices of proteins (Pauling ct fl/., 1951), the inter-
actions of proteins (Waugh, 1954), and the double helix of nucleic
acids (Watson and Crick, 1953). Recognition of the importance of
the group of short-range forces in determining specificity of inter-
action, as well as the intramolecular stabilizations just mentioned, is
due largely to the work of Pauling and co-workers.

Table 1 gives a general idea of the various types of short-range
and long-range interactions. The characteristics of the first of these,
the covalent bond, have already been mentioned. The hydrogen
bond is a permanent dipole interaction, involving a hydrogen atom
covalently bonded to one electronegative atom and the unshared
electron pair of another electronegative atom ( Pauling, 1940; Orgel,
1959). This bond has a high interaction energy, a result of the small
size of the proton which permits the close approach of dipoles. To
break a hydrogen bond requires an investment of about 5 kcal per
mole. However, if H-bond exchange with water is possible, the
investment is much reduced and may be reversed if the groups
involved make unusually strong hydrogen bonds with water. Thus
the strength of a particular H-bond is dependent on the circum-
stances permitting a simultaneous interaction with water.

The simple ionic bond, a strong interaction in the absence of
water, leads to attraction or repulsion according to Coulomb's law.
The interaction energy varies with the dielectric constant of the
intervening medium and inversely with the distance between cen-
ters. The former varies strongly with distance in an aqueous
medium and decreases as the distance between centers decreases
(Pressman et al., 1946). Assuming that water hydrates an interact-
ing pair such as an ammonium ion and carboxyl ion, close approach
will produce an interaction energy of about 5 kcal per mole ( Pauling
et at., 1946). We should remember also that charged groups lead

MACROMOLECULAR COMPLEXES

p5°

■ S 4^ Oi

G -^
O

T3

3 X

MOLECULES AND STRUCTURE FORMATION 7

to an ion-dipole interaction with water molecules and so become
hydrated, the interaction energy amounting to about 5 kcal per
mole of bound water (Dole and McLaren, 1947).

The last of the forces operating at close approach are the London-
van der Waals attractive forces, which result from the polarization
of each group of atoms in the fluctuating electrical field arising from
the instantaneous configurations of the electrons and nuclei of ad-
jacent groups. For small groups, the energy varies with the inverse
6th power of the distance between centers and directly with the
molar refractions of the groups involved (Pauling and Pressman,
1945).

Specificity Effected by Short-Range Interactions

It is of importance to note that both the hydrogen-bond and
van der Waals interactions vary, respectively, inversely with the
3d and 6th powers of the distance of separation of the groups. If
the maximum interaction energy is released when, for example, two
CHl' groups are 4 A apart, separating these groups by an additional
2 A will reduce the interaction energy by a factor of about 11. Like-
wise for the hydrogen bond, stretching the bond by 1 A essentially
terminates the interaction. We see at once an extraordinary speci-
ficity can be accounted for by the requirement that an adequate
interaction energy be obtained only when a variety of groups hav-
ing different interaction characteristics are brought into their proper
positions. These groups would include unlike charges, those form-
ing hydrogen bonds, and the bulky groups, interacting through van
der Waals forces. The latter, which will essentially define the sur-
face of a large molecule, must fit so that the protuberances on one
surface find appropriate cavities on the other surface. Remembering
that a stable interaction is produced when the total interaction
energy is about 10 to 15 X kT, and that all local or group interaction
energies are summed to obtain the total interaction energy, sug-
gests that the smaller the area of interaction, the more perfect the
fit between interacting surfaces or the more numerous the groups
giving rise to stronger interactions.

Even the most specific of interactions can now be rationalized
on the basis given; for, as is evident, minute alterations in struc-
ture may lead to imperfect steric fit, and since energies of inter-

8 MACROMOLECULAR COMPLEXES

action are so sensitive to distance, a small imperfection in fit pro-
duces a large change in interaction energy.

A product of the short-range interaction between two molecules
might well be a complex having new properties. A clear case of this
tvpe is the dissociation or fragmentation of the protein ribonuclease
into two enzymatically inactive portions ( Richards and Vithayathil,
1959 ) . Although the fragments have been produced by hydrolyzing
a covalent bond, the fragments, on mixing, combine to give a com-
plex which has full enzymatic activity. The ruptured covalent bond
is not re-foriped on mixing. A variety of exciting possibilities is
brought to our attention by this discovery. For example, the inter-
action of a steroid hormone molecule with an inactive protein might
induce enzymatic activity in the latter, this combination controlling
a host of subsequent reactions.

Specific interactions are undoubtedly also involved in establish-
ing condensed regions, which usually appear as filaments, lamellae,
or helices, at levels of size within an order of magnitude of 100 A.
Currently, the lamellar structure containing lipid and protein ap-
pears to be a conspicuous component of cells, the assumption being
made that the lipid within the lamella will be present as a double
layer or layers. Certainly such an organization of lipid would be ex-
pected from the properties of soaps and lipid emulsions (Luzzati
et ai, 1958; Lawrence, 1958; Palmer and Schmitt, 1941) and the
structure of the myelin sheath (Schmitt, 1959). The double-layer
arrangement permits the bulky non-polar portions of the lipid mole-
cules to associate, leaving surface layers of polar groups to interact
with each other and with water, protein, or other polar substances.

In many instances the development of a lamella may not involve
any high degree of fit (specificity) of the polar and non-polar por-
tions of the molecules involved. We would expect a low degree of
specificity mainly when the interaction leads to the development of
a structure which carries out some non-critical phvsical function,
a safe example being that of fat storage— although another might be
the essentially two-dimensional expansion of non-functional por-
tions of a plasma or other membrane. Certain functional aspects,
however, may require a specific arrangement of non-polar and polar
portions which will provide an energy barrier to penetration (per-
meability). Although the physical conditions establishing the basis
for interaction in monomolecular films are probably not duplicated
often in biological systems, the effects of short-range interaction in

MOLECULES AND STRUCTURE FORMATION 9

modif\'ing the properties of films are clearly shown in studies of the
evaporation of water through compressed films of stearic acid or
cetyl alcohol; the latter present an extraordinary energy barrier to
the penetration of water molecules (Langmuir and Schaefer, 1943).
This possibility, namely, that non-polar portions of molecules, par-
ticularly lipid molecules, may interact to present a high-energy
barrier to the penetration of water, is of considerable biological
interest; for, in spite of the relative abundance of water in biological
systems, local regions at the particulate level may involve bound-
aries between essentially aqueous environments and environments
where the interactions of non-polar portions of molecules are so well
integrated that water is virtually excluded. A structure of this type,
for example, may be necessarv in effecting the key reactions of
photosynthesis (Calvin, 1959).

In this introductory discussion, I would like to summarize some
of our experiences with two protein systems in which short-range
interactions, coupled with particular types of specificity, lead either
to extended fibrils or to spherical micelles, namely, the interactions
of insulin (Waugh, 1957) and of casein (Waugh, 1958).

Mechanism of Fibril Formation. The insulin fibril forms under
conditions where the dimer of M ~ 11,000 is the prevalent form,
i.e., at pH 2. Heating at 80° to 100° C causes a spontaneous trans-
formation into a population of fibrils, the most numerous, and larg-
est, of which are about 200 A in diameter and many thousands of
Angstroms in length. The reaction goes essentially to completion
but is reversible in the sense that under alkaline conditions the fibril
disaggregates to yield insulin. Evidence which gives a clue to the
mechanism of fibril formation, and which suggests that the insuhn
molecule does not undergo extensive unfolding in the process of
forming fibrils, comes from experiments in which fibril segments are
seeded into insulin solutions at pH 2. While such solutions alone are
stable for long periods of time at temperatures of 20° C or below,
the seeded fibrils or fibril segments recruit insulin from solution and
in the process grow according to first-order kinetics. Structurally,
the new portions of the fibrils obtained after fibril growth at lower
temperatures appear to be identical with those which are formed at
80° to 100° C.

When a solution of insulin is heated and the transformation of
insuhn into fibrils is plotted as a first-order reaction, the resulting
curves typically have a lag period which is followed by an essen-

10

MACROMOLECULAR COMPLEXES

tially linear rise. The extent of the lag period is determined mark-
edly by the initial insulin concentration, as is the slope of the near-
linear portion of the curve. A comparison of the reaction kinetics
obser\ed with the growth of seeded fibrils and in the absence of
seeding suggests that the fibril is first initiated by a nucleation
reaction which involves the cooperative effects of three or four in-
teracting units (dimers). Thus the rate of nucleation varies with
the 3d or 4th power of the insulin concentration. After initiation,
when the fibril has achieved a reasonable size, the fibril grows as
a function of its surface area and the 1st power of the free insulin
concentration. The cooperative effect established during nucleation
is perpetuated, and the surfaces (particularly of the ends) of the
fibril present to the entering insulin unit the correct cooperative
configuration necessary for bonding.

One particularlv interesting consequence of this type of mech-
anism is that most of the fibrils are initiated during the lag period
of the reaction. Those which are initiated during the first few
minutes of the lag period dominate the reaction, in the sense that
thev are responsible for removing most of the insulin. The fibril
population at the end of the reaction appears to be relatively

16. lA

485A

"■^1

- ~-J

Fig. 1. Diagram of a fibril nucleus,
(From Waugh, 1957.)

lustroting a cooperative interaction.

MOLECULES AND STRUCTURE FORMATION 11

homogeneous. Aggregations of the fibrils themselves may lead to
the formation of spherites.

It is clear that, at least over a portion of its development, the
fibril axial ratio increases with fibril mass. That an asymmetric
aggregate should form when charged molecules link, without speci-
fication of surface structure, has been pointed out by Rees ( 1951 ) .
This is because the electrostatic-potential barrier is lowest at the
ends of the dimer or other asvmmetric unit. However, the coopera-
tive effect itself will exert a strong directional influence. This is
illustrated in Fig. 1, which is a possible arrangement for the insulin-
fibril nucleus. The stable structure is formed when any fourth
insulin unit is added to the correct previous group of three. There-
after, insulins are expected to add most frequently in a manner
which perpetuates this stable structure; consequently, the aggregate
elongates in the direction of the cylinder axis of Fig. 1.

Stable Colloids. The interactions of the caseins (Waugh, 1958)
reveal both specificity of interaction and cooperative effects, the end
result in milk being a population of micelles varying in chemical
composition and volume, the latter over a range of about 350 fold.
Yet the micelles are in rapid equilibrium with their constituent
components in solution, some of which are relatively insoluble under
the conditions where micelles readily form. The most important
initial interaction in micelle formation occurs between a^- and
k-caseins, and in this interaction secondary valence and charge in-
teractions arrive at a remarkable mutual satisfaction. Here, a^-casein
is a phospho-protein of M ~ 23,000. It has a phosphorus content of
over 1 per cent but no disulfide. Physically, this protein appears to
be a single coil 210 A in length and 16 A in diameter. In the absence
of calcium the protein is a highly soluble polymer; the polymer be-
comes quite insoluble in the presence of calcium. The k-casein is
also apparently a single coil about 150 A long, which contains disul-
fide but little phosphorus. It forms soluble, condition-insensitive
polymers of 13.5 S in the presence or absence of calcium. When mix-
tures are made containing molecular ratios of 3 a^-casein to 1
k-casein, the original polymers disappear and a stoichiometric a,-
k-casein complex forms. This complex can form in the absence of
calcium and under conditions where both of the monomers carry a
high 'net negative charge. One must assume that here, as in other
cases, secondary valence forces are responsible for association. In
this connection, an examination of the amino acid compositions of

12 MACROMOLECULAR COMPLEXES

all the caseins reveals a high-occurrence frequency for the larger
non-polar amino acids.

The tts-k-complex, although it exists at all temperatures between
0° and 37° C, is unstable at the lower temperatures. This is ap-
parent from the results obtained on adding calcium to give 0.05M.
At the higher temperature, the complexes are stabilized and will
now engage in further aggregation among themselves and with
/3-casein to form colloidal micelles typical of milk. Once formed, the
Ca-tts-k-casein complex is unusually stable to heat, changes in pH,
etc. We propose for it a structure in which the three as-casein mole-
cules are oriented axially around a k-casein molecule. At higher
temperatures, secondary valence interactions hold (against electro-
static repulsion ) the a^-casein molecules in positions such that pairs
of phosphorus groups attached to adjacent cts-casein monomers are
in juxtaposition and can be linked through the introduction of a
calcium ion. The precision with which the phosphate groups are
positioned is probably dependent upon the conformations of the
interactants, for, at lower temperatures, the instabilitv of the com-
plex suggests that the phosphorus groups cannot be cross-linked
b\' calcium. If each phosphorus is able to accept a calcium ion,
stabilization through Ca-phosphate cross-linking would be absent,
and, since Ca-as-caseinate is relatively insoluble at low tempera-
tures, it would precipitate and thus lead to a dissociation of com-
plexes.

Long-Range Interactions

The presence of large amounts of water in certain types of bio-
logical systems poses special problems, particularly if the condensed
regions of the system in question do not form a continuous network.
Two items currently of great interest are the structure of water in
such regions and the possil)ility that long-range forces ma\- be in-
volved in establishing the structure of the system.

The last two items of Table 1 relate to long-range forces. The
first of these, listed as London-van der Waals forces, have their
origins in mutual polarization, as was stated above. However, the
specification that large planes of atoms are interacting with each
other, and the fact that these interactions are additive, modifies the
variation in attractive energy with distance so that, for molecules or
condensed regions presenting sufficiently large interacting surfaces,

MOLECULES AND STRUCTURE FORMATION 13

the interaction energy varies inversely with the square of the dis-
tance (Overbeek, 1952; Prosser and Kitchener, 1956).

The last item of Table 1 refers to large molecules carrying many
groups capable of being positively or negatively charged, or un-
charged. At the isoionic point and at low ionic strength, mutual
electrical polarizations will tend to establish charge patterns on the
two molecules which lead to an electrostatic attractive energy
which decreases in proportion to the distance. At short distances,
interaction through charge fluctuation might provide an interesting
explanation for the participation of the protein moiety of an enzyme
molecule in the mechanism of hydrolytic enzyme reactions (Kirk-
wood, 1957).

Of great interest to the biologist are those systems containing
large, usually asymmetric, particles at the correct concentration,
pH, and ionic strength which separate into two phases: one phase
in which the solute concentration is reduced below average and in
which orientation of the particles is absent, and a second phase
in which the solute concentration is increased above average and the
particles are well oriented and exist in a three-dimensional pattern.
Examples are to be found in bentonite (platelets), tobacco mosaic
protein (rodlets), and many systems in which "colloids" carrying
net charges of different sign are mixed. The last-named examples,
which must be of importance in providing an understanding of the
way in which large molecules interact over intervening aqueous
gaps, have recently been analyzed by Overbeek ( 1957 ) and by
Michaeli et al (1957). They treat the system theoretically as a
competition between charge interactions, which tend to accumulate
the charged particles, and entropy, which tends to disperse them.
Of considerable interest are the results obtained when the charged
particles, the polyanion and polycation, are present in different con-
centrations; for now the concentrated phase, the coacervate, will
be more symmetric in concentration than the starting mixture, while
the dilute phase will be more asymmetric. Separation into phases
takes place at solute concentrations which make these systems of
immediate biological interest.

The structure of water in biological systems is also a matter of
importance. Water of hydration, at least that associated with polar
group^, will be firmly bound and will be involved in defining the
contour and properties of the surface of the condensed structure to
which it is attached. In the lamellar systems mentioned previously

14 MACROMOLECULAR COMPLEXES

there is apparently space to accommodate water layers ranging up
to about 1000 A in thickness. The properties of the system will
naturally depend on whether this water is solvent water having the
properties of bulk water at the same temperature, or water which,
by virtue of the structures of the contiguous condensed regions, has
a more or less regular arrangement. Szent-Gyorgyi (1956) has
recently considered the importance of water structure and has per-
formed several intriguing experiments which suggest that water in
the biological system may have predominantly an icelike structure.
Kauzmann (1959), in a review of protein denaturation, discusses
the formation of icebergs around non-polar groups, as well as the
short-range interactions of interest here. The evidence presented by
Kauzmann suggests that the ordering of water molecules is con-
fined to the locality producing the ordering and is not transmitted
beyond a few Angstroms.

If we restrict ourselves to considerations of colloidal structure,
it is apparent from the foregoing that molecules may, by a variety
of short-range interactions, essentially come out of solution and
develop condensed regions which, from the principal directions of
development, would be classified as rodlets, filaments, lamellae,
platelets, etc. One of the systems, involving the development of
collagen filaments, not only may illustrate the relationships between
size and long-range interaction but suggests the possibility of spe-
cific long-range interactions. This system, the basement lamella
lining the skin of amphibian larvae, has been treated bv Weiss
(1957). In the region of the basement lamella there are approxi-
mately 20 layers, each 2000 A in depth. Each laver contains a
system of rods, tentativelv identified as collagen, which are 500 A in
diameter and cross-striated at 520 A intervals. Within any one
layer the rods run parallel, with their cross-striations in lateral
register. However, the orientation of each laver differs from that
of adjoining ones by an angle of 90°.

If a piece of membrane is excised, the membrane is repaired.
During this process there first appear small cross-striated fibrils of
about 200 A diameter. These are initially unoriented, but as they
grow in length and diameter the particular matched orthogonal
pattern is established. If the medium between the oriented fibrils
does not contain condensed structures to transmit directions for
orientation and for striation matching, the s\'stem stronglv suggests
the existence of specific long-range interactions. It is possible that,

MOLECULES AND STRUCTURE FORMATION 15

as the extension in one or two dimensions becomes sufficiently large,
the long-range interactions at separated loci are quite different.
Thus, possibly an interaction pattern will emerge which we will
regard as being specific.

Without proving the point, a great deal of evidence may be
selected from the behavior of biological systems which indicates
that long-range interactions may have elements of specificity. The
classical example is chromosome pairing, which has long excited
interest. I feel that the behavior of chromosomes during meiosis
suggests that a specific interaction occurs at the kinetochores, pro-
viding a primary point of orientation and possibly overcoming an
initial energy barrier to close approach. Thereafter, long-range in-
teractions might align homologous segments. Of course, a few of
the up-to-now invisible strands might provide condensed regions
over which a continuum of short-range interactions would be re-
sponsible for pairing.

We appear to be dealing with fundamental types of forces of
interaction involving long- and short-range patterns of interacting
groups which can give rise to specificity of size, shape, and chemical
differentiation of interacting particles. From the large number of
possible combinations, the problem is to isolate and understand
those particularly important types which will further our under-
standing of the way in which structure is developed and of the rela-
tionships between structure and function.

References

Bennett, H. S. 1959. Structure of muscle cells. Revs. Modern Phijs. 31: 394-401.
Bernal, J. D. 1958. Structure arrangements of macromolecules. Discussions Fara-
day Soc. No. 25: 7-18.
Calvin, M. 1959. Energy reception and transfer in photosynthesis. Revs. Modern

Phys. 31: 147.
Dole, M., and A. D. McLaren. 1947. The free energy, heat and entropy of sorption

of water vapors by proteins and high polymers. /. Am. Chem. Soc. 69: 651-657.
Fernandez-Moran, H. 1959. Fine structure of biological lamellar systems. Revs.

Modern Phys. 31: 319-330.
Hodge, A. J. 1959. Fibrous proteins of muscle. Revs. Modern Phys. 31: 409-425.
Kauzmann, W. 1959. Some factors in the interpretation of protein denaturation.

Advances in Protein Chem. 14: 1-57.
KiRKWOOD, J. G. 1957. The forces between protein molecules in solution. /. Celhdar

Comp. Physiol. 49 ( Suppl. 1 ) : 59-62.
Landsteiner, K. 1945. The Specificity of Serological Reactions. Rev. ed. Harvard

Unifversity Press, Cambridge, Mass.
Langmuir, I., and V. J. Schaefer. 1943. Rates of evaporation of water through

compressed monolayers on water. In F. R. Moulton (ed. ). Surface Chemistry.

Publ. Am. Assoc. Advance. Sci. No. 21. Pp. 17-39.

1^ MACROMOLECULAR COMPLEXES

Lawrence, A. S. C. 1958. Solubility in soap solutions. Part 10. Phase equilibrium,

structural and diffusion phenomena involving the ternary hquid crystalline phase.

Discussions Faraday Soc. No. 25: 51-58.
Lehninger, a. L. 1959. Respiratory-energy transfonnation. Revs. Modern Phys.

31: 136-146. , r , V J

LuzzATi, v., A. MusTACHOi, and A. Skoulios. 1958. The structure of the hquid-

crystal phases of some soap + water systems. Discussions Faraday Soc. No. 25:

43-58.
Michaeli, I., J. Th. G. Overbeek, and M. J. Voobn. 1957. Phase separation ot

polyelectrolyte solutions. ]. Polymer Set. 23:443-450.
Orgel, L. E. 1959. The hydrogen bond. Revs. Modern Phys. 31: 100-102.
Overbeek, J. Th. G. 1952. London-van der Waals attraction between two parallel

flat plates. In H. Kruyt (ed.). Colloid Science. Elsevier PubHshing Co., Amster-
dam. Vol. 1, pp. 266-268.
Overbeek, J. Th. G. 1957. Phase separation in polyelectrolyte solutions. Theory of

complex coacervation. /. Cellular Comp. Physiol. 49 (Suppl. 1): 7-26.
Palmer, K. J., and F. O. Schmitt. 1941. X-ray diffraction studies of lipide emul-
sions. /. Cellular Comp. Physiol. 17: 385-393.
Pauling, L. 1940. The Nature of the Chemical Bond. 2d ed. Cornell University

Press, Ithaca, N. Y.
Pauling, L., D. H. Campbell, and D. Pressman. 1943. The nature of the forces

between antigen and antibody and of the precipitation reaction. Physiol. Revs. 23:

203-219.
Pauling, L., R. B. Corey, and H. R. Branson. 1951. The structure of proteins:

Two hydrogen-bonded helical configurations of the polypeptide chain. Proc.

Nat. Acad. Set. U. S. ^7: 205-211.
Pauling, L., and D. Pressman. 1945. The serological properties of simple sub-
stances. IX. Hapten inhibition of precipitation of antisera homologous to the

0-, m-, and p-azophenylarsonic acid groups. /. Am. Chem. Soc. 67: 1003-1012.
Pressman, D., A. L. Grossberg, L. H. Pence, and L. Pauling. 1946. The reactions

of antiserum homologous to the p-azophenlytrimethylammonium group. /. Am.

Chem. Soc. 68: 250-255.
Prosser, a. p., and J. A. Kitchener. 1956. Direct measurement of long-range van

der Waals forces. Nature 178: 1339-1340.
Rees, a. L. G. 1951. Directed aggregation in colloidal systems and the formation

of protein fibers. /. Phys. Chem. 55: 1340-1344.
Richards, F. M., and P. J. Vithayathil. 1959. The preparation of subtilisin-modi-

fied ribonuclease and the separation of the peptide and protein components. /.

Biol. Chem. 234: 1459-1465.
Schmitt, F. O. 1959. Interaction properties of elongate protein macromolecules

with particular reference to collagen ( tropocollagen ) . Revs. Modern Phys. 31:

349-358.
Sjostrand, F. S. 1959. Fine structure of cytoplasm: The organization of mem-
branous layers. Revs. Modern Phys. 31: 301-318.
Szent-Gyorgyi, a. 1956. Bioenergetics. Science 124: 873-875.
Watson, J. D., and F. W. C. Crick. 1953. Molecular structures of nucleic acids.

Nature 171: 737-738.
Waugh, D. F. 1954. Protein-protein interactions. Advances in Protein Chem. 9:

325-437.
Waugh, D. F. 1957. A mechanism for the fomiation of fibrils from protein molecules.

;. Cellular Comp. Physiol. 49 (Suppl. 1 ) : 145-164.
Waugh, D. F. 1958. the interactions of as, /S- and k-caseins in micelle formation.

Discussions Faraday Soc. No. 25: 186-192.
Wei.ss, P. 1957. Macromolecular fabrics and patterns. /. Cellular Comp. Phusiol.

49 (Suppl. 1): 105-112.
Wilson, I. B. 1959. Molecular complementaritv and antidotes for alkylphosphate

poisoning. Federation Proc. 18: 752-758.

MOLECULES AND STRUCTURE FORMATION 17

DISCUSSION
A. G. Richards, D. F. Waugh, W. D. McElroy, W. J. Nickerson

Dr. Richards (University of Minnesota): During development of an in-
sect's cuticle, there is a large dehydration of the chitin-protein matrix and a
simultaneous loss of almost all the salt, as shown by microincineration. Does
this imply that the salts, which are not homogeneously distributed in the cuticle
matrix, are nevertheless present effectively only in free solution in the water
phase?

Dr. Waugh: The salts which are distributed heterogeneously are probably
not present effectively in free solution in the water phase. If we were sure
that there were no restrictions to diffusion and no active transport, the state-
ment could be made more positively. If this were so, it would appear that
during dehydration the binding sites responsible for heterogeneity effectively
disappear, leaving the ions to be swept out.

Dr. McElroy (Johns Hopkins University): Is the water between fibrils
at such a low concentration that a limited chemical reaction could remove
essentially all of it? Are there good examples from biological structtires in
which the primary forces (van der Waals, H-bond, etc.) are only important
in bringing the molecules together? Do other chemical processes then occur,
which are important in giving the final property to the structure?

Dr. Waugh: What kinds of fibrils? Usually there is too much water to be
removed by a limited chemical reaction. In this introduction, I have treated
the forces which are most likely to be operating in determining specificity of
interaction and structure formation. The final properties of the structure will
undoubtedly involve types of interactions and, particularly, chemical processes
not mentioned here; for example, the chemical processes involved in photo-
synthesis and bioluminescence.

Dr. Nickerson (Rutgers University): Following up on the second of Dr.
McElroy's questions, do you have any information on the mechanism whereby
a trace of saturated fatty acid glyceride can induce almost instantaneous
crystalline order in a Hquid, unsaturated fatty acid glyceride? This reaction
appears, from some work in our laboratory on development of yeast cells, to
have a biological reality.

Dr. Waugh: The fatty glvcerides undergo a series of phase changes with
temperature, and a sizable literature is available on this subject. It is possible
that a small amount of saturated fatty acid glyceride induces crystallization by
accelerating the formation of crystal nuclei.

The Tropocollagen Macromolecule and Its
Properties of Ordered Interaction ~

Alan J. Hodge '^ and Francis O. Schmitt "*

The discovery that collagen in sokition exists in the form of dis-
crete monomeric macromolecular units, termed "tropocollagen"
(TC) by Gross et al. (1954), and consists, in all probability, of
several polypeptide chains arranged in a specific coiled configura-
tion, has had important consequences in the investigations of the
chemistry of connective tissue, in the biomedical field, and in fields
of special application such as the chemistry of leather. The detailed
examination of the various ordered aggregation states of the TC
macromolecule in the electron microscope and, in particular, the
examination of the so-called "segment long-spacing" form (SLS),
which is in effect a "molecular fingerprint" in that it indicates the
distribution of polar side-chains, have resulted in the formulation
of a mechanism for the end-to-end linkage of the macromolecules,
a process presumably involved in fibrogenesis in vivo. The precise
packing arrangements of TC in the various ordered aggregation
states are now known in detail, and it is on the basis of the specific
stereochemical configurations of the polar side-chains in these vari-
ous forms that a mechanism for calcification has been advanced by
Glimcher et al. ( 1957 ) . The experimental evidence indicated that
only the specific steric arrangement present in the "native-type"
aggregation state was capable of initiating nucleation of hydroxy-

' The substance of this paper was also presented by one of us ( F.O.S. ) at the
Sixth Congress of the International Union of The Leather Chemists Society, Munich,
Germany, September 8, 1959 (Schmitt and Hodge, 1960). The German translation
was published in Das Leder.

- This investigation was supported by a research grant, E-1469, from the National
Institute* of Allergy and Infectious Diseases, National Institutes of Health, U. S. Public
Health Service.

^ California Institute of Technology, Pasadena, California.

■* Massachusetts Institute of Technology', Cambridge, Massachusetts.

19

20 MACROMOLECULAR COMPLEXES

apatite from solutions in metastable equilibrium with respect to
calcium and phosphate ions. There is reason to believe, also, that
the knowledge thus gained of the TC macromolecule and of its poly-
merization properties may prove useful in the elucidation of con-
nective tissue abnormalities such as characterize certain of the
rheumatoid diseases, atherosclerosis, and aging processes.

The Tropocollagen Macromolecule

The existence of a discrete type of collagen monomer extractable
from a variety of connective tissues under appropriate conditions
has been amply confirmed b)' a variety of techniques. Since both
small-angle x-ray diffraction (Bear, 1952) and electron microscope
studies on native fibrous collagen indicated a highly ordered band
structure with a fundamental repeat of about 640 A, it was assumed
bv many that the monomer must have a length of this dimension.
The first indication that this was not the case came with the ob-
serxation that soluble collagen could be reconstituted into "long-
spacing forms," which exhibited axial periods of several times the
value characteristic of native fibrils (for summaries, see Schmitt
et al, 1955; Schmitt, 1959; Hodge, 1959b). More recent measure-
ments (Hodge and Schmitt, 1960) show very sharp distributions
around length values of about 2800 A for these long-spacing forms,
corresponding to four times the native period, and comparably
sharp distributions for native-type fibrils. This value is in good
agreement with the results of Boedtker and Doty (1956), who
found by physicochemical methods that the macromolecules of
soluble collagen are highly asymmetric stiff rods with dimensions
of about 14 A X 2900 A, and with those of Hall and Doty (1958),
who utilized an improved shadow-casting technique for direct
visualization of macromolecules in the electron microscope.

The large-angle x-ray diffraction pattern of collagen is one of its
most striking characteristics and has been interpreted ( see Rich and
Crick, 1958) as indicating the presence, over large regions of the
fibrous structure, of a three-stranded helical configuration of poly-
peptide chains which involves the sequence glycine-proline-hy-
droxyproline well known from degradative experiments, the whole
structure being stabilized by hydrogen bonding. The three-stranded
model is also supported indirectly by the observe ions of Doty and
Nishihara (1958) on the denaturation of soluble collagen. On de-

INTERACTION OF COLLAGEN MACROMOLEC U LES 21

naturation, the original macromolecules, of weight ca. 360,000, yield
one chain of weight ca. 120,000 and another of weight ca. 240,000
(see also Orekovitch and Shpikiter, 1958). According to Doty and
Nishihara, the larger of these components in calfskin collagen may
comprise two chains linked by an alkali-labile ester bond.

The physical properties of collagen solutions and the other ob-
servations already cited leave little doubt that the tropocollagen
macromolecule, ca. 14 A X 2800 A in dimensions, is indeed the
monomeric unit of the various forms of ordered aggregates and must
possess a high degree of configurational order over at least the major
part of its length. A striking property of native collagen macro-
molecules is their capacity to aggregate under appropriate condi-
tions to yield a number of highly characteristic ordered structures,
the type of pattern obtained depending on the environmental con-
ditions. These various ordered aggregation states are characterized
by the highly specific band patterns they exhibit when observed in
the electron microscope, especially if the contrast is enhanced by
treatment with an "electron stain" such as phosphotungstic acid
(PTA). The ability to form such ordered structures is lost if the
macromolecules are denatured by thermal or other means. It seems,
therefore, that the polar side-chains responsible for the banding
seen in the electron microscope must be in a rather precise stereo-
chemical array, which is maintained in the native macromolecule
by hydrogen-bonding between the constituent polypeptide chains,
and which is lost when these bonds are ruptured during denatura-
tion. As we shall see, the polar groups (both basic and acidic) are
located in clusters at discrete loci along the length of the TC macro-
molecule and are separated by regions apparently poor in or lacking
polar groups. It is likely that these non-polar regions are mainly
responsible for the characteristic large-angle x-ray diffraction pat-
tern of collagens. On the other hand, there seems to be no doubt
that the small-angle pattern is the direct result of the particular
distribution of the polar side-chain clusters (see Bear and Morgan,
1957). Chemical analytical and electron microscope studies on
collagen fibrils by Kiihn et al. (1957) and Kiihn (1958) indicate
that the basic and acidic residues are in the same loci (bands) in
the native collagen structure. Their work shows that, while the
lysine-bound PTA is easily washed out, that held by arginine is
firmly bound and is responsible for the characteristic band pattern
and, further, that the rather large phosphotungstate ion is bound

22 MACROMOLECULAR COMPLEXES

hv the mutual cooperation of several polar side-chains on c djacent
macromolecules. It may reasonably be postulated, therefore, that
bands will be observed only when the macromolecules retain their
native helical configuration and will not be observed in those regions
where the polypeptide chains are uncoiled or otherwise disordered.

Aggregation States of Tropocollagen

For reasons that will become clear after consideration of the
various types of band pattern encountered, the TC macromolecule
may be considered as an 'asymmetric" or "structurally polarized"
unit; i.e., the distribution of density along its length is such that
one end is alwavs clearly distinguishable from the other. The
macromolecule may therefore be represented formally as an arrow,
the head and tail being labeled A and B, respectively (e.g., Figs.
1-3). This kind of symbolic representation has proved to be useful
for indicating the packing of TC in the various ordered aggregation
states.

NATIVE
lillllillliiilltiiillllKJIIlll

Fig. 1. Diagrammatic representation of three probable modes of aggrega-
tion of the TC macromolecule which can be formed reversibly from dilute acid
solutions of TC under appropriate conditions. See text for description.

INTERACTION OF COLLAGEN M AC ROMOLEC U LES

23

Fig. 2. Band patterns of SLS-type aggregates obtained by addition of ATP
to an acid solution of calfskin collagen. In this ordered form, the TC macro-
molecules are packed in "parallel array," with like ends in register, and their
orientation is indicated by the arrows with ends labeled A and B. (a) Stained
with PTA. (b) Stained with cationic uranium. X60,000. (From Hodge and
Schmitt, 1960.)

A ^

«B

it . •

V.

Fig. 3. SLS-type aggregates, showing the close correspondence in the
location of staining loci in the two "fingerprints" despite considerable differ-
ences in staining intensities, (a) Stained with PTA. (b) Stained with cationic
uranium. X220,O0O. (From Hodge and Schmitt, 1960.)

24.

MACROMOLECULAR COMPLEXES

The simplest "crystalline" form of collagen is that which Schmitt
et al. (1953) termed segment long-spacing (SLS), and is readily
formed on addition of adenosine triphosphate (ATP) to a dilute
acid solution of TC, although segments can also be formed under
other conditions. In this aggregation state the macromolecules are
arranged in a parallel array with like ends and other features in
register (Figs. 1-3).-^' The asymmetric or "polarized" band pattern
observed when SLS is stained with PTA is thus, in effect, a "molec-
ular fingerprint," since it directly indicates the distribution and
concentration of basic side-chain groups ^ along the length of the
TC macromolecule. A comparable "fingerprint" of the acidic side-
chains (Fig. 2) may be obtained conveniently by staining the seg-
ments with cationic uranium under appropriate conditions (Hodge
and Schmitt, 1960).

The end regions of individual segments are generally somewhat
disordered, and a detailed comparison of the staining properties of

Fig. 4. Polymeric SLS-type aggregates obtained by addition of ATP to
collagen solutions subjected to sonic irradiation (see text). In these polymeric
forms, the distortions due to drying ore minimized, particularly for staining loci
near the ends of the TC macromolecules, and thus allow a better comparison
of the various band densities, (a) Stained with cationic uranium, (b) Stained
with PTA. XI 35,000. (From Hodge and Schmitt, 1960.)

•"' Unless otherwise indicated, all the illustrations have reference to preparations
of calfskin collagen.

" It should be understood that the actual band densities observed in the electron
micrographs represent both an "intrinsic electron density," resulting from the con-
centration of the bulky polar side-chains into clusters along the length of the TC
macromolecule, and an electron density contribution arising from the binding of
heavy metal ions.

INTERACTION OF COLLAGEN MAC ROMOL EC U L ES

25

bands in these regions is more conveniently obtained from observa-
tions on certain polymeric segment-type aggregates obtained from
solutions of collagen treated with sonic irradiation (see Fig. 4). It
is of interest that the two "fingerprints" thus obtained match each
other exactly in terms of the relative positions of the forty or so

I

Fig. 5. FLS-type fibril with an axial period of ca. 2800 A. The symmetrical
intraperiod band pattern corresponds with that expected for a packing of
protofibrils (linear polymers of TC) having no preferred polarization, i.e., in
antiparallel array, and with A-B junctions in approximate register. Stained
with PTA. X 33,000. (From Hodge, 1959b.)

-^Sr« ^ «!•»

-»^%

Fig. 6. Polymeric SLS-type fibrils produced by addition of ATP to an acid
solution, of collagen at relatively high pH (co. 5). In this form, the linear
polymers of TC aggregate with all like features in register, (a) Stained with
PTA. (b) Stained with cationic uranium. X70,000. (From Hodge and Schmitt,
1960.)

26

MACROMOLECULAR COMPLEXES

o.r

r

SiiUiiJiiB

7j2^ri''»t

Fig. 7. Reconstituted native-type fibril, showing the characteristically polar-
ized intraperiod band pattern. The axial period of ca. 700 A is indicated by
the row of dots, and the longitudinally staggered packing of the protofibrils
by the arrangement of arrows along the lower edge of the figure, (a) Stained
with PTA. XI 30,000. (From Hodge, 1959b.) (b) Indicates the currently ac-
cepted band nomenclature. X230,000.

bands observable along the length of the TC, but that many of these
bands differ markedly in relative intensities in the two fingerprints.
Hence, we appear to have direct evidence that polar groups, both
basic and acidic, are localized in narrowlv defined regions at distinct
and characteristic loci along the length of the TC macromolecules.
The narrowest bands observed are in the range 15-20 A wide. The
intensity differences observed also indicate that while basic groups

INTERACTION OF COLLAGEN MAC ROMO L EC U L ES

27

predominate in some of these loci, there is a relative parity or an
excess of acidic groups in other regions. The analysis of the detailed
band pattern of SLS enables one to determine in principle, at least,
the packing arrangement of the TC macromolecules in any of the
other ordered aggregation states.

If an acid solution of collagen is dialyzed against distilled water,
the final result is the production of a water-clear gel, consisting of

^-Tij: — I

B

'■m

Fig. 8. Composite micrographs illustrating the optical synthesis (b) of the
native-type band pattern, produced by multiple printing of the image of the
single segment shown in (a), stained with PTA. The recording plate was moved
a distance corresponding to V4 of the segment length in the axial direction
prior to each consecutive exposure. The summation of band densities resulting
from this overlap (b) corresponds closely to the band pattern of native-type
fibrils (c). X 250,000. (From Hodge and Schmitt, 1960.)

28 MACROMOLECULAR COMPLEXES

A < ^B

0»S/M

Fig. 9. Dimorphic ordered aggregate of TC, produced by exposing native-
type fibrils to a solution of TC containing ATP, under closely controlled condi-
tions of pH and ionic strength. Note that all of the segments so formed are
similarly polarized, as indicated by the labeled arrows, and are located
identically with respect to the intraperiod band structure of the native-type
fibril, which is itself polarized. Stained with PTA. X70,000. (From Hodge and
Schmitt, 1960.)

very fine filaments with no striation visible in the electron micro-
scope. The viscosity of the collagen solutions during such a dialysis
rises steadily as the pH rises (Hodge and Schmitt, 1960; Hodge
et al., 1960), indicating that end-to-end polymerization of the TC
is giving rise to linear polymers or protofibrils. However, Glimcher
and Bonar (personal communication) have shown that the small-
angle x-ray diffraction pattern of fibers drawn from such gels is
characteristic of the native (700 A-repeat) type of packing of the
TC. This means that the protofibrils are packing in staggered
array (see Fig. 7) to form the very thin fibrils of the gel. If the
dialysis is carried out in the presence of suitable concentrations of
certain substances such as serum glycoprotein (see Schmitt et al.,
1955), the TC comes out of solution in a second ordered fibrous
form, the so-called fibrous long-spacing (FLS) form (Fig. 5).
Since the axial period of FLS is about equal to the length of the
TC macromolecule and the distribution of bands within each period
is centrosymmetric, it seems likely that this type of structure arises
by side-to-side aggregation of protofibrils in antiparallel array, i.e.,
with no preferred "polarity" (Figs. 1, 5). This change in the inter-

INTERACTION OF COLLAGEN MAC ROMOLEC U LES

29

^f ?!- ^3 ^

3

C.C

o.

r

Fig. 10. Higher magnification view of a region in the same preparation as
shown in Fig. 9, to illustrate the precise correspondence of the bands in the
SLS-type structure with those in the native-type structure. As indicated in the
band nomenclature (proposed by Hodge and Schmitt, 1960), each band in
the native-type structure exhibits continuity with four "equivalent bands" of
differing intensity in the SLS-type structure, and its intensity is therefore a func-
tion of the intensities of all four of the corresponding SLS bands. Thus, for
example, /^ = (/c/4) {\h + \h + /a.) + U^, where /<; is the intensity of the 6 band
in the native-type structure, /«!, U-i, \h, and U^ are the intensities of the corre-
sponding "equivalent bands" in the SLS-type structure, and /c is a constant, the
value of which will depend on the degree of additivity of the band intensities.
The close correspondence in the relative band intensities obtained by direct
optical synthesis with those in the native-type fibril (Fig. 8) indicates that the
actual value of /c must be close to unity, as would be expected if the staining
intensity is proportional only to the concentration of polar side-chain groups.
X 21 5,000. (From Hodge and Schmitt, 1960.)

30

MACROMOLECULAR COMPLEXES

NATIVE

Qj d Qi 02 d Qi Qg d a I Qj

S L S

Fig. 11. Diagrammatic interpretation of the packing arrangements of the
TC macromolecules in the dimorphic ordered forms shown in Figs. 9 and 10.
To minimize complexity, only three of the twelve or so bands actually observable
in the native-type structure are depicted. Note that in both the SLS-type and
native-type structures, all staining loci are in accurate transverse alignment,
i.e., in register. However, in the SLS-type, only like features are in register, i.e.,
"homo-register," while in the native-type structure, all bands arise by align-
ment of the four corresponding "equivalent loci" of the TC macromolecule, i.e.,
"hetero-register." In the SLS band terminology proposed by Hodge and Schmitt
(1960), the superscripts refer to the subscripts in the generally accepted band
nomenclature for the native-type fibril, the subscripts to the four corresponding
"equivalent loci" in the TC macromolecule. Thus, for example, the Oi band of
the native-type fibril arises by juxtaposition of the a\, a^-j, a'.u and a'4 sites of
the TC macromolecule when they are placed in staggered array, obeying the
selection rule that adjacent protofibrils are displaced relative to one another by
a distance equal to Va of the length of the TC macromolecule in the axial direc-
tion. (From Hodge and Schmitt, 1960.)

action properties of the TC polymers is presumably the result of an
altered charge distribution following binding of the strongly nega-
tively charged glycoprotein at specific points of the TC. The pres-
ence of one exceptionally broad and dense band per period, and
relatively few and rather diffuse intraperiod bands, suggests that
most of the glvcoprotein is adsorbed near the A and B ends of the
TC. This would be in accord with the relatively high concentration
of basic groups present in these regions as deduced from the ap-
pearance of PTA-stained segments (Figs. 2, 3).

INTERACTION OF COLLAGEN M AC ROMOL EC U L ES

31

S °

\ ^ /\ t\\i

— -► Znn^ ^11111 ^5

p z

w

< u o

\?

mn?
i,i,^|.

o

f] o z

I o

5 5

T

1

11'

i i .

I'jl

i i i

1 1

ill

'l 1

<5

e°

E-6

u c
O OJ

E a;
I- o

c

§ °

•- (1)

:^ ^

2^

•- "oj

u ■"

c a;

-c 00

"C Q) _*

0 — t

o -c

E -D "O
c c

OJ u ^

t; T "o
2 03 o

t; E ^

= i5 E
— ; o

o ;^ii:

n D <n
^ ^-
U

O

^ ^ o

'^ E b

0) (u

32

MACROMOLECULAR COMPLEXES

If the water dialysis is carried out with lower concentrations of
the glycoprotein, a mixture of FLS and native-type ' fibrils is ob-
tained (Highberger et al., 1951). At still lower concentrations, only
native-type fibrils are obtained. The system glycoprotein-collagen
thus nicel\- illustrates the manner in which binding of a second
component modifies the charge distribution at a given pH and ionic
strength and changes the type of ordered interaction favored.

Fig. 13. SLS-type aggregate of whole polymeric type from a solution
sonically irradiated 240 min, showing end-to-end linkages of the type A'-A'
and B'-B'. Stained with PTA. X 100,000. (From Hodge, 1959b.)

J^,//l

Fig. 14. SLS aggregates from a solution sonically irradiated 240 min,
showing formation of ribbon-like fragment monomeric forms of the type A'C.
Stained with PTA. X68,000. (From Hodge and Schmitt, 1958.)

■^ The term "native-type" refers to the noniial aggregation state of TC in vivo, i.e.,
with an axial period of ca. 700 A.

INTERACTION OF COLLAGEN MAC ROMOLEC U L ES 33

Under other conditions, and particularly in the presence of ATP,
the charge distribution on the protofibrils is such as to favor a
"polymeric segment" type of structure, i.e., one in which all like
features of the TC are in register across the fibril (Hodge and
Schmitt, 1960). In this type of structure (Figs. 6, 23), which we
have termed "fibrous-type segment long-spacing" (F-SLS) to dis-
tinguish it from the antiparallel FLS just described, the axial period
is equal to the length of the TC macromolecule, and the polarized
band pattern is equivalent to that obtained by placing single seg-
ments end-to-end in the sense AB-AB-AB-. If suitably oriented
fibers can be obtained from such preparations, it should be possible
to obtain direct evidence concerning the precise distribution of
density along the TC macromolecule by means of small-angle x-ray
diffraction. Such data cannot be obtained by x-ray analysis of
native-type collagen fibers, in which the TC macromolecules are
packed in a staggered fashion with respect to their neighbors.

The ordered aggregation form of TC that is stable under physio-
logical conditions and that occurs in various connective tissues will

Fig. 15. SIS aggregates of the fragment dimeric type CA'-A'C from a
solution sonically irradiated 240 min, showing extensive lateral aggregation in
the form of "ribbons." Stained with PTA. X 55,000. (From Hodge and Schmitt,
1958.)

34

MACROMOLECULAR COMPLEXES

^f Is

Fig. 16. (See next page for caption.)

c

&

,B'

A'A'

i f' \

"^

Fig. 17. (See next page for caption.)

INTERACTION OF COLLAGEN M AC ROMOLEC U LES 35

])e referred to as the native tvpe. This form mav be reconstituted
from sohitions of TC under appropriate conditions (see Schmitt
cf al., 1955), has a fundamental axial period of about 700 A, i.e.,
one-fourth the length of the TC macromolecule, and the intraperiod
liand distribution observed after electron staining is characteris-
tically asymmetric or "polarized" (Figs. 1, 7). As indicated diagram-
matically in these figures, the formation of native-type collagen
fibrils involves packing of protofibrils (linear polvmers of TC) so
that neighboring protofibrils are displaced longitudinally in relation
to one another by one-fourth of the TC length. That this must be
the case is readily apparent from the facts that ( a ) the axial period
of native-tvpe collagen is one-fourth that of the length of individual
macromolccules and (b) the additive sum of any asymmetric den-
sity distribution of length L displaced longitudinally and consecu-
tively by a distance L p, where p is an integer, results in the gen-
eration of an asymmetric density function of like "polarity" and
having an axial period of L/p. For native-tvpe collagen, p — 4. It
has been shown that other values of p pertain to other fibrous pro-
teins, e.g., paramyosin (Hodge, 1959a). In principle, it should be
possible to reconstruct the salient features of the band pattern of
native collagen bv a process of optical svnthesis from the band pat-
tern of SLS, and, in fact, such syntheses have been carried out in
this laboratory' (Fig. 8), with resolution limited only by the uncer-
tainties in the absolute positions of the various SLS bands intro-
duced by disordering during drying of the specimens for examina-
tion in the electron microscope.

However, a more definitive proof of the correctness of the stagger
theory has been forthcoming in recent work (Hodge and Schmitt,
1960). In these experiments, fibrils of native type (700 A-repeat)
were placed in an environment containing monomeric TC under
conditions favoring the formation of SLS-type aggregates. It was
found that the TC macromolccules in the surfaces of the native-type
fibrils had apparently acted as "nuclei" for the growth of individual
segments (Fig. 9). In the resultant dimorphic ordered forms, which

Fig. 16. Whole polymeric SLS form of sonicolly irradiated calfskin collagen,
showing isomorphous growth of fragment dimeric ribbons of the type CA'-A C.
Stained with PTA. X 74,000. (From Hodge and Schmitt, 1958.)

Fig. 17. Whole polymeric SLS form of sonicolly irradiated calfskin col-
lagen, showing isomorphous growth of fragment dimeric ribbons of the type
CB'-B'C. Stained with PTA. X 100,000. (From Hodge, 1959b.)

36

MACROMOLECULAR COMPLEXES

allow a direct comparison of the band patterns of both the segment
and native forms, it is observed that each band in the native-type
fibril exhibits a direct transverse continuity with one of four differ-
ent bands of the segment pattern (Figs. 10, 11), each of different
intensity and separated longitudinally from one another by L/4,
where L is the length of the TC macromolecule. It is clear, there-
fore, that the density observed in any particular band of the native-
t\pe structure represents a summation of the densities of the ap-
propriate four "equivalent bands" in the segment-type structure.
It should be noted that, although the bands observed in SLS,
whether stained for basic or acidic side-chains, obey this selection
rule (i.e., may be classified in groups of four "equivalent bands"
spaced L/4 apart), this does not imply that the TC macromolecule
consists of four identical subunits. In any case, the physical prop-
erties of collagen solutions and the distribution of density in the

Fig. 18. Composite micrograph for comparison of the band patterns of
whole monomeric forms (single segments, a and c) with that of the whole poly-
meric form (b). The use of two single segments is necessary since the band
pattern is often obscured at one or both ends by disorder resulting from drying
of the specimen. Arrow 1 indicates the new band observed in the B'-B' junc-
tions of polymeric SLS forms, which is not present in single segments, and
apparently is the result of end-to-end polymerization of the TC macromolecules.
Calfskin collagen. Stained with PTA. X 120,000. (From Hodge and Schmitt,
1958.)

INTERACTION OF COLLAGEN MAC ROMOLEC U LES 37

SLS Ixind pattern suffice to prove that the 2800-A macromolecular
unit is the true monomeric form.

Electron microscope examination of SLS-type aggregates ob-
tained from solutions of TC subjected to sonic irradiation has served
to reveal some features of interest, particularly in relation to the
end-to-end polymerization properties of TC. The physicochemical
measurements of Nishihara and Doty (1958) indicated that sonic
irradiation fragmented the TC macromolecules into shorter pieces
which, however, largely retained the helical configuration of the
native macromolecules. They concluded that the time-dependence
of the molecular weight change was compatible with a preferential
fragmentation of the macromolecules into halves and quarters.
Electron microscope examination of SLS-type aggregates produced
from such sonicated solutions confirmed this conclusion in general
(Hodge and Schmitt, 1958) and, in addition, showed that sonic
irradiation profoundlv modifies the end-to-end interactions of the
TC.

The main effects of sonic irradiation on the TC macromolecule
are illustrated diagrammatically in Fig. 12. One of the most im-
portant is a rapid alteration of "end-regions" without apparent
change in length, the modification being manifested as a change in
the ability of the TC to engage in end-to-end polymerization. This
is indicated in Fig. 12 by labeling the ends A' and B' rather than
A and B. An important consequence of this modification is an
impairment of the ability to form fibrils of native type, i.e., an in-
hibition of end-to-end polymerization of the type AB-AB-AB- in-
volved in the protofibril formation. Another result of sonic irradi-
ation is a pronounced effect on the characteristics of the SLS-type
precipitates formed on addition of ATP. The SLS-type precipitates
derived from unirradiated control solutions usually consist largely
of single segments with relatively few dimeric or trimeric forms, the
latter normally exhibiting end-to-end linkages of type A-B. How-
ever, the irradiated solutions yield progressively increasing amounts
of dimeric and polymeric forms (see Fig. 12 for explanation), in-
volving end-to-end linkages of the type A'-A' and B'-B' (Fig. 13).
Thus, in addition to fragmentation, the effects of sonication are ap-
parently twofold: (a) impairment of end-to-end interactions of the
A-B type under conditions which normally favor the formation of
protofibrils and fibrils of native type and (b) enhancement of end-
to-end linkages of type A'-A' and B'-B' under circumstances which

38

MACROMOLECULAR COMPLEXES

Fig. 19. (See next page for caption.)

Fig. 20. (See next page for caption.

INTERACTION OF COLLAGEN MAC ROMOLEC U L ES 39

usually yield single segments (whole monomelic type) together
with some polymeric forms of type A-B.

With longer times of irradiation, scission of the macromolecules
becomes evident in SLS-type precipitates (Figs. 14-17). Fragmen-
tation appears to occur initially near the middle of the TC macro-
molecule, yielding fragments A'C and B'C about 55 per cent and
45 per cent, respectively, of the original TC length. These are ca-
pable of forming monomeric (Fig. 14) and dimeric (Fig. 15) SLS-
type aggregates, and of isomorphous growth on whole polymeric
SLS (Figs. 16, 17). However, dimerization or polymerization of
the scission products apparently never involves new ends resulting
from fragmentation; i.e., end-to-end linkages of the type C-C have
not so far been observed. It would appear, therefore, that the
original ends of the TC macromolecules have properties which are
not duplicated in the new ends formed by scission. Extensive
irradiation results in still smaller fragments of the type indicated by
breakage at points D and E in Fig. 12. Of the four scission products
presumably yielded by this process, only those which include A' or
B' ends have been observed to form ordered SLS-type structures.

Detailed comparison of the band pattern of single segments
(whole monomeric SLS form) with that found in the whole poly-
meric form resulting from sonic irradiation has led to a hypothesis
concerning the mechanism by which TC macromolecules engage
in end-to-end linkage to form protofibrils (Hodge and Schmitt,

Fig. 19. At top is depicted a model of the TC macromolecule in solution,
the three polypeptide chains being represented by wires (block, gray, and
white, respectively). Within the body of the macromolecule, the three wires
are coiled about one another in on orderly helical pattern, representing the
type of structure deduced from large-angle x-ray diffraction studies of collagen.
At the ends, the postulated end-chains are shown as single, randomly coiled,
polypeptide chains. Note that parameters, such as the length and diameter of
the macromolecule, the pitch of the helix, and the lengths of the end-chains,
are not to scale. In the center is illustrated the heterologous type of end-to-end
coiling envisaged in the formation of protofibrils during normal polymerization
(end-to-end linkage of type A-B), while at the bottom is shown the homologous
type of end-chain coiling presumed to occur during formation of whole poly-
meric SLS-type aggregates, such as those shown in Figs. 13, 16, 17. (From
Hodge, 1959b.)

Fig. 20. The same model of the TC macromolecule as shown in Fig. 19,
illustrating the fragmentation by sonic irradiation (center), and the formation
of fragment dimeric forms CA -A'C and CB'-B'C (at bottom) on the addition
of ATP. Dimeric and polymeric forms involving junctions of type C— C have not
been observed. (From Hodge, 1959b.)

40 MACROMOLECULAR COMPLEXES

1958). In the A -A' junction (Fig. 18) of the polymeric form, the
two bands corresponding to the A-terminal band of the monomeric
SLS form appear as a doublet separated by an interband region
about 100 A wide. Similarly, in the B'-B' junction, the two B-ter-
minal bands are spaced about 180 A apart, and a new band ( arrow
1 in Fig. 18), not observed in monomeric SLS, is seen in the middle
of the junctional regions. If it is assumed that well-defined bands,
as discussed earlier, occur only in those regions where the poly-
peptide chains are in an orderly helical configuration, thus provid-
ing the specific array of polar side-chains necessary for the coopera-
tive binding of ions such as PTA, these observations suggest the
following conclusions (see Figs. 19, 20):

1. The TC macromolecule, the bulk of which is presumablv in
the characteristic three-strand helical configuration deduced from
x-ray diffraction studies, possesses speciahzed "end structures"
which are most probably in the form of short dangling chain ap-
pendages with maximum lengths of about 100 A and 180 A for the
A and B ends, respectively. Such terminal peptide chains would
most likely adopt a random coil configuration in solution, as indi-
cated diagrammatically in Figs. 19 and 20.

2. The appearance of a new band in the center of the B'-B' junc-
tions of whole polymeric SLS forms suggests that end-to-end poly-
merization of TC involves an orderly coiling of these terminal chains
about one another to form a junctional region with an ordered heli-
cal structure, stabilized perhaps bv hvdrogen bonds. It should be
noted that this is the simplest hypothesis. However, the possibility
is not excluded that at one end there may be two polypeptide chains,
in which case the formation of polymers involving linkages of the
type A-B would result in the formation of a three-stranded struc-
ture in the junctional regions. Such a structure would pose a prob-
lem in finding a structural basis for end-to-end interactions of
homologous type ( A'-A' or B'-B' ) .

On the basis of the picture just outlined, the presence of a new
band in the B'-B' junction and the absence of such a band in the
A'-A' junction are indicative of differences in amino acid composi-
tion, particularly of polar groups, between the A and B end-chains.
Indeed, it seems likely that the specificities of the terminal chains
reside in just such differences of amino acid composition and per-
haps to some extent depend on the presence of other components
such as carbohydrate. The marked changes in the end-to-end inter-

INTERACTION OF COLLAGEN MACROMOLECULES 41

action properties of TC resulting from sonic irradiation are pre-
sumably due to some modification of one or both of these special-
ized end-structures, possibly the splitting off of a peptide or other
small fragment.

i
■/it

m\ >

o,r^

B'B'

A'B'

A'A'

Fig. 21. Whole polymeric SLS form from o solution sonically irradiated
480 min, showing the detailed band structure within an A'-B' junction, together
with the more commonly observed A'-A' and B'-B' junctions. Stained with PTA.
XI 20,000. (From Hodge, 1959b.)

Occasional junctions of the type A'-B' have been observed in
polymeric SLS precipitates obtained from sonicated solutions of TC
(Fig. 21), and their band structure agrees well with that to be ex-
pected from an analysis of A'-A' and B'-B' junctions. The A'-B'
junctions have also proved useful in the detailed analysis of the
native-type (700 A-repeat) band pattern in terms of the SLS pattern.

Effects of Proteases on Tropocollagen

The effects of proteolytic enzymes such as trypsin, chymotrypsin,
and pepsin on the polymerization properties of tropocollagen macro-
molecules have been investigated in a recent series of experiments
(Hodge et ah, 1960). The results to date have been entirely con-
sistent with the end-chain hypothesis alreadv described.

The data indicate that many solutions of "soluble collagen" are
relatively heterodisperse in that they contain variable amounts of
linear dimers, trimers, and higher polymers in addition to large
numbers of monomeric TC macromolecules. As indicated by the
viscosity data mentioned earlier, the proportion of polymers in-
creases as the pH is raised on the acid side of the isoelectric point.
The gross macroscopic appearance of the precipitates obtained on

42

MACROMOLECULAR COMPLEXES

r

20 30

IN MINUTES

Fig. 22. Graph to illustrate the efFect of two proteolytic enzymes (trypsin,
Worthington, 2X recrystallized, and collagenase, Worthington) on the viscosity
of a typical solution of calfskin collagen in tris buffer, pH 6.9, in the presence
of 0.5M CaClj at 20 C. The enzymes were dissolved in the buffer solution and
the zero points obtained by measuring the viscosity of an aliquot of the collagen
solution when diluted with a volume of buffer equal to that used in adding the
enzyme. Note that while collagenase rapidly reduces the viscosity to a low
value, the effect of trypsin is a small but rapid reduction of the viscosity to a
high steady value and appears to be dependent on the enzyme concentration.

addition of ATP is apparently a reliable criterion of the degree of
homogeneity of such solutions; when appreciable amounts of linear
polymers are present, addition of ATP produces precipitates which
are of a flocculent, fibrous character and do not settle even on
prolonged standing. These macroscopic characteristics are sup-
ported by electron microscope observations which consistently show
"strings" of segments, apparently arising from the "crystallization"
of TC macromolecules on the linear polymers present in the solu-
tion. On the other hand, the addition of ATP to acid collagen solu-

INTERACTION OF COLLAGEN MACROMOLEC U LES

NATIVE TYPE

43

F-SLS

Fig. 23. In the center is depicted a typical acid solution of collagen, con-
taining TC monomers together with some linear dimers, trimers, and other
n-mers. Under suitable conditions (e.g., dialysis against 1 per cent NaCI), the
solution yields native-type fibrils (top center), while the addition of ATP gives
rise to a polymeric SLS-type precipitate resulting from the growth of individual
segments on the pre-existing linear polymers of TC. At left is shown the linear
polymerization resulting from dialysis against distilled water, ultimately leading
to the formation of a water-clear gel consisting of very thin fibrils of native
type (top left). The addition of ATP during late stages of the dialysis leads to
the formation of polymeric SLS-type structures (F-SLS at bottom left). At right
is illustrated the depolymerizing action of proteases such as pepsin (see Fig. 22).
Subsequent dialysis against distilled water does not induce end-to-end poly-
merization (right center and top right), and the addition of ATP produces only
single segments, i.e., whole monomeric forms (bottom right).

tions after treatment with either trypsin or pepsin (at appropriate
pH values ) produces a non-fibrous precipitate, eventually settling to
a dense mass at the bottom of the tube. When examined in the
electron microscope, such preparations exhibit only single segments,
with very few dimeric or higher polvmeric forms. It would seem,
therefore, that a hitherto unrecognized but important action of these
proteolytic enzymes on collagen in solution is a depolymerization
of protofibrils without apparent impairment of the three-stranded
helical structure of the individual macromolecules. This interpreta-
tion is supported by observations on the physical properties of col-
lagen solutions during attack by pepsin or trypsin. On exposure to
pepsin or trypsin (under suitable conditions of pH, temperature,
and ionic strength), the viscosity of collagen solutions undergoes a
small' but rapid decrease ( see also Gross, 1958 ) to a still high value
consistent with the presence of TC monomers in the solution (Fig.
22); the optical rotatory properties of the solution do not change

44

MACROMOLECULAR COMPLEXES

appreciably during this process. In contrast to this, exposure to
collagenases causes both the optical rotation and the viscosity to
decrease rapidly (Seifter et al., 1958).

These results are compatible with the hypothesis that the enzy-
matic attack of pepsin and trypsin on collagen is confined to a
limited region at the ends of the TC macromolecule, yielding modi-
fied macromolecules whose length and internal helical configuration
remain essentially unaltered, but whose end-to-end polymerization
properties have been drastically modified (Fig. 23). The enzyme-
treated solutions yield single segments (whole monomeric type)
which are normal as judged by their band pattern on staining with
PTA. In contrast to this, the same collagen solutions fail to give
ordered aggregates of segment type after treatment with collage-
nase. Treatment with pepsin or trypsin also prevents the formation
of a water-clear gel on dialysis against distilled water, and almost

I -CONTROL: NO TRYPSIN
n -0.01% TRYPSIN
m -0.05% TRYPSIN
12 - 0.1 % TRYPSIN

0.8

0.6

0.4

0.2

40 60

MINUTES

Fig. 24. The efFect of treatment with trypsin (Worthington, 2x recrystal-
lized) on the thermal gelation of a 0.1 per cent solution of calfskin collagen
(tris buffer, pH 7, with 0.45M NaCI) at 34° C, after incubation with the
enzyme 6 hr at 20 C, as determined by the optical density at 430 m/x.

INTERACTION OF COLLAGEN MAC ROMO L EC U L ES

45

completely eliminates the production of native-tvpe fibrils on dial-
ysis against 1 per cent NaCl. The phenomenon of "thermal gela-
tion" of neutral salt solutions of collagen investigated extensively by
Gross (1956, 1959) and by Bensusan and Hoyt (1958) is 'also
severely inhibited or abolished if the collagen solutions are in-
cubated with trypsin at 20° C prior to incubation in the tempera-
ture range 30°-37° C. The effect of the enzyme is illustrated in
Fig. 24. That the attack of pepsin and trypsin is largely, if not
entirely, confined to the "end regions" of intact molecules is also
suggested by the following result. When a sonicated solution of TC,
which normally yielded a large proportion of whole polymeric and
fragment dimeric SLS-type aggregates on addition of ATP (Figs.
13, 15-17 ) , was treated with pepsin at pH 3-4, prior to the addition
of ATP, it yielded only whole and fragment monomeric SLS forms
(Fig. 25). '

It seemed likely from these data that the result of protease action
on the TC macromolecule is the profound alteration, or more prob-

Fig. 25. A typical field in a preparation of SLS-type aggregates produced
by addition of ATP to a solution of sonically irradiated collagen following
incubation with pepsin. This solution, which, in the absence of enzyme treat-
ment, characteristically yielded SLS-type structures of whole polymeric and
fragment dimeric type (Figs. 13, 15-17), gave, after treatment with enzyme,
only ordered aggregates corresponding to whole and fragment monomeric
types, indicating that the enzyme inhibits end-to-end linkages of the TC macro-
molecules. Stained with PTA. X 90,000.

46

MACROMOLECULAR COMPLEXES

ably the splitting off, of one or more peptides of the end regions,
perhaps one or both of the hypothetical "end-chains," or portions
thereof. Efforts were therefore slanted toward the isolation and
characterization of any such peptide products, utilizing chiefly the
techniques of paper-curtain electrophoresis and two-dimensional
paper chromatography. Preliminary results have been reported re-
cently (Hodge et al, 1960). In eyaluating possible methods for
detecting and isolating products of protease action, we were guided
by several considerations. In the first place, it seems to be generally
accepted that the most highly purified collagens still contain a small

1900
u '^00

1 1500

a: I 300

lij

a.

I 100
(/)

§ 900

o

o

2 700

t 500
>

t 300
<

100

-A- 1 NATIVE COLLAGEN
WITH TRYPSIN

•B-l DEGRADED
COLLAGEN WITH
TRYPSIN

-C-i TRYPSIN
ALONE

4 8 12 16 20 24 28 32 36 40

FRACTION I n nr Ez: 2: 21 3ni

DISTRIBUTION OF l'"

Fig. 26. The distribution of P*^^ activity in the effluent from paper-curtain
electrophoresis at pH 4.9 in pyridine-acetic acid buffer for (A-1) native calf-
skin collagen in solution at pH 7 (tris buffer with 0.5M CaCL.) incubated with
trypsin (Worthington, 2X recrystallized) 2 hr at 20° C, followed by iodination;
(B-l) skin collagen after thermal denaturation and tryptic digestion following
the procedure of Grassmann ef al. (1956), followed by iodination; and (C-1)
trypsin alone (Worthington, 2x recrystallized) after incubation at 20° C in the
same medium as used for Run A-1, followed by iodination. A control run with
native collagen alone gave no significant activity in the region of Fraction I.
It can be seen that the peak of activity labeled Fraction I on the acid side of
the diagram appears to arise by the action of trypsin on both native and
denatured TC. Input at center. (From Hodge ef al., 1960.)

INTERACTION OF COLLAGEN MAC ROMOL EC U L ES

47

but definite amount of tyrosine (Highberger, 1956), and the ability
of the TC to form highly ordered fibrous structures is lost if the
purification treatment is such that the tyrosine is removed. In them-
selves, these results suggest that the tyrosine is located in relatively
accessible regions of the macromolecules, perhaps at the ends. This
possibility was given further force by the observation of Bensusan
( 1959 ) that iodine markedly accelerates the thermal gelation of
neutral salt solutions of collagen, especially in view of the fact that,
with the exception of histidine which reacts much more slowly with

/>/t ¥.tr

7A-

a, I ?00 -

Run B-l

Denatured collagen + Trypsin^ pH 4.6

1 00 0_

-f

l'^' Distribution
i

Boq_

-

... M

600_

-

f

i.QO_

_

r-

J V

}00_

' — 1 1 Y\ —

/

I

r

1

I

o

"5/

_ *^

2 4 6 8 10 f2 14 (6 18 20 22 24 26 28 30 32 34 36 38 40

Tabe number

Fig. 27. The distribution of 1^-^' activity in paper-curtain electrophoresis of
denatured collagen incubated with trypsin (Run B-l of Fig. 26), together with
the relative specific activities of the various fractions indicated in Fig. 26. At
top is shown the electropherogram of the same run after staining, first with
ninhydrin (positive reaction indicated by N), then with amido-black (A). Note
that Fraction I escaped detection by staining with ninhydrin and amido-black,
but has by far the highest specific activity of the various fractions examined.
(From Hodge ef a/., 1960.)

48 MACROMOLECULAR COMPLEXES

iodine, tyrosine is the only amino acid in collagen capable of iodina-
tion. A further pointer in this direction was the report by Grass-
mann et al. (1956) of a tyrosine-containing peptide, the Sf peptide,
in tryptic digests of thermally denatured collagen.''

In view of the above considerations, it seemed desirable to follow
the distribution of tyrosine in paper-curtain electrophoresis of tryp-
sin-treated collagen by iodination of the digest with iodine contain-
ing the radioactive isotope P'^V Figure 26 shows the distribution of
radioactivity observed in electropherograms of a control run with
trypsin alone, a run with trypsin and native ( undenatured ) collagen

P/^c

So/k ■ ??. ^u^e /y. SQ

m

:^-

^^^^^^

Is.

#

7^v

/iJa^ h^

no

%A:^ '

^mi/-

^7 ^ys "^^%^

2).C.^TCB)

^ 7=>^

e^oC-o- Cre so e ('/■■/) X

Fig. 28. Two-dimensional paper chromatogram of an acid hydrolysate of
Fraction I from Run B-1 shown in Figs. 26, 27, run in phenol-o-cresol (1:1)
and n-butanol-acetic acid-H^O. Note the relatively large amounts of tyrosine
(Tyr) and mono-iodo-tyrosine (MIT). (From Hodge ef o/., 1960.)

* Recent results of Grassmann (personal communication) indicate that the Sf pep-
tide arises as a result of chymotrypsin activity (an impurity in the trypsin used in
previous experiments ) rather than from the activity of trypsin itself.

INTERACTION OF COLLAGEN MAC ROMOLEC U LES 49

in solution, and a run with trypsin and collagen denatured according
to the methods described by Grassniann et al. (1956). Although the
pattern is complex, it can be seen that for both the native collagen
plus trypsin ( NC + T ) and denatured collagen plus trypsin
( DC + T ) runs an acidic peptide fraction ( Fraction I ) , not present
in either of the control runs, was obtained. Figure 27 shows the
electropherogram for the run with denatured collagen plus trypsin,
together with the distribution of radioactivity and an indication of
the relative specific activities (cpm/mg) found for the various frac-
tions. The high specific activity found for Fraction I indicated that
this peptide fraction must be rich in tyrosine, and this conclusion
was borne out by two-dimensional paper chromatography of an acid
hydrolysate of Fraction I (Fig. 28). Inspection of the chromato-
gram shows, as would be expected, a preponderance of aspartic and
glutamic acids and very little of the basic amino acids among the
polar residues. Fraction I also contains some hydroxyproline and
a considerable amount of tyrosine, some of which is present in the
hydrolysate as the mono-iodo-compound, the latter spot being radio-
active. Unfortunately, the yield of Fraction I was too small to allow
further fractionation, but experiments are under way to accumulate
a sufficient quantity of this fraction for a thorough characterization
of the peptides.

In summary, it can be said that the effect of proteases such as
trypsin and pepsin on the TC macromolecule appears to be a rela-
tively slight modification, probably in "end regions," which, while
leaving intact the helical configuration of the polypeptide chains
comprising the macromolecule, drastically modifies its end-to-end
polymerization properties. These changes in the "reactivity" of the
macromolecule are accompanied by the liberation of a peptide or
peptides relatively rich in tyrosine. The results to date are com-
pletely compatible with the "end-chain" mechanism for end-to-end
polymerization of TC and suggest that the small tyrosine content
of collagen may be importantly involved in the mechanism by which
such "end-chains" interact in orderly fashion during polymerization.

References

Bear, R, S. 1952. The structure of collagen fibrils. Advances in Protein Chem. 7:

69:160.
Bear, R. S., and R. S. Morgan. 1957. The composition of bands and interbands of

collagen. In R. E. Tunbridge (ed.). Connective Tissue. Blackwell Scientific

Publications, Oxford. Pp. 321-333.

50 MACROMOLECULAR COMPLEXES

Bensusan, H. B. 1959. Effect of iodination of procollagen on the formation of fibers.

Federation Proc. 18: 190.
Bensusan, H. B., and B. L. Hoyt. 1958. The effect of various parameters on the

rate of formation of fibers from collagen solutions. /. Am. Chem. Soc. 80:

719-724.
BoEDTKEB, H., and P. Doty. 1956. The native and denatured states of soluble

collagen. /. Am. Chem. Soc. 78: 4267-4280.
Doty, P., and T. Nishihara. 1958. The molecular properties and thermal stability

of soluble collagens. In G. Stainsby (ed.). Recent Advances in Gelatin and Glue

Research. Pergamon Press, Ltd., London. Pp. 92-99.
Glimcher, M. J., A. J. Hodge, and F. O. Schmitt. 1957. Macromolecular aggrega-
tion states in relation to mineralization: The collagen-hydroxyapatite system as

studied in vitro. Proc. Nat. Acad. Sci. U.S. 43: 860-867.
Grassmann, W., K. Hannig, H. Endres, and A. Riedel. 1956. Aminosauresequen-

zen des Kollagens. I. Zur Bindungsweise der Prolins und Hydroxyprolins. Hoppe-

Seyler'sZ. phijsiol. Chem. 306: 123-131.
Gross, J. 1956. The behavior of collagen units as a model in morphogenesis. /.

Riophijs. Biochem. Cytol. 2 (Suppl.): 261-274.
Gross, J. 1958. Studies on the fomiation of collagen. I. Properties and fractionation

of neutral salt extracts of normal guinea pig connective tissue. /. Exptl. Med.

107: 247-263.
Gross, J. 1959. On the significance of the soluble collagens. In I. H. Page (ed.).

Connective Tissue, Thrombosis and Atherosclerosis. Academic Press, Inc., New

York. Pp. 77-95.
Gross, J., J. H. Highberger, and F. O. Schmitt. 1954. Collagen structures con-
sidered as states of aggregation of a kinetic unit. The tropocollagen particle.

Proc. Nat. Acad. Sci. U. S. 40: 679-688.
Hall, C. E., and P. Doty. 1958. A comparison between the dimensions of some

macromolecules determined by electron microscopy and by physical chemical

methods. /. Am. Chem. Soc. 80: 1269-1274.
Highberger, J. H. 1956. The chemical structure and macromolecular organization

of the skin proteins. In F. O'Flaherty, W. T. Roddy, and R. M. Lollar (eds.).

The Chemistry and Technology of Leather. Reinhold Publishing Corp., New

York. Pp. 65-193.
Highberger, J. H., J. Gross, and F. O. Schmitt. 1951. The interaction of muco-

protein with soluble collagen: An electron microscope study. Proc. Nat. Acad.

Sci. U. S. 37: 286-291.
Hodge, A. ]. 1959a. Fibrous proteins of muscle. Revs. Modern Phys. 31: 409-425.
Hodge, A. J. 1959b. Principles of ordering in fibrous systems. In W. Bargmann et

al. (eds.). Proc. Fourth Intern. Congress on Electron Microscopy, Berlin, 1960.

Springer Verlag, Berlin.
Hodge, A. J., J. H. Highberger, G. G, J. Deffner, and F. O. Schmitt. 1960. The

effects of proteases on the tropocollagen macromolecule and on its aggregation

properties. Proc. Nat. Acad. Sci. U. S. 46: 197-206.
Hodge, A. J., and F. O. Schnhtt. 1958. Interaction properties of sonically frag-
mented collagen macromolecules. Proc. Nat. Acad. Sci. U. S. 44: 418-424.
Hodge, A. J., and F. O. Schmitt. 1960. The charge profile of the tropocollagen

macromolecule and the packing arrangement in native-tvpe collagen fibrils. Proc.

Nat. Acad. Sci. U. S. 46: 186-197.
KiJHN, K. 1958. iJber die Ausbildung einer hochunterteilten Querstreifung des

Kollagens nach Gerben mit Phosphorwolframsaure und basischen Chromsalz-

losungen. Leder 9: 217-222.
KiJHN, K., W. Grassmann, and V. Hofmann. 1957. Ober die Bindung der Phos-
phorwolframsaure im Kollagen. Naturwiss. 44: 538-539.
Nishihara, T., and P. Doty. 1958. The sonic fragmentation of collagen macro-
molecules. Proc. Nat. Acad. Sci. U. S. 44: 411-417.
Orekovitch, V. N., and V. O. Shpikiter. 1958. Procollagens. Science 127: 1371-

1376. ^

INTERACTION OF COLLAGEN MACROMOLECU LES 51

Rich, A., and F. H. C. Crick. 1958. The structure of collagen. In G. Stainshy

(ed.). Recent Advatices in Gelatin and Glue Research. Pergamon Press, Ltd.,

London. Pp. 20-24.
ScHMiTT, F. O. 1959. Interaction properties of elongate protein niacromolecules

with particular reference to collagen ( tropocollagen ) . Revs. Modern Phys. 31:

349-358.
ScHMiTT, F. O., J. Gross, and J. H. Highberger. 1953. A new particle type in

certain connective tissue extracts. Proc. Nat. Acad. Sci. U. S. 39: 459-470.
ScHMiTT, F. O., J. Gross, and J. H. Highberger. 1955. States of aggregation of

collagen. Symposia Sac. Exptl. Biol. 9: 148-162.
ScHMiTT, F. O., and A. J. Hodge. 1960. Das TropokoUagen-Makromolekiil und die

Eigenschaften seiner geordneten Aggregationsformen. Leder. 11: 74-91.
Seifter, S., p. M. Gallop, and E. Meilman. 1958. The degradation of ichthyocol

by bacterial collagenase. In G. Stainsby (ed.). Recent Advances in Gelatin and

Glue Research. Pergamon Press, Ltd., London. Pp. 164-172.

DISCUSSION
W. D. McElroy, A. J. Hodge, W. H. Johnson, C. L. Prosser, F. G. Sherman

Dr. McElroy (Johns Hopkins University): Could you describe the ATP
eflFect in greater detail? Does inorganic pyrophosphate or triphosphate work?
What does the addition of Mg+ + do to this process?

Dr. Hodge: ATP is not unique in its abiHty to produce SLS-type aggre-
gates from acid solutions of collagen; ITF and nucleic acids will also yield
segment-type structures under appropriate conditions. In fact, the SLS form
was first discovered by Schmitt et at. (1953) when they dialyzed a phosphate
extract of skin collagen (pH = 8, /t = 0.4) against citrate buffer (pH 4,
IX = 0.2). Such extracts show strong absorption at about 2600 A. The sodium
salts of ATP and ITP do not work, nor does inorganic pyrophosphate or tri-
phosphate. These results suggest that the binding of ATP to the TC macro-
molecules modifies the charge distribution in a way such that (a) the iso-
electric point of the TC is lowered and (b) the parallel, in-register SLS-type
of packing is favored over the normal staggered configuration of native-type
collagen. The presence of appreciable concentrations of salts prevents SLS
formation, and SLS-type structures are redissolved if salt is added, suggesting
that the linkages involved are electrostatic and relatively weak. I know of no
special effect of Mg++ on the SLS system, other than its contribution to the
ionic strength of the medium.

Dr. Johnson (University of Illinois): Do you think that spatial separation
of charged sequences along the molecule would account for periodicities seen
in electron micrographs of precipitated preparations of other proteins, such as
paramyosin, tropomyosin, and light meromyosin? We have evidence for the
presence of polyglutamic acid along the paramyosin molecule. Could =this
account for periodicity in paramyosin?

Dr. Hodge: Yes. The band patterns observed in the various ordered aggre-
gation kates of proteins such as paramyosin, tropomyosin, and light mero-
myosin are almost certainly due to a precise distribution of "clusters" of polar
residues separated axially by regions of non-polar residues. As is the case for
collagen, it is the precise spatial separation of these polar regions and their

52 MACROMOLECULAR COMPLEXES

charge characteristics which are responsible for (a) the fact that several differ-
ent stable aggregation states can be formed under appropriate conditions (see
Hodge, 1959a, 1959b, for summaries) and (b) the sharpness of the various
band patterns. The presence of short sequences of polyglutamic acid in the
paramyosin macromolecule would be in accord with this concept, especially
since this is an a-type protein, in which such polar "clusters" must be formed
by the coiling of a single polypeptide chain; in the tropocollagen macro-
molecule, on the other hand, such polar "clusters" could result from the mutual
coiling of three separate polvpeptide chains, and the probability of finding
polypolar sequences would be less.

Dr. Prosser (University of Illinois): I am still not clear regarding the
ATP effect on collagen polymerization. If it is a matter of charge distribution,
the polyanions should be effective. Is energy liberated? Is the ATP split?

Dr. Hodge: The effect of ATP on the ordered aggregation of TC macro-
molecules is unquestionably a matter of charge distribution, and polyanions,
such as nucleic acid, are effective in inducing the formation of SLS-type struc-
tures. However, the system is rather delicately balanced with respect to pH
and ionic strength, so that the pK of the inducing agent is fairly critical. As far
as I know, SLS formation does not involve liberation of energy from splitting
of ATP.

Dr. Sherman (Brown University): Is Mg++ required for the ATP effect?

Dr. Hodge: Mg++ is not required for the formation of SLS-tvpe aggre-
gates, and indeed, would probably inhibit the precipitation because of its
relatively high contribution to the ionic strength of the medium.

The Role of the Macromolecular Aggregation
State and Reactivity of Collagen
in Calcification^

Melvin J. Glimcher -

The mineralization of certain specialized tissues of living organ-
isms is a widespread biological phenomenon ( Table 1 ) which serves
a number of highly important mechanical and chemical functions.

It has long been recognized, and recently emphasized by the
newer techniques of x-ray diffraction and electron microscopy, that
an intimate relationship exists between the mineral crvstallites and
the organic matrix of these tissues (Schmidt, 1935; Robinson, 1952;
Robinson and Watson, 1952, 1955; Finean and Engstrom, 1953;
Carlstrom and Finean, 1954; Carlstrom et al., 1955; Speckman and
Norris, 1957; Gregoire, 1957; Tsujii et al, 1958; Watabe et al, 1958).
Such observations have naturally led investigators interested in the
mechanism of tissue mineralization to study the role of the organic
matrix in initiating the formation of the inorganic crystals ( Freuden-
berg and Gyorgy, 1921; Gutman and Yu, 1950; Rubin and Howard,
1951; Gersh, 1952; Neuman and Neuman, 1953; Engel et al, 1953;
Sobel and Burger, 1954; Sobel, 1955; Strates et al, 1957; Glimcher

' These studies were aided by research grants RG 6391 from the Division of
General Medical Sciences, RG (Mult) 6391 from the National Institute of Dental
Research, H-4777 from the National Heart Institute, and RG 6391-Sl from the Na-
tional Institute of Arthritis and Metabolic Diseases of the National Institutes of
Health, Public Health Service, U. S. Department of Health, Education, and Welfare;
and by research grants from the John A. Hartford Foundation Inc., the Easter Seal Re-
search Foundation of the National Society for Crippled Children and Adults, Inc.,
the Liberty Mutual Insurance Company, and the Orthopedic Research and Education
Foundation. Some of the work described was carried out at the Massachusetts Insti-
tute of Technology, Cambridge, Massachusetts, and was supported by research grants
E-1469, from the National Institute of Allergy and Infectious Diseases, and A-2317,
from the 'National Institute of Arthritis and Metabolic Diseases, National Institutes of
Health, Public Health Service, United States Department of Health, Education and
Welfare, to the Massachusetts Institute of Technology (Prof. Francis O. Schmitt).

- Research Fellow of the Medical Foundation of Boston, Boston, Massachusetts.

53

54

MACROMOLECULAR COMPLEXES

et al, 1957; Strates and Neurnan, 1958). In our laboratory, we have
been primarily concerned with one particular mineralizing system:
the deposition of the calcium phosphate crystals of apatite in an
organic matrix of collagen such as occurs in bone, cartilage, and
tooth.

TABLE 1
Examples of Biologically Mineralized Tissues

(Glimcher, 1959)

Major

Organic-

Tissue

Crystal

Mineral

Matrix

Species

Mineralized

Chemistry

Form

Components

Plants

..Cell wall

CaCOa

Calcite

Cellulose,
pectins,
lignin

(?)

Radiolarians

. .Exoskeleton

SrSo4

Celestite

Diatoms . . .

. .Exoskeleton

Silica

(?)

Pectins

Mollusks . . .

. .Exoskeleton

CaCOa

Calcite,
aragonite

Conchiolin
( protein )

Arthropods .

. .Exoskeleton

CaCOs

Calcite

Conchiolin
( protein )

Vertebrates .

. . Endoskeleton

Bone

Ca,o(P04)o(OH),

Hydroxyapatite

Collagen

Cartilage

Ca,o(PO.)o(OH),

Hydroxyapatite

Collagen, acid

mucopoly-
saccharides

Tooth

Dentin

Caio(POO..(OH).

Hydroxyapatite

Collagen

Cementum

Cau,(P04),>(0H).

Hydroxyapatite

Collagen

Enamel

Ca,o(POOo(OH).

Hydroxyapatite

Eukeratin

As a result of a series of experiments carried out at the Massachu-
setts Institute of Technology in collaboration with Professor Francis
O. Schmitt and Dr. Alan J. Hodge, a general concept of the basic
physicochemical mechanism initiating calcification has been postu-
lated, based on the observations of the specific interrelations be-
tween native and reconstituted collagens and metastable calcium
phosphate solutions, and on a companion x-ray diffraction and elec-
tron optical study of both developing and adult bone (Glimcher
et al, 1957; Glimcher, 1958; Glimcher, 1959; Glimcher et al, 1960a,
1960b). While only the collagen-hydroxyapatite system has thus
far been given careful experimental study, the basic hypothesis may
apply to the problem of biological mineralization in general. Com-
prehensive reviews of these concepts and experiments have recently

MACROMOLECULAR AGGREGATES IN CALCIFICATION 55

been presented (Glimcher et al., 1957; Glimcher 1958; Glimcher,
1959; Glimcher et al, 1960a, 1960b) and therefore need only be
summarized here.

The basis of the hypothesis rests on an understanding of the
phvsicochemical phenomena involved in phase transitions, of which
crvstallization is just one specific case (Turnbull, 1956; Glimcher,
1959). A common example of a phase transition is the formation of
ice cr\'stals from liquid water or from supersaturated water vapor.
The use of Agl crystals in "seeding" supersaturated water vapor to
initiate ice crystal formation is well known. The formation of a new
phase (i.e., ice from supersaturated vapor) by the use or introduc-
tion of such a seeding material is called heterogeneous nucleation.
Two important points should be noted : ( 1 ) The water vapor or
liquid water must be supersaturated, that is, in metastable equilib-
rium; systems in metastable equilibrium may remain stable indefi-
nitely yet still retain the ability to form a new phase under the proper
conditions, "seeding," for example. ( 2 ) The "seed" crystal must have
an atomic lattice structure similar to that of the crystal being nu-
cleated (Turnbull, 1950a, 1950b, 1952; Turnbull and Vonnegut,
1952 ) ; in the case of Agl and ice, for example, there is a remarkably
close fit between their lattice parameters.

Examination of the biological conditions as they relate to calcifi-
cation of bone, cartilage, and tooth reveals some interesting correla-
tions. The extracellular fluids of many vertebrates have been shown
to be in metastable equilibrium with respect to the formation of
apatite crystals (Neuman, 1950; Strates et al, 1957; Strates and
Neuman, 1958; Glimcher, 1959). Furthermore, the major organic
component of bone, of cartilage, and of the dentin and cementum
of tooth is the fibrous protein, collagen. This complex macromol-
ecule, which is thought to be composed of three polypeptide chains
in a coiled-coil configuration (Rich and Grick, 1955, 1958), is so
structurally regular that it gives rise to a characteristic x-ray diffrac-
tion pattern and therefore for our purpose may be considered
"crystalline" (Fig. 1). In addition, the native collagen fibril, which
is composed of such macromolecules aggregated both linearly and
longitudinally in a highly specific and characteristic fashion, is there-
fore also a well-ordered structure and may also be considered "crys-
talline" (Fig. 2). When to such data is added the observation of
the intimate association between the apatite crystals and collagen
(the inorganic apatite crystals in bone are located primarily within

56

MACROMOLECULAR COMPLEXES

¥\

A BCD

Fig. 1. (A), (B), (C), (D) Diagrammatic derivation of the coiled-coil structure
of the collagen macromolecule. Each of the three polypeptide chains is wound
about its own axis as a left-handed helix in a threefold screw fashion. The
axes are located at the vertices of an equilateral triangle and are parallel. The

the collagen fibrils; see Glimcher, 1959, Glimcher et al., 1960a, and
Figs. 3, 4, 5), the possibility that either the macromolecular struc-
ture or the macromolecular aggregation state of collagen is respon-
sible for the induction of calcification by acting as a catalytic hetero-
geneity (similar to Agl in the case of supersaturated water vapor)

MACROMOLECULAR AGGREGATES IN CALCIFICATION

57

r.

screw repeat in the case of polyglycine, for example, is 9.3 A, and the rise per
residue (each of which is represented here by a black dot) is 3.1 A. (E) X-ray
difFraction pattern of oriented collagen fibers.

is apparent. The concept that one or more components of the
organic matrix of bone, cartilage, and tooth catalyzed the formation
of apatite crystals was first proposed by Nemnan and Neuman
(1953).

From solutions of collagen extracted from a wide variety of con-

58

MACROMOLECULAR COMPLEXES

||.| HI0H. HI

\

il

il

MACROMOLECULAR AGGREGATES IN CALCIFICATION

59

?= 6. i -° :6 ^ S .E

o o cp >r E ■- = -^

*7 — C O. O -^

-D 3:e 2^ - u o

^c ^_^ E i g ^ ■§

.y ./i -C U C ? 3 O)

>- 0 c >.^^ a; 8
° O O) u ^ t^ "fc

°- E oy^Z °-D °

Q)0=a^OcOC

i - ~ ^ -^ E Q-
^ ^ 2 Jit °-°.E ^
-5 .2. o ^ o 5 ^ --C

S I 8 2 ^ •- ■" °

CD.tz

_9 ^

Ol .tr Ic O

_u oj "^ o ^ '1' "^ 0

o cnJJ o D c o £ ~
^ t; § o)^ ^-2 .g'-S

u c D 0) O 't -C

D ><

>--°3E

'^^^^ ° § ^ :: 2!-
i^^.i B^ sz ^ . s

<« OJ t.- •- c « J:; (U

(h .2 m cc -£

60

MACROMOLECULAR COMPLEXES

MACROMOLECULAR AGGREGATES IN CALCIFICATION

61

$S^ I

'J^MT'^

62 MACROMOLECULAR COMPLEXES

nective tissues it is possible to reconstitute in vitro not only the
native-type (640-700- A axial repeat) collagen fibrils, but also a
number of other fibrillar forms (Schmitt et al, 1942; Highberger
et al, 1951; Gross et al, 1954, 1955; Schmitt et al, 1955a, 1955b).
These various fibrillar forms represent different aggregation states
of the same macromolecules and therefore reflect differences be-
tween the stereochemical relations of the amino acid sidechain
groups of adjacent macromolecules. Since it is also possible
to isolate the macromolecules and linear aggregates of the macro-
molecules (protofibrils), we were thus provided with a unique sys-
tem which would not only test the nucleation hypothesis, but which
also could distinguish whether this property (if such existed) was
a function of the structure of the macromolecule itself or was de-
pendent on steric factors and a characteristic electrical charge dis-
tribution provided by the aggregation of the macromolecules into
fibrils (Fig. 6).

The experiments were conducted by exposing the various types
of fibrils, macromolecules, and protofibrils, as well as previously
demineralized bone collagen (shown by low-angle x-ray diffraction
to have maintained its 640-A long-range fibrillar order) to meta-
stable calcium phosphate solutions.

Under identical physicochemical conditions, only the native-type
(640-700- A axial repeat) reconstituted collagen fibrils and the de-
mineralized bone collagen were able to nucleate apatite crystals
from metastable calcium phosphate solutions. The failure of the
various other fibrils, representing different aggregation states of the
same macromolecules, to initiate this phase change demonstrated
the rather remarkable steric specificity required in the nucleation
phenomenon. Since it was also demonstrated that the individual
tropocollagen macromolecules and the protofibrils were unable to
initiate crystallization, it was evident that the specificity did not lie
in the macromolecule itself, but in the specific manner in which the
collagen macromolecules were packed to make native fibrils, i.e.,
their macromolecular aggregation state. The structural matching
in the case of collagen depends in the first instance upon the integ-
rity of its molecular structure, which assures a specific linear array
of amino acid residues in the covalent polypeptide chains. Yet even
when this condition is met it is not sufficient to provide a nucleation
site. Nucleation occurs only when the tropocollagen particles are
aligned both laterally and longitudinally in the specific array char-

MACROMOLECULAR AGGREGATES IN CALCIFICATION

63

acteristic of the nativc-t\'pe collagen fibril, leading to a precise
juxtaposition of amino acid side-chain groups from adjacent macro-
molecules.

Additional confirmation of this steric requisite was obtained by
subjecting aliquots of reconstituted native-type collagen fibrils and

Fig. 6. The steric diflFerences between the reactive side-chain groups of
adjacent macromolecules in the native and SLS forms of collagen, diagram-
matically portrayed. (From Giimcher, 1959.)

of demineralized bone collagen to a number of physical and chem-
ical agents which destroyed the ordered structure of the fibrils
without altering the reactivity of the amino acid side-chain groups.
These fibrils lost the ability to induce crystallization ( Fig. 7 ) .

From the macromolecular organization of the collagen fibrils

64

MACROMOLECULAR COMPLEXES

and from the fact that a precise stereochemical configuration of a
number of amino acid side-chains was necessary to form a nucleus,
it was reasonable to expect that there should be a limited number
of such preferred centers and that they should occur periodicalh'
both laterally and longitudinally along the axial repeat structure of

NATIVE TISSUES

(DECALCIFIED BONE)

calcification:

CALCIFICATION

LOSS OF LONG-RANGE
Fl BRI LLAR ORDER

CALCIFICATION

Fig. 7. Some of the experiments which demonstrated the specificity of the
macromoleculor aggregation state of collagen in in vitro calcification, diagram-
matically illustrated. (From Glimcher, 1958.)

the collagen fibrils. Confirmation of the existence of such discrete
nucleation centers has come from a time-study of the nucleation
process, both by electron microscopy and by low-angle x-ray diffrac-
tion. Using the former technique, we haye been able to visualize
the initial stages of mineralization in vitro. These appear yery
similar to the initial stages of mineralization in embryonic bone
(Figs. 8, 9, 10), and emphasize the regularity of the initial crystal-

MACROMOLECULAR AGGREGATES IN CALCIFICATION

65

66

MACROMOLECULAR COMPLEXES

MACROMOLECULAR AGGREGATES IN CALCIFICATION

67

.♦.

m ' ^- .-■ •

*4

. (»■..

* / . .

*^

t * *» ,

.V%, >it .^*t

■■

O -Q iJ

l"!

0 C D

<■

■- O O

« -^

ii- >-

O O -Q

Z .E-o

4

-D 0

.•*,■■;

1

oil

■t C 3

«'■■■'

D o >«

ai

<+: "* ±-

la

l-'?-^

M

*-^

0 "— ' •-

W

G -i<:

M.

• ^

w

ges o
fibril:
iginal

v^

, •

2 ? o

01 0)

^ C31 E

*

S^ 2

— o '*-

D '-' -D

m

^ <D S

0) -C 3

^

^:^

c •— '-

liri!

- ^ a.

m

d) • - <u

W:

c ^ -

> -^0-^

* j^n

: u 2 o^

r C >.

o >- -

1 >. '-' =^

C o

J3 <D <«

^

E -.^ u

1

q; o 0

1

a—.

1

§ ° c

J

> 35 °

.

° £^

j

E ^*- <D

* °^C0

1/1 0

,*

« l-i

■ D ^ t:

h. O "-

U) '-' ^

0 +-

1- c

c D)0"

0 c O

A

.il D --

!r^ t X

^ o

LU

.y c

if

ig. 10.
period
s-sectio

^^

^ LJ- M

''* 0

J "? 2

-^-4

i ou

68 MACROMOLECULAR COMPLEXES

lization sites in the collagen fibrils. Preliminary studies by low-
angle x-rav diffraction, in which the changes in the intensities of the
diffraction spacings were measured, have also indicated that there
are specific sites in the collagen fibrils which act as nucleation cen-
ters.^

Obviously, the next step in understanding the nucleation process
would be to discover which amino acid side-chain groups interact
with which mineral ions, the nature of the intermolecular forces be-
tween collagen and the mineral ions, and the kinetics and atomic
sequence of the nucleation process itself. Unfortunately, even in
very simple systems, little is known about such phenomena, and
more attention has been paid to nucleation rates than to how the
nuclei themselves are formed. However, since it is possible that
the initiation and sequence of nucleation may be enzymatically
controlled and regulated in biological mineralization, such informa-
tion is of crucial importance.

As to the specific organic groups and mineral ions involved in
calcification, most investigators in the past have felt that the initial
step involved the combining of an anionic organic group with cal-
cium ion (Rubin and Howard, 1951; DiStefano et ah, 1953; Sobel
and Burger, 1954; Sobel, 1955). However, there are some consider-
ations which cast doubt on the hypothesis of the primary role of
calcium-binding in crystal induction.

The case for the importance of the phosphate ion has been sum-
marized excellently by McLean ( 1958 ) . To such points as the im-
portance of organic phosphate compounds for energv transfer
( high-energy phosphate bonds ) and the possible enzymatic control
of such reactions, might be added some structural considerations as
well. Namely, the apatites are essentiallv phosphate salts with struc-
tural characteristics primarily attributable to the phosphate groups
and not to calcium atoms; the latter can be replaced bv a number of
other cations, such as strontium and lead, without changing the
major features of the crystal structure or svmmetrv. In addition,
if cellular-controlled, enzymatic transfers of mineral ions to specific
groups in the collagen are involved in the nucleation process, phos-
phate is again the most likely candidate. No calcium-splitting
enzymes or calcium transferases have as yet been described, al-

^ Many of the low-angle x-ray diffraction studies were carried out in cooperation
with Dr. A. Posner, American Dental Association, Research Division, National Bureau
of Standards, and with Dr. L. Bonar, Department of Biophysics, Massachusetts In-
stitute of Technology.

MACROMOLECULAR AGGREGATES IN CALCIFICATION 69

though there are a host of such enzymes for phosphate. Further-
more, if Ca++ were transferred enzymaticallv to the collagen, any
covalent chelating or co-ordinating structure formed would be very
unreactive, i.e., would not be very accessible to reaction with the
other ions making up the crystal lattice; such bonds would tend to
demineralize (similarly to other chelates like ethylenediaminetetra-
cetic acid, EDTA), rather than to initiate nuclei formation. On the
other hand, covalent phosphate compounds such as -N-P or -O-P
alkyl amides or esters would still be very reactive, with Ca + + for
example. Since the phosphate groups are not only the struc-
tural "backbone" of the apatite lattice, but are also more Hkely to
be involved from the biological standpoint, their role in the forma-
tion of the initial fragments of the crystal lattice or nucleus would
appear to be equally as important as that of the calcium ions, and
probably more important. Indeed, the proposal that phosphoryla-
tion of free amino groups constitutes the initial step in calcification
has been made in the past (Polonovski and Cartier, 1951; Cartier,
1951, 1952; Cartier and Picard, 1955; Zambotti, 1957), and it has
been suggested that pyrophosphate radicals are the initially bound
phosphate moiety (Cartier, 1959).

Two general methods of attack are possible in an effort to ascer-
tain the reactive amino acids involved in the nucleation phenomena.
In the first, which we may term the "availability" method, one may
determine the difference ( if any ) between the number of side-chain
groups of a particular amino acid available in the fully mineralized
matrix and the number available in demineralized matrix, the as-
sumption being that groups which are not reactive in the mineral-
ized tissue are not available because of interaction with the mineral
ions. In the second method, specific amino acid side chains of de-
mineralized collagens are blocked or altered, and the effect of such
alterations evaluated by comparing the amount of mineral deposited
during in vitro calcification with untreated controls.

Solomons and Irving ( 1958 ) have studied the availability of the
epsilon amino groups (e-NH^) of lysine and hvdroxylysine in bone
and dentin by reacting such tissues with l:fluoro:2:4 dinitrobenzene
(FDNB). They noted that, as bone and dentin were demineral-
ized in stages by a mild chelating agent (EDTA), there was a direct
relation between the availability of such groups to FDNB and the
amount of total mineral present in the sample (calcium and phos-
phate) (Fig. 11).

70

MACROMOLECULAR COMPLEXES

However, when one carefully considers the interpretations of
such experiments, it appears likely that, rather than demonstrating
an inorganic-organic linkage, the findings are more consistent with
the interpretation that the phenomenon represents a diffusion-limit-
ing process. Some of the detailed arguments and experiments for
such an interpretation have already been presented (Glimcher,
1958).

100

^

60

40

20

0 25 50 75

Demineralization (%)

Fig. 11. Availability of e-DNP-lysyl and hydroxylysyl amino groups of hard-
tissue collagens to FDNB during demineralization. O, • denote successive
demineralization and dinitrophenylation of ox dentin and human dentin,
respectively; +, A, D denote simultaneous demineralization and dinitrophenyla-
tion of ox dentin, ox bone, and human dentin, respectively. (From Solomons
and Irving, 1958.)

In brief, these experiments showed ( 1 ) that the number of e-NHj
groups in a variety of different bones varied from approximately 35
micromoles per gram of collagen, in oxbone, to approximateh' 200-
250 micromoles per gram of collagen, in fish bone, and was directly
related to the packing and density of the tissues and not to the
amount of mineral per gram of organic weight, which was essentialh'
the same in all cases (Tables 2A, 2B); (2) that both the number of

MACROMOLECULAR AGGREGATES IN CALCIFICATION 71

TABLE 2A

Availability of €-NH2 Groups of Bone Collagen to FDNB

as a Function of Their Mineral Content (Hand-cut, or

"coarse," particles demineralized with EDTA)

Per Cent

Per Cent

IX mole <

r-NHo Groups

Organic Weight

Demineralization

Availabl

le/g Collagen

Fish Bone

34

0

193

39

8

243

52

27

288

74

60

305

98

97
Calf Bone

311

37

0

137

39

3

179

42

8

205

53

25

242

72

56

261

92

87

291

96

94

298

98

97

304

TABLE 2B

Availability of e-NHi.. Groups ^o FDNB in Native Undecalcified Bones

of Various Densities, in Order of Decreasing Density

(Hand-cut, or "coarse,'" particles reacted for 48 hours)

Bone Source

Per Cent
Mineral

IX mole
Availa

\ e-NHj Groups
ble g Collagen

Ox . .

66.23

35

Hen

68 20

82

Chick

65 53

114

Calf

Fish

Embryonic

chicken

63.10

.... 64.57
.... 56.0

161
216
263

carboxyl groups (glutamic and aspartic acids) and the number of
guanidino groups (arginine) which were available also increased
with demineralization (Tables 3A, 4); (3) that the number of such
groups available in native bone of various kinds also varied in a
fashion similar to the e-NH. groups, and was also apparently de-
pendent on the density and packing of the specimen; and (4) that
the humber of reactive groups available in native bone varied with
the particle size: the finer the particle size of the bone samples, the
more groups that were available for reaction ( Tables 2C, 2D, 3B ) .

72 MACROMOLECULAR COMPLEXES

TABLE 2C

Availability of e-NH-j Groups to FDNB in Native Bone of Various Species

OS a Function of Particle Size

(Reacted for 48 hours)

fjL mole 6-NHo Groups Available/g Collagen

Bone Source Coarse Finely Powdered

Ox 35 88

Hen 82 116

Calf 166 204

Fish 216 267

Embryonic chicken 263 323

TABLE 2D

Availability of e-NHo Groups to FDNB in Native Bone as a Function
of Time of Exposure and Particle Size

/J. mole e-NHo Groups Available/g Collagen
Coarse Finely Powdered

Bone Source 2 Days 2 Weeks 2 Days 2 Weeks

Calf 161 200 203 229

Fish ( a ) 216 267 247 273

(b) 193 244 313 312

TABLE 3A

Availability of Guanidino Groups of Arginine in Bone Collagen to
Phosphotungstic Acid, as a Function of Mineral Content

( Note that the mineral contents of the native undecalcified bone are

approximately the same although they have markedly different densities,

ox > fish.)

Per Cent

Per Cent

/J. mole Arginine

Organic Weight

Demineralization

Available g Collagen

Fish Bone

36.5

0

144

41.1

8

162

51.4

24

217

89.9

84

300

99.0

98
Ox Bone

416

36.8

0

42

58.6

. 34

124

95.0

92

309

-100.0

-100

419

MACROMOLECULAR AGGREGATES IN CALCIFICATION 73

TABLE 3B

Availability of Guanidino Groups of Arginine to PTA as a Function
of Particle Size in Native Mineralized Bone

/u mole of Arginine Available/g Collagen
Bone Source Coarse Powdered

Ox

Fish

23

77

43
129

TABLE 4

Availability of Carboxyl Groups to Methylation in Tv/o Species
of Bone as a Function of Mineral Content *

( Note that the mineral contents of the native undecalcified bones are

approximately the same although they have markedly different densities,

fish < calf. )

Per Cent
Organic Weight

Per Cent
Demineralization

M mole Carboxyl Groups
Available/ g Collagen

Fish

44.0
98.4

ll.Of
97.5

Calf

132
590

46.0
98.5

14.0f
97.5

«70{
595

* The number of groups available (for methylation) depends,
among other things, on the volume of solution used, normality of acid,
and time of exposure. Therefore, the figures do not represent the total
number of groups available, but only that number available under ex-
perimental conditions used. Since the two species of bones were
treated identically, the values obtained are comparable. Other experi-
ments have shown thai m hand-cut, coarse fish bone particles over 70
per cent of the theoretical number of carboxyl groups can be methylated.

f Due to the HCl in the methylating media and not to EDTA.

\ Analytical values varied but at the most were well under 70
fj. mole/g.

In any event, determining the number of e-NH^, groups which are
available for reaction with FDNB will not show whether there is or
is not an electrostatic interaction between the e-NH. groups and the
mineral ion(s). In both instances (free or electrostatically linked
e-NHv groups), the FDNB would react with the e-NH. groups, in
the latter case by displacing the relatively weak electrostatic bond.
On the other hand, failure of the e-NH. groups to react with FDNB

74

MACROMOLECULAR COMPLEXES

despite the removal of the mineral (and assuming no steric hin-
drance) could mean that a stable (covalent or coordinate) bond
existed between the e-NH. groups and one or more of the mineral
ions.

On the other hand, group-blocking and -substitution experiments
appear to give more conclusive evidence. We are continuing to
study the problem by this latter method; after blocking specific
amino acids, the abihty of the collagen to mineralize in vitro is ana-
lyzed, and the quantity of Ca-salts deposited is measured. It is well

DECALCI Fl ED CALF BONE
37% £"- AMINO GROUPS BLOCKED
AVAIL- NH; IT) _ flP (T) _

AVAIL. NH, (C) ~^^ AP (C) " ^^

^= CONTROL(C)
^=TREATED(T)

1 h^

aODITIONAL

DECALCIFIED FISH BONE
59% 5'-AMIN0 GROUPS BLOCKED
AVAIL NH, (T) . ,, flP (T) . ,^

AVAIL. NH2 (C)

m

Fig. 12. The effect of blocking the e-NH^ groups of bone collagen on
remineralization in vitro. Collagen treated with carbobenzoxychloride. (From
Glimcher, 1958.)

known that it is difficult to pinpoint the active groups in biological
molecules which are necessary for a physiological function; in order
to be meaningful, experiments designed for this purpose must meet
a number of general and specific criteria. These have been suffi-
ciently and thoroughly discussed in the past in relation to the block-
ing of specific groups in proteins in general (Olcott and Fraenkel-
Conrat, 1947; Putnam, 1953) and therefore need not be detailed
here.

While it is perfectly clear that our experimental results must
therefore be viewed with great caution, they have been encouraging

MACROMOLECULAR AGGREGATES IN CALCIFICATION 75

enough and sufficienth' suggestive to warrant discussion here. Our
attention has been primarily concentrated on the e-NH^ groups of
h'sine and hvdroxvlysine partly because this group is quite easily
l)locked without altering either the molecular structure of collagen
or its macromolecular aggregation state and partly because of the
likelihood that this group would be involved in phosphate-binding,
which we feel is an important step in nucleation. Experiments in
which the e-NH. groups were specifically blocked in varying
amounts by several different reagents showed a definite correlation
between the number of e-NHo groups available and the total amount
of mineral deposited. Fig. 12 and Table 5 give examples of the re-

TABLE 5

Effect of Blocking the e-NH_. Groups of Deminerolized Fish Bone

(Collagen) with l:Fluoro:2:4 Dinitrobenzene (FDNB)

on Its Subsequent Recolcificotion in vitro

( Recalcified at 25° C for approximately 48 hours)

fi mole e-NHo

Groups Blocked by

Mg Ca, g Collagen

Mg

P/g Collagen

FDNB, g Collagen

45.70

28.63

O**

33.49

19.43

53

23.51

13.01

84

20.52

10.89

226

18.67

9.72

286

4.55

1.88

334

* A sham control which was exposed to the same reagents under
identical conditions of pH, etc., but without the addition of FDNB.

suits of such experiments. One of the most convincing pieces of
evidence thus far was the demonstration that the e-NH^ groups
could be reversibly blocked, and that there was a concomitant
change in the amount of mineral deposited (Table 6).

While such experiments indicate that the group substitution has
been carried out without destroying the molecular structure of the
collagen, they do not rule out the possibility that the observed re-
sults are, in fact, due to the shielding ( either electrostatic or steric )
of groups adjacent to the e-NH. groups, rather than to the e-NH^
groups themselves.

Since availability studies of fully demineralized bone and dentin
indicate at least 95 per cent of the e-NH;, groups are not cova-

76 MACROMOLECULAR COMPLEXES

TABLE 6

Reversible Blocking of the e-NH:. Groups of Reconstituted Collagen

(Ichthyocol) by Carbobenzoxychloride and the Concomitant

Change in Its Recalcification in vitro

fi mole e-NHo Groups Mg Ca/g Mg P/g

Specimen Available/g Collagen Collagen Collagen

Calcified Approximately 60 Hours at 25° C

Control 198 68.2 34.7

Treated with carbobenzoxy-
chloride : 39 3.5 2.3

After catalytic removal

of carbobenzoxy groups 180 40.4 23.8

Calcified Approximately 36 Hours at 25° C

Control 242 26.5 14.9

Treated with carbobenzoxy-
chloride 42 1.3 0.65

After catalytic removal of

carbobenzoxy groups 97 5.5 1.8

lently bound to the mineral ions ( presumably phosphate ) , and sub-
ject to the misgivings about the steric or electrostatic hindrance as
noted earlier in interpreting the e-NH., blocking experiments which
showed the importance of such groups in in vitro calcification, we
may interpret this to mean the interaction between the vast majority
(95 per cent or more) of the e-NH^ groups and mineral ions is
electrostatic. This does not rule out the possibility that the initial
event, particularly in vivo, is a phosphorylation of collagen to form
either -N-P-alkyl amides or -O-P-ester linkages. Only a small num-
ber of such phosphate esters or amides need be formed to start the
nucleation process, and the remaining phosphate ions as well as
calcium ions could then simply be electrostatically attracted to the
rest of the polar amino acid side chain groups of the collagen and to
each other.

Further experiments are now in progress in which a number of
other amino acid side-chain groups are being blocked singly and
in combination, and in which not only the final result of nucleation
(i.e., the amount of mineral formed) but also the binding of the
various constituent mineral ions is being studied. These experi-
ments should lead to a clearer understanding of both the exact roles
of specific amino acids and mineral ions and the sequence of events
leading up to nuclei formation.

MACROMOLECULAR AGGREGATES IN CALCIFICATION 77

On the basis of the experimental evidence presented, it would
appear that the collagens of all tissues which have fibrils in the
specific macromolecular aggregation state characterized by the 640-
700-A axial repeat structure should be inherently capable of nucle-
ating apatite crystals from metastable calcium phosphate solutions.
In the case of the vertebrates, for example, the collagens of all the
normall)' unmineralized tissues such as the skin, ligaments, and ten-
dons are organized in this fashion, and yet under normal conditions
onlv the collagens of skeletal system and tooth are mineralized; it is
well known, however, that the former may become mineralized in
pathological conditions. To explain these findings, one might sup-
pose that bone, cartilage, and tooth collagens are somehow differ-
ent from the collagens of non-mineralized connective tissues. This,
however, would not explain how ligaments, tendons, or skin col-
lagens, for example, may be calcified pathologically.

On the other hand, it seemed more reasonable to us to postulate
that in normally unmineralized tissues there are other substances
present which under ordinary circumstances prevent or inhibit
calcification (Glimcher ct a/.,' 1957; Glimcher, 1958, 1959). Data
indicating that the latter is the correct interpretation come from
both in vitro and in vivo studies. In vitro experiments conducted in
our laboratories have shown that, whereas native (as opposed to
reconstituted ) collagen of tissues such as rat-tail tendon, guinea-pig
skin, calfskin, etc., were not able to mineralize, these same collagen
fibrils were able to mineralize after appropriate treatment of the
tissue by procedures designed to extract and depolymerize certain
components of the ground substance (Glimcher, 1958, 1959; Fig.
13). This, coupled with the fact that the collagen from these nor-
mally unmineralized tissues, when extracted, purified, and recon-
stituted into native-type fibrils essentially free of ground substance,
is able to mineralize, appears to implicate other inhibitory sub-
stances in such tissues.

Very strong additional support for the hypothesis that native-
type collagen fibrils are inherently able to mineralize comes from the
studies of Likens et al (1958), Nylen et al. (1958), and Johnson
( 1958 ) , who showed by chemical as well as by electron microscope
and x-ray microscope methods that the original collagen fibrils of
turkey tendons, and not a special kind of bone collagen, initially
calcified in vivo.

78

MACROMOLECULAR COMPLEXES

It should be clearlv kept in mind that what we have been dis-
cussing is the phvsicochemical basis of the induction of mineiahza-
tion. The minerahzation of the endoskeleton of vertebrates and the
exoskeleton of certain invertebrates is a vital process which demands
exquisite homeostatic control mechanisms which are highly specific
in their operation. As in so many other cases in the organism where

NATIVE TISSUES

I

Fig. 13. Diagrammatic illustration of the possible role of certain components
of the ground substance (long threadlike material surrounding the collagen
fibrils) in helping to inhibit calcification in normally unmineralized tissues. The
"?" indicates that it is not yet certain whether depolymerization (represented
by the short segments surrounding the fibrils) is itself effective, or whether
depolymerization and subsequent removal from the tissue are necessary. (From
Glimcher, 1958.)

control and specificity must be of the highest order, in both nor-
mally mineralized and normallv unmineralized tissues a number of
highly specific regulating mechanisms obviously must exist. There
is every reason to expect that mineralization in both kinds of tissues
is imder the control of many delicately balanced regulatory factors,
providing the organismic and cellular regulation of the mechanism
which initiates the formation of the inorganic crystals and which
controls their further growth.

MACROMOLECULAR AGGREGATES IN CALCIFICATION 79

References

Block, R. J. 1951. Chemical classification of keratins. Ann. N. Y. Acad. Sci. 53:

608.
Block, R. }., M. K. Horwitt, and D. Bolling. 1949. Composition of protein of

human teeth and fish scales. /. Dental Research 28: 518.
Carlstrom, D., a. Engstrom, and J. B. Finean. 1955. The influence of collagen

on the organization of apatite crystallites in bone. Symposia Soc. Exptl. Biol.

9: 85.
Carlstrom, D., and J. B. Finean. 1954. X-ray diffraction studies on ultrastructure

of bone. Biochim. et Biophtjs. Acta 13: 183.
Cartier, p. 1'951. Biochimie de I'ossification. La Role de I'acide adenosine triphos-

phorique dans la mineralization du cartilage ossifiable. /. physiol. (Paris) 43: 677.
Cartier, P. 1952. Le Mechanisnie enzymatique de Tossification. Exposes ann. bio-
chim. med. 14: 73.
Cartier, P. 1959. La Mineralization du cartilage ossifiable. VIII. Les Pyrophos-
phates du tissu osseux. Bull. soc. chim. biol. 41: 573.
Cartier, P., and J. Picard. 1955. La Mineralization du cartilage ossifiable. III.

Le Mechanisme de la reaction atpasique du cartilage. Bull. soc. chim. biol. 37:

1159.
DiStefano, v., W. F. Neuman, and G. Rouser. 1953. The isolation of a phosphate

ester from caicifiable cartilage. Arch. Biochem. Biophys. 47: 218.
Engel, M. B., N. R. Joseph, and H. R. Catchpole. 1953. Equilibrium of calcium

and other ions in connective tissues. 5th Conf. on Metabolic Interrelations, Trans.

Josiah Macy, Jr., Foundation Publications, New York. P. 105.
Finean, J. B., and A. Engstrom. 1953. Low-angle x-ray diffraction of bone. Nature

171: 564.
Fitton-Jackson, S. 1957. The structure of developing bone in the embryonic fowl.

Proc. Roy. Soc. (London) B 146: 270.
Freudenberg, E., and P. Gyorgy. 1921. Ueber Kalkbindung durch Tierische

Gewebe III. Biochem. Z. 118: 50.
Gersh, I. 1952. Some functional considerations of ground substance of connective

tissues. 2nd Conf. on Connective Tissues, Trans. Josiah Macy, Jr., Foundation

Publications, New York. P. 11.
Glimcher, M. J. 1958. The specificity of the molecular structure of organic

matrices in mineralization. In li. Sognnaes (ed.). Symposium on Calcification

in Biological Sy.stems, Washington, D. C, Dec, 1958. Am. Assoc. Advance. Sci.

P. 421.
Glimcher, M. J. 1959. The molecular biology of the mineralized tissues with

particular reference to bone. Revs. Modern Phys. 31: 359.
Glimcher, M. J., A. J. Hodge, and F. O. Schmitt. 1957. Macromolecular aggre-
gation states in relation to mineralization: The collagen hydroxyapatite system as

studied in vitro. Proc. Nat. Acad. Sci. U. S. 43: 860.
Glimcher, M. J., A. J. Hodge, and F. O. Schmitt. 1960a. Electron optical and

x-ray diffraction studies of bone and in vitro calcified collagens. (In press.)
Glimcher, M. J., A. J. Hodge, and F. O. Schmitt. 1960b. The specificity of the

macromolecular aggregation state of collagen in in vitro calcification. ( In press. )
Gregoire, C. J. 1957. Topography of the organic components in mother-of-pearl.

/. Biophys. Biochem. Cytol. 3: 797.
Gross, J., and B. Dumsha. 1958. Elastoidin: A two-component member of the

collagen class. Biochim. et Biophys. Acta 28: 268.
Gross, J., J. H. Highberger, and F. O. Schmitt. 1954. Collagen structures con-
sidered as states of aggregation of a kinetic unit: The tropocollagen particle.

Proc. Nat. Acad. Sci. U. S. 40: 697.
Gross, J., J. H. Highberger, and F. O. Schmitt. 1955. Extraction of collagen from

connective tissues by neutral salt solutions. Proc. Nat. Acad. Sci. U. S. 41: 1.

80 MACROMOLECULAR COMPLEXES

GuTMAN, A. B., and T. F. Yu. 1950. Further studies of the relation between glyco-
genolysis and calcification in cartilage. 1st Conf. on Metabolic Interrelations,
Trans. Josiah Macy, Jr., Foundation Publications, New York. P. 11.

HiGHBERGER, J. H., J. Gross, and F. O. Schmitt. 1951. The interaction of muco-
protein with soluble collagen; an electron microscope study. Proc. Nat. Acad.
Set. U. S. 37: 286.

Johnson, L. C. 1958. Mineralization of turkey leg tendon. I. Histology and histo-
chemistry of mineralization. In Pi. Sognnaes (ed.). Calcification in Biological
Systems, a Symposium presented at the Washington Meeting of the American
Association for the Advancement of Science, Dec. 29, 1958. Publication Number
64 of the American Association for the Advancement of Science, Washington,
D. C., 1960. P. 117.

Likens, R. C., K. A. Piez, and M. L. Kunde. 1958. Mineralization of turkey leg
tendon. III. Chemical nature of the protein and mineral phases. In R. Sognnaes-
(ed.). Calcification in Biological Systems, a Symposium presented at the Wash-
ington Meeting of the American Association for the Advancement of Science, Dec.
29, 1958. Publication Number 64 of the American Association for the Advance-
ment of Science. Washington, D. C., 1960. P. 143.

McLean, F. C. 1958. Phosphorus Metabolism. In K. Rodahl et al., (eds.). Bone as
a Tissue, Proceedings of a Conference held at The Lankenau Hospital, Phila-
delphia, Pa., October 30-31, 1958. McGraw-Hill Book Co., Inc., New York, 1960.

Neuman, W. F. 1950. Bone as a problem in surface chemistry. 2d Conf. on Meta-
bolic Interrelations, Trans. Josiah Macy, Jr., Foundation Publications, New York.
P. 32.

Neuman, W. F., and M. W. Neuman. 1953. The nature of the mineral phase of
bone. Chem. Revs. 53: 1.

Nylen, M. U., D. B. Scott, and V. M. Mosley. 1958. Mineralization of turkey leg
tendon. II. Collagen-mineral relations revealed by electron and x-ray microscopy.
In R. Sognnaes, (ed.). Calcification in Biological System, a Symposium presented
at the Washington meeting of the American Association for the Advancement of
Science, Dec. 29, 1958. Publication Number 64 of the American Association for
the Advancement of Science, Washington, D. C, 1960. P. 129.

Olcott, H. S., and H. Fraenkel-Conrat. 1947. Specific group reagents for pro-
teins. Chem. Revs. 41: 151.

Polonovski, M., and P. Gartier. 1951. Sur le premier stade biochemique de
I'ossification. Compt. rend. 232: 119.

Putnam, F. 1953. Tlie chemical modification of proteins. In H. Neurath and
K. Bailey (eds.). The Proteins. Academic Press, Inc., New York. Vol. I, part
B, p. 893.

Rich, A., and F. H. C. Crick. 1955. The structure of collagen. Nature 176: 915.

Rich, A., and F. H. C. Crick. 1958. The structure of collagen. In G. Stainsby
(ed.). Recent Advarices in Gelatin and Glue Research. Pergamon Press, Ltd.,
London. P. 20.

Robinson, R. A. 1952. An electron-microscopic study of the crystalline inorganic
component of bone and its relationship to the organic matrix. /. Bone and Joint
Surg. 34A: 389.

Robinson, R. A., and M. L. Watson. 1952. Collagen-crystal relationships in bone
as seen in the electron microscope. Anat. Record 114: 383.

Robinson, R. A., and M. L. Watson. 1955. Crystal-collagen relationships in bone
as observed in the electron microscope. III. Crystal and collagen morphology as
a function of age. Ann. N. Y. Acad. Sci. 60: 596.

Rubin, P. S., and J. E. Howard. 1951. Histochemical studies in the role of acid
mucopolysaccharides in calcifiability and calcification. 2d Conf. on Metabolic
Interrelations, Trans. Josiah Macy, Jr., Foundation Publications, New York. P.

Schmidt, W. J. 1935. Polarisationsoptische Analyse des submikroskopischen Baues
von Sellen und Geweben. Handbuch biol. Arbeitsmethoden Abt. 5, TIO, p. 435.

MACROMOLECULAR AGGREGATES IN CALCIFICATION 81

ScHMiTT, F. O., J. Gross, and J. H. Highberger. 1955a. Tropocollagen and the

properties of fibrous collagen. Exptl. Cell Research 3 (Suppl.): 326.
ScHAriTT, F. O., |. Gross, and J. H. Highberger. 1955b. States of aggregation of

collagen. In Fibrous Proteins and Their Biological Significance. Sifinposia Soc.

Exptl. Biol. 9: 148.
ScHMiTT, F. O., C. E. Hall, and M. A. Jakus. 1942. Electron microscope investiga-
tions of the structure of collagen. /. Cellular Comp. Physiol. 20: 11.
SoBEL, A. E. 1955. Local factors in the mechanism of calcification. Ann. N. Y. Acad.

Sci. 60: 713.
SoBEL, A. E., and M. Burger. 1954. Calcification. XIV. Investigation of the role of

chondroitin sulfate in the calcifying mechanism. Proc. Soc. Exptl. Biol. Med.

87: 7.
Solomons, C. C., and J. T. Irving. 1958. The reactions of some hard- and soft-tissue

collagens with l-fluoro-2:4-dinitrobenzene. Biochem. J. 68: 499.
Speckman, T. W., and W. P. Norris. 1957. Bone crystalHtes as observed by the

electron microscope. Science 126: 753.
Strates, B. S., W. F. Neuman, and G. J. Levinskas. 1957. The solubility of bone

mineral. II. Precipitation of near-neutral solutions of calcium phosphate. /. Phys.

Chem. 61: 279.
Strates, B., and W. F. Neuman. 1958. On the mechanism of calcification. Proc.

Soc. Exptl. Biol. Med. 97: 688.
Tsujii, T., D. G. Sharp, and K. M. Wilbur. 1958. Studies on shell formation. VII.

The submicroscopic structure of the shell of the oyster Crassostrea virginica.

J. Biophys. Biochem. Cytol. 4: 275.
Turnbull, D. J. 1950a. Kinetics of heterogeneous nucleation. /. Chem. Phys. 18:

198.
Turnbull, D. J. 1950b. Formation of.crvstal nuclei in liquid metals. /. Appl. Phys.

21: 1022.
Turnbull, D. J. 1952. Kinetics of solidification of supercooled liquid mercury

droplets. /. Chem. Phys. 20: 411.
Turnbull, D. 1956. Phase changes. Advances in Solid State Physics 3: 225.
Turnbull, D., and B. Vonnegut. 1952. Nucleation catalysis. /n(^. Eng. Chem.

44: 1292.
Underwood' D. L., and H. J. White, Jr. 1954. The absorption of sodium sulphate

and sulphuric acid by hair. In The Physical Chemistry of Dyeing and Tanning.

Discussions Faraday Soc. No. 16: 66.
Watabe, N., D. G. Sharp, and K. M. Wilbur. 1958. Studies on shell formation.

VIII. Electron microscopy of crystal growth of the nacreous layer of the oyster

Crassostrea virginica. J. Biophys. Biochem. Cytol. 4: 281.
Zambotti, V. 1957. The biochemistrv of preosseous cartilage and of ossification.

Sci. Med. Ital. 5: 614.

DISCUSSION

K. E. Van Holde, M. J. Glimcher, J. W. Green, S. B. Barker,
L. V. Beck, C. L. Prosser, A. G, Richards

Dr. Van Holde (University of Illinois): Have you studied the binding of
phosphate ion by collagen fibrils or by soluble collagen?

Dr. Glimcher: We are currentlv studving the binding of both calcium and
phosphorus by collagen macromolecules in solution and bv the "solid" collagen
fibrils i^i small decalcified bone particles and plan to repeat the studies using
purified reconstituted collagens as well as collagen in solution. In these studies
we have encountered several difficult technical problems, as well as some
interesting phenomena which are difficult to interpret. In the first place, only

82 MACROMOLECULAR COMPLEXES

a relatively small amount of the phosphorus is bound by the collagen, making
quantitative determinations difficult. Secondly, the amount of phosphorus-
binding is so small that the correction for the amount of phosphorus in the
solvent water of the fibrils is proportionately large. Since there is no way as
yet of determining accurately the amount of "bound" or "free" water in the
fibrils, we have been unable to correct for this factor quantitatively. Thirdly,
part of the phosphorus appears to be irreversibly bound to the collagen; in
fact, there may be a spectrum of phosphorus-binding with different binding
energies, and when this is coupled with diftusion phenomena, the final results
are as yet impossible to interpret. Similar difficulties have been encountered
by workers using S-^-^ to study the sulfate-binding of solid keratin structures
such as hair (Underwood and White, 1954).

In relation to the mechanism of nucleation, when phosphate-binding was
measured in solutions the concentrations of which were similar to those used
in calcifying solutions (calcium and phosphate solutions), the amount of phos-
phate bound was very small compared to the amount of phosphorus bound or
otherwise adsorbed in the initial stages of nucleation before detectable amounts
of crystallization had occurred. This has suggested that the "building up" of
the nucleus is not a simple addition of phosphate and calcium ions independent
of each other, but rather a cooperative phenomenon in which the calcium ions,
phosphate ions, and certain amino acids interact collectively to influence the
state of aggregation of the mineral ions. To this must be added the possibility
that specific clusters of calcium and phosphate ions are the initially bound
moiety.

Dr. Green (Rutgers University): Have you examined the collagens in
elasmobranch fishes? Have you any information bearing on the failure of these
collagens to be mineralized?

Dr. Glimcher: We have onlv done a few experiments with the so-called
elastoidin of shark fins. As you know, this collagen has some of the low-angle,
as well as the wide-angle, x-ray diffraction properties of other native collagens,
but is cross-linked and very insoluble. The tissues also contain another as-yet-
unidentified protein (Gross and Dumsha, 1958). Since we were not able to
get the elastoidin free of other constituents, or to solubilize and reconstitute
the fibrils, we were unable to investigate this system adequately.

I can only speculate as to why such tissues are not normally mineralized.
We do not know whether, in fact, the collagen in elastoidin has the same
molecular structure and macromolecular aggregation state as other native
collagens. According to the concept which I presented today, however, if
the structural characteristics were the same, one would have to investigate:
(1) the metastability of the extracellular fluids; (2) the possibilitv of the
presence of inhibitory substances; and (3) the lack of regulatory mechanisms,
presumably enzyme-controlled, which are probably needed to initiate the
nucleation process in vivo. We could presumably get some useful information
about these questions if it were possible in vitro to test the mineralizing
capacity of both the native tissue and the purified elastoidin in a fashion similar
to the native collagen-rich tissues, and of the purified collagen from these
tissues in a fashion similar to the experiments which I have already described.

MACROMOLECULAR AGGREGATES IN CALCIFICATION 83

Dr. Barker (University of Alabama Medical Center): Should we not be
as critical of FDNB's blocking free amino groups other than e-amino groups
of lysine or hydroxylysine of reconstituted collagen as we are of the lack of
specific freeing of e-amino groups upon decalcification of bony structures?

Dr. Glimcher: I would be the first to agree that we must be extremely
cautious in interpreting any group-blocking experiments, particularly if only
one blocking agent such as FDNB is used. Because of the extremely large size
of the collagen macromolecule, the number of alpha amino groups is relatively
small; thus, we can be reasonably sure that the major effect of using FDNB is
to block the €-NH2 groups. As I pointed out, however, it is quite possible
that the actual sites involved in nucleation are adjacent to the e-NHo groups,
and that when one blocks the €-NH2 groups by substituting such large radicals
as FDNB, the really important adjacent groups are sterically (or electrostat-
ically) shielded and prevented from interacting with the mineral ions. We
would then be misled and erroneously assign a role to the 6-NH2 groups. It is
for these reasons, and not the lack of the specificity of FDNB (and other
blocking agents), that I am most concerned about in the interpretation of the
results of experiments such as the ones that I described.

Dr. Beck (University of Pittsburgh): Is it too early to speculate on the
role(s) of parathormone, Vitamin D, and dihvdroergosterol?

Dr. Glimcher: Yes!

Dr. Prosser (University of Illinois): Is there any correspondence between
apatite crystal formation in bone and calcification of chitin in crustaceans?
Also calcification in clams, where the protein does not show the periodicity of
collagen?

Dr. Glimcher: Although we have only studied the collagen hydroxyapa-
tite system in vitro, we believe that the underlying basic principle, wherein
the organic matrix nucleates the inorganic crystals as a result of a specific
stereochemical array of certain reactive groups, may apply to biological min-
eralization in general. In the case of oyster shells, for instance, calcium
carbonate in the form of either calcite or argonite is deposited in an intimate
relationship to the organic matrix, conchiolin, which is composed of conchiolin,
an unidentified protein. The fact that there is no repeating structure visible
with the electron microscope in the organic matrix of such material does not
mean that the matrix is not a well-organized and well-ordered structure at the
molecular level. In the case of collagen we are fortunate in being able to "see"
this ordering by electron microscopy and by low-angle x-ray diffraction be-
cause of certain characteristics dependent on the periodic concentration of cer-
tain reactive polar amino acids.

Dr. Richards (University of Minnesota): Do the apatite microcrystals in
bone have the typical crystal faces of larger crystals grown from solution in
vitro? If so, do you have any idea why in some cases, such as the subcutaneous
anchors of holothurian worms, one finds normal calcite crystals in weird shapes
such as (anchors?

Dr. Glimcher: Unfortunately, the apatite crystals of bone are so small
(15-30 A X 200-400 A) that it is impossible to be sure either of their gross
external habit or of their individual crystal faces.

MACROMOLECULAR COMPLEXES
84

Neither the physical nor the biological factors controlling crystal growth
are well enough known to answer the second part of your question. It may be
of interest, however, that the apatite crystals in the enamel of tooth are very
much larger than those in the adjacent dentin. Presumably, part of this may
be due to rates of crystal growth; i.e., the slower the rate, the larger and more
perfect the crystals. I have been told that the "spikes" of certain sea urchins
which are 7 inches or longer are composed of a single crystal of calcite.

The Chloroplast: Its Lamellar Structure
and Molecular Organization

J. J. WOLKEN -

A basic problem in biology is the conversion and transfer of en-
ergy, e.g., light energy to chemical energy or to electrical energy.
Recently, research workers have been investigating how the chloro-
plast, a differentiated cytoplasmic structure of plant cells, efficiently
performs this conversion in the process of photosynthesis. The
number of research publications, reviews, and symposia in the past
few years on the chloroplast structure and energy-transfer are evi-
dence of interest in this exciting problem (Boulder Conference on
Biophysics, 1959; Brookhaven Svmposia, 1958; Frey-Wyssling, 1957;
Gaffron, 1957; Granick, 1955; ' Miihlethaler, 1955; Thomas, 1955,
1958; Wolken, 1959). The purpose, then, of this discussion is to
describe the structure and composition of the chloroplast as revealed
by microscope studies and chemical analyses, and from the structure
and composition to suggest a model for the pigment molecules
within the chloroplast; then, if possible, from such a model to specu-
late about the molecular structure and function of the chloroplast
in energy transfer.

The photoreceptors of plant cells exist in a variety of shapes and
sizes, from those of the photosynthetic bacteria to those of the
higher plants. These structures have been given a varied nomen-
clature, i.e., chromatophores, megaplasts, plastids, and grana, de-
pending on their phylogenetic position and internal organization.
Because of inconsistencies in these definitions, even additional
nomenclature has been suggested. All of these structures except

' Aided in part by grants from U. S. P. H. S. Institute Neurological Diseases and
Blindness ( B-397 C5), National Council to Combat Blindness (G-199 C7), and the
McClintic Endowment.

^ Biophysical Research Laboratory, Eye and Ear Hospital, and the University of
Pittsburgh Medical School, Pittsburgh, Pennsylvania.

85

86

MACROMOLECULAR COMPLEXES

those in which the photosynthetic pigments are located in small
granules, chromatophores, will be here referred to as chloroplasts.
It is technically difficult to isolate chloroplasts from other cellular
components, and precise data on their chemical composition are not
easily obtainable. A summary of the data on isolated chloroplasts
^rom a variety of plants is given in Rabinowitch (1945, 1951, 1956).
In all of these studies of composition, particularly of the pigments
and their concentration, the state of the plant cells, age, nutrition,
temperature, and light conditions are extremely important.

r

1

D

O

1 TT

^— - 0-

1

G

]-@

Fig. 1. Diagram of microspectrophotometer. L, light source; G, mono-
chromotor; M, microscope with reflecting optics; C, photoconductive cell;
A, amplifier; D, oscilloscope.

A range of values on a dry-weight basis from 35 to 55 per cent for
the protein, 18 to 37 per cent for the lipids, and 5 to 8 per cent for
the inorganic materials is given for a variety of plant chloroplasts.
Studies are being made to determine the kinds of proteins and their
amino acid composition (Sissakian, 1958). Two cytochromes, cyto-
chrome f and cytochrome b^, together, can make up as much as 20
per cent of the chloroplast's total protein content. Nucleic acids
(both RNA and DNA) are also reported to be in the chloroplasts;
experimental values range from 0.3 to 3.5 per cent on a dr\'-weight
basis.

The chlorophyll pigments constitute about 5 to 8 per cent of the
chloroplast (although concentrations as high as 20 per cent have
been reported), and the carotenoids are about 2 per cent. Chloro-
phyll a occurs in all plant chloroplasts, but other isomers are also

THE CHLOROPLAST

87

found; in the higher plants, chlorophylls a and b occur in a ratio of
about 3:1. The number of chlorophyll molecules per chloroplast is
of the order of 10"'. The carotenoids, the xanthophylls, particularly
lutein and zeaxanthin, and the carotene, ^S-carotene, are intimately

400 500 600

WAVE LENGTH IN mp

700

Fig. 2. In vivo spectra from two different Euglena chloroplasts taken with
the microspectrophotometer (8/x- area).

associated with chlorophyll in the chloroplast. Other photosynthetic
pigments, phycoerythrin and phycocyanin, are present in the red
and bliie-green algae.

In order to analyze more directly the kinds of pigments and their
concentrations in a single chloroplast, we have designed and built a

MACROMOLECULAR COMPLEXES

microspectrophotometer (Strother and Wolken, 1959). The photo-
sensitive element is a cadmium selenide photoconductive cell with
a light-sensitive surface measuring approximately 0.5 mm X 1 mm.
The photocell output is amplified by a transistorized dc amplifier

0.7

0.6

r

y\

0.5

■

0.4

1
•

1

>

!

\

\

^\

-. 0.3

4

\

\\

CO

1

\ \

// I

z

Lul

1
1

\\

Q

i

(/ ♦

_l

1

O 0.2

-

^

^

h-

V\

1

a

V^

j 1

O

n

\\

\

1 \

•

1

\

0.1

\

r

.05

1 1 1

1 1

\\^

1 1 J L.

1 1

400 500 600

WAVE LENGTH IN mjJ

700

Fig. 3. Spectrum of Euglena chloroplastin (digitonin extract of chloroplasts),
from Beckman DK-1 recording spectrophotometer, compared to spectrum of

Euglena chloroplast taken with microspectrophotometer , chloro-

plast spectrum. , chloroplastin spectrum.

and displayed on an oscilloscope used as a dc voltmeter. A grating
monochromator is used for illumination, and the microscope is fitted
with reflecting optics (Fig. 1). Absorption spectra are obtained by
plotting readings of the dc output of the photocell at selected wave

THE CHLOROPLAST 89

lengths. The instrument, with quartz components and a xenon arc
Hght source, can l)e used over the wave-length range 200 to 990
m^u, at a low magnification for specimen areas of 16;Lt- to 2|U,-, at an
entering half -band width of 2.6 m/^. In Fig. 2 are shown the ab-
sorption spectra obtained with this instrument for the chloroplasts

MICRONS ACROSS CHLOROPLAST

Fig. 4. Absorption across a single chloroplast at 0.5/x intervals as measured
at 675 m/j. relative to 550 myu, with the microspectrophotometer, indicating the
distribution of chlorophyll in the chloroplast.

in two different euglenas. The magnification was 250X; an area of
8|U- is represented. These spectra begin to show fine structure. The
major absorption peaks are near 675 and 435 m/u, with other peaks
at 480, 590, and 625 yu/jl. Since these are in vivo spectra, some shifts
are to be anticipated from the spectrum of the pigments dissolved
in organic solvents. These spectra (Fig. 2) closely approximate

90 MACROMOLECULAR COMPLEXES

pievioush obtained spectra for the in vivo Euglena chloroplast and
for digitonin extracts of their chloroplasts (Fig. 3). The concentra-
tion of chlorophyll calculated from the absorption data corroborates
the previous analysis that there are of the order of 10^ chlorophyll
molecules per chloroplast. By scanning across the chloroplast at 0.5/>t
intervals at the absorption maximum 675 mfj. and at the absorption
minimum 550 m/>i, it can be shown that the chloroph\'ll is evenly
distributed throughout the whole chloroplast (Fig. 4). Because
of the limitation in the resolving power of the instrument it is not
possible to locate chlorophyll precisely in the chloroplast lamellae.
Together with microspectrophotometry, the techniques of polariza-
tion, fluorescence, phase, and interference microscopy are the most
direct approach for exploring cellular structure. These methods per-
mit the study of the chloroplasts in their natural state.

The electron microscope now permits us to see through the fixed
chloroplast in situ instead of seeing its shadow or replica. The in-
creased resolution of the electron microscope has greatly aided our
attempts to obtain a molecular picture of the chloroplast. In the
preparation for electron microscopy, the cells were fixed with 1 per
cent osmium tetroxide buffered with acetate-veronal pH 7 to 8. The
osmotic changes were partially controlled by adding sucrose
(0.15M) to the fixative. The fixed cells were embedded in a resin
( /j-butyl methacrylate, methyl methacrylate, or mixtures of these ) .
Thin sections, less than 0.05/x in thickness, were cut with a glass
knife.

Chloroplast Structure

There is now good evidence from a great variety of plant species
that, although chloroplasts may differ in shape and size, they have
in common a basic internal organization. We have investigated the
structure of chloroplasts in several algae and higher plants with the
electron microscope. These studies reveal that the chloroplasts are
composed of lamellae.

We have found the algal flagellate Euglena to be uniquely adapt-
able for experimental studies of the chloroplast structure, since
Euglena behaves as a photosynthetic plant cell in the light and as
an animal cell in the dark. Changes in the environmental conditions
( light, darkness, temperature, drugs ) are reflected in the organisms'
chemistry and morphology. Using Euglena, we have been able to

THE CHLOROPLAST 91

study the chloroplast by xarious microscope techniques and have
attempted to correlate structural changes of the chloroplast with
pigment synthesis and with photosynthesis in light ^ dark adapta-
tion (Wol'ken axid Palade, 1953; Wolken and Sciwertz, 1953; Wol-
ken et al, 1955; Wolken and Mellon, 1956; Wolken, 1956a).

E. gracilis when light-grown contains manv green elongated
cylindrical chloroplasts 5 to 10/>t in length and 0.5 to 2/>i in diameter.
When the chloroplasts are viewed by interference-phase, by absorp-
tion, and by polarizing microscopv, an ordered structure is indicated.
Previously, it was proposed from studies of negative birefringence
that the chloroplast is a lamellar structure of 20 to 30 lipid layers of
the order of 50 A, each separated from layers of aqueous protein of
the order of 250 A by monomolecular films of chlorophyll molecules
( Frey-Wyssling, 1957). There is now excellent proof of this lamel-
lar structure from electron microscope studies, and x-ray diffraction
techniques have corroborated the electron microscope observations,
showing a repeating unit of the order of 250 A.

In order to determine more precisely the lamellar structure and
to test the pigment monolayer hypothesis, the diameter, length,
number, and thickness of the lamellae of the chloroplasts were meas-
ured and statistically evaluated (Wolken and Schwertz, 1953). For
example, the chloroplast of E. gracilis consists of 21 dense lavers of
the order of 250 A in thickness, with less dense interspaces of 300 to
500 A in thickness. Each dense layer appears to be covered on both
sides by a thinner and denser layer (lamellae) of the order of 50
to 100 A in thickness. The average thickness measured for a variety
of plant chloroplasts ranges from 20 to 100 A for the electron-dense
lamellae (Leyon, 1956; Sager and Palade, 1957; Sager, 1958). Frey-
Wyssling ( 1957 ) had noted in previous studies that the thin lamellae
consist of spherical particles 65 A in diameter; these were considered
to be the protein or lipoprotein macromolecules (Fig. 6).

There is experimental evidence to indicate that the chlorophyll
molecules are preferentially oriented parallel to the lamellae and are
within the very dense (lipoprotein) lamellae (Thomas, 1955). The
electron-dense layers are believed to be lipoprotein and lipid mate-
rial. The areas of least density are considered to be aqueous pro-
teins, enzymes, and dissolved salts. From the geometry of the
chloroplast, the number of dense layers, and the chlorophyll con-
centration per chloroplast (Table 1), we have calculated the cross-
sectional area occupied by each chlorophyll molecule to be 222 A"

92 MACROMOLECULAR COMPLEXES

for the Euglena chloroplast and 246 A- for the Poteriochromonas
chloroplast (Wolken and Schwertz, 1953). Elbers et al. (1957) have
since collected data on chlorophyll concentrations and geometry of
chloroplasts in a variety of plant species, and by similar analogy
have calculated the mean area available for the chlorophyll molecule
in the monolayer to be of the order of 200 A-. Studies of the di-
chroism, birefringence, and polarization of fluorescence in Motigeo-
tia chloroplasts also indicate that chlorophyll resides as a monolayer
on the lamellar surface, and the area available per chlorophyll
molecule was calculated to be ~250 A" (Goedheer, 1955, 1957).

TABLE 1
Geometry of Chloroplasts

Organism N d

Thick-
ness

n

A»

Euglena gracilis 1.02 X 10^ 6.50m

242 A
390 A

21
10

222 A-

Poteriochromonas stipitata 1.10 X 10^ 3.26m

246 A^

■* A, area accessible to each chlorophyll molecule, is Ci;
^ {8/n)nd'

lie,

Lilated from

where n is the average number of dense layers (double lamellae), providing 2n sur-
faces for the pigment molecules, d is the average observed length of the chloroplast,
and N is the average number of chlorophyll molecules per chloroplast (Wolken and
Schwertz, 1953).

Since the cross-sectional area of the porphyrin head of the chloro-
phyll molecule is known from x-ray studies to be about 225 A" to
242 A", these results indicate that all the available chlorophyll mole-
cules could be packed into the interfacial area and cover all of the
dense surface of the lamellae as a monolayer.

On the basis of these calculations, a simplified schematic molec-
ular model was proposed (Fig. 5). The suggestion of Bass-Becking
and Hanson (1937), that four chlorophyll molecules are united to
form tetrads in which the reactive isocylic rings turn toward each
other, was employed (Wolken and Schwertz, 1953). Interaction be-
tween the phytol tails was eliminated in the model by arranging the
tetrads in such a way that one, and only one, of the phytol tails is
located at each virtual intersection in the rectangular network. If the
chlorophyll were packed as a monolayer as shown in the schematic
molecular network in Fig. 5(b), there would still be space available

THE CHLOROPLAST

93

a 2

.:: a.

n 11

Up]

t PA I

'//

4/v '.

o o o
_o o _2
-F. E-c

3 (U ^

f> *- a

D i2

Q. Q)

0-2

m

(U

D O (U 0) ^
^ = Z 0^:5
S E I ^ 2

ti ° D ^ 2

Q..E g'l "
-0.2 1-^^

^ "O ^ T! "O
U C w 0 C

.C </) C 0 —

D "o Q) 0 °

w C u .E

0 O <U <U -^

E O) o = Q.

.y o c -D 0

c •£ O) c
t C 0 D)

<^ sj

to

6>

o ^

94

mmassmm

i- ,.,-gigi?ig.. ■

MACROMOLECULAR COMPLEXES
. ♦ » . • •• . . •

PROTEIN

T

~«5;^ h

•1,^

. • • ..*. • • ^"^i/* • •,

Fig. 6. (a) Schematic structural identification of the chloroplast layers.
(b) Electron micrograph of chloroplastin, not fixed and not shadowed, (c) Elec-
tron micrograph of the chloroplast of Euglena granulata.

for the carotenoid molecules at the interstitial positions between
the chlorophyll molecules. If these spaces are occupied as shown,
there will be one carotenoid molecule for at least every three chloro-
phyll molecules in the network. This kind of close packing of the
chlorophyll and carotenoid molecules in the pigment monolayers
would also permit energetic interaction between them. Other
models slightly modified from the one presented here indicate that

THE CHLOROPLAST 95

the chloioplnll molecules are also turned inward, as well as being
oriented on the surface (Hodge ct ah, 1955). Calvin has recently
suggested a similar model in which the porphyrin heads lie at an
angle of 45°; one protein layer has CO^-reducing enzymes and an-
other protein la\'er has Oj-evolving enzymes (Calvin, 1958, 1959a,
1959b).

There are several possible ways in which the chlorophyll mole-
cules could be oriented in the lamellae. If the porphyrin heads of
the chlorophyll molecules lav at 0° as flat plates as indicated in Fig.
5(b), their greatest cross-section would be available. However, if
they were oriented within the lamellae at increasing angles up to
90°, the cross-sectional area available would be decreasing, as would
the photosynthetic efficiency. Studies of chlorophyll monolayers on
various liquid surfaces suggested that the chloroph\'ll molecules
would probably lie at an angle of 35° to 55° within the chloroplast,
thus reducing the above calculation for the cross-section of the
chlorophyll molecule to 100 A- (Trurnit and Colmano, 1958). It is
very likely that the absorption oscillators of these pigment molecules
are arranged with an orderly orientation in a way that a maximum
absorption will occur for an incident light polarized in a given direc-
tion.

The question of how the chloroplast evolves from a rudimentary
proplastid into an organized structure is a difficult one to answer.
Euglenas grown in darkness for long periods become colorless. We
have noted that there is an obvious change in the chloroplasts; they
fragment, and their lamellar structure is no longer recognizable.
Such dark-adapted euglenas, brought back to the light and fixed for
electron microscopy at regular time intervals, showed in electron
micrographs, after as little as 4 hours' light exposure, recognizable
elongated bodies with the characteristic laminations of chloroplasts.
At the beginning, their lamellae were very thin and few in number
and were not tightly or regularly packed. The lamellae grew in size
and number progressively with continuing light exposure along with
the increase in the synthesis of chlorophyll, and, by 72 hours of
light exposure, the chloroplasts had the form and organization
described for actively photosynthesizing euglenas (Wolken and
Palade, 1953). As soon as we could detect chlorophyll spectrophoto-
metribally, the chloroplasts were already lamellar. The alga Chlamij-
domonos, whether grown in light or in darkness, shows a chloroplast
organization of lamellae, pyrenoid, and eyespot within a limiting

Fig. 7. (See next page for caption.
96

THE CHLOROPLAST 97

membrane. In the absence of chlorophyll, the lamellae are not
formed, although the eyespot, starch grains, and pyrenoid are still
found within the chloroplast membrane (Sager and Palade, 1957;
Sager, 1958).

In the development of the chloroplast in flowering plants, the
first visible stages lack all the structural properties typical of the
fully developed chloroplast. Heitz (1955) and Leyon (1956) con-
clude from electron microscope studies that there is a primarv
granum that has a crvstalline structure. Whether such crystalline
bodies exist in the living state or whether their crvstallinity is en-
hanced by preparative techniques for electron microscopy is difficult
to decide. The crystalline structure of the primary granum, as
found in Chlorophijtum, Aspidistra, and other monocotyledons, is
also present in dicotyledons. This suggests that there must exist
some association between the crystalline lattice and the develop-
ment of the lamellae ( Heitz, 1955 ) . Miihlethaler ( 1959 ) has shown
that the lamellar structure develops from "tubules" or "cristae" anal-
ogous to mitochondria. Upon illumination, the cystalline structure
develops into a lamellar system as seen in the full-grown chloroplast
of Elodea canadensis ( Fig. 7a and b ) . It is indicated that here too
the formation of the lamellae proceeds parallel with the pigment
synthesis.

Von Wettstein investigated the relation of gene action to the sub-
microscopic structure of the chloroplasts by employing chlorophyll
lethals of barley (von Wettstein, 1957a, 1957b, and 1958). In the
development of the barley chloroplast, there is an undifferentiated
proplastid with a center core; starch grains accumulate, lamellae are
formed, and the plastid as a whole becomes trans versed by lamellae.
In an albino type, essentially normal early proplastids enclosed by
a double membrane were found. There is no differentiation of
ordered layers, lamellae, or grana. Differentiation does not proceed
beyond the earliest stages, but the plastid increases in size to that
of a full-grown chloroplast. The gene change has blocked the de-
velopment of the submicroscopic chloroplast structure at an early
stage, but not the growth of the plastid as a whole. Instead of the

Fig. 7. Electron micrographs of (a, b) developing chloroplast in Elodea
canadensis, a higher plant, showing (a) crystalline structure and (b) fully
developed chloroplast (Courfesy, Dr. K. Muhlethaler, Eidgenossische Technische
Hochschule, Zurich), (c) Chloroplast of Poteriochromonas stipitafa, a chryso-
moncd.

98 MACROMOLECULAR COMPLEXES

formation of lamellae, globules accumulate in large masses, and the
chloroplast function appears to be arrested. The pigment concen-
tration in the mutant indicates that there is a greater accumulation
of chlorophyll b, and that the pigments are probably located in the
globules. In older stages of the plastids in this mutant, the globules
are broken down with further breaks in the inner structure, along
with a decrease in pigment concentration.

The picture of chloroplast development still remains inconclusive,
and more experimental work is necessary. However, in order for the
plant to carry on photosynthesis, an ordered lamellar structure is
necessary. As already noted, the chloroplast contains nucleic acid.
It is very suggestive that nucleoproteins are involved in the synthesis
of the lamellar lipoproteins of the chloroplast, and that the synthe-
sis of chlorophyll and the lamellae proceed simultaneously.

Chloroplast Function as Related to Structure

What does the chloroplast structure tell us about how the chloro-
plast functions in photosynthesis? That is, how does the conversion
of light energy to chemical energy take place via chlorophyll? The
chloroplast is a polyphasic system comprising a stacked array of
layers of lipids and proteins separated by monolayers of chlorophyll.
Experimentally in Eii galena it has been found that any physical or
chemical forces (i.e., light, temperature, drugs) that affect the syn-
thesis of either chlorophyll or the chlorophyll-complex will disrupt
the lamellar structure (Wolken et ah, 1955; Wolken, 1956a). This
change was associated with the removal of Mg++ from the chloro-
phyll molecule. Is the chloroplast, then, like an emulsion, which is
stabilized by the chlorophyll molecules? Levitt (1954) suggests
that the magnesium atom is the necessary part of the molecule and
that any theory of the conversion of light energy into chemical en-
ergy via chlorophyll has to take into account the valence changes of
the magnesium ion.

Resonance studies indicate that the chloroplast has properties of
a crystal and, together wih the evidence of an ordered structure
from electron microscopy, permit an analogy to a photobattery—
a semiconductor (Calvin, 1958). The assumption is that we are
dealing with an electron-transfer phenomenon through many mole-
cules by a conduction-band mechanism. Films of both chlorophyll
a and methyl chlorophyllide a deposited from organic solvents have

THE CHLOROPLAST 99

been shown to be photoconductive (Nelson, 1957). Monolayers of
chlorophyll, chlorophyll plus ^-carotene, or yS-carotene alone, spread
on various surfaces, have been demonstrated to be photoconductive
(Arnold and Maclay, 1958). The role of the carotenoids in the
chloroplast is more controversial. Lynch and French (1957) found
that when the chloroplasts were extracted with petroleum ether, the
chloroplasts were inactivated, whereas the addition of ^-carotene
restored their activity; hence, /3-carotene must be an active com-
ponent of the photochemical system. This study neglected the role
of the lipids in the complex. Fujimori and Livingston (1957), from
flash-photoh'sis studies, also indicated that carotene plavs a direct
part in the primary act of photosynthesis. On the other hand, Stanier
(1958), working with mutants of the purple bacterium Rhodospiril-
lum, and Sager (1958), with mutants of the alga Chlamijdomonas,
indicate that the carotenoids may act only in catalvtic amounts and
protect the chlorophyll molecules against high light intensities, tem-
peratures, and oxygen concentrations. Piatt (1959) suggests that
if the carotenoids are part of the primary sequence, there is a
"photosynthetic unit" that can transfer their absorbed energy to a
single reaction site.

Chlorophyll and carotenoid in their natural state within the
chloroplast do not exist as free pigments but are bound to proteins
or lipoproteins. There is an obvious analogy to hemoglobin, where
heme is bound to globin, i.e., a chloroglobin. In order to build some
models for the chloroplast structure, we have undertaken several
experimental approaches.

Chlorophyll-Protein Complex. Investigations of chlorophvll
complexed with proteins indicated that some of the complexes have
properties similar to those of active chloroplasts (Rodrigo, 1955).
We have tried to learn something of the nature of these chlorophyll-
protein complexes formed with a variety of proteins. The chloro-
phyll was extracted from the chloroplasts in 85 per cent acetone
(including the carotenoids and other extractables ) and was added
to various protein solutions (20 mg/ml in 0.9 per cent NaCl), until
no further precipitation occurred. The heavy green chlorophyll-
protein precipitate was then washed and resuspended in salt solu-
tions of various concentrations. The stability of the complex was
deteitnined from absorption spectra, electrophoresis, and analytical
ultracentrifugation. It was found that the extracted chlorophyll
complexed more readily with native globulins containing lipid and

TOO MACROMOLECULAR COMPLEXES

lipoproteins. No complexing could be achieved with the albumins,
gelatin, or peptones. In most cases there was difficulty in resolubil-
izing the precipitate, and pheophvtin was formed. The complex was
not soluble in reagents which are known to solubilize lipoproteins,
but was readily soluble in digitonin. Attempts to demonstrate that
these complexes are photoactive were on the whole unsuccessful.
The difficulty with this procedure may be that the acetone causes
denaturation of the proteins. An analysis of two such complexes,
the chlorophyll-horse serum and the chlorophyll-y globulin, is shown
in Table 2.

TABLE 2
Pigment-Protein Complex

M

p ( Mol wt calcu-

( Moles chloro- n lated from

Complex phyll/1 ) ( Mg Nitrogen/ml ) p and n )

Chlorophyll

with horse sermn 11.4 X 10-' 1.2 69,300

Chlorophyll

with 7-globulin (bovine) .. 2.42x10"'' 0.62 173,000

Chloroplastin. Another possibihty was to extract the whole
chloroplastic material from the chloroplasts by surface-active agents.
Detergents in solution form colloidal aggregates, micelles, consisting
of many molecules of detergent clumped together. Such micellar
bodies have a very strong attraction for many of the more complex
dye molecules. Commercially available non-ionic and ionic surface
agents were used to extract the pigment complex from the chloro-
plasts. The non-ionic recrystallized detergents, Nacconal NRSF ^
and digitonin,^ were found superior to all other surface-active agents
tested. These are nitrogen-free detergents and therefore do not
interfere with the protein analyses. Although the ionic detergents
Alkonal B, glycocholate, and taurocholate were noted as good
solubilizing agents, they did not preserve the spectral and other
properties of the chlorophyll- complex. The digitonin-extracted pig-
ment-complex therefore serves as the best experimental material.

■' Nacconal NRSF, anionic organic detergent, alkyl aryl sulfonate, Allied Chemical
and Dye Corporation.

•* Digitonin, anionic detergent ( C5.-,H[,o02o ) D-58, Fisher Scientific Co.

THE CHLOROPLAST 101

This method of obtaining a pigment-complex is similar to that
used for the visual complex, rhodopsin, from retinal rods. From the
chloroplasts of Etiglcna and spinach we have extracted in 1.8 per
cent recrystallized digitonin a pigment-protein complex containing
both chlorophyll and carotenoids (Wolken, 1956b; Eversole and
Wolken, 1958). Other such chloroplastins have been extracted
previously by Smith (1941a, 1941b, 1941c). We have been able
to determine the physical and chemical properties of chloroplastin
and to compare them with those of the chloroplast in vivo. The
absorption spectrum of chloroplastin closely approximates the ab-
sorption spectra of the pigments in the in vivo chloroplasts ( Fig. 3 ) .
Chloroplastin ^ appears to be a homogeneous complex, judging from
its sedimentation in the analytical ultracentrifuge and its electro-
phoretic pattern (Wolken, 1956b, 1958).

Temperature and Wave Length. The bleaching of chlorophyll
in Euglena can be induced by growing the organisms in darkness or,
in light, by raising the temperature above 32° C. Chloroplastin
(pH 7.2) can be bleached by light or by heat, analogous to the
organisms in vivo. Chloroplastin bleaches at a rate proportional to
the amount of light or heat energy absorbed, causing a steady de-
crease in optical density with the disappearance of the maximum
absorption peak at 675 m^u,; it is bleached to pheophytin and unknown
products. Below 560 m/x, bleaching is accomplished by light energy
alone; above 560 m/x, the combined effects of light and heat are
required. The total activation energy for bleaching is 48.3 kcal/mole
(Wolken and Mellon, 1957).

Photochemistry. Photochemical reactions analogous to photo-
synthesis were carried out with Euglena chloroplastin, e.g., the rate
of photoreduction of the dye, 2,6-dichlorobenzenoneindophenol, the
evolution of oxygen, and the conversion of inorganic phosphate to
labile phosphate (ATP).

The rate of photoreduction of the dye, 2,6-dichlorobenzenone-
indophenol (3 X 10~'' M), was measured at 600 myu, after illumina-
tion for 1-4 minutes with 300 foot-candles. In 20 per cent of the
preparations, photoreduction of dye was complete within 2 minutes
with no accompanying reduction in darkness. The molecular turn-
overt with a typical preparation indicated that 7 X 10^ molecules of

'' Chloroplastin is not a purified macromolecule as it contains a cytochrome c type
haem protein as well as other impurities.

102 MACROMOLECULAR COMPLEXES

dye could be photoreduced per molecule of chlorophvll per minute;
this is preparation-dependent, and the number may not be quanti-
tatively significant. When the reaction was followed at 488 m/x
(a carotenoid peak), changes in optical density with light or dark-
ness suggested that the carotenoid participates in the reaction
(Eversole and Wolken, 1958).

Extracts which actively photoreduced the dye were tested for
their ability to cause photolvsis or evolution of oxygen. Photolxsis
was measured manometrically in completely anaerobic Warburg
vessels with KOH in the center well, and the dye was tipped from
a side arm at zero time. The reaction system for photolysis measure-
ments was made oxygen-free in order to estimate the oxygen evolved
by bacterial bioluminescence. In some preparations of chloroplastin
photolysis occurred, yielding 20-30 fA ox\'gen in 2 minutes, with a
distinct luminescent glow persisting for almost a minute after a
suspension of anaerobic Photobacterium phosphoreum was injected
into the experimental vessel in darkness. Controls in the absence of
light and those without dye did not evolve oxygen.

In the experiments in which photolysis occurred, a light-catahzed
conversion of inorganic phosphate into labile phosphate ( ATP ) was
measurable. The reaction vessels contained 2 ml of chloroplastin
(having a chlorophyll concentration of 10""' M), 20 ^umole Mg"^"^,
30 yumole keto-glutarate, 0.3 yumole riboflavin-5-phosphate, 0.6 ^amole
menadione, 2 /^mole ascorbate, 5 ;ug cytochrome c, 55 /xmole adenosine
monophosphate, and 4 ^g inorganic phosphate. These experiments
were immediately repeated with the addition of glucose and hexo-
kinase, and the glucose-6-phosphate formed was determined by
triphosphopyridine nucleotide reduction at 340 m^ in the presence
of glucose-6-phosphate dehydrogenase. In this way 80-90 per cent
of the inorganic phosphate disappearing was accounted for as labile
phosphate. The phosphate conversion occurring in the dark control
was only 3-4 per cent of that found in the light.

Electron Microscopy. When a film of chloroplastin was evap-
orated from a digitonin solution on an electron microscope screen,
not fixed, and viewed directly in the electron microscope, there was
sufficient electron density to reveal particles of an average diameter
of 100 A. A digitonin solution alone when examined in the same
manner shows no such electron-dense particles. The electron den-
sity of the chloroplastin particles is attributable to the Mg^+ of the

THE CHLOROPLAST 103

chloiophvll molecule and other metals associated with chloroplast.
These chloroplastin particles are of the same order of size as those
found in the fixed dense lamellae of some chloroplasts; also they
could be considered similar to chromatophores. Therefore, it is
suggestive that chloroplastin is physiologically active, since the
chloroph\ll complex remains oriented in the digitonin micellar
particles. These particles could contain from 16 to 32 chlorophyll
molecules on the surface, as depicted by Frey-Wyssling ( 1957 ) .
The number of molecules that could be packed into this space would
be doubled if the chlorophylls were tilted on an angle of 45°, as
previously suggested. Such pigment molecules oriented in the dig-
itonin micelle particles could then be the active "photosynthetic
units." (See Fig. 6 for electron micrograph of Euglcna chloroplast
and chloroplastin. )

X-RAY Diffraction. Chlorophyll a has been shown to crystallize
out in thin sheets ^50 A thick, perhaps corresponding to bimolecular
layers of chlorophyll. The crystallized chlorophyll molecules occupy
an area of ~ 106 A-, and x-ray diffraction patterns obtained from
these crystals show strong bands at 7.6 A and 4 A (Jacob ct al.,
1954). X-ray diffraction pictures were obtained by Belavtseva
(1957) from crystallized chlorophyll a plus h, prepared by petro-
leum ether extraction in Krasnovskii's laboratory. These patterns
showed a strong band at 7.68 A, with average bands at 4.22, 4.06,
3.66, 2.78, and 2.44 A, and weak bands at 2.05, 1.83, and 1.62 A.

X-ray diffraction patterns were prepared at the Mellon Institute
for chloroplastin solutions and for chloroplastin films on glass slides.
For the solution, strong bands were found at 3.4 to 2.9 A and another
at 2.3 to 2.1 A; for the film, a broad band from 5.3 to 3.9 A was ob-
tained, with weak lines at 3.35, 2.13, 2.02, 1.82, 1.77, and 1.57 A.
These numbers are interplanar spacings. It is difficult at present to
interpret these bands; however, the 4 A band and some of the other
lines of chloroplastin in dried films approximate those for the crystal-
lized chlorophyll.

Molecular Weight. Previously suggested molecular weights for
chloroplastin have been calculated for leaf extracts (spinach and
Aspidistra) prepared in a similar manner with digitonin. These
calculations, made from sedimentation rates in the analytical ultra-
centrifuge, give a molecular weight of the order of 265,000 ( Smith,
1941a, 1941b, 1941c). This is a high estimate, due in part to the

104 MACROMOLECULAR COMPLEXES

contribution of the digitonin micelle. Digitonin alone forms micelles
of minimum molecular weight equal to 75,000. Three such micelles
of digitonin are probably linked, yielding a molecular weight of
225,000 (Hubbard, 1954). If the digitonin micellar weight is sub-
tracted from the molecular weight of 265,000 for the complex, the
molecular weight of chloroplastin would be of the order of 40,000
(Table 3). Takashima (1952) crystallized from leaf extracts a

TABLE 3
Analytical Ultracentrifuge Data

Sedimenting
Complex

Average
Sedimentation
(S.0 X 10-'3)

Average
Complex
Micelle
Weight

(M')

Average

Molecular

Weight

(M)'

7.1

155,000
290,000

155,000

Euglena chloroplastin

( digitonin + chloroplastin )

13.5

40,000

* M is calculated from M' using the dry weight and per cent nitrogen.

chlorophyll-lipoprotein complex with a molecular weight of 19,000,
determined from diffusion studies. The crystalline complex indi-
cated that there were two molecules of chlorophyll per lipoprotein
molecule. A chlorophyll "holochrome" isolated by Smith and Young
(1956) from bean seedlings {Phaseoloiis vulgaris) in glvcerine-
KOH at pH 9.6 suggests a molecular weight of the order of 400,000.^
For the chloroplasts of the algal flagellates Euglena and Poterio-
chromonas, the average molecular weights calculated from the
geometry of the chloroplast and the pigment concentration were
21,000 and 37,000, respectively. Using the interference microscope,
molecular weights were calculated to be of the order of one-half of
these values for the in vivo chloroplast. The molecular weight for
chloroplastin, calculated from its sedimentation, nitrogen, pigment
concentration, and dry weight, is of the order of 40,000 (Wolken,
1956b). The chloroplastin data implv that there is one chlorophyll
molecule to one protein molecule, although the possibility of two or
more chlorophyll molecules cannot be excluded (Table 4). Frey-
Wyssling predicts that there are 16 molecules of chlorophyll to one

"The molecular weight of this complex is now reported to be 1 X 10*^ (Smith,
1959).

THE CHLOROPLAST 105

macromolecule (on the assumption that the macromolecule is 65 A in
diameter ) . Therefore, the chlorophyll-protein complex would have a
molecular weight of the order of 68,000. It must be remembered,
though, that chlorophyll could be complexed with different proteins
in the various plant species, and that the molecular weights would
vary, depending upon the kind of protein. The nature of the pro-
tein, or how chlorophyll is complexed to the macromolecule, has
not yet been elucidated.

TABLE 4
Analysis of Euglena Chloroplastin

w dry wt, mg/m! 27.3

n mg nitrogen/ml 0.36

/; moles chlorophyll 1 6.3 X 10"^

p' moles chlorophyll/1 calculated from w/M' 9.4 X 10"^

M mol wt calculated from M' 38,000

M mol wt calculated from w, n, and p 30,000-60,000

Crystallization and Periodicity

The next steps in our exploration were to see if there were any
possible wav of accounting for such a lamellar structure for the pig-
ment complex in the chloroplast. Here I would like to refer to some
experiments on the Liesegang phenomena (Hedges, 1932). Liese-
gang observed the formation of periodic structures in the course of
staining histological specimens by the Golgi technique (i.e., the
impregnation of tissue with potassium dichromate and silver ni-
trate ) . The Liesegang phenomenon has been often cited as a model
for study of growth and differentiation in living cells. Antigen-anti-
body reactions in gels also have been compared to Liesegang ring
phenomena.

The formation of the rings can be observed if a drop of 15 per
cent silver nitrate is placed on a sheet of gelatin which has been
impregnated with about 0.4 per cent potassium dichromate. The
silver slowly diffuses into the gelatin, there reacts with the potassium
dichromate, and silver dichromate is precipitated in the gelatin.
The precipitation is not continuous but forms a series of concen-
tric rings separated by clear spaces in the gel. ( Fig. 8d. ) There are
many other such examples; if ferric chloride be added to gelatin
and a drop of potassium ferrocyanide solution placed in the center,
blue rings of ferriferrocyanide will be formed. Light can modify

106 MACROMOLECULAR COMPLEXES

these periodic precipitations if the precipitated molecules are light-
sensitive. Many experiments on the Liesegang ring formation have
been performed; data on the concentration of reactants, tempera-
ture, interspace distances between ring formations, and the time of
the reactions have been collected. How much the experimental data
can tell us about the chlorophvll-protein specificity or the orienta-
tion of chlorophyll in the chloroplast is at present questionable, but
it has led us to some interesting considerations on periodic crystal-
lization in colloids and proteins.

Periodic Crystallization. Molecules in solution take up con-
figurations of lowest energy. This leads to crystallization when the
number of molecules in the solution exceeds a certain minimum
value characteristic of those particular molecules. Such molecular
interaction described in terms of crystallization can perhaps be
extended to the kinds of molecules that are present in lamellar sys-
tems of the chloroplast. An example of such periodic crystallization
is that of potassium dichromate in gelatin. The procedure is to place
a drop of saturated potassium dichromate in gelatin solution on a
microscope slide, warm gently, and quickly transfer the slide to
the microscope. Crystallization begins around the periphery of the
drop and proceeds in a periodic manner, bv alternating periods of
rapid and slow growth during which a few relatively large geo-
metrical crystals grow. The distance between the rings decreases
with the thickness of the film and with increasing rate of crystalliza-
tion. (Fig. 8c). The crystallization of sodium chloride is also
affected by extremely small concentrations of most colloids (e.g.,
serum ) ; as the ions are absorbed on the surface, the sodium chloride
molecules are deposited in fine crystals on the glass surface, and
periodic rings are formed at certain salt concentrations (Du Noiiy,
1926). Digitonin, when it evaporates as a drop on a glass surface,
forms periodic rings; however, when a digitonin solution of chloro-
plastin evaporates as a drop, similar rings develop, but now the
chlorophyll concentrates in these rings ( Fig. 8a and b ) . The chloro-
phyll molecules are carried with the digitonin and become oriented
within these rings. When the rings were scanned from the center
out with the microspectrophotometer at 675 mix, the major absorp-
tion peak for chloroplastin, and at 550 m/x, the minimum absorption
peak, it was found that the maximum chlorophyll absorption occurs
in the rings, and although the concentration of chlorophyll decreases
as it moves out, its concentration remains relatively constant within

THE CHLOROPLAST

r

#

- *•■

F 0 J

,- f

/-

t

■J' i

f *

4

f J

^ 4

J

9^

/,/

Fig. 8. (a) Periodic ring formation from a 1.8 per cent solution of digitonin
evaporated from a drop on a glass surface. xlOO. (b) Periodic rings from
chloroplastin similarly evaporated from solution. X130. (c) Periodic crystalliza-
tion of potassium dichromate from gelatin. '140. (d) Liesegang ring forma-
tion obtained from saturated solution of potassium dichromate in gelatin reacted
with silver nitrate. XI 40.

108

MACROMOLECULAR COMPLEXES

NUMBER OF RINGS

Fig. 9. (a) Periodic rings of chloroplastin prepared as in Fig. 8b (XlOO).
(b) Chloroplastin rings scanned at 675 m/x with the microspectrophotometer,
showing that chlorophyll is located within the ring structures.

each ring. (Fig. 9). The chlorophvll complex in digitonin there-
fore behaves in a fashion similar to salts crystallizing out from
colloids, gels, and proteins, and to the Liesegang phenomenon.

Now let us turn to some other experimental studies on the
properties of liquid crystals. Liquid crystals generally exist in either
the smectic or the nematic paracrystalline state. The more highly

THE CHLOROPLAST 109

ordered of these is the smectic, where the molecules are arranged in
equidistant parallel layers. Robinson (1958) has shown that poly-
adenylic acid and polyuridylic acid, when mixed under the right
conditions, show a specific interaction, forming crystals. Robinson
and Ward ( 1957 ) have also shown that when the L and D forms of
the synthetic polypeptide, polybenzoylglutamate, are mixed in
dioxan, they behave as a liquid crystal. Although the solution was
birefringent, no regular orientation was observed with the polar-
ization microscope; but after a short time, orientation appeared on
the walls of the capillary and spread toward the center with a very
regular orientation.

If digitonin is caused to flow through a capillary, it becomes
birefringent when observed through crossed polaroids; a similar
phenomenon is observed with chloroplastin. This is due to the
orientation of the particles in the flowing stream. The chlorophylls
then become oriented within the digitonin micelles. If chloroplastin
is allowed to stand, no birefringence is observed. However, a liquid
crystalline phase will separate out; under certain conditions, bire-
fringent rodlike fibers (tactoids) form. These observations are simi-
lar to those made on tobacco mosaic virus (Lauffer et at, 1949).

Quantitative theories have been proposed to explain liquid crys-
tals as being the result of anisotropic interactions between long, rod-
like molecules. These theories have the essential feature that long
rods can be packed economically in a given space only if they are
aligned. Hence, a concentrated solution of rodlike molecules tends
to form liquid crystal structures, but in dilute solutions, isotropic
structures occur (Zimm, 1959). For a more complete discussion
of liquid crystals, reference should be made to the Discussions of
the Faraday Society (1958) and Transactions of the Faraday So-
ciety (1933').

These studies suggest that chloroplastin possesses properties of
a liquid crystal and because of these properties, structural as well as
photochemical integrity is maintained. Its behavior then is similar
to that of the in vivo chloroplast. The lamellar structure may be an
efficiency mechanism rather than a critical functioning device. How-
ever, this does not exclude the probability that a liquid crystal type
of matrix is necessary for orientation of the active pigment complex.

no MACROMOLECULAR COMPLEXES

References

Arnold, W., and H. K. Maclay. 1958. In The Photochemical Apparatus: Its Struc-
ture and Function. Brookhaven Stjmposia in Biol. No. 11: 1-9.

Bass-Becking, L. G. M., and E. A. Hanson. 1937. Note on the mechanism of
photosynthesis. Proc. K. Nederland. Akad. Wetensch 40: 752-755.

Belavtseva, E. M. 1957. Electron diffraction and electron microscope studies of
chlorophyll a and b and Gramicidin. Biophysics (U.S.S.R.) (English Transhi-
tion) 2: 617-622.

Boulder Gonference on Biophysics. 1959. Revs. Modern Phys. 31: Nos. 1, 2.

Brookhaven Symposia in Biology, No. 11. 1958. The Photochemical Apparatus:
Its Structure and Function.

Calvin, M. 1958. In The Photochemical Apparatus: Its Structure and Function.
Brookhaven Symposia in Biol. No. 11: 160-180.

Calvin, M. 1959a. Energy reception and transfer in photosynthesis. Revs. Modern
Phys. 31: 147-156.

Calvin, M. 1959b. Free radicals in photosynthetic svstems. Revs. Modern Phys.
31: 157-161.

Discussions of the Faraday Society. 1958. Configurations and Interactions of
Macromolecides and Liquid Crystals. No. 25.

Du NoLiY, P. Lecompte. 1926. Surface Equilibria of Biological and Organic Colloids.
The Chemical Catalog Co., New York.

Elbers, p. F., K. Minnaert, and f. B. Thomas. 1957. Submicroscopic structure of
some chloroplasts. Acta Botan. Neerl. 6: 345-350.

EvERSOLE, R. A., and J. ]. Wolken. 1958. Photochemical activity of digitonin ex-
tracts of chloroplasts. Sc/erice 127: 1287-1288.

Frey-Wyssling, a. 1957. Macromolecules in Cell Structure. Harvard University
Press, Cambridge, Mass.

Fujimori, E., and R. Livingston. 1957. Interactions of chlorophyll in its triplet
state with oxygen, carotene, etc. Nature 180: 1036-1038.

Gaffron, H. (ed.). 1957. Research in Photosynthesis. Interscience Publishers, Inc.,
New York.

Goedheer, J. C. 1955. Orientation of the pigment molecules in the chloroplast.
Biochim. et Biophys. Acta 16: 471-476.

Goedheer, J. C. 1957. Optical properties and in vivo orientation of photosynthetic
pigments. Thesis. State University of Utrecht, Netherlands.

Granick, S. 1955. In W. Ruhland (ed.). Handbuch der Pflanzenphysiologie.
Springer Verlag, Berlin. 1: 507-564.

Hedges, E. J. 1932. Liesegang Rings and Other Periodic Structures. Chapman and
Hall, Ltd., London.

Heitz, E. 1955. tJber die Struktur der Chromosomen imd Chloroplasten. Nova
Acta Leopoldina 17: 517-540.

Hodge, A. J. 1959. Fine structure of lamellar systems as illustrated by chloroplasts.
Revs. Modern Phys. 31: 331-341.

Hodge, A. J., J. D. McLean, and F. V. Mercer. 1955. Ultrastructure of the lamellae
and grana in the chloroplasts Zea Mays L. J. Biophys. Biochem. Cytol. 1: 606-614.

Hubbard, R. 1954. The molecular weight of rhodopsin and the nature of the
rhodopsin-digitonin complex. /. Gen. Physiol. 37: 381-399.

Jacob, E. E., A. E. Vatter, and A. S. Holt. 1954. Crystalline chlorophyll and bac-
teriochlorophyll. Arch. Biochem. Biophys. 53: 228-238.

Lauffer, M. a., W. C. Price, and A. W. Petre. 1949. The nature of viruses. Ad-
vances in Enzymol. 9: 171-240.

Levitt, L. S. 1954. The role of magnesium in photosynthesis. Scietice 120: 33-35.

Leyon, H. 1956. The structure of the chloroplasts. Svensk Kern. Tidskr. 68: 70-89.

Lynch, V. H., and G. S. French. 1957. B-carotene, an active component of chloro-
plasts. Arch. Biochem. Biophys. 70: 382-391.

THE CHLOROPLAST 111

MiJHLETHALER, K. 1955. The structure of cliloroplasts. Intern. Rev. Ci/tol 4: 197-
220.

MiJHLETHALER, K. 1959. Private communication.

Nelson, R. C. 1957. Some photoelectric properties of chlorophyll. /. Chcni. Phi/s.
27: 864-867.

Platt, y. R. 1959. Carotene-donor-acceptor complexes in photosynthesis. Science
129:"372-374.

Rabinowitch, E. I. 1945. Photosynthesis and Related Processes. Interscicnce Pub-
lishers, Inc., New York. Vol. 1. Pp. 355-480.

Rabinowitch, E. I. 1951. Photosynthesis and Related Processes. Interscience Pub-
lishers, Inc., New York. Vol. 2, part 1. Pp. 603-736.

Rabinowitch, E. I. 1956. Photosynthesis and Related Processes. Interscience Pub-
lishers, Inc., New York. Vol. 2, part 2. Pp. 1714-1758.

Robinson, C. 1958. In J. R. Danielli, K. G. A. Pankhurst, and A. C. Riddiford
(eds.). Surface Phenomena in Chemistry and Biology. Pergamon Press, Ltd.,
London.

Robinson, C, and S. C. Ward. 1957. Liquid-crystalline structures in polypeptides.
Nature 180: 1183-1184.

Rodrigo, F. a. 1955. Experiments concerning the state of chlorophyll in the plant.
Thesis. State University of Utrecht, Netherlands.

Sager, R. 1958. In The Photochemical Apparatus: Its Structure and Function.
Brookhaven Symposia in Biol. No. 11: 101-117.

Sager, R., and G. E. Palade. 1957. Structure and development of the chloroplast
in Chlamydomonas. I. The normal green cell. /. Biophys. Biochem. Cytol. 3:
463-488.

Sissakian, N. M. 1958. Enzymology of the plastids. Advances in Enzymol. 20: 201-
236.

Smith, E. L. 1941a. The chlorophyll-protein compound of the green leaf. ]. Gen.
Physiol. 24: 565-582.

Smith, E. L. 1941b. The action of sodium dodecyl sulphate on the chlorophyll-pro-
tein compound of the spinach leaf. /. Gen. Physiol. 24: 583-596.

Smith, E. L. 1941c. The effects of detergents on the chlorophyll-protein compound
of spinach as studied in the ultracentrifuge. /. Gen. Phy.siol. 24: 753-764.

Smith, J. H. C. 1959. Proc. Intern. Botan. Congr. 9th Congr., Montreal. In press.

Smith, J. H. C., and V. M. K. Young. 1956. Chlorophyll formation and accumula-
tion in plants. In A. Hollaender (ed.). Radiation Biology. McGraw-Hill Book
Co., Inc., New York. Vol. 3. Pp. 393-442.

Stanier, R. Y. 1958. In The Photochemical Apparatus: Its Structure and Function.
Brookhaven Symposia in Biol. No. 11: 43-53.

Strother, G. K., and J. J. Wolken. 1959. A simplified microspectrophotometer
using photoconductive cells. Science 130: 1084-1088.

Takashima, S. 1952. Chlorophyll-lipoprotein obtained in crystals. Nature 169: 182-
183.

Thomas, J. B. 1955. Structure and functions of the chloroplast. Prog, in Biophys.
and Biophys. Chem. 5: 109-139.

Thomas, J. B. 1958. Chloroplast structure and function. Endeavour 17: 156-161.

Transactions of the Faraday Society. 1933. Theory of Liquid Crystals. 29, part
2: 883-1085.

Trurnit, H. J., and G. Colmano. 1958. Chloroplast studies. I. Absorption of
chlorophyll monolayers at liquid interfaces. Biochim. et Biophys. Acta 30: 435-
447.

von Wettstein, D. 1957a. Chlorophyll-Letale und der submikrosopische Form-
wechsel der Plastiden. Exptl. Cell Research 12: 427-506.

VON Wettstein, D. 1957b. Genetics and the submicroscopic cytology of plastids.
//0rff/i7fls 43: 303-317.

VON Wettstein, D. 1958. In The Photochemical Apparatus: Its Structure and Func-
tion. Brookhaven Symposia in Biol. No. Il: 138-159.

112 MACROMOLECULAR COMPLEXES

WoLKEN, J. J. 1956a. A molecular morphology of Euglena gracilis var. hacillaris.

}. Protozool. 3: 211-221.
WoLKEN, J. J. 1956b. Photoreceptor structures. I. Pigment monolayers and molecu-
lar weight. ;. Cellular Comp. Physiol. 48: 349-370.
WoLKEN, J. J. 1958. In The Photochemical Apparatus: Its Structure and Function.

Brookhaven Symposia in Biol. No. 11: 87-100.
WoLKEN, J. J. 1959. The structure of the chloroplast. Ann. Rev. Plant Physiol. 10:

71-86.
WoLKEN, J. J., and A. D. Mellon. 1956. The relationship between chlorophyll

and the carotenoids in the algal flagellate, Euglena. J. Gen. Physiol. 39: 675-685.
WoLKEN, ]. ]., and A. D. Mellon. 1957. Light and heat in the bleaching of chloro-

plastin, Euglena. Biochim. et Biophys. Acta 25: 267-274.
Wolken, J. J., A. D. Mellon, and C. L. Greenblatt. 1955. Environmental factors

affecting growth and chlorophyll synthesis in Euglena. 1. Physical and chemical.

2. The effectiveness of the spectrum for chlorophyll synthesis. /. Protozool. 2:

89-96.
Wolken, J. J., and G. E. Palade. 1953. An electron microscope study of two

flagellates. Chloroplast structure and variation. Ann. N. Y. Acad. Sci. 56: 873-

881.
Wolken, J. J., and F. A. Schwertz. 1953. Chlorophyll monolayers in chloroplast.

/. Gen. Physiol. 37: 111-120.
ZiMM, B. H. 1959. Concentrated macromolecular solutions. Revs. Modern Phys.

31: 123-129.

DISCUSSION
A. Hodge, ]. ]. Wolken, A. C. Giese

Dr. Hodge (Massachusetts Institute of Technology): Would you please
indicate the conditions under which the chloroplastin rings form?

Dr. Wolken: The simplest procedure is to place a drop of chloroplastin
on a clean microscope slide, warm gently, and then quickly transfer the slide
to the microscope.

Dr. Giese (Stanford University): Is there any suggestion of the deriva-
tion of chloroplasts from cilia, as there is in the case of rods and cones?

Dr. Wolken: I do not think that the chloroplasts are derived from the
cilia. However, it is quite likely that the chloroplast lamellae are formed from
fibrillar protein macromolecules.

Lamellar Systems in Myelin and Photoreceptors

as Revealed by High-Resolution

Electron Microscopy'

H. Fernandez-Moran -

Introduction

Ultrastructure studies (Engstrom and Finean, 1958; Fernandez-
Moran, 1959b; Fernandez-Moran and Brown, 1958; Schmidt, 1937;
Schmitt, 1944; Schmitt et ah, 1935) of the regularly arranged sub-
microscopic layers which build up the nerve myelin sheath, photo-
receptors, chloroplasts, and numerous other types of "lamellar sys-
tems," have furnished the basis for a broad approach to related
problems of fundamental interest in the field of molecular biology.
Current biophysical and biochemical investigations are therefore
concerned not only with the nature of these multilayered structures
considered as specialized derivatives of the cell membrane, but also
with the distinctive repetitive features of their periodic arrav which
seem to underlie the specific processes of energy reception and trans-
fer in living organisms.

Many lamellar structures may be regarded as model systems
which are particularly favorable for detailed investigation of cell-

' This work was supported by Atomic Energy Commission Contract AT(30-1)-
2278 and by a grant (C-3174) from the National Institutes of Health. The author
is particularly indebted to Dr. William H. Sweet, Associate Professor of Neurosurgery,
Harvard Medical School, and to Dr. Raymond D. Adams, Bullard Professor of Neuro-
pathology, Harvard Medical School, for their generous assistance and support of this
project. It is also a pleasure to thank Professor Samuel C. Collins, Director of the
Cryogenic Engineering Laboratory, M. I. T., for his guidance and kind help in extend-
ing to us the facilities of his laboratory during the course of the liquid helium experi-
ments; and his assistant, Robert Cavileer, for his technical contribution in the design
and instrumentation of the liquid helium experiments. Sincere thanks are also due to
Frederick B. Merk, Ernest Shmid, and Joanne T. Frederick for their able technical
assistance; and to Gertrude Dole and Irene Brierley for their valuable help in pre-
paring ithe manuscript.

^ Mixter Laboratories for Electron Microscopy Neurosurgical Service, Massachu-
setts Ceneral Hospital; and Department of Neurology and Psychology, Harvard Med-
ical School, Boston, Massachusetts.

113

114 MACROMOLECULAR COMPLEXES

membrane organization in the living state because they are com-
posed ahiiost exckisively of numerous unit membranes multiply
wrapped or folded in highly ordered complexes. These compact
aggregates of oriented membranes are well suited for analysis by
polarized light and x-ray diffraction, which can often be performed
on the intact components under physiological conditions (Schmitt,
1944; Schmittei«/., 1935).

In the embryogenesis of peripheral nerve mvelin, Geren and
Schmitt ( 1955 ) have shown that the Schwann cells "wrap them-
selves" many times around an outgrowing axon by a continuous in-
folding of the outer Schwann-cell surface membrane ( Schmitt, 1959,
p. 460). After numerous double membranes are thus spirally
wrapped about the axon, the layers become condensed to form the
compact myelin sheath (Geren and Schmitt, 1955; Robertson, 1959),
which exhibits strong birefringence (Schmidt, 1937), indicating a
high degree of orientation of the constituent lipid-protein units, and
other physical properties characteristic of the smectic fluid-crystal-
line state (Schmidt, 1937; Schmitt, 1959).

The analysis of the myelin sheath ultrastructure is a classic dem-
onstration of the advantages inherent in an approach combining
direct and indirect methods (Schmidt, 1937; Fernandez-Moran,
1959b). Based on the polarized-light investigations (Schmidt, 1937;
Schmitt, 1944) begun over a century ago, and on subsequent low-
angle x-ray diffraction studies (Schmitt et al., 1935), the concentric
laminated structure of the sheath was accurately predicted, and the
thickness of the layers deduced, before the electron microscope
provided direct confirmation of the postulated multilayer arrange-
ment (Fernandez-Moran, 1950; Robertson, 1959).

Electron microscopy is the only method that permits direct
visualization of structural details of molecular dimensions within a
selected region and thus furnishes data well above the level of
statistical uncertainty commonly associated with indirect analytical
methods. However, electron microscope investigations are severely
limited by numerous preparation artifacts resulting from complex
perturbations of the living state when biological systems are sub-
jected to fixation, dehydration, embedding, and ultrathin-sectioning
procedures, and finally to the deleterious effects of high-vacuum and
electron-beam irradiation during observation.

Practically all of our present direct knowledge of lamellar sys-
tems, and of cellular fine structure in general, is based on electron

LAMELLAR SYSTEMS 115

microscope examination of osmium-fixed specimens (Feinandez-
Moian and Brown, 1958; Porter, 1957). Although this type of prep-
aration has ah'eady yielded much valuable information, it is com-
monly acknowledged that we are merely disclosing the refractory
macromolecular framework of a far more intricate dynamic system.
Thus, the concentric array of dense bands and light interspaces
revealed by electron microscopy in the nerve myelin sheath repre-
sents merely the "osmium-stabilized skeleton" (Fernandez-Moran,
1957), or the pattern of selective deposition of osmium at certain
sites, without permitting identification of specific regions containing
lipids, lipoproteins, or protein constituents (Fernandez-Moran and
Finean, 1957). Within this altered matrix, deprived of the funda-
mental water component, there is scant possibility of localizing or
even detecting the numerous enzymes, electrolytes, trace metals,
and other important constituents associated with the myelin sheath
(Fernandez-Moran, 1959b; Lumsden, 1957).

The shortcomings of our present preparation techniques are more
acutely felt now that modern electron microscopes consistently
achieve resolutions of the order of 7 to 10 A, and are thus inherently
capable of directly visualizing molecular structures in the size range
of certain enzymes and hydrated lipid-protein complexes, particu-
larly when these are incorporated in the periodic multilayer arrange-
ment characteristic of lamellar systems. The development of ade-
quate preparation methods is therefore a major problem which must
be solved before high-resolution electron microscopy can be more
effectively applied in the study of biological systems during growth
and function.

Low-temperature preparation techniques provide one of the most
promising approaches, since rapid freezing of biological specimens
suspends all physiological activity, immobilizing and preserving
tissue constituents (Harris, 1954). That coohng to temperatures
close to absolute zero does not appreciably impair critical life
processes has now been amply demonstrated by the survival of a
wide variety of living organisms, including bacteria, spermatozoa,
and many other sensitive cells and tissues, which are first treated
protectively with glycerol, then frozen with liquid nitrogen or liquid
helium, and subsequently thawed in a controlled manner (Harris,
1954;' Lovelock, 1953; Parkes, 1951 ) . By ultrarapid cooling to low
enough temperatures, it may be feasible to preserve the original
position and relationship of the main organic and inorganic con-

116 MACROMOLECULAR COMPLEXES

stituents of cellular organization in tissues, including the predomi-
nant water component, the transient intermediates with unpaired
electron-spin (which participate in enzymatic reactions and meta-
bolic electron transfer), and other unstable chemical species gen-
erally referred to as free radicals (Broida, 1957; Minkoff, 1959).
However, in order to achieve this optimum preservation and effec-
tively "fix" the highly reactive free radicals in biological systems,
it would be necessary to work at temperatures of liquid helium
(Allen, 1952; Mendelssohn, 1956), which are about 100° C lower
than those currently obtained with isopentane-liquid nitrogen
coolants.

Improved low-temperature preparation techniques for electron
microscopy of biological tissues (Fernandez-Moran, 1959a) have
recently been developed which yield better morphological and
histochemical preservation of lamellar systems and other cell com-
ponents than do the standard freeze-drying or freeze-substitution
methods. These "cryofixation techniques" (Fernandez-Moran,
1959c) are based on rapid freezing of fresh or glycerinated tissues
with liquid hehum II at 1° to 2° K, followed by freeze-substitution
and embedding in plastics at low temperatures, under conditions
which minimize ice-crystal formation, artificial osmotic gradients,
and extraction artifacts.

The following review deals with salient features of the fine struc-
ture of lamellar systems in thin sections of the myelin sheath and
selected photoreceptors, as revealed bv high-resolution electron
microscopy, using both standard preparation techniques and the
new low-temperature procedures. Although the latter are still in a
preliminary stage of development, the experimental approaches will
be outlined and the underlying operational concepts discussed in
order to illustrate the potentialities of ultrastructure research at low
temperatures (Fernandez-Moran, 1959d). Lamellar systems are
particularly suitable for this type of work, since all steps of the
preparation procedures can be followed and the artifact sources
analyzed directly at low temperatures bv combined application of
x-ray diffraction techniques and electron microscopy. Moreover,
by cooling the object during observation in the microscope, and
using an electron microbeam of very low intensitv, irradiation
damage and specimen contamination can be minimized ( Fernandez-
Moran, 1959a, 1959d). With this experimental arrangement for
low-temperature electron microscopv, a direct investigation of ice-

LAMELLAR SYSTEMS 117

crystal structure and growth has already been undertaken ( Fernan-
dez-Moran, 1959d). Further developments along these lines may
eventually permit direct electron-optical studies of thin biological
specimens in which the native hydrated state has been partially
preserved through vitrification at low temperatures.

Low-Temperature Preparation Techniques
for Electron Microscopy

Standard Freeze-Drying and Freeze-Substitution Techniques.
The classical freeze-drying techniques (Gersh and Stephenson,
1954) of established value in morphological and histochemical
studies of tissues by light microscopy have generally proved to be
inadequate for electron microscope preparations. Even when freeze-
drying is performed under favorable conditions (Sjostrand and
Baker, 1958), the resulting thin sections show extensive vacuoliza-
tion due to ice-crystal formation, and generallv deficient preserva-
tion of tissue structure. This ice-crystal artifact plays a major role
in all low-temperature procedures, and can be largely accounted for
in terms of the characteristic phase transformations which water
undergoes at different temperatures. If a sufficiently small tissue
sample is cooled rapidly enough (in less than -/looo second) below
— 100° C with isopentane-liquid nitrogen (Stephenson, 1956), the
water in the tissue will solidify in a metastable "vitreous" or glassy
state ( Gersh and Stephenson, 1954 ) . In the case of pure water, this
vitreous ice turns abruptly into crystalline ice above the critical
"glassy transformation temperature," around — 130°C (Meryman,
1956). The phase transformations of water in tissues are different
from those in pure water (Luyet, 1957; Stephenson, 1956), but even
in tissues a transition temperature of about —100° C has been as-
sumed by Stephenson (1956). Meryman (1956) has shown that a
pure ice crystal can develop in 30 seconds from the glassy state to
a length of l/>t at —70° C. Since the subhmation of ice in freeze-
drying is usually carried out at temperatures not below —80° G, the
rapid growth of ice crystals within the tissue matrix and the ensuing
damage occurring during the relatively long periods of vacuum de-
hydration are readily understandable.

In the freeze-substitution process introduced by Simpson (1941),
the ice formed within the tissue is slowly dissolved or "substituted"
in a fluid solvent at temperatures of about — 70° G (Patten and

118 MACROMOLECULAR COMPLEXES

Brown, 1958). Although addition of chemical fixatives to the sub-
stituting fluids (Feder and Sidman, 1958; Fernandez-Moran, 1957)
improves the quality of tissue preservation, the artifacts associated
with the formation of ice crystals during prolonged immersion at
these temperatures— well above the critical transition at — 100°C—
cannot be avoided. In both freeze-drying and freeze-substitution,
embedding of the tissues produces further extraction and rearrange-
ments which, though indiscernible under the light microscope,
nevertheless obliterate important structural detail at the submicro-
scopic level. The partially undenatured, freeze-substituted tissue
components are particularly vulnerable to the lipid-extracting action
of the usual transfer and embedding media. All of these artifact
sources severely limit the usefulness of standard freeze-substitution
techniques for electron microscopy.

In order to adapt freeze-substitution for the study of tissue fine-
structure, it would be necessary to dissolve the ice matrix at tem-
peratures preferably below the critical glassy transition in tissues,
tentatively assumed to be at —100° C. The entire process of im-
pregnation and embedding would also have to be carried out at
temperatures of — 100°C to approximately — 50° C, at which the
solubility of the labile lipoprotein membrane complexes would be
greatly reduced, and the stiffness of the frozen tissue matrix would
prevent gross structural rearrangements from occurring. Muller
( 1957), who first successfully applied photopolymerization of meth-
acrylate at —10° C for embedding of frozen-dried preparations,
noted marked improvements in the preservation of chloroplast fine-
structure, and considerably reduced extraction of lipoprotein com-
plexes. Freeze-substitution procedures are well suited for infiltra-
tion with a monomer and photopolymerization at low temperatures,
since, as Feder and Sidman ( 1958 ) have pointed out, the transitions
do not involve the disruptive effects of a liquid-gas interface sweep-
ing through the specimen, as is the case when frozen-dried tissues
are immersed in the embedding fluids. As shown by previous studies
(Fernandez-Moran, 1950, 1957), treatment of the tissues with glyc-
erol or other protective agents prior to freezing is essential to pre-
vent destructive ice-crystal formation in the more commonly en-
countered larger tissue specimens.

Cryofixation and Related Low-Temperature Techniques.
Guided by these considerations, and ])ased on the results of earlier
work (Fernandez-Moran, 1950, 1957) and the recent technical

LAMELLAR SYSTEMS 119

contributions by Feder and Sidman (1958) and by Miiller (1957),
improved low-temperature preparation methods liave been devel-
oped for the study of tissues by high-resolution electron microscopv
(Fernandez-Moran, 1959a, 1959c, 1960). The designation "cryo-
fixation techniques" (Fernandez-Moran, 1959c) was suggested to
distinguish these procedures from the related freeze-sulDstitution
methods (Feder and Sidman, 1958; Patten and Brown, 1958; Simp-
son, 1941 ) because of the different type of approach, involving freez-
ing with liquid helium II and low-temperature polymerization,
which will become of increasing operational value in the course of
further developments.

The basic cryofixation techniques comprise sequential application
of the following procedures : ( 1 ) rapid or ultrarapid freezing of
thin, fresh specimens or of glycerinated tissues with liquid helium II
at 1° to 2° K; (2) staining at — 150° C with halogens or organo-
metallic compounds dissolved in isopentane; (3) substitution of the
ice matrix with solutions of heavy-metal salts in alcohol-acetone and
alkyl halide mixtures, or alternatively in certain organometallic com-
pounds, at temperatures of —130° to —80° C; (4) infiltration of the
specimens at —100° to —80° C with acryhc monomers; and (5)
final embedding of the specimens by photopolymerization with ul-
traviolet light at temperatures of -80° to -20° C (Fox et al, 1958).

In an important technical variant used mainly for examination of
specimens which have been subjected to a minimum of chemical
fixation, all intermediate staining procedures are eliminated, and
only freeze-substitution is carried out with acetone-ethyl chloride
mixtures at temperatures of about —130° to —110° C, which are still
below the critical transition range for tissues.

A detailed account of the crvofixation and related low-tempera-
ture preparation procedures is now being prepared for publication ^
and only representative methodological aspects will be reviewed
here.

Freezing of Biological Specimens With Liquid Helium II.
The use of liquid helium as an indispensable refrigerant for attain-
ing temperatures close to absolute zero is based on important recent
developments in the fields of low-temperature chemistry ( Klein and
Scheer, 1958) and free-radical research (Broida, 1957; Minkoff,
1959). The temperature domain below the glassy transformation of

^ H. Fernandez-Moran. Ultrastructure of the Nervous System. Part I, Techniques-
(To be published by Academic Press, Inc., New York. )

120 MACROMOLECULAR COMPLEXES

water at — 130°C, which had been set aside bv the biologist as a
"sanctuary" protecting all life processes from further change, corre-
sponds in fact to the upper limit of the vigorous new field of low-
temperature chemistry. This discipline concerns itself mainly with
the wide variety of low-temperature reactions which occur below
150° K and are characterized by low activation energies of 5 kcal/
mol or less. The studies of Klein and Scheer (1958) have demon-
strated that numerous chemical reactions involving the addition of
hydrogen atoms to solid olefins readily occur at —195° C with
measurable activation energies, and also evidence of diffusion proc-
esses which are of considerable importance at low temperatures.
Recently, these authors have shown that even at 20° K h\'drogen
atoms will react with solid oxvgen. Free radicals and many other
unstable chemical species will exhibit considerable reactivity at
70° K (Broida, 1957), and the trapping of these transient interme-
diates by freezing into an inert solid at liquid helium temperatures
(Minkoff, 1959) has been the subject of considerable investigation.
Free radicals and other intermediates with unpaired electron-spin
play an important role in many enzymatic reactions and in photo-
synthetic processes (Calvin, 1959a). The possibility of stabilizing
these highly reactive intermediates in intact biological systems,
which can be adequately achieved only at liquid helium tempera-
tures, would therefore amplv justify its application.

Normal liquid helium I shows poor heat conductivity and is
therefore less suitable than its low-temperature phase, known as
liquid helium II, for direct, rapid cooling of biological specimens.
Helium II, which is obtained by evaporating ordinary liquid helium
under reduced pressure until the temperature falls below the critical
X-point (2.19° K), exhibits the unique properties of heat supercon-
ductivity and superfluidity (Allen, 1952; Mendelssohn, 1956). Un-
der certain conditions, the bulk liquid helium II will conduct heat
10,000 times better than will copper, and it can flow rapidlx- through
the finest capillaries without any viscous drag. These phenomena
are restricted to a narrow temperature range, and transfer of heat
from a specimen into the bulk liquid is actually impeded by a
poorly conducting boundary layer (Allen, 1952; Mendelssohn,
1956). However, despite the existence of this thermal boundary re-
sistance, helium II appears to be the most effective refrigerant for
direct cooling of thin biological specimens to temperatures which
are 100° C lower than those obtained with the standard isopentane-

LAMELLAR SYSTEMS 121

liquid nitrogen mixtures (Fernandez-Moran, 1959c). Attainment of
these extremely low temperatures, at which all other substances
solidify, and at which diffusion or recombination of unstable chemi-
cal species within the tissue matrix can be practically eliminated, is
therefore the main reason for the use of liquid helium II in con-
nection with ultrastructure studies. Although better preservation of
fine structure has already been achieved at this preliminary stage,
it must be emphasized that the full benefits of this approach are to
be expected only if the trapped free radicals can be effectively im-
mobilized and made electron-optically visible within the intact
tissues by suitable interaction with heavy atoms and (addition)
polymerization at lower temperatures. Liquid helium is chemically
inert and far safer to handle than isopentane-propane-liquid nitro-
gen mixtures or liquid hydrogen. Moreover, the widespread use of
liquid helium in industrial cryogenics research now makes it more
readily available, and the practical difficulties connectied with its
application can be easily solved once the basic experimental tech-
niques have been mastered.

In collaboration with Professor Samuel C. Collins and his asso-
ciates at the Cryogenic Engineering Laboratory of the Massachu-
setts Institute of Technology, a series of preliminary experiments
with liquid helium II have been carried out during the past year.
As shown in Fig. 1, the basic equipment consists of a specimen stage
attached to a shielded helium Dewar which can be evacuated with
a large forepump of 311 cu. ft./min. capacity. For processing even
a limited number of specimens it is necessary to prepare relatively
large amounts of bulk liquid helium II because of its small heat of
vaporization. This is accomplished by filling the inner Dewar with
liquid helium I and rapidly lowering the vapor pressure to approxi-
mately lOfi, equivalent to —272.2° C. The simple type of stage
consisting essentially of a large glass stopcock makes it possible to
introduce fresh or glycerinated tissues into the liquid hehum II bath
without breaking the vacuum (Fernandez-Moran, 1959d). High-
speed photographs, recorded in collaboration with D. Eldridge of
the Electrical Engineering Department of M. I. T., show that im-
mersion of the specimen into the bulk liquid does not produce boil-
ing (Figs. 2, 3), suggesting that equilibration of the temperature
difference takes place very rapidly. If necessary, the specimen can
be precooled with propane-liquid nitrogen or liquid hydrogen and
immediately plunged into liquid helium II to prevent transitional

122

MACROMOLECULAR COMPLE'XES

Fig. 1. Equipment for rapid freezing of biological specimens in liquid
helium II with attached gas thermometer, high-speed camera, and microscope
for direct observation of specimens within helium Dewar at —272 C. Arrow
indicates specimen stage for introducing fresh or glycerinated tissues into
liquid helium II without breaking vacuum. (Installed at the Cryogenic Engi-
neering Laboratory, M. I. T., in cooperation with Professor Samuel C. Collins.)

changes from occurring. However, very rapid cooling leads to
severe distortion and cracking of larger tissue sections, and the opti-
mum cooling velocity must therefore be determined for each type
of specimen.

A simplified version of this equipment has proved to be verv
useful for achieving rapid and ultrarapid cooling of tissues with
other types of refrigerants. Thus, if liquid nitrogen is cooled close
to its fusion temperature (63.15° K) by evacuating the Dewar, small
tissue specimens can be introduced through the attached vacuum
stage, and appear to freeze more rapidlv in this quiescent fluid than
in the boiling liquid nitrogen at atmospheric pressure. Since the
use of liquid nitrogen under these conditions yields excellent tissue
preservation, it can be recommended as an inexpensive alternative
method or training procedure for helium cryofixation.

LAMELLAR SYSTEMS

123

|)|ll

>-

Fig. 2. Liquid helium II (arrow) in shielded Dewar at T" K, showing charac-
teristic quiescent appearance. Because of the exceedingly high heat conduc-
tivity of helium II, a temperature difference between top and bottom o* the
liquid sufficiently great to allow the formation of vapor bubbles cannot be
produced.

Fig. 3. Serial high-speed photographs showing immersion of specimen in
liquid helium II without producing boiling. (Courfesy, Professor Harold E.
Edgerton, Department of Electrical Engineering, M. I. T.)

Technical Requirements and Equipment for Cryofixation.
In order to avoid ice-crystal and extraction artifacts during the
processing cycle, which may take several weeks in certain specimens,
a special "cryofixation assembly" (Fig. 4) was developed for semi-
automatic freeze-substitution and photopolymerization embedding
within a modified Harris refrigeration unit designed to maintain
— 100° C. The sealed specimen chamber (Fig. 5), formed by a
Teflon gasket bounded by thin Plexiglas sheets and sandwiched
between aluminum plates, permits regulated fluid transfer through
inserted valves. This type of chamber is also useful for other types of
preparations, since whole tissues can be embedded in one block and
the segments removed without distortion for precise sectioning in
any desired orientation. All substitution and embedding fluids were

124

MACROMOLECULAR COMPLEXES

Fig. 4. Cryofixation assembly for semiautomatic freeze-substitution, infiltra-
tion of specimens, and photopolymerization embedding. After rapid freezing
in liquid helium 11 or Freon 22-liquid nitrogen, the specimens remain in special
chambers (arrow) within refrigeration unit (at —130° to —90° C), emerging
ready for sectioning.

prepared in anhydrous (less than ten parts per milHon water con-
tent ) form by passage through Linde 4A molecular sieves, followed
by ultrafiltration to remove suspended particles. Protective treat-
ment (Fernandez-Moran, 1952; Lovelock, 1953; Parkes, 1951) by
controlled glycerination (30 to 60 per cent glycerol-veronal buffer
solutions) of larger tissue specimens prior to freezing, with subse-
quent low-temperature transfer over alcohol or methylcellosolve
mixtures to the acrylic monomers, was found to be essential for
adequate preservation of fine structure (Fernandez-Moran, 1959a,
1959c). Osmium cryofixation used mainly in this study involved
( 1 ) freezing of the tissues in liquid helium II ( or in Freon 22-liquid
nitrogen at —150° C for control purposes); (2) freeze-substitution
at —130° to —80° C in 2 per cent osmium tetroxide in acetone-ethyl
chloride mixtures; (3) impregnation at — 75° C in mixtures of
methyl acrylate and butyl-methyl methacrylate monomer; and (4)

LAMELLAR SYSTEMS

125

Fig. 5. Specimen chamber for cryofixation preparations, consisting of Teflon
gasket bounded by thin Plexiglas sheets sealed between aluminum plates.
Polymerized specimens are easily removed and mounted for ultrathin sectioning
(arrow).

final low-temperature photopolvmerization with ultraviolet light,
using benzoin in 0.3 per cent concentration as a catalyst.

The control preparations included standard osmium-fixed and
araldite-embedded nerve fibers and retinas, as well as specimens
fixed in situ with osmium, iodine, or bromine vapors and embedded
at low temperatures. Serial ultrathin sections of 100 to 300 A were
prepared with a Moran-Leitz ultramicrotome equipped with a dia-
mond knife (Fernandez-Moran, 1953). The sections of "unfixed"
retinas were collected on the cold liquid surfaces of certain fluorine
compounds (e.g., Freon 11) to avoid contact with water. These
sections were examined with an electron microbeam of very low in-
tensity, using the double condenser of the Siemens Elmiskop I
operating at 40 to 80 kv. A Siemens-Liesegang liquid nitrogen cool-
ing device with modified shielding apertures was used in conjunc-
tion with multiple-objective apertures for low-temperature electron
microscopy (Fernandez-Moran, 1959c).

126 MACROMOLECULAR COMPLEXES

Application of Histochemical and Autoradiography Tech-
niques. Thin sections of the "unfixed" and largely undenatured spec-
imens can be further subjected to controlled staining, extraction,
enzymatic digestion, and numerous other histochemical procedures.
The development of adequate high-resolution autoradiography
techniques for electron microscopy would be of considerable
significance for the field of ultrastructure research, since the
structures revealed at the macromolecular level could then be more
specifically identified and evaluated in terms of chemical composi-
tion and activity. Low-temperature preparation techniques appear
to be particularly suitable for this purpose, since the radioactive
tracers could be immobilized by rapid freezing in their natural
position within the well-preserved tissue matrix ( Fernandez-Moran,
1959c ) . The introduction of compounds leading to the formation of
silver bromide within the cells during the process of freeze-substi-
tution represents one of the most promising approaches. In this
high-resolution autoradiography method which is now being de-
veloped, the cellular matrix substitutes for the gelatin base of ordi-
nary photographic emulsions. Low temperatures markedly reduce
the fog and background without impairing the cumulative effects of
the radioactivity processes on the silver halides. After exposures
of the order of 2 to 8 weeks, the latent image centers are developed
and enlarged at low temperatures by using gold thiocyanate ( Hoer-
lin and Hamm, 1953) solutions. Interpretation of the images ob-
tained with this method is still highly tentative, since the possibility
that we might be dealing with "autochemograms" and other spurious
effects must first be conclusively ruled out.

Results

To establish a reliable basis for evaluation of the fine structures
revealed by different types of preparation procedures, the lamellar
systems in the nerve myelin sheath and retinal rod outer segments
of the frog and guinea pig were examined, in addition to character-
istic paracrystalline granules found in the retinal pigment epithe-
lium. Using thin sections of standard osmium-fixed and embedded
preparations as controls, the effects of freezing, freeze-substitution,
staining, and low-temperature polymerization were systematically
investigated.

LAMELLAR SYSTEMS 127

In general, satisfactor>' preservation of the integrity and fine struc-
ture of lamellar systems could be achieved by rapid freezing of thin,
fresh specimens or of glycerol-treated larger tissue samples, prefer-
ably with liquid helium II, followed by controlled freeze-substitution
at temperatures of —130° to —80° C, and low-temperature poly-
merization. However, thin sections of this "unfixed" material
showed such lability and low contrast when examined directly in
the electron microscope that staining with heavy-metal salts was
found to be necessary, except in the case of nuclear components
which exhibit sufficient contrast and structural detail even in un-
stained preparations. In order to draw a more direct comparison
of the new procedures with standard electron microscopy tech-
niques, the findings presented will be confined mainly to those ob-
tained with osmium tetroxide staining, either introduced during
freeze-substitution or applied in vapor form to the "unfixed" ultra-
thin sections.

However, in addition to the well-preserved regions, there are still
many areas in the specimens which exhibit ice-crystal artifacts,
artificial rearrangements, and other complex modifications intro-
duced by the new low-temperature methods. Evaluation of these
preparation artifacts has not only served to emphasize the limita-
tions and pitfalls of the new techniques, but also provided valuable
indications for improved preservation of the peculiarly labile hy-
drated state of the lipoprotein components in lamellar systems.

Fine Structure of the Nerve Myelin Sheath. Selection of the
nerve myelin sheath as a standard reference system for comparative
studies of lamellar fine structure was clearly indicated in view of the
extensive data already available on normal and modified myelin
(Fernandez-Moran, 1957, 1959b; Finean, 1958; Schmitt, 1959;
Schmitt et ah, 1935). The correlative investigation of the prepara-
tion procedures and experimental modifications, achieved by sys-
tematic application of x-ray diffraction methods and electron micros-
copy (Fernandez-Moran, 1959b; Fernandez-Moran and Finean,
1957; Finean, 1958; Schmitt et al, 1935), has been of particular
value in defining the basic structural parameters of the fresh myelin
sheath. Thus, the fundamental radial repeating unit of 170 to 174 A
shown in the low-angle x-ray diffraction pattern recorded from fresh
amphibian nerve (Fig. 8) corresponds to the layer spacing of the
myelin sheath with an average "period of 130 to 140 A, as seen

128 MACROMOLECULAR COMPLEXES

directly in electron micrographs of osmium-fixed thin sections (e.g.,
Figs. 9, 10). Shrinkage effects introduced by the preparation tech-
niques account for the difference of 20 to 40 A between the two
values; and from a detailed analysis carried out on the same speci-
men at each stage of the process, quantitative data were obtained
on the dehydration and embedding artifacts ( Fernandez-Moran
and Finean, 1957).

Against this background, which provides a reliable and detailed
correlation between the organization of lamellar structures in the
fresh myelin sheath and the corresponding electron microscope
images, it is evident that mvelin constitutes one of the best systems
for a step-by-step validation of the results of the new preparation
methods. In the following, we shall first consider the main artifact
sources derived from dehydration and embedding at room tempera-
ture, before dealing with the ice-crystal artifacts.

Effects of Dehydration and Embedding. When the sciatic
nerve from a living frog or guinea pig is rapidly frozen in Freon
22-liquid nitrogen at approximately —160° C, subjected to freeze-
substitution in a 1 per cent osmium tetroxide-acetone mixture at
—75° C according to the procedure of Feder and Sidman (1958),
and then embedded in methacrylate or araldite at 40° to 55° C for
ultrathin sectioning, certain modifications are regularly observed.
Although the myelin sheaths of a few small fibers appear to be well
preserved, most nerve fibers show extensive vacuolization of the
sheath, gross distention, and other rearrangements leading to partial
or complete disruption of the layers. The poor preservation of fine
structure revealed by the high resolving power of the electron micro-
scope stands in marked contrast to the well-preserved appearance
under light microscope examination of a thicker section taken from
the same block. The obliteration of fine structure, although caused
in part by ice-crystal artifacts, is largely due to the extraction of the
hpoprotein components of the layers during treatment with the
methacrylate monomer, which acts as a lipid-extracting agent, par-
ticularly at embedding temperatures of 40° to 55° C. Bv contrast,
remarkable improvement in the preservation of the m\elin sheath
structure is immediately noted when impregnation with the mono-
mer and subsequent photopolymerization are carried out at tem-
peratures of —80° to —30° C. These observations suggest that os-
mium tetroxide does not fix the lipoprotein constituents as effectively
at low temperatures as at 0° C. Low-temperature embedding is

LAMELLAR SYSTEMS 129

therefore essential for all freeze-substitiition procedures, and has
proved to be the most important single factor, actually making
possible the systematic application of low-temperature preparation
techniques to the stud\' of the labile lamellar systems.

Earlier x-ray and electron microscope studies of the standard
preparative procedures (Fernandez-Moran and Finean, 1957) had
already disclosed complex modifications, including, first, a shrink-
age of the myehn layer spacing during osmium fixation and alcohol
dehvdration, then a compensating expansion during methacrvlate
embedding. The significant extraction of material known to occur
during fixation and dehydration (Miiller, 1957) was therefore being
partially masked by the methacrylate embedding. Although such a
detailed analysis of the modifications introduced during fixation,
dehydration, and embedding at low temperatures remains to be
carried out, possible mechanisms which might be operative under
these conditions are already suggested. Since the macromolecular
matrix of the myelin lamellae, frozen and stiff, is thus maintained
essentially immobilized during impregnation with the monomer and
subsequent photopolymerization at low temperatures, the lipid ex-
traction effects are not only considerably reduced, but the whole
system is permanently "set" or fixed in its natural position by the
polymerizing embedding medium before it is brought up to room
temperature. In this way, major rearrangements of fine structure
which might otherwise take place during the process of thawing
are largely avoided; and investigation of ice-crvstal artifacts can
now be profitably undertaken.

Effects of Ice-Crystal Formation. The comprehensive x-ray
diffraction studies carried out by Finean (1958) have shown that
definite structural modifications occur after rapid freezing and thaw-
ing of nerve. In peripheral nerve, this usually results in a complete
halving of the radial repeating unit and partial disorganization of
the layers. The corresponding electron micrographs demonstrated
characteristic changes ranging from an increased general structural
breakdown of the layers to a pronounced accentuation and broaden-
ing of the intermediate line, which could be correlated with the
halving of the radial repeat and abolishing of the difference factor
(Fernandez-Moran and Finean, 1957). These changes are brought
about primarily by transient formation of submicroscopic ice crys-
tals and associated freezing effects in the compact myelin sheath.
They can therefore serve as a reference for evaluation of the results

130 MACROMOLECULAR COMPLEXES

of ice-crystal formation and growth within ordered multilayered
systems during protracted processing at low temperatm^es. When
fresh nerve trunks are rapidly frozen and freeze-substituted at
—75° C and subjected to low-temperature embedding, the resulting
thin sections reveal numerous large internal patches of stRictural
obliteration which correspond mainly to regions of extensive ice-
crystal growth. The associated extraction, diffusion, and rearrange-
ment artifacts are more marked after substitution in methanol than
in acetone. Bearing in mind the rapid ice-crystal growth at —70° C
observed by Meryman (1956), it is evident that at temperatures
above the critical glassy transition point of water, considered to be
around — 100° C in tissues (Stephenson, 1956), ice-crystal growth
can assume significant proportions during the prolonged freeze-sub-
stitution periods of days or weeks in certain specimens. More direct
observations can be performed with a special cold stage and a
fluorescence microscopy attachment, on model systems consisting
of thin gelatin strips containing riboflavin-phosphate-sodium. As
demonstrated by Szent-Gyorgyi ( 1957 ) , the characteristic yellow-
ish-green fluorescence emitted by a watery riboflavin solution under
ultraviolet light is replaced by an orange phosphorescence on the
formation of ice during freezing. This phenomenon can be used as
a sensitive indicator of ice under certain conditions, and the ribo-
flavin enclosed in the ice matrix serves also as a visible marker by
which to follow the freeze-substitution process. Moreover, as the
riboflavin dye is released from the dissolving ice, its progressive
spread throughout the alcohol-solvent column gives an indication
of the diffusion processes still operative at temperatures below
-75° C.

Upon performing freeze-substitution at temperatures of —120°
to —90° C, using acetone-ethyl chloride mixtures, substantial im-
provement in the preservation of the microstructure was noted, due
to absence of larger ice-crystal formations and to the diminished
extracting effects of the substituting fluids at these lower tempera-
tures. However, at temperatures below —90° C, the time factor
enters as an important consideration. Only thin specimens can be
freeze-substituted at these temperatures within reasonable time
periods; for thicker specimens, the complete processing c\cle ma\
require several weeks.

After elimination of gross ice-crystal artifacts, it is now possible
to recognize some of the direct and indirect effects of submicro-

LAMELLAR SYSTEMS 131

scopic ice crystals within the layered structure. The images ob-
tained often resemble the structural modifications in sciatic nerve
described earlier (Fernandez-Moran and Finean, 1957) after rapid
freezing and thawing, standard osmium fixation, and methacrvlate
embedding. When fresh nerves were rapidly frozen in liquid helium
I (—269° C) and then thawed, a characteristic multiple splitting of
the dense and intermediate lines was noted (Fernandez-Moran and
Finean, 1957), in contrast to the more homogeneous appearance
and broadening of the layers encountered after rapid freezing in
liquid nitrogen. In cryofixation preparations, there appears to be a
more difl^erentiated interaction of these freezing effects with the
ordered lamellar substrate, leading to varying degrees of enhanced
structural definition. Thus, the better preservation of tissue fine-
structure generally observed in specimens frozen with liquid helium
II, as compared with control preparations cooled with other refrig-
erants, might be related to the initial formation of finer ice crystals,
and to other as-yet-unknown auxiliary factors which could con-
tribute to the stabilization of the macromolecular matrix. However,
elucidation of these essential problems must await further analysis
of myelin ultrastructure performed directly at liquid helium tem-
peratures by x-ray diffraction techniques, and pursued systemat-
ically, in combination with electron microscopy, throughout the
successive preparative manipulations.

Osmium Cryofixation. When osmium cryofixation is carried out
on fresh frog sciatic nerve under conditions minimizing the de-
scribed artifacts, the myelin sheath in most of the outer fibers is
unusually well preserved. As shown in Fig. 6, the highly regular
aiTangement of the concentric layers is clearly visible over extensive
areas of the sheath, giying an over-all impression of precise align-
ment, which derives from immobilization of the straight fiber bun-
dles in their stiffened state at low temperatures. The fine structures
revealed are essentially similar to those detected in the best stand-
ard osmium-fixed preparations. The "unfixed" thin sections stain
intensely with a wide variety of reagents, including uranyl acetate,
phosphotungstic acid, lead hydroxide, and iodine and bromine com-
pounds, disclosing a similar type of laminated structure in the
sheath, with characteristic variations which are being investigated in
greater detail. Evidence of fine structure within the layers or split-
ting of the dense and intermediate lines (Figs. 7-11), in addition to
a more compact and dense appearance of the interspace, were con-

Fig. 6. Electron micrograph of myelin sheath segment from transverse sec-
tion of frog sciatic nerve, showing uniform preservation of fine structure and
regular arrangement of concentric layers consistently obtained with low-
temperature preparation techniques. Osmium-cryofixation preparation, low-tem-
perature embedding (-20' C) in mixture of methyl acrylate and butyl-methyl
methacrylate. X 1 20,000.

132

LAMELLAR SYSTEMS

133

sistently found in most preparations. In the axonal region, close
attachment, or actual fusion, of the myelin layers with the mito-
chondria is occasionally observed. Unfortunately, this exceptional
degree of structural integrity is confined to thin specimens and to
peripheral regions of larger tissue blocks. The remaining areas show
unmistakable signs of ice-crvstal formation, and all transitions can
be found between partial disruption of the myelin sheath structure
and the zones of excellent preservation. This constitutes a serious
limitation of all low-temperature techniques for normal histological

— 35A-
— A3 —

I — 58
■87^

PS

i Hi

©

1000 A

^

220

HlO
-73^
-AA A

Fig. 7. Myelin sheath segment from transverse thin section of frog optic
nerve, showing regular fine structure (arrow) in dense and intermediate regions
of concentric layers. Nerves equilibrated in glycerol before rapid freezing in
liquid helium II. Osmium-cryofixation preparation. X380,000.

Fig. 8. Low-angle x-ray diffraction patterns, showing the effects of glycerol
treatment on the myelin sheath structure. (Above) Pattern of fresh giant toad
sciatic nerve. (Below) Pattern of giant toad sciatic nerve after equilibration
with 30 per cent glycerol in Ringer solution (24 hr), showing swelling of struc-
ture. Upon rapid freezing of the glycerol-treated nerve with liquid nitrogen or
liquid helium I (—269° C), followed by thawing, a substantially unaltered
x-ray diffraction pattern was obtained, in contrast to the untreated nerve, in
which marked modifications of the pattern were observed, indicating a partial
breakdown of structure. (Experiments carried out in collaboration with Dr. J. B.
Finean in 1957.)

134

MACROMOLECULAR COMPLEXES

Fig. 9. Myelin sheath segment from transverse thin section of frog sciatic
nerve, showing granular fine-structure (arrows) in dense and intermediate lines
of concentric layers. Nerves equilibrated in glycerol before rapid freezing in
liquid helium II. Osmium-cryofixation preparation. X270,000.

studies, and a systematic survey of the usefulness of certain protec-
tive agents like glycerol for the preservation of tissue ultrastructure
therefore became one of the major practical objectives of our in-
vestigations.

Effects of Protective Glycerol Treatment on Myelin Ul-
trastructure. Pursuing earlier studies of the histological applica-
tions of glycerol (Fernandez-Moran, 1952, 1957), a series of interest-
ing observations has been made, demonstrating the general effects
of glycerol treatment on the ultrastructure of the myelin sheath.
The remarkable effects of glycerol which protect living cells against
the damaging action of freezing and thawing (Harris, 1954; Parkes,
1951 ) have been primarily ascribed to the water-binding properties
and salt-buffering potentiality (Lovelock, 1953) of this agent, in
addition to its lack of toxicity and other unique features. Although
numerous investigations of the effects of glvcerol treatment on
extra- and intracellular crystallization have been carried out with

LAMELLAR SYSTEMS

135

Fig. 10. High-resolution electron micrograph of myelin sheath segment
from transverse thin section of frog sciatic nerve, shov/ing splitting of interme-
diate lines (arrows) in concentric laminated structure, v^^ith an average period
of 140 A. Osmium-cryofixation preparation. X 500,000.

the light microscope at low temperatures (Harris, 1954), tliere is
no relevant information available on the interaction of glycerol solu-
tions with the submicroscopic organization of tissues.

Preliminary (unpublished) investigations, carried out in collab-
oration with Dr. J. B. Finean at the Venezuelan Institute for Neu-
rology and Brain Research in 1957, disclosed characteristic modifica-
tions of the low-angle x-ray diffraction pattern of fresh nerve after
treatment with glycerol. The changes produced in the diffraction
pattern of peripheral nerve myelin from giant toad after equilibra-
tion with 30 per cent glycerol in Ringer solution are shown in Fig. 8.
Compared with the pattern of fresh nerve, which features a funda-
mental period of 174 A, the glycerinated nerve shows general swell-
ing of the structure with an enlarged period of 220 A and intensifica-
tion ;of the first- and third-order reflections. The relative intensities
of the x-ray reflections can be restored to approximately normal ap-
pearance by immersion in Ringer solution, but the structure remains

136

MACROMOLECULAR COMPLEXES

Fig. n. High-resolution electron micrograph of myelin sheath segment
from transverse thin section of frog sciatic nerve, showing granular structure
of concentric layers (arrov^s), with a modified regular period of 50 to 60 A
which probably results from enhancement of the intermediate line by treatment
with bromine. Bromine-cryofixation preparation. X450,000.

expanded. The protective effect of glycerol on the myelin structure
could be clearly established by showing that the diffraction pattern
was not changed after rapidly freezing the glycerol-treated nerve
with liquid nitrogen (-195° C) or liquid helium I (—269° C), then
thawing. By contrast, the pattern of untreated nerve, which was
subjected to the same type of freezing and thawing, exhibited
pronounced modifications, indicating a partial breakdown of struc-
ture. In the course of this work it was also found that glycerinated
nerves could be preserved in an essentially unaltered state for
periods of several weeks at room temperature, whereas fresh nerve
kept in normal physiological saline degenerated within a few days.
Only when nerve is kept at 4° C is the disintegration process re-
tarded, but even then the state of preservation cannot compare with
that of glycerinated nerve. This objective evidence of the extensive
protective action of glycerol on myelin ultrastructure emphasized

LAMELLAR SYSTEMS 137

its potential value for electron microscopy. Glycerinated nerve
could not be examined directly with the techniques which were
available at the time, because glycerol is not miscible in methacry-
late, and the required treatment with alcohol followed by embed-
ding at 45° C would have destroyed the unfixed lipoprotein struc-
tures.

However, with the development of the cryofixation techniques,
the glycerol permeating the tissues could be substituted at ade-
quately low temperatures with alcohol or, preferably, methyl cello-
solve, and the tissues impregnated with the acrylic monomers and
finally photopolymerized under favorable conditions which would
minimize extraction and embedding artifacts. The highly encourag-
ing results obtained with this technique have established it during
the past year as the method of choice when dealing with larger
tissue blocks or even whole small organs. Since equilibration with
buffered glycerol solutions of increasing concentration can be per-
formed gradually at temperatures of —5° to —35° C, the usual de-
hydration and extraction effects are markedly reduced. Once a
glycerol concentration of 40 to 60 per cent has been attained, the
tissue can be cooled to — 75° C or lower, and the entire preparation
cycle completed at these "safe" temperatures. When the embedded
tissue is brought up to room temperature for sectioning, the re-
markable preservation of the original configuration and translucent
natural colors stands in marked contrast to the opaque and artificial
appearance of other preparations. Electron microscopy confirms
that a high degree of structural integritv has been achieved, and it
is instructive to see a cross-section through a whole "unfixed" retina
perfectly intact, and free from ice-crystal artifact. In glycerinated
low-temperature preparations, the lamellar systems and the nuclear
structures are better preserved than the loose cytoplasmic matrix;
the remaining artifacts may in fact be reduced when the equilib-
rium with glycerol can be refined or supplemented with a more
physiological protective agent. Examination of thin sections of
glycerinated nerve fibers (Figs. 7, 9) which have been stained with
osmium discloses an expanded myelin layer spacing of the order of
150 to 160 A, with clear-cut indications of a granular fine structure
of the intermediate and dense layers. In several areas this particu-
late fine structure exhibits a certain degree of regularity in the
radial direction. A satisfactory correlation of the x-ray diffraction
patterns with the electron micrographs must await further studies;

138 MACROMOLECULAR COMPLEXES

but it is conceivable that the expansion of the layers may result from
association of the glycerol with the water layers at the aqueous
interfaces of the fundamental repeating unit of myelin. Although
the glycerol effects on the myelin sheath are not completely revers-
ible, this may still be compatible with partial recuperation of physio-
logical activity. Thus Pascoe ( 1957 ) has shown that the rat's isolated
superior cervical ganglion preparation could be protected from
damage by cooling to temperatures as low as —76° C, as judged by
the size of the transmitted action potential. Glycerol ( up to 20 per
cent) progressively slowed nervous conduction, but the effect was
reversible (Pascoe, 1957).

Bromine Cryofixation. In contrast to the osmium-cryofixation
preparations (Fig. 10), treatment with iodine or bromine reveals
the presence of a more compact, granular multilayered structure
(Fig. 11) with a regular period of approximately 50 to 60 A. Al-
though this periodicity seems to result from enhancement of the
intermediate line, as already noted in nerve fixed with KMn04
( Fernandez-Moran and Finean, 1957), there appears to be an addi-
tional staining of the less dense intermediate layer which is normally
not brought out in osmium-fixed preparations. In this connection it
is of interest to point out that, on the basis of his comprehensive
studies of cell membrane structures, Robertson (1959) has postu-
lated the possible existence of a polysaccharide monolayer incor-
porated in the outer layer of the Schwann-cell surface membrane
which contributes to the formation of the intraperiod line in com-
pact myelin. Application of more specific reagents may eventually
permit reliable localization of this highly hydrated, polysaccharide
component.

Fine Structure of the Myelin Layers. Previous investigations
(Fernandez-Moran, 1950, 1952; Fernandez-Moran and Finean,
1957) had revealed that the dense myelin layers dissociate into
granular or rod-shaped particles after a wide variety of treatments,
including enzymatic digestion, KMnOj fixation, and freezing and
thawing of nerve. X-ray diffraction studies (Fernandez-Moran and
Finean, 1957) also furnished indirect evidence for the presence of
a regular organization within the plane of the layers. In vitro nerve
degeneration studies (Fernandez-Moran, 1959b) likewise disclosed
various forms of granular dissociation of the lamellae closely re-

LAMELLAR SYSTEMS 139

sembling the structures observed after enzymatic digestion. Based
on these collateral data, the distinct granular structure of the dense
and intermediate layers in the myelin sheath as revealed by the
new low-temperature preparation techniques merits special atten-
tion, and will be considered in terms of our present concepts of the
molecular organization of myelin.

It is assumed that the fundamental radial unit of the sheath is
formed by two lipoprotein layers consisting of bimolecular leaflets
of 67 A with interposed protein and water layers, each contributing
approximately 25 A (Schmitt, 1959). The two bimolecular leaflets
of mixed lipids are distinguished by a "difference factor" (Finean,
1958), which can be accounted for by assuming that in the process
of myelin formation the asvmmetric Schwann-cell membrane's wrap-
ping itself around the axon (Geren and Schmitt, 1955; Robertson,
1959) will produce a symmetry difference in successive layers. Ac-
cording to Robertson (1959), each repeating unit would therefore
essentially comprise two Schwann-cell membranes in contact along
their outside hydrophilic surfaces to form the intraperiod line. Since
rapid freezing and the described low-temperature preparation tech-
niques directly affect the aqueous interfaces and other hydrated
structures, the characteristic changes can be related in part to the
regular distribution of the water layers in the myelin sheath. The
intensification of the intermediate line observed after freezing and
thawing (Fernandez-Moran and Finean, 1957; Finean, 1958) would
therefore probably result from ice formation and associated freezing
effects along the highly hydrated intraperiod line. By analogy, the
swelling of the myelin layers found after glycerol treatment would
be produced by infiltration and binding of the glycerol within the
same hydrophilic planes. In order to account for the permeability
properties of the membrane, the radially oriented lipid molecules
with a cross-section of 4.7 A are considered to be sandwiched be-
tween monolayers of protein with a fenestrated type of structure
( Engstrom and Finean, 1958; Fernandez-Moran, 1959b ) . Consider-
ing the postulated sieve mechanism and the fluid crystalline state of
the membranes, the water constituent would again have to play an
important role in the organization within the plane of the lipopro-
tein layers. The characteristic dissociation of the layers into rod-
shaped granules produced by freezing, enzymatic digestion, and
glycerol treatment can be understood in terms of an underlying

140

MACROMOLECULAR COMPLEXES

drastic alteration of the hydrated matrix, which is responsible for
the ordered lateral aggregation of the radially oriented lipoprotein
components.

Axon Structures. While surveying the sheath-axon relation-
ships, it was found that a large number of the normally dense axon
filaments, about 100 A in diameter, showed a characteristic "hollow"
annular appearance (Fig. 12, a and b) with a central light area

•^:r:i.i^-z-':.

:t:^'-mr7^x^. ^JW,

i:

/

■w

(n) 1000& Q

?)iv- •

^4' "

Fig. 12. Electron micrographs of myelin sheath-axon segment from trans-
verse section of frog sciatic nerve, showing (a) fine structure of concentric
layers and (b) characteristic annular appearance of cross-sectioned axon fila-
ments which is regularly encountered in low-temperature preparations. Osmium-
cryofixation preparation, (a) X 250,000. (b) X 450,000.

approximately 30 A in diameter. The edges of these ring-shaped
elements exhibit indications of granular fine-structure which could
not be clearly resolved. Although this hollow appearance of the
axon filaments might be due to ice-crystal artifacts, the regularity
of the cross-sectioned structures and the satisfactory preservation of
the adjacent myelin sheath would favor the assumption that we are
dealing with a reproducible structural pattern, corresponding to an

LAMELLAR SYSTEMS

141

equivalent configuration of the native filaments. The images re-
corded are similar to transverse sections of dendrites, or to the
cross-sectioned tobacco mosaic virus particles described earlier
( Fernandez-Moran and Schramm, 1958). These observations may
be of interest in connection with the extensive chemical and physico-
chemical studies of the macromolecular constituents of axoplasm
which are now being carried out by Davison, Taylor, and Schmitt

Fig. 13. Myelin sheath-axon segment from transverse section of frog sciatic
nerve, showing (a) dense body frequently observed in the axon, v/ith (b) in-
ternal layered structure exhibiting a period of 40 to 50 A. Osmium-cryofixation
preparation after rapid freezing in Freon 22-liquid nitrogen, (a) X 80,000.
(b) X 450,000.

(Schmitt, 1959). Another component which is encountered in
osmium-cryofixation specimens more frequently than in the stand-
ard preparations, is represented by dense bodies of 0.2 yu, to over 1 /x
in diameter. As shown in Fig. 13, many of these bodies exhibit a
regular layered fine-structure with a period ranging from 30 to 40 A.
The bodies are more numerous in certain regions of the nerve fiber,
particularly in the vicinity of the nodes.

142 MACROMOLECULAR COMPLEXES

Fine Structure of Photoreceptors. The lamellar systems in photo-
receptors are of particular interest because they represent the pri-
mary sites of visual excitation where light is converted into chemical
or electrical energy. From earlier polarized-light studies (Schmidt,
1937, 1951) it had been deduced that the rod outer segments of
the vertebrate retina consist of transversally arranged, thin protein
layers alternating with longitudinally oriented layers of lipid mole-
cules. Electron microscopy confirmed this conception by demon-
strating that the entire rod outer segment is built up of several
hundred unit disks about 150 A thick, in regular compact arrange-
ment ( Fernandez-Moran, 1954; Sjostrand, 1949, 1953). As first
shown by Sjostrand (1949), the isolated unit disks of the guinea-pig
rod outer segment are about 140 A in thickness and approximately
2 n. in diameter. In thin sections of osmium tetroxide standard prep-
arations (DeRobertis, 1956; Sjostrand, 1953), each disk is seen to
correspond to two dense layers fused at their ends and enclosing a
lighter space, to give a total thickness of approximately 140 A, The
light interspaces between the disks are considered to be filled mainly
with an aqueous, ionic medium, according to Sjostrand, who has also
tentatively assumed that the 30 to 40 A-thick dense osmiophilic
layers are essentially of protein nature, while the 70 to 80 A, less
dense interspace could accommodate a double layer of lipid mole-
cules (Sjostrand, 1959).

Thin sections of light-adapted guinea-pig retinas which have been
subjected to the osmium-cryofixation procedures (Figs. 14, 16) dis-
close a more differentiated picture. The rod outer segments now
appear exceptionally compact, with few indications of the discon-
tinuities frequently observed in standard preparations. The less
dense interspace of each unit disk, which in standard preparations
appears essentially devoid of fine structure, contains a distinct in-
termediate line 15 to 20 A thick (arrows. Figs. 14, 16). This inter-
mediate line is associated with a compact granular material which
gives each unit disk the appearance of greater density than is ob-
served in the corresponding images of standard preparations (Fig.
15). The interspaces between the disks have also been reduced to
barely visible gaps between the dense lines of each disk. This close
apposition of two structurally asymmetric units now bears a close
resemblance to the arrangement of the asvmmetric unit membranes
as seen in the myelin sheath and in chloroplasts ( Hodge, 1959 ) . The
photoreceptor elements, which had apparently constituted an excep-

LAMELLAR SYSTEMS

143

tion to the general structural pattern, are now seen to be invested,
through the presence of an intermediate line, with the same feaures
common to all types of compound unit membranes (Robertson,
1959). These findings have been further confirmed by cryofixation
of other types of vertebrate and invertebrate photoreceptors, using

iMfMK

o -^- ■

PPgWC-i'f'tjitoig'-^A

®w^

tfTii ifai^^^i^' 1 fi

l^1^v^^^^^^

i^<^.-

^ . IBOA

Fig. 14. High-resolution electron micrographs of rod outer segments from
light-adapted guinea-pig retina frozen in liquid helium II, showing intermediate
lines (arrows) and compact fine-structure of unit disks which are not observed
in standard osmium-fixation preparations (see Fig. 15). (a) X600,000. (b)
X 350,000.

Fig. 15. External segment of light-adapted guinea-pig retinal rod, showing
profiles of comparatively "empty" unit disks without intermediate lines, as
commonly observed in standard osmium-fixed preparations. X400,000.

a wide variety of heavy-metal salts and other fixation procedures
( Fernandez-Moran, 1959a). Although the possibility of artifacts
cannot be discoimted here, it nevertheless appears likely that the
low-temperature preparation techniques, with their absence of ex-
traction, and other favorable features, reveal a structural array which
approximates the natural state more closely than do the standard

144 MACROMOLECULAR COMPLEXES

fixation and embedding procedures, with tlieir concomitant extrac-
tion artifacts.

Since the photopigments form an integral part of each unit disk,
it appeared worthwhile to apply the new low-temperature prepara-
tion techniques to a comparative study of dark- and light-adapted
photoreceptors. Rapid freezing preserves the labile photopigments.

Fig. 16. External segment of light-adapted guinea-pig retinal rod, showing
intermediate lines and compact fine-structure of the unit disks. Liquid helium II,
osmium-cryofixation preparation. X600,000.

Fig. 17. External segment of dark-adapted guinea-pig retinal rod, showing
granular and rod-shaped structures (arrows) with indications of orientation in
the planes of the unit disks. Platinic chloride, osmium-cryofixation preparation
after liquid helium ll-freezing. X600,000.

and it was therefore hoped to detect characteristic differences in the
ultrastructure of dark- and light-adapted rod disks as a basis for
further studies on photopigment localization. In preliminary experi-
ments (Fernandez-Moran, 1959a), dark-adapted fresh or glycerin-
ated guinea-pig retinas were rapidly frozen in liquid helium II or in
Freon 22-liquid nitrogen, then subjected to freeze-substitution and
low-temperature polymerization. When the frozen retinas were

LAMELLAR SYSTEMS 145

examined at approximately —150° C in a special low-temperature
stage, the characteristic salmon-pink color of the dark-adapted
retinas could be observed essentially unaltered over periods of
several hours. However, in the subsequent process of freeze-sub-
stitution, the red color regularly changed to a pink-yellow. Upon
completion of the low-temperature photopolymerization process, the
dark-adapted retinas exhibited a distinct yellow color and could be
readily distinguished from the corresponding light-adapted retinas.
If ultrathin sections of dark-adapted retinas are collected on inert
fluids instead of water and examined in the electron microscope,
using a liquid-nitrogen cold stage (Fernandez-Moran, 1959a, 1959d),
a characteristic electron-dense constituent is detected associated
with the double layers. This component appears to be very sensitive
to the effects of the electron beam, and even when a microbeam of
very low intensity is used, it rapidly disintegrates. It was therefore
found necessary to stain the dark-adapted retinas with platinic
chloride and other heavy-metal salts introduced during the process
of freeze-substitution. Under these conditions, the dark-adapted
retinas exhibited a distinct yellow color which was preserved with
little alteration throughout the entire preparation process. Exami-
nation of thin sections of these specimens (Fig. 17) disclosed the
presence of a granular material, associated with the dense layers and
the intermediate line, which seemed to exhibit a certain degree of
orientation as compared with the corresponding preparations of
light-adapted retinas (Fig. 16). However, these observations are
still very preliminary, and much more work remains to be done be-
fore definite conclusions can be drawn. The described changes oc-
curring in dark-adapted retinas during the process of freeze-sub-
stitution emphasize the lability of the photopigments and the need
for caLition in the interpretation of these images. Through a corre-
lation of the electron microscope studies with polarization-optical
analysis of the characteristic dichroism observed in dark-adapted
retinal rods (Schmidt, 1951), it may eventually be possible to local-
ize the photopigments and establish their structural relationships
to the lipoprotein layers.

The photoreceptor elements of the insect eye, which consist of
closely packed, thin-walled tubules approximately 400 to 500 A in
diameter (Fernandez-Moran, 1958), have also been examined with
the new techniques. The walls of each of the tubular compartments
are formed by a thin osmiophilic line, 20 to 30 A wide, associated

146

MACROMOLECULAR COMPLEXES

with a dense layer approximately 60 A thick. In dark-adapted insect
eyes, a dense, granular component with indications of orientation is
seen in the outer tubular compartments, particularly after platinic
chloride-osmium cryofixation. Continuation of these studies, which
have already demonstrated characteristic differences between dark-
and light-adapted photoreceptor elements, will undoubtedly prove
to be of interest in connection with the postulated models of photo-
receptor structure (Wald, 1958; Wolken, 1958b).

Paracrystalline Components of the Retinal Pigment Epithe-
lium. The pigment epithelium is intimately associated both struc-
turally and functionally with the photoreceptor elements of the
vertebrate retina (Wald, 1958). Previous electron microscope
studies (Porter, 1957; Yamada et ah, 1958) had demonstrated the
presence of characteristic lamellated components in the cytoplasm
of the pigment epithelium cells. With improved preparation tech-
niques, it has now been possible to resolve the highly regular, para-

Fig. 18. Dense granules and modified mitochondria in pigment epithelium
of guinea-pig retina. Certain forms of these granules are surrounded by multiple
membranes (arrow) and show no distinct internal fme-structure in standard
preparations. Osmium fixation-araldite embedding. X30,000.

LAMELLAR SYSTEMS

147

1M

Fig. 19. Modified mitochondria in pigment epithelium cell of light-adapted
guinea-pig retina, showing characteristic condensation of granular material in
the layered structure. Numerous intermediate stages are found which show
gradual transformation of the layered structure into the highly regular com-
ponent of the paracrystalline granules. Osmium-cryofixation preparation.
X 55,000.

crystalline structure of certain types of dense granules within these
cells and to study their interrelationship with the lamellar com-
ponents of the cytoplasm. The slender, microvilli-like processes of
the pigment epithelium cells envelop the retinal rod outer segments
and establish intimate contact. In addition to specialized lamellar
components (Porter, 1957; Yamada et al., 1958), numerous dense
granules approximately 0.3 //, to 1 or 2 /x in diameter are found in
these cells. As indicated by Yamada et al. (1958), these "fuscin
granules" are extremely dense, and their internal structure is difficult
to study because of their mechanical hardness, so that only small
particulates or fine tubular networks could be seen.

In standard osmium-fixed and araldite-embedded preparations of
the guinea-pig retinal pigment epithelium, these granules can be
sectioned with the diamond knife, and exhibit an extremely com-
pact, granular fine-structure (Fig. 18) with only occasional indica-

148

MACROMOLECULAR COMPLEXES

tions of fine structure. In osmium-cryofixation preparations, most
of these granules show a characteristic highly ordered paracrystalline
arrangement (Figs. 20, 21). Depending on the plane of sectioning,
a typical cross-grating pattern formed by dense round or polygonal
structures 30 to 40 A in diameter with a regular spacing of 50 to 60 A
is seen; or line patterns with the same spacing. This type of organ-
ization, which closely resembles the paracrystaUine structure de-

Fig. 20. High-resolution electron micrograph of ultrathin section of para-
crystalline granule from pigment epithelium of guinea-pig retina, showing
highly regular arrangement of the constituent dense particles, 30 A in diameter,
with an average spacing of 50 to 60 A. Osmium-cryofixation preparation.
X 300,000.

scribed in the inclusion bodies of certain insect viruses and other
biological paracrystalline components, is clearly different from the
compact lamellar systems of the myeloid bodies and other lamellated
corpuscles with a period of 40 to 100 A, as described by other
authors (Porter, 1957; Yamada et at., 1958). Examination of a large
number of sections discloses characteristic transitions between the
mitochondria and these dense pigment granules (Fig. 18). The
lamellar structure of the mitochondria shows dense patches of gran-

LAMELLAR SYSTEMS 149

ular material (Fig. 19) which gradually infiltrates the entire matrix
and finally separates out into these highly ordered structures. The
intermediate stages are surrounded by multiple membranes which
are gradually lost. These transition stages demonstrate how a lamel-
lar structure can transform into a more highly ordered paracrystal-
hne system by a gradual process of condensation. Correlative bio-
chemical studies of the paracrystalline granular components would

Fig. 21. Paracrystalline granule from pigment epithelium of guinea-pig
retina, showing regular line or cross-grating pattern of the dense particles,
depending on their orientation in the plane of sectioning. Osmium-cryofixation
preparation. X 400,000.

be of interest in view of the studies carried out by Wald and co-
workers (Wald, 1958) which emphasize the active contribution of
the pigment epithelium in the visual processes. The described para-
crystalline components bear a strong resemblance to the protein
crystals in the eggs of Limnea stagnalis described by Elbers ( 1957) ;
and to the vitelline bodies in the eggs of Planorbis corneus. The
high contrast in the macromolecular constituents of these protein
crystals with a similar periodicity of 60 A is attributed to the iron
content. Of particular interest are the numerous transition forms

150 MACROMOLECULAR COMPLEXES

described by Favard and Carasso (1958), showing the mitochon-
drial origin of these paracrvstalhne granules. Further studies will
contribute to a better understanding of the functional significance of
this type of paracrystalline component in the cytoplasm of the dif-
ferent kinds of cells.

Discussion

After this survey of representative lamellar structures as revealed
by different preparation methods, it may be profitable to discuss the
suggested molecular organization of lamellar systems in relation to
general concepts of energy transfer processes which are emerging
from recent biochemical and biophysical studies.

Molecular Organization of Lamellar Systems. From a consid-
eration of the common structural pattern revealed at the macro-
molecular level in the myelin sheath, photoreceptors, mitochondria,
cytoplasmic lamellar complexes, and other lamellar systems, the
following general picture of their molecular organization is obtained.
The highly ordered lipoprotein layers derived from close packing
of structurally asymmetric unit cell membranes, permeated with
hydration water of possil^le icelike character, serve as a "paracrys-
talline" matrix in which assemblies of specific enzymes, associated
photopigments, or specialized electron-transfer systems ( Lehninger,
1959; Wolken, 1958a; Kropf and Hubbard, 1958; Szent-Gyorgyi,
1957) are precisely arranged in regular patterns.

Of special interest are the recent studies of Calvin and associates
( Calvin, 1958, 1959a, 1959b ) on energy-transfer processes in photo-
synthesis. These studies indicate that the initial process of energy
reception and transfer is dependent on the specific organization of
the chlorophyll and associated pigments in the highlv ordered
lamellar systems. Absorption of a light quantum by chlorophvll in
the ground state would raise it to its lowest singlet excited state,
allowing it to move around among the chlorophvll molecules by
resonance transfer. The "exciton," visualized as a charge-pair which
cannot move separately, migrates as an electron and "hole" through
the oriented array of chlorophyll molecules. The conjugated carot-
enoid molecules would act as conductors of electrons, while the
lipids in the chloroplasts would function as an insulator permitting
a separation of charges when the electrons are "trapped" in suitable
places of low potential energy on one side of the layer and the

LAMELLAR SYSTEMS 151

"holes" on the other. According to the proposed scheme (Calvin,
1959a), "the exciton is converted into conductive carriers, these into
chemical radicals, and the chemical radicals lead to stable chem-
icals" (p. 161). Direct experimental evidence for the existence of
trapped electrons and holes has been provided bv recording elec-
tron-spin resonance signals from illuminated whole chloroplasts at
temperatures as low as —150° C (Calvin, 1959a, 1959b).

Electron-spin resonance techniques are being extensivelv applied
to detect the presence of free radicals in enzymatic reactions, and of
other intermediates with unpaired electron-spin in biological sys-
tems. However, these techniques lack the necessary sensitivity for
adequate detection of the small, steady-state concentrations of free
radicals in many biological processes, and for a specific localization
within the complex lamellar systems. As pointed out earlier, the
possibility of trapping and detecting free radicals within highly
organized tissue components by rapid freezing with liquid helium II
represents one of the major justifications for the use of extremely
low temperatures in ultrastructure research.

It has been suggested that lamellar systems might have many
properties in common with semiconductors (Fernandez-Moran,
1957; Szent-Gyorgyi, 1957), although the concepts of solid-state
physics are not directly applicable to the fluid-crystalline state
(Fernandez-Moran, 1954, 1959b). Recent physicochemical studies
of water and ice by Eigen and his associates indicate that "The
('excess' and 'defect') proton-conducting H-bond systems (which
are also of biological interest) show certain parallels to electronic
semiconductors" (Eigen and DeMaever, 1959, p. 84). The semi-
conductor concept is now introduced from the unexpected quarter
of water, the very constituent which had previously restricted the
solid-state analogy. Attention is thus focused on a novel aspect of
this fundamental "matrix of life" as Szent-Gyorgyi ( 1957 ) has so
aptly designated water.

The Role of Hydration Water In Lamellar Systems. Water is a
major component of all lamellar complexes, constituting, for example,
at least 35 per cent of the myelin sheath. The water layers at the
aqueous interfaces of the fundamental repeating unit in the compact
myelin are considered to be about 12 to 15 A thick (Schmitt, 1959).
It has 'been shown (Robertson, 1959; Schmitt, 1959) that, under
certain experimental conditions, much larger amounts of water can
be incorporated between these aqueous interfaces, leading to con-

152 MACROMOLECULAR COMPLEXES

siderable swelling. It is assumed that water is coordinated on the
protein layer located at the aqueous surface of each bimolecular
leaflet, the thickness of the water layer being largely determined by
the electrical charge density at these interfaces, and by the ionic
strength. As Schmitt (1959) has pointed out, these aqueous spaces
between the membranes play an important role as channels for the
products of metabolism, including ions, estabhshing contact be-
tween internal and external cellular spaces. The classic work of
Teorell (1958) has demonstrated that periodic transport of water
across the membranes and electroendosmotic phenomena are also
of fundamental importance in relation to the nerye mechanism.

The effects of the electrically charged lipid groups on the struc-
ture of the hydration water in these aqueous interfaces may also
induce the formation of a "crystalline" aiTangement resembling the
"frozen" hydration sheaths postulated in protein molecules (Fer-
nandez-Moran, 1959b). According to Klotz (1958) and Klotz and
Luborsky ( 1959 ) , the induction of a crystalline lattice of hydration
water may be ascribed to the local and long-range cooperatiye elec-
tric field effects of the non-polar side-chains in proteins ( Frank and
Wen, 1957). The general problem of proton hydration and transfer
in aqueous solution is also of fundamental interest to the biologist.
According to the model proposed by Eigen and DeMaeyer (1959),
the proton exists as an H3O+ ion, connected to at least three addi-
tional HjO molecules by means of hydrogen bonds to form a H<)04 +
complex (Beckey, 1959) (primary hydration shell), with a proton
of excessiye mobility. The proton in ice behayes more like an elec-
tron than like an ion, having a mobility which is only 10 to 50 times
smaller than the mobility of an excited electron in iHuminated alkali
halide crystals. Eigen and DeMaeyer (1959) have pointed out that
these remarkable kinetic properties of protons may prove to be of
important biological significance. If we assume the existence of ice-
like hydration shells permeating the entire lipoprotein matrix, then
this high proton mobility would become an important factor to
reckon with in addition to the postulated electron mobility.

Experimental Approaches for Correlation of Structure and
Function. It is against this background of a far more d}'namic pic-
ture involving migration of electrons, production of free radicals,
and the unique kinetic properties of protons that we must view the
experimental possibilities for establishing a closer correlation be-
tween ultrastructure and function in the ordered lamellar systems.

LAMELLAR SYSTEMS 153

Our criteria for preservation of biological systems must therefore
be considerably extended to take into account the new range of
phenomena being revealed by refined physicochemical techniques.
From a hypothetical point of view, the "ideal" preparation tech-
nique for electron microscopy of biological systems would first have
to achieve complete immobilization of all tissue constituents in their
natural position by rapid freezing to temperatures close to absolute
zero, so that even free radicals and other transient intermediates
are trapped in an inert matrix and stored under conditions restrict-
ing diffusion and radical recombination. Once the submicroscopic
organization of the tissue has been "frozen in" to this characteristic
"solid state," a critical transformation must then be effected at the
molecular level in order to change the metastable equilibrium,
which was essentially determined by the water matrix, into a new,
more stable phase. This induced phase transfoiTnation, which can
be likened to a process of precisely controlled, heterogeneous nucle-
ation, would ensure adequate preservation of the tissue ultrastruc-
ture over a wide enough range of temperature and other physical
conditions to permit adequate manipulation during subsequent ul-
trathin sectioning and electron microscope examination. Such a
highly idealized form of preservation does not appear to be un-
attainable in principle, and eventually it may be possible to make
use of recent advances in polymer chemistry involving addition
polymerization reactions with certain ionic catalysts for this type
of controlled synthesis of ordered polymers fitted specifically within
the tissue matrix at low temperature. In order to obtain the re-
quired fundamental data on the hydration state and properties of
water in lamellar systems, combined application of nuclear mag-
netic-resonance spectrometry ( Denis et al., 1957; Fernandez-Moran,
1957), low-temperature electron microscopy (Fernandez-Moran,
1959d), and neutron diffraction analysis may prove to be essential.
Of immediate interest in connection with the potentialities of cryo-
fixation techniques (Fernandez-Moran, 1959a, 1959c, 1959d) would
be an attempt to arrest sequentially the different states of activity
in nerve and photoreceptors (Katz, 1959; Kropf and Hubbard,
1958). As already shown by the pioneering work of von Muralt
(1946, 1958), rapid freezing of excited nerve fixes the chemical
status 'and produces an accumulation of excitation waves. The re-
cent investigations by Hill and his associates (Abbott et al., 1958)
on heat production in nerve demonstrated a complex cycle of posi-

154 MACROMOLECULAR COMPLEXES

tiv'e and negative phases, indicating that the chemical changes
involved are so extensive that, according to Nachmansohn (1959):
"they ought to be susceptible to biochemical analysis" (p. 7). In
view of the characteristic coupling between the enzymatic reac-
tions and these ordered lamellar structures, a systematic investiga-
tion of excited nerve fibers by means of cryofixation techniques
might be rewarding. In this way, it may be possible to detect at
the molecular level the structural correlates of the role of the acetyl-
choline system in the permeability changes associated with nerve
conduction, as postulated by Nachmansohn ( 1959 ) on the basis of
his classic studies of the biochemical mechanisms of nerve activity.
Tissue culture techniques (Hild, 1957; Peterson and Murray, 1955)
should be of great operational value in connection with ultrarapid
freezing because of the critical limitations imposed by the specimen
size. Chemical and biochemical investigations (Folch et ah, 1958)
of the myelin sheath and other lamellar systems will continue to
provide the frame of reference for all research in this field, making
full use of the specific advantages inherent in the new techniques
for the study of enzymatic activity by electron microscopy ( Barrnett
and Palade, 1959), high-resolution autoradiography (Fernandez-
Moran, 1959d; Hoerlin and Hamm, 1953), and related cytochemical
methods. With an integrated physiological, biochemical, and bio-
physical approach, we may thus hope to gain a better understand-
ing of these Huid crystal elements which are the true "energy trans-
ducers of life."

Summary

A review is presented of the salient structural features of repre-
sentative lamellar systems in the nerve myelin sheath and in photo-
receptors, as revealed bv high-resolution electron microscopv,
using standard preparation techniques and improved low-tempera-
ture preparation methods based on rapid freezing with liquid helium
II. The information obtained by correlative application of direct
and indirect methods has furnished a reliable basis for evaluation
of the findings obtained with the new methods. Characteristic
paracrystalline granules of the retinal pigment epithelium in the
guinea pig are described. The molecular organization of lamellar
systems is discussed in relation to general concepts of energy trans-
fer processes which are emerging from recent biochemical and bio-
physical studies.

LAMELLAR SYSTEMS 155

References

Abbott, B. C, A. V. Hill, and J. V. Howahth. 1958. The positive and negative
heat production associated with a nerve impulse. Proc. Roy. Sac. (London) B
148: 149.

Allen, J. F. 1952. Liquid Helium. In F. E. Simon, N. Kurti, and E. Mendels-
sohn (eds.). Low Temperature Phifsics. Pergamon Press, Ltd., London. Pp. 66-93.

Barrnett, R. J., and C. E. Palade. 1959. Enzymatic activity in the M band. /.
Biophij.s. Biochem. Ctjtol. 6: 163-170.

Beckey, H. D. 1959. Massenspektrometrische Untersuchungen mit Hilfe einer
Feldemissions-Ionenquelle. Z. Natui-forsch. 14a: 712-721.

Broida, H. p. 1957. Stabilization of free radicals at low temperatures. Ann. N. Y.
Acad. Sci. 67(9): 530-545.

Calvin, M. 1958. From microstructure to macrostructure and function in the photo-
chemical apparatus. In The Photochemical Apparatus: Its Structure and Func-
tion. Brookhaven St/mposia in Biol. No. 11: 160-180.

Calvin, M. 1959a. Free radicals in photosynthetic systems. Revs. Modern Phifs.
31: 157-161.

Calvin, M. 1959b. Energy reception and transfer in photosynthesis. Revs. Modern
Phys. 31: 147-156.

Denis, P., A. Csaki, M. Delco, J. Sprenger, H. Fernandez-Moran, and W. Rawy-
LER. 1957. Spectrometre a resonance magnetique nucleaire utilisant des transis-
tors. Arch. sci. (Geneva) 10: 223.

DeRobertis, E. 1956. Electron microscope observations on the submicroscopic or-
ganization of retinal rods. /. Biophys. Biochem. Cytol. 2: 319-329.

Eigen, M., and L. DeMaeyer. .1959. Hydrogen bond structure, proton hydration,
and proton transfer in aqueous solution. In W. J. Hamer (ed.). The Structure oj
Electrolytic Solutions. John Wiley & Sons, Inc., New York. Pp. 64-85.

Elbers, p. F. 1957. Electron microscopy of protein crystals in ultrathin sections of
the egg of Limnea stagnalis. Koninkl. Ned. Akad. Wetenschap Proc. B 60: 96-98.

Engstrom, a., and J. B. Finean. 1958. Biological Ultrastructure. Academic Press,
Inc., New York.

Favard, p., and H. Carasso. 1958. Origine et ultrastructure des plaquettes vitellines
de la planorbe. Arch, d'anatomie microscopique et de morphologic experimentale
47(2): 211-234.

Fedeb, N., and R. L. Sidman. 1958. Methods and principles of fixation by freeze-
substitution. /. Biophys. Biochem. Cytol. 4: 593-600.

Fernandez-Mor.\n, H. 1950. An electron microscope study of rat and frog sciatic
nerves. Exptl. Cell Research 1 : 309-340.

Fernandez-Moran, H. 1952. The submicroscopic organization of vertebrate nerve
fibres. Exptl. Cell Research 3: 282-350.

Fernandez-Moran, H. 1953. A diamond knife for ultrathin sectioning. Exptl. Cell
Research 5: 255.

Fernandez-Moran, H. 1954. The submicroscopic structure of nerve fibres. Progr.
in Biophys. and Biophys. Chem. 4: 112.

Fernandez-Moran, H. 1957. Electron microscopy of nervous tissue. In E. Richter
(ed.). Metabolism of the Nervous System. Pergamon Press, Ltd., London.

Fernandez-Mor.\n, H. 1958. Fine structure of the light receptors in the compound
eyes of insects. Exptl. Cell Research Suppl. 5: 586.

Fernandez-Moran, H. 1959a. Electron microscopy of retinal rods in relation to
localization of rhodopsin. Science 129: 1284.

Fern.'\ndez-Moran, H. 1959b. Fine structure of biological lamellar systems. Revs.
Modern Phys. 31: 319-330.

FernaWdez-Moran, H. 1959c. Cryofixation and supplementary low-temperature
preparation techniques applied to the study of tissue ultrastructure. Proc. 17th
Ann. Meeting, Electron Microscope Society of America. /. Appl. Phys. 30(12):
2038.

156 MACROMOLECULAR COMPLEXES

Fernandez-Moran, H. 1960. Low-temperature preparation technique for electron
microscopy of biological specimens based on rapid freezing with liquid helium
II. In Conference on Freezing and Drying of Biological Materials. N. Y. Acad.
Sci. 85: 689-713.

Fernandez-Mor.\n, H., and R. Brown. 1958. Submicrcscopic organization and
function of nerve cells. Exptl. Cell Research Suppl. 5: 1-644.

Fernandez-Moran, H., and J. B. Finean. 1957. Electron microscope and low-angle
x-ray diffraction studies of the nerve myelin sheath. /. Biophys. Biochem. Cytol.
3(5): 725-748.

Fernandez-Moran, H., and G. Schramm. 1958. The structure of tobacco mosaic
virus as revealed in ultrathin sections by electron microscopy. Z. Naturforsch.
13b: 68-71.

Finean, J. B. 1958. X-ray diffraction studies of the myelin sheath in peripheral and
central nerve fibres. Exptl. Cell Research Suppl. 5: 18.

Finean, J. B., F. S. Sjostrand, and E. Steinman. 1953. Submicroscopic organiza-
tion of some layered lipoprotein structures. Exptl. Cell Research 5: 557.

FoLCH, J., M. Lees, and S. Carr. 1958. Studies of the chemical composition of the
nervous system. Exptl. Cell Research 5: 58-71.

Fox, T. G., B. S. Garrett, W. E. Goode, S. Gratch, J. S. Kincaid, A. Spell, and
J. D. Stroupe. 1958. Crystalline polymers of methyl methacrylate. /. Am. Chem.
Soc. 80: 1768-1769.

Frank, H. S., and W-Y, Wen. 1957 Structural aspects of ion-solvent interaction in
aqueous solutions. A suggested picture of water structure. Discussions Faraday
Soc. 24: 133-140.

Geren, B. B., and F. O. Schmitt. 1955. Electron microscope studies of Schwann
cell and its constituents with particular reference to their relation to the axon.
In Symposium on Fine Structure of Cells. Interscience Publishers, Inc., New
York. P. 251.

Gersh, I., and J. L. Stephenson. 1954. Freezing and drying of tissues for morpho-
logical and histochemical studies. In R. J. C. Harris (ed. ). Biological Applica-
tions of Freezing and Drying. Academic Press, Inc., New York. Pp. 329-384.

Goldsmith, T. H., and D. E. Philpott. 1957. The microstructure of the compound
eyes of insects. /. Biophys. Biochem. Cytol. 3(3): 429.

Harris, R. J. C. (ed. ). 1954. Biological Applications of Freezing and Drying. Aca-
demic Press, Inc., New York.

HiLD, W. 1957. Myelogenesis in cultures of mammalian central nervous tissue. Z.
Zellforsch. u. mikroskop. Anat. 46: 71-95.

Hodge, A. |. Fine structure of lamellar systems as illustrated bv chloroplasts. Revs.
Modern Phys. 31(2): 331-341.

HoERLiN, H., and F. A. Hamm. 1953. Electron microscopical studies of the latent
image obtained bv exposures to alpha particles, x-rays, and light. /. Appl. Phys.
24(12): 1514.

Katz, B. 1959. Nature of the nerve impulse. Revs. Modern Phys. 31(2): 466-474.

Klein, R., and M. D. Scheer. 1958. The reaction of hydrogen atoms with soHd
olefins at -195° C. ]. Phys. Chem. 62: 1011.

Klotz, I. M. 1958. Protein hydration and behavior. Science 128: 815.

Klotz, I. M., and S. W. Luhorsky. 1959. The binding of organic ions by proteins.
Effect of changes in solvent and their implications as to the nature of the com-
plexes. /. Am. Chem. Soc. 81: 5119-5124.

Kropf, a., and R. Hubbard. 1958. The mechanism of bleaching rhodopsin. Ann.

N. Y. Acad. Sci. 74: 266-280.
Lehninger, a. L. 1959. Respiratorv-energy transfonnation. Revs. Modern Phys.

31: 136-146.
Lovelock, J. E. 1953. The mechanism of the protective action of glycerol against

haemolysis by freezing and thawing. Biochim. ef Biophys. Acta 11: 28-36.
LuMSDEN, C. E. 1957. Aspects of the chemistry of the myelin and sheath cell com-
plex. In Modern Trends in Neurology. Butterworth Publications, Ltd., London.
Ser. II: 130.

LAMELLAR SYSTEMS 157

LuYET, B. J. 1957. On the growth of the ice phase in aqueous colloids. Proc. Roy.

Soc. (London) B 147: 434-451.
Mendelssohn, K. 1956. Liquid helium. In S. Flugge (ed. ). Handbuch der Physik.
Encyclopedia of Physics. XV. Low Temperature Physics. Springer Verlag, Berlin.
Vol. 2. Pp. 370-461.
Meryman, H. T. 1956. Mechanics of freezing in living cells and tissues. Science

124: 515-521.
Miller, W. H. 1957. Morphology of the ommatidia of the compound eye of Limulus.

J. Biophys. Biochem. Cytol. 3(3) : 421-427.
MiNKOFF, G. J. 1959. Frozen free radicals. In K. Mendelssohn (ed.). Progress

in Cryogenics. Academic Press, Inc., New York. Vol. 1. Pp. 139-178.
MiJLLER, H. R. 1957. Gefriertrocknung als Fixienrngsmethode an Pflanzenellen.

/. Ultrastruct. Research. 1: 109-137.
Nachmansohn, D. 1959. Chemical and Molecular Basis of Nerve Activity. Aca-
demic Press, Inc., New York.
Parkes, a. S. 1951. Preservation of spermatozoa, red blood-cells, and endocrine
tissue. In R. J. C. Harris (ed.). Freezing and Drying The Institute of Biology,
London. Pp. 99-105.
Pascoe, J. B. 1957. The survival of the rat's superior cervical ganglion after cooling

to -76° C. Proc. Roy. Soc. (London) B 147: 510-519.
Patten, S. F., Jr., and K. A. Brown. 1958. Freeze-solvent substitution technic.

Lab. Inv. 7: 209.
Peterson, E. R., and M. R. Murray. 1955. Myelin sheath formation in cultures of

avian spinal ganglia. Am. J. Anat. 96(3) : 319-346.
Porter, K. R. 1957. The submicroscopic morphology of protoplasm. Harvey Lec-
tures Ser. 51 (1955-1956): 175.
Robertson, J. D. 1959. The ultrastructure of cell membranes and their derivatives.

Biochem. Soc. Symposia (Cambridge, England) 16: 3-43.
Schmidt, W. J. 1937. Die Doppelbrechung von Karyoplasma, Zytoplasma und

Metaplasma. Gebriider Borntraeger, Berlin.
Schmidt, W. J. 1951. Polarisationsoptische Analyse der Verkniipfung von Protein-
und Lipoidmolekeln, Erlautert am Aussenglied der Sehzellen der Wirbeltiere.
Pubbl. staz. zool. Napoli 23: 158-183.
ScHMiTT, F. O. 1944. Tissue ultrastructure analysis. In O. Glasser (ed.). Medical

Physics. Year Book Publishers, Chicago. P. 1586.
ScHMiTT, F. O. 1959. Molecular organization of the nerve fiber. Revs. Modern

Phys. 31(2): 455-465.
ScHMiTT, F. O., R. S. Bear, and G. L. Clark. 1935. X-ray diffraction studies on

nerve. Radiology 25: 131.
Simpson, W. L. 1941. An experimental analysis of the Altmann technic of freezing-
drying. Anat. Record 80: 173.
SjosTRAND, F. S. 1949. An electron microscope study of the retinal rods of the

guinea pig eye. /. Cellular Comp. Physiol. 33: 383.
SjosTRAND, F. S. 1953. The ultrastructure of the outer segments of rods and cones
of the eye as revealed by the electron microscope. /. Cellular Comp. Physiol.
42: 15-44.
SjosTRAND, F. S. 1959. Fine structure of cytoplasm. Revs. Modern Phys. 31(2) : 301-

318.
SjosTRAND, F. S., and R. F. Baker. 1958. Fixation by freezing-drying for electron

microscopy of tissue cells. /. Ultrastruct. Research 1: 239-246.
Stephenson, J. L. 1956. Ice crystal growth during the rapid freezing of tissues.

/. Biophys. Biochem. Cytol. Suppl. 2(4): 45-52.
Szent-Gyorgyi, a. 1957. Bioenergetics. Academic Press, Inc., New York.
Teorel^, T. 1958. Transport processes in membranes in relation to the nerve mech-
anism. Exptl. Cell Research Suppl. 5: 83-100.
voN Muralt, a. 1946. Die Signaliibermittlung im Nerven. Birkhauser, Basel.
von Muralt, A. 1958. The role of thiamine in nervous excitation. Exptl. Cell Re-
search Suppl. 5: 72-79.

158 MACROMOLECULAR COMPLEXES

Wald, G. 1958. Photochemical aspects of visual excitation. Exptl. Cell Research
Suppl. 5: 389.

WoLKEN, J. J. 1958a. The chloroplast structure, pigment, and pigment-protein com-
plex. In The Photochemical Apparatus: Its Structure and Function. Brookhuven
Symposia in Biol. No. 11: 87-100.

WoLKEN, J. J. 1958b. Photoreceptor structures. Ann. N. Y. Acad. Set. 74: 164-181.

Yamada, E., K. Tokuyasu, and S. Iwaki. 1958. The fine structure of retina studied
with electron microscope. II. Pigment epithelium tmd capillary of the chorio-
capillary layer. /. Electronmicroscopy (Chiba) 6: 42-46.

DISCUSSION
A. C. Giese, H. Ferndndez-Mordn

Dr. Giese (Stanford University): For what biological problems other than
the ones you have mentioned are the techniques you describe likely to be
especially useful?

Dr. Fernandez-Moran: The described cryofixation and related low-
temperature preparation techniques have been used primarily for the study of
biological lamellar systems and of nuclear structures. Although they are still
in an early stage of development, the following fields of application appear to
be the most immediate ones:

(1) The study of nuclear stmctures. At low temperatures, and with the
described techniques, nuclear stiuctures appear to be better preserved than
by conventional procedures. Combined staining, selective extraction, and
enzymatic digestion can be carried out on these sections which have been
subjected to a minimum of chemical fixation.

(2) Of particular interest are the autoradiography techniques. The pre-
liminary attempts to increase the resolving power of autoradiography by using
tritium-labeling in connection with electron microscopy should be carried
further. If successful, they would permit detection of radioactive tracers at the
macromolecular level; that is to say, at a resolution of 200 to 400 A, and per-
haps less in the case of track formation. This would be particularly useful in
connection with radiotracer studies of nucleic acid components in bacterio-
phages and viruses for the contact radiography method. A collateral investi-
gation could, of course, also be carried out with dilterential ultracentrifugation
and radioautography studies. In this case, the pellet itself would be sectioned,
and radioautography performed, thus considerably increasing the scope of this
technique, which has already proved so powerful.

Beyond these, there are, of course, numerous other applications. The main
value of the new techniques derives from the fact that we are now in a position
to considerably extend the applications of high-resolution electron microscopy
in the study of biological tissues. Moreover, low-temperature stage observa-
tions permit us to enter a domain which has hitherto remained unexplored.
Taking into consideration that practically all observations published to date
have been made with the specimen heated to relatively high temperatures
within the microscope, the value of being able to cool it down becomes evi-
dent. Specific examples would be:

LAMELLAR SYSTEMS 159

( 1 ) The study of labile components such as photopiginents at relatively
low temperatures. As mentioned earlier, the possibility of making thin films
of photopigments such as extracted rhodopsin, and then examining them at
low temperatures with the electron microscope, appears to be quite promising.

(2) The study of tissues with their water components (hydration shells
and extracellular water) retained in their native states.

Many other applications can be foreseen in connection with x-ra\ diffrac-
tion studies, including investigations of the parameters connected with long-
term preservation of biological systems, and the possibility of exploring new
phenomena which might occur at these extremely low temperatures.

Fibrillar Systems in the Mitotic Apparatus

Hans A. Went - and Daniel Mazia ^

Macromolecules are intimately associated with all phases of cell
function and, indeed, are absolutely indispensable for many. Some,
such as enzymes, can apparently perform properly as individual
molecules in vitro, although within a cell they may be coupled to
structural elements imparting rigorous spatial orientation to the
enzymes. Small amounts can be studied profitably by virtue of their
catalytic activity. Other macromolecules occur in highly oriented
aggregates forming specific subcellular structures such as chromo-
somes, chloroplasts, mitochondria, plasma membrane, nuclear en-
velope, and mitotic apparatus. The lack of easily measured activity
specific for these macromolecules frequently presents difficulties to
experimental elucidation of their function within the aggregates.
One of the first steps that must be taken to study a macromolecular
aggregate is to isolate it in quantity and disperse it into its con-
stituent molecular subunits in a uniform solution. A potential ob-
stacle arises when observations on a dispersed system are extra-
polated and applied as an explanation for the structure and function
of the intact, oriented aggregate. However, we will have to content
ourselves with this limitation pending improved experimental pro-
cedures.

Many cells entering division construct an elaborate structure
whose unique function appears to be that of insuring equal dis-
tribution of the hereditary material to the daughter cells. The
modern term "mitotic apparatus" was applied to this structure by
Mazia and Dan (1952) and embraces both the achromatic and the
chromatic portions of the nuclear division figure as described by
Flemming (1882). The centrosomes, the spindle, and the asters
constitute the achromatic portion, and the chromatic portion is now
}

^ This work was in part supported by University of California research funds.
^ Department of Zoology, Washington State University, Pullman, Washington.
^ Department of Zoology, University of California, Berkeley, Cahfornia.

161

162 MACROMOLECULAR COMPLEXES

called the chromosomes. It is not unnatural that a structure of the
beauty of the mitotic apparatus should receive so much attention
from the early cytologists, and many of the classic observations
made by them have not been surpassed to this day.

One of these observers who speculated about its origin was
Flemming ( 1882), who offered two alternatives for the source of the
material incorporated into the spindle fibers of the mitotic appara-
tus. The first alternative states that the spindle fibers arise from the
pale fibers of the achromatic region of the prophase nucleus, which
in turn originate from the reticulum and nucleoli of the interphase
nucleus. The second alternative imagines that they are derived from
the cytoplasm, growing from this into the nucleus. Other distin-
guished scholars have held similar views. Boveri (1887), Stras-
burger (1880), Wilson (1925), and Hertwig (1909) have all
considered as valid the formation of the achromatic figure from
pre-existing structures visible in their cytological preparations.
There was general agreement that the asters were of cytoplasmic
origin but considerable dissension over whether the spindle fibers
were of nuclear or of cytoplasmic origin. However, either interpre-
tation could be valid, depending upon the material under consider-
ation. In its essential form, the precursor concept for the origin of
the mitotic apparatus germinated during the fertile "Golden Era"
of cytology and, after about 50 years of relative dormancy, it is being
revived. Elucidation of the origin of the mitotic apparatus could
provide a made-to-order system in which to examine the completely
reversible aggregation of molecular units into a transient structure
of unquestionable significance to the cell. Important to this studv is
the synchronization that exists between the appearance of the
mitotic apparatus and other cell activities, which provides the ex-
perimenter with a convenient clock.

The problem of the molecular origin of the mitotic apparatus
could not be investigated directly until Mazia and Dan (1952)
demonstrated that this structure could be freed of its investing
cytoplasm in large numbers from a population of dividing sea-urchin
eggs. One could now inquire experimentally into the possible modes
of origin of the mitotic apparatus. Two basically different ones at
opposite extremes were envisaged. One of these views the formation
of the mitotic apparatus as the result of the de novo synthesis from
small, non-specific units such as amino acids or small polypeptides.
These may arise from the earlier breakdown of some storage pro-

FIBRILLAR SYSTEMS IN THE MITOTIC APPARATUS 163

teins or be present in a pool. At the other extreme, we may imagine
the existence in the cytoplasm of preformed molecular subunits
which are assembled into the definitive mitotic apparatus with little
or no modification in their structure. The latter will be referred to
as the "precursor concept." It does not differ basically from views
held by the early cytologists, except at the level of organization of
the precursor units. They thought in terms of the rearrangement of
microscopically detectable structures present in their cytological
preparations, while we look upon the precursor units as being of
macromolecular dimensions. The problem, therefore, was to dis-
tinguish a net synthesis, on the one hand, from the spatial rearrange-
ment of pre-existing structures, on the other. The advent of the
technique for obtaining isolated mitotic apparatus suggested an ex-
perimental design which could decide between these two extremes.
Immunology offered potentially a suitable method for attempting
to identify some component of the unfertilized egg with a structural
component of isolated mitotic apparatus. If the mitotic apparatus
were formed de novo, it would be reasonable to expect the complete
absence of its structural constituents from the unfertilized egg. The
precursor concept, as stated earlier, would predict the existence in
the mitotic apparatus of molecules identical to a species present in
the unfertilized egg.

The antigen-antibody reaction is one of the most specific chemi-
cal reactions known ( Landsteiner, 1945), but this in itself did not
provide a suitable means to identify with each other the same molec-
ular species from different sources. An added complication was the
expected heterogeneous nature of the reactants. The gel-diffusion
technique, as originated by Oudin ( 1946 ) and modified by Ouch-
terlony ( 1949), provided the ideal solution to these dilemmas. Two-
dimensional gel-diffusion techniques (Oudin, 1952; Ouchterlony,
1958) can resolve very complex antigen-antibody reaction systems
into a series of discrete bands of precipitate immobilized by the gel
medium, and further provide an extremely sensitive means for es-
tablishing an identity or non-identity between antigens from differ-
ent sources. When bands of precipitate assignable to different
antigen solutions fuse or merge, instead of intersecting, the respon-
sible antigens are considered to be immunologically identical.

What restrictions must be imposed upon an identity between two
proteins as established by the fusion of two bands in gel-diffusion
analysis? It seems reasonable to expect that identical antigenic

164 MACROMOLECULAR COMPLEXES

patches are involved in the reaction between the two immunolog-
ically identical proteins and the specific antibody species participat-
ing in the precipitate. But the antigenic patches represent only
very small areas on the antigen molecule relative to the entire mole-
cule (Haurowitz, 1956; Kabat, 1957). Therefore, the combination
of the antibody with its specific antigenic patch in itself could not
be expected to reveal much direct information about the configura-
tion of the remainder of the antigen molecule. Assuming that suit-
able precautions have been taken against coincident band forma-
tion, a very important fact is provided by the fusion of the bands.
Presumably, each band develops in the gel medium along the locus
characteristic for the equivalence-ratio for the specific antigen-anti-
body system in the gel medium. When two bands fuse, therefore,
the equivalence-ratios for the reactants involved in the two bands
must have been identical. This tells us that, in all probability, the
same number of antigenic patches were present on each antigen
molecule; for, had this number been different, the equivalence-ratios
for each antigen-antibody complex would not have been the same.
Consequently, the locus of each band of precipitate would have
been different, and they would not have fused. Therefore, gel-
diffusion methods do more than inform us that identical antigenic
patches occur on the molecules from different sources; they imply
that the same number of antigenic patches are present on each
molecule and that the spatial configuration of these molecules is
sufficiently congruent to yield identical equivalence-ratios. In view
of the previous discussion, it would appear that gel-diffusion tech-
niques present us with the most critical criterion currently available
for establishing an identity or non-identity between two proteins.

Very recent evidence (Feinberg and Grayson, 1959) provides
some latitude for doubt about an identity between two proteins
based solely on the fusion of bands of precipitate. It revealed that
grass-pollen antigens cross-reacting with two different antisera could
yield band patterns in conventional Ouchterlony gel-diffusion tech-
niques suggesting an identity between two antigens, while a slight
modification of this technique indicated that only a similarity existed
rather than an identity. Any conclusions concerning an identity
between two proteins, based on the fusion of bands, must be re-
stricted within the limitations, not completely understood, inherent
in the procedure itself. The term "identity" will be used throughout
the remainder of the paper with this in mind.

FIBRILLAR SYSTEMS IN THE MITOTIC APPARATUS

165

The investigation into the molecidar origin of the mitotic appara-
tus was begun bv preparing two basically different antisera. One
antiserum was prepared against unfertilized sea-urchin egg antigens
soluble in O.IM KCl, and the other used dissolved, digitonin-isolated
mitotic apparatus as the homologous antigen. These two antisera

A B

Fig. 1. Comparing the immunochemical behavior of antiserum to unferti-
lized-egg antigens (ANTI EGG) with that of antiserum to dissolved, digitonin-
isolated mitotic apparatus (ANTI DMA) when both were made to react with
dissolved, digitonin-isolated mitotic apparatus (DMA) and an extract of unfer-
tilized eggs (EGG EXT.). The precursor-1 band is identified by "P" in the line
drawing corresponding to each photograph. The fusion of the only band, the
P-band, ascribable to the dissolved mitotic apparatus with one assignable to the
extract of unfertilized eggs indicates the presence of the precursor-! component
in both solutions.

were then made to react with soluble antigens from unfertihzed
eggs and from dissolved, isolated mitotic apparatus. Figure 1 shows
photographs of two Ouchterlony plates, comparing the immuno-
chemical behavior of each antiserum toward its homologous and
heterologous antigens. Two important facts are established by these

166 MACROMOLECULAR COMPLEXES

results. One is that the immunochemical behavior of the antiserum
to dissolved mitotic apparatus is indistinguishable from that of the
antiserum to unfertilized-egg antigens when they are both made to
react with solutions of mitotic apparatus. The other is the fusion
of the only band assignable to the dissolved mitotic apparatus with
one of several assignable to a solution of antigens from unfertilized
eggs. The antigen responsible for this band has been termed the
precursor-1 component. These results are in complete agreement
with the premise that the structural constituents of the mitotic ap-
paratus are on hand in the unfertilized egg prior to the onset of the
mitotic cycle. On a qualitative basis, at least, there appears to be
no need for the egg to synthesize new protein species necessary for
the mechanical events of mitosis, once it has left the ovary. The
machinery essential for the equipartition of the hereditary material
resides unassembled in the unfertilized egg awaiting the signal,
brought by the penetration of a sperm, to become aggregated into
the definitive mitotic apparatus.

Another very interesting inference can be drawn from the fusion
of the precursor-1 com^ionent band with a band ascribable to an
antigen from unfertilized eggs. When the mitotic apparatus was
put into solution, it was, in all probability, disassembled by breaking
the same links that were formed to assemble the precursor-1 com-
ponents into the mitotic apparatus. This is the simplest interpreta-
tion for the consistent correlation of the appearance of the precur-
sor-1 component with the loss of structural integrity by the mitotic
apparatus when the latter is dissolved. It would thus seem that the
dissolving of the mitotic apparatus, in terms of the end products
observed, may be viewed as the reversal of the assembly process.
The very mild conditions under which isolated mitotic apparatus
can be dissolved (Mazia, 1959) to vield the precursor-1 component
in solution preclude the peptide bond from consideration as the in-
termolecular link responsible for the structural integrity of the
mitotic apparatus.

What type of intermolecular cross-links can account for this be-
havior? Mazia (1958, 1959) presents evidence, based primarily
upon solubility studies of mitotic apparatus isolated by different
methods, that intermolecular bridges involving sulfur— thought to be
disulfide— are responsible for the polymerization of the molecular
subimits into the division figure, although hydrogen-bonding can-
not be entirely excluded. The involvement of sulfur-containing

FIBRILLAR SYSTEMS IN THE MITOTIC APPARATUS 167

proteins in the structural organization of the mitotic apparatus is
more strongly indicated by the in vivo studies of Mazia (1958) and
Mazia and Zimmerman (1958). They observed that proper concen-
trations of mercaptoethanol (HSCH.CH^OH), a simple, penetrat-
ing SH-compound of low toxicity, applied to sea-urchin and sand-
dollar eggs any time before metaphase, blocked the first cleavage
division. If it was applied during and subsequent to metaphase, no
interference with the first cleavage division was detected. The in
vivo experiments using mercaptoethanol argue strongly for the in-
volvement of protein-SH groups during the buildup of the mitotic
apparatus prior to cleavage, but do not give much insight into the
actual mechanism involved. The in vitro observations on the solu-
bilitv characteristics of mitotic apparatus isolated from ethanol-
preserved material by the digitonin method (Mazia, 1958) favor
the interpretation that some intermolecular disulfide bonds may be
responsible for maintaining the spatial orientation of the "precursor"
molecules in the mitotic apparatus.

The speculation that the reversible conversion of sulfhydryl
groups to intermolecular disulfide linkages performs an essential role
in the construction and function of the division figure finds support
from other investigators. Some cytochemical studies for sulfhydryl
groups in sea-urchin eggs during the first cleavage division were
made by Kawamura and Dan ( 1958 ) . Using Bennett's reagent, they
observed that the cytoplasm and nucleus of the unfertilized egg did
not stain, but 3 minutes after fertilization, they detected an in-
creased general stainability. At metaphase, the entire mitotic figure
of one species of sea urchin was intensely and uniformly stained,
contrasting sharply with the less strongly stained cytoplasm. In
early anaphase, the traction fibers were very intense, while the
interzonal region and centrospheres took up much less of the stain.
Little change in the astral rays was evident. Immediately after
cleavage, the staining character of the cytoplasm of the blastomeres
was essentially the same as that observed for the unfertilized egg;
that is, it was unstained. Shimamura et al. ( 1957 ) have stained the
mitotic apparatus for both sulfhydryl and disulfide groups, in paral-
lel, simultaneous experiments. Their photographs clearly reveal
that the regions stained for sulfhydryl groups usually complement
the areas stained for disulfide groups at given stages. This was most
evident in the anaphase and early telophase figures, in which the
asters stained only lightly for sulfhydryl groups but very intensely

168 MACROMOLECULAR COMPLEXES

for disulfide groups. At anaphase, the traction fibers stained
strongly for both groups, while the interzonal fibers only appeared
to contain the sulfhydryl groups in detectable amounts.

Studies designed to reveal quantitative fluctuations in the sulf-
hydryl and disulfide groups associated with protein and non-protein
compounds have been performed on populations of cells at various
stages of mitotic and meiotic divisions. The findings most often
quoted are those of Rapkine (1931) on the TCA-soluble fraction of
sea-urchin eggs. He observed a large steady drop in soluble sulf-
hydryl concentration after fertilization, followed by a rapid increase
to a maximum just prior to cleavage. He felt that his results re-
flected variations in the glutathione content of the cells, and pre-
sented an ingenious scheme in which glutathione was oxidized, with
the concomitant reduction of intramolecular protein disulfide bonds.
The latter were then reoxidized into intermolecular bonds, and
reduced glutathione was recovered. This was presumably related
to the buildup of the mitotic apparatus. Recently, Neufeld and
Mazia (1957) reinvestigated the problem of the non-protein sulf-
hydryl cycle in Strongylocentrotus purpuratus, but were unable to
corroborate Rapkine's data. Since then some very careful work by
Sakai and Dan (1959), again on sea urchins, uncovered a possible
explanation for this discrepancy. By carefully duplicating the ex-
traction procedure of Rapkine, they substantiated his original ob-
servations, but found that the TCA-soluble material responsible for
the fluctuations in sulfhydryl was not glutathione, as Rapkine had
supposed, but a protein that was soluble in TCA under the condi-
tions used by him. Plant material has also been analyzed by Stern
( 1956, 1958 ) for protein and non-protein ( soluble ) sulfhydryl and
disulfide. He observed fluctuations in protein and soluble sulfhydryl
that correlated with meiosis and microspore mitosis in both Lilium
and Trillium.

Some preliminary solubility studies on the mitotic apparatus iso-
lated directly from living material have been conducted recently
in Mazia's laboratory. The data make it more difficult to accept
disulfide bridges per se as the primary intermolecular links involved
in maintaining the structural integrity of the mitotic apparatus.
With suitable precautions, these isolated division figures can be dis-
solved under extremely mild conditions. They are completely
dispersed in water in the pH range between 8.5 and 9, while KCl
solutions, at neutrality, of 0.5M concentration or greater will dis-

FIBRILLAR SYSTEMS IN THE MITOTIC APPARATUS 169

solve only the fibrous components of the mitotic figure, leaving
intact the centrosomes and chromosomes (Mazia, 1958). These
are not the conditions usually considered appropriate for reducing
disulfide bonds. Yet the success of isolating the mitotic figure from
living material depends upon treating it as though it were held
together by disulfide bridges (Mazia, 1959). The apparent incon-
sistency between the ease with which the isolated mitotic apparatus
may be dissolved and the need, during the isolation procedure itself,
to use an agent ( dithiodiglycol ) that presumably stabilizes disulfide
links, remains to be resolved. There is, thus, no clear-cut experi-
mental evidence that directly corroborates the involvement of inter-
molecular disulfide bridges in the structural organization of the
mitotic apparatus, nor do we have convincing proof that sulfhydryl-
disulfide interchanges play an important role in the functional
activity of the mitotic apparatus. However, we have enough data to
say that protein molecules containing reactive sulfur groups par-
ticipate in some manner in the structure of the mitotic figure, and
we can only speculate about the manner in which the sulfur groups
react with each other to confer strict spatial orientation to the mole-
cules bearing them. The possible mechanisms of sulfhydryl-disulfide
interchanges and how these may be invoked to bring about the
polymerization of the precursor molecules into the mitotic figures
are adequately discussed by Jensen ( 1959 ) and Mazia ( 1959 ) .

Immunochemical results allow us to consider the formation of
the mitotic apparatus in terms of the aggregation of precursor sub-
unit molecules that are present in the unfertilized egg, and the
endeavors of other investigators provide indirect evidence that pro-
teins bearing reactive sulfur groups perform important roles in the
orientation of the precursor molecules in the mitotic apparatus.
Gross ( 1954 ) observed that extracts of Arhacia eggs contained cal-
cium-insoluble protein and speculated on its importance to sol-gel
transformations. A similar protein was subsequently detected in the
soluble fraction of a O.IM KCl extract of unfertilized S. purpuratus
eggs. This protein appeared as a clear fibrous precipitate when
CaCL to a final concentration of 0.05M was added to the unferti-
lized egg extract. It could be redissolved by dialysis against distilled
water. Some of the physical properties of this protein, which has
beerf termed the calcium-insoluble fraction, were studied by Kane
and Hersh ( 1959). Their results indicated a strong resemblance be-
tween this protein and the protein present in solutions of dissolved

170 MACROMOLECULAR COMPLEXES

mitotic apparatus. They shared similar molecular weights; each
contained a ribonucleotide component; they were both relatively
asymmetrical molecules; and the calcium-insoluble fraction repre-
sented about 10 per cent of the total protein of the unfertilized egg.
Mazia and Roslansky (1956) have estimated that the cell invests
about 12 per cent of its total protein in the mitotic figure. This im-
mediately gave rise to the speculation that the calcium-insoluble
fraction was the long-sought-after precursor component, and now
we had at our disposal the means to test this more directly The
calcium-insoluble fraction could be purified to the extent that, when
analyzed by gel-diffusion methods, it gave rise to only one distinct
band in reaction with antiserum to unfertilized egg antigens. Thus
it was a simple matter to see if the precursor- 1 component band was
the same as the only band assignable to the calcium-insoluble frac-
tion.

Figure 2 summarizes the essential results. When dissolved mi-
totic apparatus and calcium-insoluble fraction are each compared to
a solution of unfertilized egg antigens, one observes that the band
associated with the dissolved mitotic apparatus merges with the one
assignable to the unfertilized egg antigens, but it is readily seen that
this is not the same band that merges with the only one attributable
to the calcium-insoluble fraction. This clearly demonstrates that the
calcium-insoluble fraction bears no direct structural relationship
to the mitotic apparatus and demolishes the attractive possibility
that it represents the precursor protein. Although there can be little
doubt about the existence of the precursor- 1 component, it could not
be identified with the calcium-insoluble fraction, which was the only
protein fraction obtained in a homogeneous form from unfertilized

eggs-

The precursor- 1 component was invariably present in all solutions
of mitotic apparatus isolated from preserved material by the digi-
tonin method. These solutions were obtained by bringing an
aqueous suspension of mitotic apparatus to pH 10-10.5. At these
pH values, no mitotic figures were recognizable under phase con-
trast, while the contaminating cytoplasmic particles remained rela-
tively unaff^ected by the treatment. On the other hand, pH values
in the range 8.5-9.0 were adequate to dissolve completely the
mitotic figures isolated directly from living material, while leaving
many of the undesirable cytoplasmic particles intact. These could
be removed by centrifugation. Such solutions also gave rise to the

FIBRILLAR SYSTEMS IN THE MITOTIC APPARATUS

171

preciirsor-1 component ])and when made to react with antiserum
to unfertihzed egg antigens. In addition, there verv frequently ap-
peared a second band in the Ouchterlony gel-diffusion plates, sug-
gesting that there was a second component participating in the
structural aspects of the mitotic apparatus. As a routine procedure,
antiserum was prepared against solutions of mitotic apparatus iso-
lated directly from living material, and a very unexpected result

Fig. 2. Summary of the evidence indicating the absence of any direct rela-
tionship between the precursor-1 component and the calcium-insoluble fraction
(CIF). The latter antigen is responsible for the C-band. (A) The C-bands were
seen to fuse in the original Ouchterlony plate, but this relationship was lost in
subsequent photographic procedures. (B) It can be seen that the C-band does
not fuse with the only band assignable to dissolved mitotic apparatus, revealing
the lack of any simple relationship between the calcium-insoluble fraction and
the precursor-1 component. The abbreviations ANTI EGG, P, EGG EXT., and
DMA are the same as in Fig. 1 .

was observed when this antiserum was reacted with various anti-
gens. It could not react with the precursor-1 component but was
completely specific for the second component, which will be termed
the precursor-2 component. For some obscure reason, the rabbit
had focused its antibody-forming capacities on only the precursor-2
component, although the immunizing solutions had, in addition,
contained the precursor-1 component. What had happened to the

172 MACROMOLECULAR COMPLEXES

precursor-2 component in solutions of digitonin-isolated mitotic
apparatus? Experiments established that the precursor-1 component
was relatively insensitive to alkaline environments up to pH 10.5-11,
the conditions under which digitonin-isolated mitotic apparatus
could be put into solution, while the antigenic properties of the
precursor-2 component were nearly completely destroyed in this pH
range. Until further information is available, it will be necessary to
consider the precursor-2 component essential to the structural con-
tinuity of the mitotic figure. In no way does the existence of this
component necessitate any modification of the precursor concept,
for it is also found in the unfertilized egg.

An important aspect of the precursor concept is the intracellular
localization of the precursor-1 component in the unfertilized egg,
which in turn may provide an insight into the sequence for its
mobilization into the mitotic apparatus. Absorption studies were
conducted, using the particulate fraction, which is arbitrarily de-
fined as everything that can be sedimented at 60,000 g for 30 min-
utes from a homogenate of ethanol-preserved eggs. Extraction of
the particulate fraction by various procedures, and its subsequent
use to absorb antiserum, revealed that tenaciously associated with
the particulate fraction was an antigen that could combine specif-
ically with the antibodies homologous for the precursor-1 compo-
nent. Evidently the precursor-1 component occurs in an insoluble
form in addition to the readily extractable form, which will be re-
ferred to as the "soluble" form.

Attention was then directed toward living material, for this would
lend itself more favorably to critical fractionation procedures neces-
sary to resolve the intracellular distribution of the "precursor" com-
ponents. Living, unfertilized eggs were gently dispersed in isotonic
dextrose medium in a manner that avoided disrupting the yolk
particles. The resultant suspension was then centrifugally fraction-
ated, roughly into three fractions : the volk particles, the microsome-
mitochondria fraction, and the supernatant to the microsome-mito-
chondria fraction. The yolk particles, after careful washing, were
osmotically lysed and the contents analvzed by gel-diffusion meth-
ods. Numerous antigens were present, including the precursor-1
and precursor-2 components. The supernatant was qualitatively
very similar to the yolk-particle lysate. The situation with the
microsome-mitochondria fraction, on the other hand, was verv dif-
ferent. This had been extracted with 0.5M KCl, and the soluble

FIBRILLAR SYSTEMS IN THE MITOTIC APPARATUS

173

fraction contained, as the only antigens detectable, large amounts of
both the precursor- 1 and precursor-2 components. Undoubtedly
other protein species were present but eluded detection since the
antiserum did not contain antibodies specific for them. After the
yolk particles had been lysed, the residual solid matter was washed
and also extracted in 0.5M KCl. The resultant supernatant, termed

FixT^IJ
Ilyp^B

i

1

ANTI
.EGG

M

■ SUPl

1^ I J

^urI

M

N

#

.

Fig. 3. Intracellular distribution of the precursor-! and precursor-2 com-
ponents. SUP. I, SUP. II, and SUP. Ill refer to the original supernatant, the
yolk particles, and the two subsequent wash solutions, respectively. YPL rep-
resents the yolk-particle lysate, and EXT. LYP refers to the 0.5M KCl extract of
the structural material remaining after osmotic rupture of the yolk particles.
EXT. M&M indicates the 0.5M KCl extract of the mitochondria-microsome frac-
tion. Arrow 1 points to the precursor-! component band, and arrow 2 indi-
cates the position of the precursor-2 component band. It can be seen that
these two components were the main antigens detected in the extracts of the
mitochondria-microsome fraction and the lysed yolk-particle fraction. It would
appear that the precursor-! component is present as a "soluble" antigen within
the yolk particles as well as in the surrounding cytoplasm.

the lysed yolk-particle extract, was immunochemically identical with
the microsome-mitochondria extract. These relationships can be
seen in Fig. 3. Both components of the mitotic apparatus have been
found in all fractions studied from the unfertilized egg. However,
two fractions, the lysed yolk-particle extract and the microsome-
mitochondria extract, yield solutions in which the precursor- 1 com-

174 MACROMOLECULAR COMPLEXES

ponent and the precursor-2 component were the only or major
antigens detectable.

At this point, some speculation about the ubiquitous intracellular
distribution of the precursor- 1 component may be profitable. The
much less studied precursor-2 component will be ignored, but not
forgotten, in this discussion. There is little doubt that the precur-
sor-! component can exist in three "states." It is present in a "solu-
ble" as well as in an insoluble form in the unfertilized egg, and is
an important constituent of the mitotic apparatus. One can imagine
a dynamic equilibrium to exist among the three "states" of the
precursor- 1 component. The insoluble fraction would represent the
major pool of precursor- 1 component, being converted into the solu-
ble form as an intermediate step before its incorporation into
the mitotic apparatus. An interesting consequence could be that the
level of the "soluble" pool would remain unchanged during the
formation of the mitotic apparatus and might or might not be
greatly reduced when the mitotic apparatus is fullv formed. There-
fore, analysis of the "soluble" fraction for the precursor-1 component
during development from the unfertilized egg to the two-celled
stage would not reveal the fluctuations necessary to account for the
amount incorporated into the mitotic apparatus.

Kane and Hersh (1959) describe some experiments that mav
have a bearing on the proposed interrelationships among the three
"states" of the precursor-1 component. Ultracentrifugal studies on
the soluble fraction of different stages from the unfertilized egg to
the two-cell stage, in Arhacia, revealed a component whose concen-
tration decreased with time and was undetectable at metaphase. It
reappeared after cleavage. At the time, it was speculated (Mazia,
1957 ) that this behavior reflected the incorporation of soluble struc-
tural units into the developing mitotic apparatus. These analyses
had been conducted on extracts obtained from ethanol-preserved
material. When the experiment was repeated on S. purpuratus, no
change in any component was detected that could be correlated
with the buildup of the mitotic apparatus. Again, ethanol-preserved
material had been used. For this reason, their experiments cannot
be compared directly with the immunochemical data presented on
the distribution of the precursor-1 component among the xarious
fractions obtained by fractionation of living eggs in isotonic non-
electrolyte medium. However, their results are sufficiently con-
sistent within themselves, so that the observed differences between

FIBRILLAR SYSTEMS IN THE MITOTIC APPARATUS 175

the two species of sea urchin may be considered real. Quantitative
measurements on various fractions from living material, during the
time course from fertilization to cleavage, combined with parallel
qualitative immunochemical determinations, could be expected to
improve our understanding of the proposed interrelationships among
the states of the precursor-1 component.

By way of summary, we can say that the unfertilized egg is
endowed with the molecular subunits that have been identified as
important structural elements in the mitotic apparatus. The forma-
tion of the mitotic apparatus may then be considered the result of
the spatial rearrangement of the precursor-1 and precursor-2 com-
ponents into the highly oriented aggregate, the mitotic apparatus.
It has been further speculated that the main reservoir of these com-
ponents in the unfertilized egg resides in existing subcellular moie-
ties—the microsomes, mitochondria, and the structural elements of
the yolk particles— and that the soluble fraction represents an inter-
mediate form in the mobilization of the insoluble fraction into the
mitotic apparatus.

Before we can fully accept the precursor concept for the origin
of the mitotic apparatus in the sea urchin, it must be demonstrated
that tlie precursor molecules that appear in the mitotic figure are
indeed the same ones that occur in the unfertilized egg.

References

Anderson, N. G. 1956. Cell division II: A theoretical approach to chromosomal

movements and the division of the cell. Quart. Rev. Biol. 31: 243-269.
BovERi, T. 1887. Zellen-studien. G. Fischer, Jena.
Feinberg, J. G., and H. Grayson. 1959. A critical test of antigenic relationships.

Nature 183: 987.
Flemming, W. 1882. Zellsuhstanz, Kern und Zelltheilung. F. C. W. Vogel, Leipzig.
Gross, P. R. 1954. Alterations in the proteins of sea urchin homogenates treated

with calcium. Biol. Bidl. 107: 364-385.
Haurowitz, F. 1956. The nature of the protein molecule: Problems of protein

structure. /. Cellular Comp. Physiol. Suppl. 1. 47: 1-16.
Hertwig, O. 1909. The Cell; Outlines of General Anatomy and Physiology. Sonnen-

schein & Co., London.
Jensen, E. V. 1959. Sulfhydryl-disulfide interchange. Science 130. 1319-1323.
Kabat, E. a. 1957. Size and heterogeneity of the combining sites on an antibody

molecule. /. Cellular Comp. Physiol. Suppl. 1. 50: 79-102.
Kane, R. E., and R. T. Hersh. 1959. The isolation and preliminary characterization

of a major soluble protein of the sea urchin egg. Exptl. Cell Research 16: 59-69.
Kavuamura, N., and K. Dan. 1958. A cytochemical study of the sulfhydryl groups

of sea urchin eggs during the first cleavage. /. Biophys. Biochem. Cytol. 4: 615-

620.
Landsteiner, K. 1945. The Specificity of Serological Reactions. 2d rev. ed. Harvard

University Press, Cambridge, Mass.

176 MACROMOLECULAR COMPLEXES

Mazia, D. 1957. In W. D. McElroy and B. H. Glass (eds.). Symposium on Chem.

Basis Heredity. Johns Hopkins Univ., McCollum-Pratt Inst. Contrib. No. 153.
Mazia, D. 1958. Cell division. Harvey Lectures Set. 52 (1957-58): 130-170.
Mazia, D. 1959. In R. Benesch et al. (eds.). Sulfur in Proteins. Academic Press,

Inc., New York.
Mazia, D., and K. Dan. 1952. The isolation and biochemical characterization of the

mitotic apparatus of dividing cells. Proc. Nat. Acad. Sci. U. S. 38: 826-838.
Mazia, D., and J. Roslansky. 1956. The quantitative relations between total cell

proteins and the proteins of the mitotic apparatus. Protoplasma 46: 528-534.
Mazia, D., and A. M. Zimmerman. 1958. SH compounds in mitosis. II. The effect

of mercaptoethanol on the structure of the mitotic apparatus in sea urchin eggs.

Exptl. Cell Research 15: 138-153.
Neufeld, E. F., and D. Mazia. 1957. Nonprotein sulfhydryl compounds in the divi-
sion of eggs of Strongylocentrotus purpuratus. Exptl. Cell Research 13: 622-624.
OucHTERLONY, O. 1949. Autigen-autibody reactions in gels. Acta Pathol. Microbiol.

Scand. 26: 507-515.
OucHTERLONY, O. 1958. Diffusion in gel methods for immunological analysis. Prog-
ress in Allergy 5: 1-78.
OuDiN, J. 1946. Methode d'analyse immuno-chimique par precipitation specifique

en miheu gelifie. Compt. rend. 222: 115-116.
OuDiN, J. 1952. Specific precipitation in gels and its application to immunochemical

analysis. Methods in Med. Research 5: 335-378.
Rapkine, L. 1931. Sur les processus chimiques au cours de la division cellulaire.

Ann. physiol. physicochim. Biol. 7: 382-418.
Sakai, H., and K. Dan. 1959. Studies on sulfhydryl groups during cell division of

sea urchin eggs. Exptl. Cell Research 16: 24-41.
SHIMA^cuRA, T., T. Ota, and T. Hishida. 1957. Cytochemical studies on the mitotic

spindle. Symposia Soc. Cellular Chein. 6: 21-32.
Stern, H. 1956. Sulfhydryl groups and cell division. Science 124: 1292-1293.
Stern, H. 1958. Variations in sulfhydryl concentrations during microsporocyte

meiosis in the anthers of Lilium and Trillium. J. Biophys. Biochem. Cytol. 4: 157-

161.
Strasburger, E. 1880. Zellbildung und Zelltheilung. G. Fischer, Jena.
Wilson, E. B. 1925. The Cell in Development and Heredity. 3d ed. The Mac-

millan Co., New York.

DISCUSSION
M. V. Edds, Jr., H. A. Went, A. C. Giese

Dr. Edds (Brown University): Would you care to speculate on the mech-
anism responsible for concentrating and ordering the precursor molecules in
the spindle?

Dr. Went: There is evidence, as I have indicated, that the precursor
molecules are rather widely distributed throughout the cytoplasm of the un-
fertilized egg, but we know nothing regarding the forces that align the pre-
cursor molecules into an orderly array. It is fairly clear that the centriole is, in
some unknown manner, responsible for this orientation. Experiments with
model systems (Anderson, 1956) show that one can get precipitation of pro-
teins in solution radiating from a point which can be analogized to the cen-
triole. Aggregation and ahgnment of macromolecules in model systems have
been thought of in terms of surface charges and changes in their distribution
or strength, conferring preferential aggregation patterns on the macromole-
cules. However, any resemblance between precipitation patterns observed in

FIBRILLAR SYSTEMS IN THE MITOTIC APPARATUS 177

in vitro and in vivo systems may be only coincidence, with no fundamental
mechanism in common. I don't feel justified in speculating further on what
is responsible for bringing the precursor molecules sufficiently close to one
another for short-range forces to bind them together into the definitive mitotic
apparatus.

Dr. Giese (Stanford University): Has any study been made of the con-
stituents of the centriole and their effect on spindle formation?

Dr. Went: No, not that I am aware of. However, this problem is now
under investigation in Professor Mazia's laboratory. It is in its infancy, for a
method still has to be perfected whereby centrioles, or their constituents, can
be freed from undesired cytoplasmic matter. Then it is a question of whether
or not enough material can be obtained, for the centriole is a very small entity.

The Submicroscopic Basis of Bacterial
Photosynthesis: The Chromatophore'

J. A. Bergeron " and R. C. Fuller -

Introduction

The preceding symposium papers have considered submicro-
scopic aspects of objects with which biologists are aheady famihar
at the microscopic or gross anatomical level, i.e., the collagenous
fiber, the chloroplast, the mitotic spindle, and the myelin sheath. By
contrast, the subject of this paper is less familiar because the bac-
terial chromatophore is itself submicroscopic and, as a consequence,
was not identified until recently. Since the subject is new, it seems
apropos to indicate the value ot the bacterial chromatophore for
studying photosvnthesis.

Sixteen years ago, Ruben (1943) suggested that the energy of
the pyrophosphate bond of adenosine triphosphate (ATP) could
supply the energy needed for the photosynthetic assimilation of
carbon dioxide. The following year, Emerson et al. ( 1944 ) proposed
that the function of light energy in photosynthesis is the formation
of energy-rich phosphate bonds. The concept of ATP as the fuel
of photosynthesis gradually gained acceptance as the path of car-
bon during photosynthesis became clearer and as knowledge of
carbohydrate phosphate interconversion improved. The search for
a direct relation between light absorption and ATP formation cul-
minated with the demonstration of photosynthetic phosphorylation
by chloroplast preparations (Anion et at, 1954a, 1954b) and by sub-
cellular preparations of the non-sulfur purple bacterium, Rhodo-
spirillum rubrum (Frenkel, 1954). Currently, the mechanism by

' Research carried out at Brookhaven National Laboratory under the auspices of
the U. S. Atomic Energy Commission. We are indebted to Mr. Mark Gettner, Mr.
Walter Geisbusch, and Misses Gayle Schaeff, Catherine Kostuk, and Geraldine David
for technical assistance.

- Biology Department, Brookhaven National Laboratory, Upton, New York.

179

180

MACROMOLECULAR COMPLEXES

which Hght energy is used to create the energy-rich bond of the
terminal phosphate of ATP is receiving considerable attention.

In higher plants, all of the many reactions of photosynthesis are
integrated in a complex organelle, the chloroplast. The modern
evidence in support of this old concept has been summarized by
Arnon ( 1958 ) . In order to study the energy-supplying step directly,
it is desirable to ehminate all of the non-pertinent variables and use

CHLOROPLAST

MITOCHONDRION

0.5M

CHROMATOPHORE
• 0.03 m

Fig. 1. Relative size of organelles. There are great differences in the level
of organization of the organelles which supply chemical energy. The sub-
microscopic chromatophore converts light energy into the readily available
chemical energy of the pyrophosphate bond of ATP. The mitochondrion, which
lies just within the microscopic range, performs successive oxidations to generate
ATP from substrates which act as energy reservoirs. The microscopic chloroplast
converts light energy into pyrophosphate bond energy and uses the ATP to
forge carbon dioxide and water into stable energy reserves.

the simplest possible system. The principal subject of this paper, the
chromatophore of the purple sulfur bacterium 'Chromatium,' meets
these conditions remarkably well since it is smaller than the chloro-
plast by two orders of magnitude ( Fig. 1 ) and appears to have only
one photosynthetic role, namely, the light-dependent formation of
adenosine triphosphate (ATP) from adenosine diphosphate (ADP)
and inorganic phosphate.

BACTERIAL PHOTOSYNTHESIS 181

The concept of the chromatophore as a photochemical organelle
is beginning to assume definite form. Until recently it was believed
that the pigments of photosynthetic bacteria are freely dispersed
throughout the cell in the form of protein complexes. The chromato-
phore, in name at least, came into being in 1952, when Pardee,
Schachman, and Stanier found that the photosynthetic pigments in
extracts of R. rubrum sedimented much more rapidly than antici-
pated. The electron micrographs of shadowed specimens of these
chromatophores showed thin disks about 1100 A in diameter. It was
assumed that the disks were derived from spheres with an average
diameter of about 600 A. This value agreed roughly ( Schachman
et al., 1952) with the diameter (400 A) calculated by Stokes' rela-
tion from the sedimentation constant (200 S) of a purified prepara-
tion with a partial specific volume of 0.73 (assumed). The esti-
mated molecular weight of these R. rubrum chromatophores was
thirty million. At the same time, Thomas (1952) also published
electron micrographs of shadowed specimens prepared from crude
extracts of several photosynthetic bacteria. These preparations also
contained thin disks. Reasoning by analogy with similar prepara-
tions from chloroplasts, Thomas described these objects as "lamellae"
and as "grana," or stacks of lamellae. Despite the obvious differences
in interpretation, these reports had in common the suggestion that
the pigments of photosynthetic bacteria are localized in specific
structures. Subsequently, when Frenkel (1954) described photo-
phosphorylation by subcellular preparations of R. rubrum, it became
logical to assume that the pigments of photosynthetic bacteria
are integrated structurally and functionally at the submicroscopic
level.

With this point of view, we began a systematic study of the struc-
ture and function of the photochemical apparatus of the purple
sulfur bacterium, Chromatium strain D. Several properties of this
organism made it especially attractive. Many photosynthetic organ-
isms have the ability to form ATP bv non-photosynthetic pathways
and can grow aerobically in the dark. Chro7natium is an obligate
phototroph and an obligate anaerobe (van Niel, 1931, 1935, 1936);
thus, in this organism ATP formation seems to be completely light-
dependent (see general discussion by van Niel, 1956). The syn-
thetic'abilities are well developed, and Chromatium, given light,
can live and grow anaerobically in simple salt solutions containing
carbonate and hydrogen sulfide.

182

MACROMOLECULAR COMPLEXES

The systematic study has progressively elevated the status of the
chromatophore from that of a submicroscopic, pigmented particle
which appears when the cells are broken, to the status of a photo-
chemical organelle with a well-defined structure and a specific func-
tion. Using this point of view we shall consider the chromatophore
at three conceptual levels: (1) as a pigmented particle, (2) as the
structural and functional unit of photophosphorylation, and (3) as
the photochemical organelle. Then we will present our interpreta-
tion of the electron-optical image of the chromatophore and a hy-
pothesis of the molecular architecture.

The Chromatophore as a Submicroscopic
Pigmented Particle

Isolation. The structures which we identify (Bergeron et al.,
1957) as the chromatophores of Chromatiiim can be obtained in
quantity by the procedure which is outlined in Fig. 2. The buffered
0.5M sucrose medium is a recent refinement which was designed to
maintain the physiological relationship between the carotenoid pig-
ments and the bacteriochlorophyll (Bergeron, 1958; Bergeron and
Fuller, 1959; Anderson et al, 1958). After the chromatophores have
sedimented, the preparation is washed by resuspension and recen-

BACTERIA ARE SUSPENDED IN 40 ML OF 0. 5M SUCROSE

AT PH 7.8, O.IM TRIS BUFFER, AND 4 CM OF ALUMINA

SONICATE FOR 90 SEC AT 0-5° C IN 10 KC RAYTHEON

OSCILLATOR

35,000 XG 20 MIN

WHOLE CELLS

AND DEBRIS

35,000 XG

20 MIN

1
DEBRIS

144,000 XG

90 MIN

CHROMA

TOPHORES COLORLESS

SUPERNATANT

Fig. 2. The procedure used for isolating the Chromatium chromatophores.
The final preparation is washed two or three times by recentrifugation in the
buffered medium.

BACTERIAL PHOTOSYNTHESIS

183

trifugation for 90 minutes at 144,000 g in the same medium. For
some purposes the chromatophores can be washed by repeated
centrifugation in dikite buffer or distilled water. The isolated
chromatophores can be lyophilized and stored over phosphorus
pentoxide at reduced temperature (—20° C) for long periods with-
out significant loss in enzymic activity or alteration in the over-all
phvsicochemical properties.

Physicochemical Properties. About 90 per cent of the cellular
pigments are recovered in the final fraction. The absorption spec-
trum between 350 m^u and 1000 m/>i of the chromatophores prepared
in the sucrose medium corresponds with the absorption spectrum of
the original cells (Fig. 3). It might be expected, a priori, that the
chromatophores are rather fragile, but the reverse is true. Prolonged
oscillation at 10 kilocycles in dilute buffer does not produce appre-
ciable fragmentation. The chromatophore is hydrophilic and has
an isoelectric point between pH 3 and 4; this is in accordance with

0.8-

0.2

WHOLE CELLS AND CHROMATOPHORES
IN SUCROSE

400 500 600 700 800

X (MILLIMICRONS)

900

Fig. 3. A spectrophotometric comparison of chromatophores, isolated in the
buffered 0.5M sucrose medium, with the intact organisms, suspended in Hendley's
medium. The spectra were measured (Cary Model 14) through opal glass to
reduce the divergence caused by light-scattering at the lower wave lengths.
The close correspondence indicates that the photosynthetic pigments are not
disturbed by this isolation procedure.

184 MACROMOLECULAR COMPLEXES

the observations on the minimum sokibihty and charge reversal of
crude extracts (Katz and Wassink, 1939; French, 1940). The pro-
nounced color of the chromatophores facilitates their identification
in the analytical ultracentrifuge. The color is related to only one
peak; thus the chromatophore preparation is monodisperse. The
sedimentation constant, measured at extreme dilution in de-ionized
water at neutrality, is 143 S corrected to 20° C. Since the chromato-
phore is not disrupted by lyophilization, it is possible to measure the
density of aqueous suspensions in terms of the gram fraction. The

_

I 1

1 1

_

_

AQUEOUS SUSPENSIONS
"CHROMATOPHORES"

OF

_

•"OLD" CULTURE

Y --

.9975 + .199 X

~

o "YOUNG" CULTURE

v =

.8009 cc/g

~

.0030

-

D =

.2486

-

.0010

-

^ 0-^^

-

.9990

-

..^...'^t-^^^^^^^^'^

""o^

-

.9970

^-0

1 1

1 1

-

10

GRAM FRACTION x 10

20

Fig. 4. The relation between the gram fraction and the density of aqueous
suspensions of lyophilized chromatophores. The equation of the line, Y = 0.9975
+0.199x, can be used to calculate the partial specific volume (v) and density
(D) of the chromatophore.

resulting plot (Fig. 4) is Hnear and can be used to calculate the
partial specific volume (0.8 cc/g) at extreme dilution in pure water.
Since the chromatophores are spherical (see below), Stokes' relation
is applicable and an average particle diameter of 320 A is obtained.
The molecular weight is about thirteen million.

Thus the chromatophore preparation contains the photosynthetic
pigments in a physiological relationship and is composed of a rather
uniform population of submicroscopic particles.

Composition. The chemical composition of the chromatophore
is known reasonably well; about 89 per cent of the dry weight is

BACTERIAL PHOTOSYNTHESIS 185

present in components which are not free to exchange. These "fixed"
components— protein, phosphohpid, bacteriochlorophN'll, and the
carotenoids— are integral parts of the structure. Our data (Table 1)
for the chromatophore (Bergeron, 1958), if converted to a milligram
protein basis, agree very well with the data of Newton and Newton
( 1957 ) for the fraction which they called the "small particles"
(chromatophores, see below). Since these authors reported that the

TABLE 1
Chromatophore Composition

Per Molecular Mole- Mole

Substance Cent G/"Mole" Ch Weight cules Ch Ratio

Carotenoid 1.5 2.0 X 10= 700 300 1

Bacteriochlorophyll . . 4.2 5.5 X 10= 900 600 2

Phospholipid 22.3 30 X 10= 900' 3000 10

Protein 61.0 80 X 10= 120f/A.A. 67,000 A.A. 220 A.A.

DNA 0 — — — —

PNA 0 — — — —

Other 11.0 12.5 x 10= — — —

The per cent composition on a dry-weight basis has been converted to grams
per "mole" Ch by employing the "molecular weight" of thirteen miUion. By assigning
an average molecular weight to each substance, one can obtain an estimate of the
number of molecules of each substance in a chromatophore. The mole ratio repre-
sents a "minimal unit" of composition. This imit represents a molecular weight of
about 40,000.

* Cephalin with Cm fatty acids.

} Round value for average amino acid residue.

phospholipid contains ethanolamine, phosphorus, and glycerol in a
1:1:1 ratio, we have treated the phospholipid as a cephalin. The
"molecular weight" of the chromatophore has been employed to
calculate the amount of each "fixed" substance in a "mole" of chro-
matophores. By assigning an average molecular weight to each
substance, one can estimate the numlDer of molecules of each type
in a chromatophore. By this method of calculation, the chromato-
phore contains 300 carotenoid molecules, 600 chlorophyll molecules,
3000 cephalin molecules, and protein equivalent to 67,000 amino
acids. The hypothetical "minimal unit of composition" (empirical
formula) is 1 carotenoid molecule, 2 bacteriochlorophyll molecules,
10 phospholipid molecules, and protein equivalent to 220 amino
acids. The "formula weight" is about 40,000.

186 MACROMOLECULAR COMPLEXES

It should be noted that the carotenoids occur as a variable mix-
ture of several types, particularly lycopene, spirilloxanthin, and ly-
coxanthin ( Goodwin and Land, 1956 ) . It is also possible to interfere
with carotenoid synthesis by treatment with diphenylamine (Good-
win and Osman, 1954) and obtain chromatophores which appear to
totally lack the colored carotenoids (Anderson and Fuller, 1958;
Bergeron and Fuller, 1959); however, poorly colored precursors
appear to be present instead (reviewed by Stanier, 1958). Since the
carotenoid :bacteriochlorophyll ratio observed above agrees with
the ratio reported by Newton and Newton (1957), and with the
maximum value observed by Stanier ( 1958 ) in Rhodopseudomonas
spheroides, it seems safe to assume that these chromatophores also
contained about the maximum amount of colored carotenoids.

The status of polysaccharide as a "fixed" component is debatable.
Newton and Newton ( 1957 ) describe polysaccharide in their "small
particle" fraction ( 100,000 g, 90 min ) . Their data show that in
passing from the "chromatophore fraction" ( 25,000 g, 60 min ) to the
"small particle," the phospholipid and chlorophyll content double on
a protein basis while the polysaccharide content decreases by a fac-
tor of three and one-half. These authors suggested that the chroma-
tophore consists of pigmented macromolecules in a carbohydrate
envelope. Newton (1958) has elaborated this point of view on the
basis of the immunochemical properties of intact cells and pigmented
fractions.

Our data, particularly the electron microscope observations on
thin sections ( see below ) , indicate that the chromatophore ( 144,000
g, 90 min ) is in the "small particle" fraction of Newton ( 100,000 g,
60 min ) . Irregular cell fragments are common in the 25,000 g frac-
tion, which was identified with the chromatophore by Newton.
For this reason, we interpret the data on carbohydrate release as
evidence that it is a contaminant and not an integral part of the
chromatophore.

The relation to the chromatophores of several biologicaljv im-
portant compounds, i.e., pyridine nucleotides, cvtochrome, co-
enzyme Q-, and vitamin K, is of special interest but remains obscure.
One complicating factor is the rather low isoelectric point ( pH 3-4) ;
as a consequence, the chromatophore has considerable net charge
at neutrality. This appears to be a factor in explaining a net uptake
by chromatophores of added cytochrome c (Newton and Newton,
1957).

BACTERIAL PHOTOSYNTHESIS

187

The Chromatophore as the Structural and Functional
Unit of Photophosphorylation

Experiments with crude extracts of Chromatiwn by Williams
(1956) suggested that photophosphorylation might have occurred
as in the R. ruhrum preparations used by Frenkel (1954). Experi-
ments with the purified chromatophore fraction ( Fuller and Ander-
son, 1957, 1958) have unequivocally demonstrated the light-de-
pendent formation of ATP from ADP and inorganic phosphate. The

Fig. 5. A plot of the rate of ATP formation against an index of chromato-
phore concentration. The proportionality which results indicates that photo-
phosphorylation does not depend upon interaction between chromatophores.

reaction also requires Mg++ and is stimulated by small amounts of
the uncolored supernatant fluid. Two distinct factors are involved:
one is heat-labile, non-dialyzable, and can be precipitated by 35-45
per cent ammonium sulfate; the other is heat-stable, dialyzable, and
can be replaced by catalytic amounts of succinate or reduced diphos-
phopyridine nucleotide (DPNH). Photophosphorylation by Chro-
rnatium fractions has also been studied by Newton and Kamen
(1957).

The rate of photophosphorylation by the purified chromatophores
is a linear function (Fig. 5) of the chromatophore concentration
(indirectly expressed by the bacteriochlorophyll content). A linear
dependence indicates that the chromatophores are not random frag-
ments of a photophosphorylating system; otherwise a falling rate

188 MACROMOLECULAR COMPLEXES

would be expected with progressive dilution. Since the pigment
content of the chromatophore is known, it can be calculated that, in
these experiments, each chromatophore was forming about 500 high-
energy phosphate bonds per minute ( cf . Bergeron, 1958 ) .

The enzymes involved in the assimilation of carbon dioxide are
not integrated in the chromatophore and are found in the uncolored
supernatant fluid. If the chromatophores are recombined with the
supernatant fluid, the reconstituted system fixes COo when the sys-
tem is illuminated (Table 2). When the light: dark ratios of CO^.
incorporation are compared, the reconstituted system is 30 per cent
as effective as the intact cells. It should be noted in passing that,
unlike higher plants, Chromatium does not synthesize large quan-
tities of reserve carbohydrates, and the rate of incorporation of COo
into amino acids such as aspartic is proportionately more rapid.

TABLE 2

COo-Fixation by Various Cell Fractions
of Chromatium. 1 Hr *

Dark
or Total CO. Fixed

Cell Fraction Light as C* in Counts/Min

Whole cells Light 4,500,000

Whole cells Dark 50,000

Chromatophores Light 750

Supernatant Light 1,750

Supernatant and chromatophores Dark 1,950

Supernatant and chromatophores Light 69,000

" The cell-fractionation procedure separates the energy-yielding and energy-requir-
ing steps of photosynthesis. The chromatophore fraction catalyzes photophosphoryla-
tion, but the enzymes which use ATP for the synthetic reactions are in the supernatant
fraction. When the two fractions are combined, their fixation of COj proceeds in the
light.

The Chromatophore as the Photochemical Organelle

The procedure used to isolate a given set of cellular properties
serves as an operational definition of the fraction; correlated data
from both direct and indirect methods of observation are needed to
characterize the fraction and relate it to the cells from which it is
obtained. In this investigation, the requirement for direct observa-
tion was met by the electron microscope study of intact cells, crude
pigmented fractions, and the purified chromatophore fraction (Ber-

BACTERIAL PHOTOSYNTHESIS 189

geron et al., 1957; Bergeron, 1958). The most profitable observa-
tions were made with thin sections of methacrylate-embedded pel-
lets of preparations which had been fixed for 1 hour at 25° C in 1
per cent osmium tetroxide solutions. The fixatives were buffered at
pH 7.4 with Veronal-HCl and matched as closely as possible in ion
balance and osmolarity with the culture or suspension medium. All
substances which reacted with the osmium tetroxide were excluded.

In such sections (Fig. 6), the bacteria have smooth contours and
are bounded by two distinct membranes, the cell wall and the cyto-
plasmic membrane. The obvious and predominant intracellular
structures are the chromatophores. This abundance, though puz-
zling, agrees with the indirect measurements; it was calculated, by
comparing the bacteriochlorophyll content of lyophilized cells and
of lyophilized chromatophores, that the chromatophores represent
about two-thirds of the total cellular solids. Sometimes the arrange-
ment of chromatophores is so regular that it suggests a crystalline
pattern; consequently, the details of the chromatophore are ob-
served more easily in thin sections of the chromatophore fraction
(Fig. 7).

The chromatophores occur as individuals, clumps, and occasion-
ally as chains. These configurations can be regarded as secondary
associations produced during fixation since the fixatives cause the
progressive flocculation and precipitation of dilute suspensions of
either the cells or the chromatophores. There is no evidence that
the chromatophore is a fragment of a more highly organized, larger
organelle.

The sections show that even the finest preparations are not per-
fectly clean. They contain a small number of larger vesicular struc-
tures which resemble the chromatophore; in addition, small particles
( '— 100 A ) of high intrinsic electron density are present in variable
numbers. It is interesting to note that these impurities were not
noticed in the analytical ultracentrifuge. This illustrates the sensi-
tivity achieved by using thin sections of sedimented fractions as an
assay of homogeneity.

In studying the crude cell fractions we were surprised to find
large cell fragments in which the contents appeared to be quite
undisturbed. We assume that the cytoplasm has "gelled" as a re-
sponse to stimuli received during the sonic oscillation at 10 kilo-
cycles in the cold (0-6° C), or because of the formation of nucleic
acid gels during rupture. One consequence of this phenomenon is

190

MACROMOLECULAR COMPLEXES

Fig. 6. An electron micrograph of a thin section of the purple sulfur bac-
terium, Chromatium. The organism is smooth-contoured and is bounded by a
dual membrane which can separate into two distinct structures, the cell wall (W)
and the plasma membrane. The cell is filled with the chromatophores (Ch),
which are minute vesicles. These appear as annular images with an outer
diameter of about 300 A and a cortical thickness of about 70 A. Large vesicles
(V), about 1000 A in diameter, are also visible. The irregular areas of low
density (N) are considered to be parts of an irregularly shaped nuclear com-
partment. The closely packed chromatophores conceal the small particles which
are observed in the fractions. X48,000.

that the cliromatophores hterallv have to be shaken out of the
broken cells. The different rates of sedimentation of the pigmented
cell fragments and of the individual chromatophores led us earlier
to the artificial and erroneous distinction between "laree" and

BACTERIAL PHOTOSYNTHESIS 191

1 i^^

■p. ■

'J
4

V

¥*

'^1'

^4

--0.5^-

Fig. 7. An electron micrograph of o thin section of a typical chromatophore
preparation. The chromatophores occur singly, in clumps, and occasionally as
chains. This aggregation seems to be produced by the fixative. Larger vesicles
(V) may be present. It is not known whether or not these structures are large
chromatophores. Small dense particles (Sp) are often present as contaminants.
X 78,000.

"small" chromatophores, just as Newton and Newton (1957) have
distinguished "chromatophores" and "small particles."

We have already reviewed the biochemical reasons for doubting
the idea (Newton and Newton, 1957) that the structure which we
identify as the chromatophore is a fragment of the "real" chromato-
phore. When we consider, in addition, the electron microscopic
observations on intracellular components and remember that two-

192 MACROMOLECULAR COMPLEXES

thirds of the celkilar sohds are chromatophores, it is clear that,
although chromatophores are present at the surface of the cyto-
plasm, the structure with which they are associated is the cell itself.
In our opinion, the data obtained in this systematic study are
consistent with the view that the chromatophore is the photochemi-
cal organelle of Chromatium (Bergeron, 1958). Since the electron
micrographs of thin sections of several other photosynthetic bacteria,
Rhodospirillum nibriim, Rhodopseudomonas spheroides, and Chlo-
robium limicola, also show vesicular inclusions in abundance ( Vatter
and Wolfe, 1958), it is possible that this view has some general
validity. Since the vesicles in the non-sulfur photosynthetic bacteria
are much larger than the Chromatium chromatophore, they may be
more complicated. For example, the Rhodospirillum ruhrum chro-
matophore, which has been studied extensively, though not as sys-
tematically, appears to be more complex (see review by Frenkel,
1959). It is also possible, considering the variety of these micro-
organisms, that lamellated pigmented structures exist. The electron
micrographs of Rhodomicrohium vannielii (Vatter et ah, 1959) sug-
gest this possibility. It remains to be established, however, that
these lamellae are pigmented. It is also conceivable that in some
photosynthetic bacteria the pigments will not be integrated into
vesicular structures which can perform the complete sequence of
events involved in photophosphorylation.^

Interpretation of the Electron-Optical Image

Both the physicochemical and the electron microscope data indi-
cate that the isolated chromatophore is a sphere. The electron mi-
croscope observations on the thin sections of the organism also reveal
discrete spherical structures; anastomoses or tubular profiles have
not been observed. The annular form of the image of the sectioned
chromatophore shows that the central region does not scatter 50-
kilovolt electrons as efi^ectively as does the cortical region. Since the
chromatophores are frequently seen in direct contact, we can infer
that, within the limits of resolution, the outer boundarv of the image

^ The suggestion that photosynthesis can proceed at an even simpler level of
structural organization than the chromatophore has been supported by study of the
green sulfur bacterium, Chlorohium thiosulfatophilum. In this instance, the photo-
synthetic pigments are recovered in a macromolecule with a molecular weight of
about 1 million (J. A. Bergeron and R. C. Fuller in "Biological Structure and Func-
tion," lUB/IUBS Symposium, 1960, Academic Press, Inc., in press).

BACTERIAL PHOTOSYNTHESIS 193

coincides with the surface. The average diameter observed in tlie
thin sections of the chromatophorcs is 300 A, which agrees very well
with the value (320 A) calculated by Stokes' relation from the sedi-
mentation constant and density. We employ the latter as the diam-
eter of the unfixed chromatophore. This agreement between the
different measurements suggests that shrinkage has been minimized
in the specimens used for electron microscopy.

The objective lens of the electron microscope has considerable
depth of focus, and points which are separated by considerable dis-
tances in the direction of the axis of the lens can be superimposed in
the final image. Since the chromatophorcs tend to pack very closely
in the cell, image-overlap due to the depth of focus will obscure
the chromatophorcs and produce optical artifacts if the sections
contain more than one laver of chromatophorcs. This effect also
makes it difficult to assign a value to the thickness of the cortex of
the chromatophore. There are at least four intrinsic factors which
influence the apparent thickness of the cortex in a section: the actual
thickness of the cortex, the efficiency with which the electrons are
scattered by this region, the thickness of the section, and the diam-
eter of the chromatophore. The apparent thickness of the cortex in
the thinnest sections is about 70 A, but the actual thickness of this
region must be less than this value. The magnitude of the deviation
becomes clear ( Fig. 8 ) when we compare the effect on the image of
varying the thickness of sections through the center of a sphere 300 A
in diameter with a completely opaque cortex. An apparent cortical
thickness of 70 A can be obtained for actual values ranging from 25
to 60 A as the section thickness decreases from 200 to 100 A. Since
the cortex of the chromatophore is only moderatelv opaque to 50-
kilovolt electrons and the thickness of the sections appears to be
less than 200 A, a cortical thickness of about 60 A seems reasonable.

Since the intrinsic electron density of most cellular constituents
is rather low, procedures which enhance electron densitv are nor-
mally employed for electron microscopy. Practice has established
osmium tetroxide, particularlv in the buftered solution advocated by
Palade (1952), as the standard or reference fixative. A classical
aphorism, namely, that osmium tetroxide fixes and blackens fats, has
been responsible for a tendency to assume that the patterns of elec-
tron density produced bv this reagent reflect the deposition of the
metal or its oxides in the lipid constituents of cellular structures.
Nevertheless, the data have often suggested that protein rather than

194

MACROMOLECULAR COMPLEXES

lipid is involved in defining the image of cytoplasmic membranes
and lamellae (see review by Sjostrand, 1959).

It is clear that the chromatophore is not a rigid structure, for the
electron micrographs of shadowed preparations show that the chro-
matophore becomes a thin disk when it is dried upon a support. By
the same token, water accounts for a considerable fraction of the

Fig. 8. A simple illustration of several factors which determine the image
of the sectioned chromatophore. If the cortex were opaque, the effect of geom-
etry and depth of focus could produce an apparent cortical thickness of 70 A
for actual values between 25 and 60 A as the section thickness varied between
200 and 100 A.

volume of the organelle. Since the chromatophore behaves and
"looks" like a vesicle, it is probable that most of the structural mate-
rials are concentrated in a cortical region.

The properties of the chromatophore indicate that protein rather
than lipid is present at the surface; for example, the chromatophore
is hydrophilic, has an isoelectric point, and acts as an enzyme. In
addition, the distortion or denaturation of protein is a prerequisite

BACTERIAL PHOTOSYNTHESIS 195

for the release of the hpids. Since the carotenoids, upon (\\traction,
are accompanied Idv a proportionate amount of the phospliohpid
(Newton and Newton, 1957), these pigments can serve as indicators
of the leaching of lipids in the solvents used in preparing specimens
for electron microscopy. Using such indications, we observed that
the image of the chromatophore was not appreciably different if
obvious leaching had occurred or if leaching had been minimized
bv using the solvents at the lowest practical temperature and poly-
merizing the monomeric methacrylate at dry ice temperature.

It is a common practice to fix tissues rather briefly because ex-
posure beyond an ill-defined optimum period in the customary
buffered osmium tetroxide fixatives produces a pronounced deterior-
ation of structure. In view of this practical limitation on fixation
time, the loss of lipid could have been due to inadequate osmication
or to an inherent lack of reactivity. These possibilities were tested
by experiments (1) with lyophilized chromatophores, (2) with the
proteinaceous residue obtained by extracting the lipids in a sequence
of boiling solvents (ethanol, 24 hours; ether, 24 hours) in a Soxhlet
apparatus, and (3) with the lipid films produced by evaporation of
the combined extracts. The rate and extent of osmication of each
of the three systems was determined by the change in dry weight as
a function of time. Numerous samples containing about 10 mg of
material were exposed to the fixative in a pre-equilibrated moist
chamber which contained a large volume of 1 per cent osmium
tetroxide. At intervals, samples were removed and reweighed after
dessication to constant weight.

The lipid films incorporated osmium rapidly and became black;
in 1 hour the weight increased by 40 per cent. At this point, the
bulk of the film was insolubilized, but about 10 per cent of the mass
could be extracted by ether or methacrylate monomer; 20 per cent
could be removed by ethanol. The incorporation of osmium was
finished in 8 hours with a net weight increase of 50 per cent. If we
use 900 as the average molecular weight of the lipids and assume
that the osmium tetroxide is reduced to the metal (at. wt. 190.8),
then at saturation each lipid molecule is related to two or three
atoms of osmium.

Under comparable conditions, the defatted protein actually in-
corporated osmium more rapidly than the lipid films, but the net
uptake was much less. The weight increment was 18 per cent in 1
hour. The final value was 20 per cent. The osmium content at

196

MACROMOLECULAR COMPLEXES

Fig. 9. An electron micrograph of a thin section of organisms which had
been lyophilized and extracted in boiling solvents before fixation. The organ-
isms were distorted by this procedure, but the bulk of the electron-dense material
is still observed despite the absence of lipids. The metaphosphote inclusions
(M) also persist. X 80,000.

saturation represents a ratio of about one osmium atom for ever\-
eight amino acid residues.

The total uptake of osmium by the Ivophihzed chromatophores
produced a 30 per cent increase in weight. This is essentially a
summation of the uptake expected for the sum of the two fractions;
however, the course of the reaction was not a simple summation of
concomitant reactions by the protein and the lipid. In 1 hour the
chromatophores had increased in weight by 15 per cent, or half of

BACTERIAL PHOTOSYNTHESIS 197

the final \aliie: nevertheless, 70 per cent of the hpid could be ex-
tracted in ethanol. The incorporation of osmium continued grad-
ually and required 3 days for completion. These data show that the
rate of osmication of the lipids is greatlv reduced when the lipids
are integrated in the chromatophore. If, as we propose, the lipids
are coated with protein (see below), then it might be expected that
osmication of the lipid would not proceed readih" until the protein
had been saturated. Be that as it mav, it is reasonably clear that
in our usual preparations the lipid does not contain much osmium
and will not ha\e a high electron densit\ .

The role of osmicated protein in image formation is shown un-
equi\ocall\- b\" electron micrographs of hophilized bacteria from
which the lipids were removed by Soxhlet extraction before fixation.
The organisms are distorted bv this violent treatment, but the bulk
of the electron-dense material is still present (Fig. 9). Taken col-
lectixely, these observations support the \iew that osmicated protein
is imaged as the cortex of the chromatophore in our preparations.

It should be noted, howe\ er, that even if the lipid were totallv
osmicated it might not be re\ealed. Stoeckenius (1959) has studied
thin sections of myelin forms prepared from phospholipid and fixed
in osmium tetroxide. The periodicit\- of the lamellae agrees with
the 40 A-spacing expected for bimolecular leaflets, but the dense
line is only 18-20 A thick. If the phospholipids in the chromatophore
are arranged in a monolaver (see below), the electron-dense line
might be only 10 A in diameter, which is well within the limits of
our measurement errors.

Hypothesis of the Ultrastructure of the Chromatophore

All of the protein can be accommodated in the region which is
imaged as the cortex of the chromatophore. The \olume of this zone

is (1.3 X 10')A^ There are about 1.3 X lO"^"

y (leO.V-lOOA^

grams of protein or, alternati\ely, some 67,000 axerage amino acid
residues (see above). The volume required bv the protein is about
10' .V when typical (cf. Frey-W'yssling, 1953) \ alues are chosen
for either the partial specific \olume (0.78 cc g) or the volume
(16lA'^) per residue. In \ iew of the agreement of such volume
considerations with the phvsicochemical and electron micro-
scope data, it is reasonable to infer that the bulk of the protein is

198 MACROMOLECULAR COMPLEXES

actually distributed in a laver about 60 A thick at the surface of the
chromatophore.

It is generally assumed that the photosynthetic pigments exist in
a condensed state, such as a monolayer, in vAvo. This concept has
been inferred repeatedly (see Rabinowitch, 1945, 1951, 1956) from
considerations of the absorption spectrum, fluorescence, energy
transfer, etc. In Chromatin m, energy is transferred from the carot-
enoids to bacteriochlorophyll with an efficiency of 30 to 40 per cent
(Duysens, 1952); neyertheless, these accessory pigments are not
necessary as functional or structural components (Fig. 10) of the
chromatophore. In addition, there is supposed to be a fundamental
difference between the relationships of the two types of pigment
with the protein. The bacteriochlorophyll is a clayate molecule;
that is, the phytol "tail" is hydrophobic, but the porphyrin "head"
is hydrophilic and can enter into intimate relation with protein.
The carotenoids are pure lipids and cannot form strong bonds with
protein.

The pigments of the chromatophore will fit in a monolayer about
25 A thick at the internal face of the proposed protein layer. The
area available at this interface [4n ( 100 A)-] is (1.26 X 10') A". The
carotenoids of purple bacteria are aliphatic C4,, compounds about
25 A in length. Since the yan der Waals radius of the methyl group
is 2 A, each of the 300 carotenoid molecules will require about 13 A-
at the interface, for a total of 4000 A". The remaining area, ( 1.22
X 10'' )A-, allows 200 A- for each of the 600 molecules of bacterio-
chlorophyll. This yalue falls between the area of the porphyrin
itself ( 242 A- ) and the area occupied by the "head" of the chloro-
phyll molecule in the tilted position assumed in monolayers (Trur-
nit and Colmano, 1959 ) . Similar calculations indicate that the 3000
cephalin molecules can be ordered with the polar groups directed
into the central aqueous phase and with the non-polar fatty acid
moieties forming a lipid phase in conjunction with the carotenoids
and the phytol "tails." This asymmetric organization, though specu-
latiye, is consistent with the known physicochemical properties, the
composition data, and the electron microscope observations.

The de novo origin of bacterial chromatophores seems to have
been established by observations on RhocIospiriUum ruhnim (Pardee
et al, 1952; Vatter and Wolfe, 1958). Such observations suggest
that the chromatophores are assembled from smaller particles. In
terms of the "fixed" components, the simplest representative build-

BACTERIAL PHOTOSYNTHESIS

ing block contains 1 carotenoid molecule

199

2 bacteiiochlorophyll
molecules, 10 cephalin molecules, and protein containing 220 amino
ds. Although a structural subunit witli a molecular weight of

aci

Fig. 10. An electron micrograph of thin sections of organisms in which the
development of colored carotenoids has been prevented by diphenylamine
treatment. The absence of these pigments does not alter the characteristic
appearance of the chromatophores which are dispersed throughout the cyto-
plasm. M: Metaphosphate inclusions; N, V, W: see Fig. 6. X22,000.

about 40,000 seems ridiculously small, this agrees with the molecu-
lar weight of the chloroplastin obtained by digitonin extraction of
Euglena chloroplasts (Wolken, 1956; Wolken and Schwertz, 1956).

200

MACROMOLECULAR COMPLEXES

The area available at the surface ( R = 160 A ) for the protein of
any subunit is about twice as large as the area available at the lipid-
protein interface ( R = 100 A ) . In the case of the hvpothetical
minimal unit, the area available at the lipid interface is only 420 A".
The protein could be folded, with the terminal groups brought to

Fig. 11. A working hypothesis of the ultrastructure of the chromatophore.
The Chromatium chromatophore is described as a hollow sphere about 320 A
in diameter with a cortex about 90 A thick. The pigment molecules (B) aligned
in a monolayer are bounded internally by a phospholipid monolayer (A) and
externally by a 60-A-thick protein layer. The "minimal unit" of composition has
been used as a structural subunit. The protein has been folded and is related
directly to two chlorophyll molecules. On the average, the protein is related
indirectly to one carotenoid molecule and ten phospholipid molecules.

the lipid surface as a-helices. Such an arrangement provides an
opportunity for visualizing a specific relationship between the ter-
minal amino acid sequence and the polar groups of the porphyrin.
In addition, a particular spatial orientation could be specified by the
pitch of the helix. Disulfide bridges between the subunits would
account for the observed stability of the chromatophore.

BACTERIAL PHOTOSYNTHESIS 201

A working hypothesis should represent a balance between sim-
phcity and concrete suggestions. Our simplified model (Fig. 11)
satisfies this criterion, but testing by prediction is needed to assess
its utility.

Note Added in Proof

W. Arnold and R. K. Clayton have obtained evidence with chro-
matophore films ( Proc. Nat. Acad. Sci. U.S. 46: 769-776; 1960) that
the first step in photosynthesis appears to be the separation of an
electron and a hole in a chlorophyll semiconductor. The data sup-
port our hypothesis of the ultrastructure of the chromatophore.

References

Anderson, I. C, and R. C. Fuller. 1958. Photophosphorylation by isolated chro-

matophores of the purple sulfur bacteria. Arch. Biochem. Biophys. 76: 168-179.
Anderson, I. C, R. C. Fuller, and J. A. Bergeron. 1958. Environmental effects on

photophosphorylation by bacterial chromatophores. Plant Physiol. 33: Suppl.

XVII.
Arnon, D. I. 1958. Chloroplasts and photosynthesis. Brookhaven Symposia in Biol.

No. 11: 181-235.
Arnon, D. I., M. B. Allen, and F. R. Whatley. 1954a. Photosynthesis by isolated

chloroplasts. Nature 174: 394-396.
Arnon, D. I., F. R. Whatley, and M. B. Allen. 1954b. Photosynthesis by isolated

chloroplasts. II. Photosynthetic phosphorylation, the conversion of light into

phosphate bond energy. /. Am. Chem. Soc. 76: 6324-6328.
Bergeron, J. A. 1958. The bacterial chromatophore. Brookhaven Symposia in Biol.

No. 11: 118-131.
Bergeron, J. A., I. C. Anderson, and R. C. Fuller. 1957. Studies on the chro-
matophore of the purple sulfur bacteria. Plant Physiol. Suppl. XVI. Vol. 32.
Bergeron, J. A., and R. C. Fuller. 1959. The influence of carotenoids on the infra-
red spectrum of bacteriochlorophyll in Chromatium. Nature 184: 1340-1341.
Duysens, L. N. M. 1952. Transfer of excitation energy in photosynthesis. Thesis.

State University of Utrecht, Netherlands.
Emerson, R. L., J. F. Stauffer, and W. W. Umbreit. 1944. Relationship between

phosphorylation and photosynthesis in Chlorella. Am. J. Botany 31: 107-120.
French, C. S. 1940. The pigment-protein compound in photosynthetic bacteria. I.

The extraction and properties of Photosynihin. J. Gen. Physiol. 23: 469-481.
Frenkel, a. W. 1954. Light induced phosphorylation by cell-free preparations of

photosynthetic bacteria. /. Am. Chem. Soc. 76: 5568-5569.
Frenkel, A. W. 1959. Light induced reactions of bacterial chromatophores and

their relation to photosynthesis. Ann. Rev. Plant Physiol. 10: 53-70.
Frey-Wyssling, a. 1953. Submicroscopic morphology of protoplasm. 2d English

ed. Elsevier Publishing Co., Amsterdam.
Fuller, R. C, and I. C. Anderson. 1957. Photophosphorylation in purple sulfur

bacteria. Plant Physiol. 32: Suppl. XVI.
Fuller, R. C, and I. C. Anderson. 1958. Suppression of carotenoid synthesis and

its effect on the activity of photosynthetic bacterial chromatophores. Nature

181: 252-254.

202 MACROMOLECULAR COMPLEXES

Goodwin, T. W., and D. G. Land. 1956. The carotenoids of photosynthetic bacteria.
Arch. Mikrobiol. 24: 305-312.

Goodwin, T. W., and H. G. Osman. 1954. Studies in carotenogenesis. 10. Spiril-
loxanthin synthesis by washed cells of Rhodospirilbim ruhrum. Biochem }. 56:
222-230.

Katz, E., and E. C. Wassink. 1939. Infrared absorption spectra of chlorophyllous
pigments in living cells and in extra-cellular states. Enzijmologia 7: 97-112.

Newton, J. W. 1958. Immunochemical reactions of the photosynthetic apparatus
in purple bacteria. Brookhcwen Symposia in Biol. No. 11: 289-295.

Newton, J. W., and M. D. Kamen. 1957. Photophosphorylation by sub-cellular
particles from Chromatium. Biochim. et BiopJujs. Acta 25: 462-474.

Newton, J. W., and G. A. Newton. 1957. Composition of the photoactive sub-
cellular particles from Chromatium. Arch. Biochem. Biophys. 71: 250-265.

Palade, G. E. 1952. A study of fixation for electron microscopy. /. Exptl. Med.
95: 285-298.

Pardee, A. B., H. K. Schachman, and R. Y. Stanier. 1952. Chromatophores of
RhodospiriUum ruhrum. Nature 169: 282-283.

Rabinowitch, E. I. 1945. Photosynthesis and Related Processes. Interscience Pub-
hshers, Inc., New York. Vol. 1.

Rabinowitch, E. I. 1951. Photosynthesis and Related Processes. Interscience Pub-
lishers, Inc., New York. Vol. 2, part. 1.

Rabinowitch, E. I. 1956. Photosynthesis and Related Processes. Interscience Pub-
lishers, Inc., New York. Vol. 2, part 2.

Ruben, S. 1943. Photosynthesis and phosphorylation. /. Am. Clicm. Soc. 65: 279-
282.

Schachman, H. K., A. B. Pardee, and R. Y. Stanier. 1952. Studies on the macro-
molecular organization of microbial cells. Arch. Biochem. Biophys. 38: 245-260.

Sjostrand, F. S. 1959. Fine structure of cytoplasm: The organization of mem-
branous layers. Revs. Modern Phys. 31: 301-318.

Stanier, R. Y. 1958. Formation and function of the photosynthesizing pigment sys-
tem in purple bacteria. Brookhaven Symposia in Biol. No. 11: 43-53.

Stoeckenius, W. 1959. An electron microscope study of myelin figures. /. Biochem.
Biophys. Cijtol. 5: 491-500.

Thomas, J. B. 1952. A note on the occurence of grana in algae and in photosyn-
thesizing bacteria. Koninkl. Ned. Akad. Wetenschap. Proc. Ser. C. 55: 207-208.

Trurnit, H. J., and G. Colmano. 1959. Chloroplast studies I. Absorption spectra
of chlorophyll monolayers at liquid interfaces. Biochim. et Biophys. Acta 31: 434-
447.

VAN NiEL, C. B. 1931. On the morphology and physiology of the purple and green
sulphur bacteria. Arch. Mikrobiol. 3: 1-112.

VAN NiEL, C. B. 1935. Photosynthesis of bacteria. Cold Spring Harbor Symposia
Quant. Biol. 3: 138-150.

VAN NiEL, C. B. 1936. On the metaboHsm of the Thiorhodaceae. Arch. Mikrobiol.
7: 323-358.

VAN NiEL, G. B. 1956. In A. J. Kluyver and C. B. van Niel, The Microbes con-
tribution to Biologij. Harvard University Press, Cambridge, Mass. Chap. iii.

Vatter, a. E., H. C. Douglas, and R. S. Wolfe. 1959. Structure of Rhodomicro-
bium vannielii. J. Bacterial. 11: 812-813.

Vatter, A. E., and R. S. Wolfe. 1958. The structure of photosynthetic bacteria.
/. Bacterial. 75: 480-483.

Williams, A. M. 1956. Light-induced uptake of inorganic phosphate in cell-free
extracts of obligately anaerobic photosynthetic bacteria. Biochim. et Biophys.
Acta 19: 570.

WoLKEN, J. J. 1956. Photoreceptor structures. I. Pigment monolayers and molecular
weight. /. Cellular Comp. Physiol. 48: 349-369.

WoLKEN, J. J., and F. A. Schwertz. 1956. Molecular weight of algal chloroplastin.
Nature 111: 136-138.

BACTERIAL PHOTOSYNTHESIS 203

DISCUSSION
D. Waugh, J. A. Bergeron, A. C. Giese, M. V. Edds, Jr.

Dr. Waugh (Massachusetts Institute of Technology): The properties
found for the chroinatophores themselves prompt me to ask about the physical
and chemical properties of the protein component alone.

Dr. Bergeron: Our knowledge hasn't advanced sufficientlv for such a
study. However, the results of our initial attempts to obtain subunits by lysing
the chromatophores with disulfide-cleaving agents have been promising. If
we succeed in obtaining preparations which fulfill the criteria for subunits,
then we will be in a better position to study the protein itself.

Dr. Giese (Stanford University): Does the proportion of one pigment to
another in the chromatophore vary with the quality of the light, or with other
conditions under which the culture is grown?

Dr. Bergeron: As you know, the colored-carotenoid:bacteriochlorophyll
ratio can vary considerably. Since Stanier's evidence (1958) indicates that
under certain conditions colorless precursors are transformed in situ, it is con-
ceivable that there is a fixed ratio between bacteriochlorophyll and the C40
compounds, without specifying the degree of unsaturation. We believe that
the protein-bacteriochlorophyll relationship probably sets a limiting value, with
local conditions determining the kind and amount of carotenoids which are
built into the chromatophore.

Dr. Edds (Brown University): Are you able to tell us anything about the
development of chromatophores?

Dr. Bergeron: We know very little about the development of bacterial
chromatophores. There is the evidence that chromatophores form de novo in
cells of Rhodospirillum nihrum when these organisms are induced to photo-
synthesize after a period of total dependence upon aerobic metabolism in the
dark. We cannot perform similar experiments with Chromatium, which is an
obligate anaerobic phototroph.

If chromatophores can appear in a cytoplasm which has been devoid of
these structures, then we can assume either that the new chromatophores are
derived by proliferation from other structures, such as the plasma membrane,
or the larger vesicles, or that the chromatophores are the morphological con-
sequence of the "homing instinct" of molecular subunits. Although, as oui
model indicates, we favor the idea of the subunit, it should be pointed out
that immunochemical data have been employed by Newton (1958) to infer a
specific relationship between the surface of the organism and the chromato-
phores, and we have observed evaginations from the larger vesicles. These
evaginations, if severed, would be morphologically indistinguishable from the
chromatophores.

Polysaccharide-Protein Complexes of
Yeast Cell Walls'

Walter J. Nickerson,- G. Falcone,'^ and Gian Kessler ^

Rapid advances in our knowledge of the chemical composition of
the cell walls of microorganisms have resulted from the develop-
ment of techniques for obtaining cell-wall preparations free from
other cellular components, and from the establishment of criteria of
homogeneity of cell- wall preparations. Microbial cells may be rup-
tured by mechanical or enzymatic procedures, and the broken cells
may be separated from intact cells by centrifugation at low speed.
By differential centrifugation and repeated washing of the cell-wall
fraction, clean cell-wall preparations may be obtained that appear
morphologically uniform in the electron microscope. The dialyzed
and lyophilized wall preparations may be fractionated and analyzed
with a fair degree of confidence that the results will not be negated
by contamination from cytoplasmic debris.

The cell wall comprises about 30 per cent of the dry weight of
bakers' or brewers' yeast, and includes a fraction which is remark-
ably resistant to chemical or enzymatic attack. In fact, the insoluble
residue obtained on toluene autolysis of yeast, and the insoluble
residue remaining after treating yeast with 1 N NaOH at 100° C,
retains much of the shape of a yeast cell. Early investigations drew
attention to the polysaccharide nature of this residue, and deduced
that the cell wall of yeast comprised "yeast cellulose" and "yeast
gum."

' Contribution from the Institute of Microbiology, Rutgers, The State University,
New Brunswick, New Jersey; experimental work described supported in part by
grants from the U. S. Public Health Service. The authors express their appreciation
to Pauline E. Holbert (Institute of Microbiology, Rutgers, The State University) for
her collaboration in the preparation of electron micrographs, and to Dr. D. E.
Williams ( Merck & Co., Rahway, N. J. ) for his collaboration on sedimentation
analyses.

^ Rutgers, The State University, New Brunswick, N. J.

^ Istituto di Patologia Generale, Universita, Naples, Italy.

" Department of Biochemistry, University of California, Berkeley, Calif.

205

206 MACROMOLECULAR COMPLEXES

Extraction of Wall Components from Intact Cells

In view of the substantial body of information developed up to
about 1950 on the polysaccharides obtained by extraction of intact
yeast cells with hot water, hot alkali, or cold concentrated acid, it
seems advisable to present a short survey of information available
prior to the present era of studies on the composition of clean cell
walls isolated from mechanically disrupted yeasts. A comparison
of results obtained by application of the several methods of extrac-
tion enumerated above has been presented by Garzuly-Janke
(1940), who concluded that hot alkaline extraction procedures lead
to more or less profound modifications of the alkali-soluble poly-
saccharide component. Certainly the application of more or less
drastic extraction procedures designed to remove a polysaccharide
material from a cell as a protein-free, nucleic acid-free moiety has
led us to some insight into the nature of the polysaccharide in ques-
tion, but it has also given rise to considerable controversy about the
natural structure of the polysaccharide. Furthermore, efforts di-
rected at "throwing away the protein" have not led to an apprecia-
tion of the importance of protein -polysaccharide complexes as the
macromolecular fibers of the cell-wall fabric.

Yeast Glucan. The water-insoluble polysaccharide fraction of
yeast that resists boiling in dilute alkali was studied by Salkowski
(1894b) and designated yeast "achroocellulose." This fraction was
subsequently studied by Zechmeister and Toth (1934), who found
that, like cellulose, the resistant polysaccharide could be hydrolyzed
in 40 per cent HCl, although it did not exhibit many of the other
characteristics of cellulose (yeast glucan is insoluble in Schweizer's
reagent, does not produce a color with zinc-chlor-iodide, and does
not yield cellobiose on acetolysis). An uncommon 1,3-glucosidic
linkage was indicated for yeast glucan from methylation studies:
2,4,6-trimethylglucose was the sole product isolated. Further work
by Hassid et al. (1941) on the glucan fraction confirmed 2,4,6-tri-
methylglucose as the sole methylation product, and supplied evi-
dence of the /^-configuration for the anhvdroglucose units in the
polymer (i.e., low specific rotations of the alkylated derivatives and
upward mutarotation during their hydrolysis). From viscosity de-
terminations a molecular weight of 6500 was suggested, and the

POLYSACCHARIDE-PROTEIN COMPLEXES 207

glucan was characterized as a presumably closed-chain polvmer of
yS-l,3-glucosido-gliicopvranose units.

The structure of yeast glucan was reinvestigated by Bell and
Northcote (1950). Mildlv acidic conditions (0.5 N acetic acid), in
place of strong mineral acid, were employed to effect isolation of the
polymer, and chromatographic procedures were used to detect hy-
drolysis products following exhaustive methylation (former methods
relied on refractive indices and methoxyl content of the fractions
obtained by distillation of the hvdrolvzed methvlated glucan ) . With
these refinements, 2,3,4, 6-tetramethvl-D-glucose and 4,6-dimethyl-D-
glucose were detected, in addition to 2,4,6-trimethyl-D-glucose, as
products of hydrolysis of the methylated glucan. The methylated
derivatives of D-glucose were present in a ratio of tetra:di:tri
= 1:1:9, thus necessitating a drastic revision of the structural for-
mula that could be considered for glucan. Bell and Northcote
considered the polymer to be a highly branched structure in which
each unit chain consists, on the average, of nine anhvdroglucose
radicals.

Glucomannans of Yeast. Early studies on the so-called yeast
gum showed it to contain glucose and mannose. Purified prepara-
tions with [ajo"^' — +88.5° were obtained by Salkowski (1894a),
Meigen and Spreng (1908), and von Euler and Fodor (1911).

The preparation obtained by Meigen and Spreng had a ratio of
mannose : glucose = 2:1, whereas that of von Euler and Fodor
showed a ratio of 1:1. In the latter case, the gum was obtained by
extracting fat-free yeast with boiling water, or by precipitating the
polysaccharide with ethanol from an autolysate of brewers' yeast.
The polysaccharide was purified by repeated precipitation of the
copper complex, according to the technique of Salkowski, and sub-
sequent dialysis. It should be noted that treatment with hot alkali
was avoided. Quantitative analyses for mannose in the presence of
glucose in acid hydrolysates of the polysaccharide were carried out
by differential precipitation of the mannose phenylhydrazone.

Subsequent investigations by Harden and Young (1912) and by
Ling et at. ( 1925 ) employed an extraction step involving boiling 2
per cent NaOH, and subsequent treatment of the alcohol-precipi-
table material with 60 per cent KOH for 2 hours in a vigorously
stirred boiling-water bath.

The structure of a yeast mannan that was presumed to yield only
D-mannose on acid hydrolysis was studied by Haworth and collab-

208 MACROMOLECULAR COMPLEXES

orators (1937, 1939, 1941). Exhaustive methylation of alkaline
solutions of the polysaccharide resulted in the formation of 2,3,4,6-
tetramethyl-, 3,4-dimethyl-, and 2,4,6-trimethylmannopyranose de-
rivatives. The occurrence of a 3,4-dimethyl derivative indicated a
double linkage to other components of the polymer. The evidence
was assessed as indicating that the polymer consists of a main chain
of 1,6-a-mannopyranose units to which are joined short side-chains
at the number 2 carbon of alternate rings. It should be pointed out
that the mannan employed in these studies was extracted from
bakers' yeast by boiling in 6 per cent NaOH solution for 8 hours.
Furthermore, it must be noted that the yield of mannose phenyl-
hydrazone from acid-hydrolyzed purified mannan was actually only
50 per cent of the theoretical amount. When a correction (based on
the recovery of only 82 per cent of the theoretical amount of man-
nose phenylhydrazone from pure mannose) was applied, the claim
was made that a minimum yield of 92 per cent of the theoretical
quantity of mannose was obtained on the hydrolysis of mannan.
Actually, even allowing correction for failure to recover the theo-
retical quantity of phenylhydrazone from a known solution of man-
nose, it would seem that the hydrolvsate of mannan did not contain
more than two-thirds of the theoretical amount of mannose.

Wall Components of Mechanically Disintegrated Yeasts

Publications appeared independently by Northcote and Home
(1952) and by Houwink et ah (1951) reporting the localization of
glucan, mannan, and chitin as components of isolated cell-wall frag-
ments. After mechanical rupture of cells, Northcote and Home iso-
lated washed cell walls of bakers' yeast. This material was free of
whole cells and cell debris and comprised the outer membranes of
the cell; it was completely Gram-negative in character, in contrast to
whole yeast cells which normally are intensely Gram-positive. The
cell wall was found to include nitrogenous substance and lipid and
to be composed principally of two polysaccharides— a mannan and a
glucan. Glycogen was not associated with cell walls isolated by this
procedure. Concurrent studies with the electron microscope showed
that the isolated wall material consisted of at least two layers, the
outer composed principally of glucan and the inner consisting of a
nitrogenous substance and mannan. The over-all composition of the
cell wall was taken to be: glucan 29 per cent, mannan 31 per cent,
protein 13 per cent (based only on determination of total nitrogen).

POLYSACCHARIDE-PROTEIN COMPLEXES

209

lipid (mainly neutral fat) 8.5 per cent, and ash 3 per cent, leaving
16 per cent unaccounted for by this procedure. Comparable data,
in agreement with the above, were reported by Roelof sen ( 1953 ) .
Not all of the nitrogen found in yeast cell-wall preparations is asso-
ciated with protein; many reports indicate the presence of chitin in
veast cell walls. Employing the chitosan sulfate microchemical tech-
nique, Roelofsen and Hoette ( 1951 ) demonstrated the presence of
chitin in bakers' yeast and other yeast species, thereby confirming
observations of Schmidt ( 1936) and Nabel ( 1939), who had claimed
the presence of chitin in yeast.

Preparation of Isolated Clean Cell Walls of Yeast. By mechan-
ical disruption of bakers' yeast and Candida albicans, followed by

Fig. 1, Cell-wall fragment obtained by mechanical disintegration and dif-
ferential centrifugation of Candida albicans.

210 MACROMOLECULAR COMPLEXES

differential centrifugation in media of differing density, preparations
of isolated cell walls were obtained by Falcone and Nickerson
( 1956 ) that were entirely free of intact cells and other particulate
cellular matter. Cells from starch-free pound cakes of bakers' yeast
(Anheuser-Busch) or laboratory-grown C. albicans were suspended
in water or in a 5 per cent aqueous solution of dithioglycol ( 150 g
yeast: 50 ml solution) and were disrupted in the cold by agitation
in a Waring Blendor with glass beads, according to the method of
Lamanna and Mallette (1954). More than 90 per cent breakage of
cells was achieved after 90 minutes' agitation.

IJarg

a

VAL-MET

O

PHE

LEU-ILEU

<

z
o

<

o

z

UJ

X

LYS J

If] ^'''

(JQthr

0-

O".

Q-

Ocvs

n-BUTANOL-ACETIC ACID-WATER
(a)

Fig. 2. Two-dimensional chromatograms of amino acids obtained on acid
hydrolysis of isolated, clean cell walls (a) of Candida albicans, normal strain
582, and (b) of C. albicans, selenite-resistant mutant strain RM806, grown on a

POLYSACCHARIDE-PROTEIN COMPLEXES

211

Cell-wall fragments were isolated by differential centrifugation
and repeated washing with distilled water, 8.5 per cent sucrose
solution, and phosphate buffer (pH 7.4, O.IM). Washing with
sucrose and with buflFer served to eliminate completely the small
particles which, in water, sediment along with the cell-wall frag-
ments. After about 50 repetitions of washing and differential cen-
trifugation, a fraction consisting exclusively of cell-wall material was
obtained (Fig. 1). This was lyophilized, and the analvtical data are
based on material prepared in this manner. The nitrogen content of
cell-wall preparations from two different batches of bakers' yeast

lARG

o

(^^VAL-MET

PHE

LYS

HXE

LEU-
ILEU

n-BUTANOL-ACETIC ACID-WATER
(b)

medium containing lO-"^ M Na2Se03. Note spot (X) in (b), corresponding to
cysteic acid of (a), presumed to be a selenium-containing analogue.

212

MACROMOLECULAR COMPLEXES

was 1.28 and 1.24 per cent. These values are lower than that (2.1
per cent) reported bv Xorthcote and Home ( 1952), who stated that
their preparation contained no unbroken cells but was contaminated
bv small particles which were difficult to eliminate.

The analytical data show that the clean cell-wall material con-
tains approximateb' 7 per cent protein, 85 per cent pohsaccharide,
and 3 per cent chitin. The high sulfur: protein ratio (2.1 per cent)
indicated that the protein may be a pseudokeratin t\"pe. Bv means
of two-dimensional paper chromatographv of h\drol\sates of cell-
wall material (6 X HCl for 16 hours at 110= C), the presence of 16
amino acids was detected. Representati\-e chromatograms from cell
walls of two strains of C. albicans are shown in Fig. 2 a and b.

Fig. 3. Sedimentation pattern of primary glucomannan-protein (isolated
from bakers' yeast cell wall) in analytical ultracentrifuge. Duration of centrifu-
gotion noted on each exposure. Note indication of low molecular weight mate-
rial and apparent homogeneity of major peak.

From preparations of clean cell wall, the major fraction ( approxi-
mately 75 per cent) was solubilized in 1 X KOH on warming to
about 50° C. The solubilized material was dialvzed against running
tap-water for 24 hours; on lyophilization, a white powder was ob-
tained, of which the bulk was readih" soluble in water. The water-
soluble fraction was found to contain pohsaccharide and protein,
and was assumed to be a mannan-protein complex on the basis of
the absence of the characteristic copper precipitate with Fehling's
solution. This view was reinforced bv ultracentrifugal studies which
revealed the material to be monodisperse at several concentrations
in water and in buffer. Representative sedimentation patterns are
showTi in Fig. 3, from which a sedimentation constant So,, = 4.3
X 10~^^ was calculated.

Separation of Glucomannan-Proteins from Isolated Cell Walls.
The initial work on fractionation of cell-wall components led to
recognition of their polysaccharide-protein nature and made clear
the necessitv for removal of lipid components; subsequent work

POLYSACCHARIDE-PROTEIN COMPLEXES

213

214

MACROMOLECULAR COMPLEXES

( Kessler and Nickerson, 1959) followed the scheme of fractionation
shown in Fig. 4. Lipid-free cell-wall material was suspended in 1 N
KOH for 1 hour under nitrogen at 25° C; the solution obtained was
passed through an ultrafine sintered glass filter. The filtrate was
dialyzed for 24 hours against running tap-water and lyophilized.
The resulting white powder was suspended in water at 25° C for 1
hour, then centrifuged. The centrifugate was saturated with am-
monium sulfate; a precipitate was obtained that was removed bv
high-speed centrifugation. The precipitate and the supernatant
were dialyzed, concentrated, and Ivophilized. Both of the resulting
white powders proved to be polysaccharide-protein complexes, with

Fig. 5. Electron micrograph of glucomannan-protein I (GMP-I). Note aggre-
gation of small spherical structures into globular masses. Deposited from
solution onto Formvar film.

POLYSACCHARIDE-PROTEIN COMPLEXES

215

Fig. 6. Electron micrograph of GMP-I; as for Fig. 5. Note organization of
small spheres into shape of inner portion of bud scar.

216 MACROMOLECULAR COMPLEXES

mannose and glucose present in a ratio of approximately 1:1 in the
fraction (termed glucomannan-protein I, or GMP-I) precipitated
by saturated ammonium sulfate, and approximately 2:1 in the frac-
tion (GMP-II) that remained in solution. The alkali-insoluble
residue proved to be a third polysaccharide-protein fraction, the
polysaccharide being composed very largely of glucose. This frac-
tion, termed glucan-protein, comprised approximatelv 40 per cent
of the weight of the wall material.

Examination of Wall Components by Physical Methods. The
polysaccharide-protein fractions obtained in the manner just de-
scribed were examined by a variety of physical and chemical means
to ascertain their form, relative homogeneity, and composition.
Electron micrographs of GMP-I, evaporated from solution onto
Formvar membranes, are shown in Figs. 5, 6. The pronounced
globular appearance of aggregates of GMP-I can be seen to arise
from the clumping of much smaller spherical shapes that seem to
possess an electron-dense center. In Fig. 6, a "structure" that has
the size and shape of the inner portion of a bud scar is seen. This
"scar" portion is clearly distinct from the "bud scar" of the outer

Fig. 7. Appearance of budding cells of bakers' yeast by aniline blue fluo-
rescence microscopy. Note localization of oriented dye molecules in toroidal
zone between bud and mother cell. (Courtesy, Currier, 1957.)

POLYSACCHARIDE-PROTEIN COMPLEXES

217

glucaii component of the wall; in the latter, the scar results from
the circular ordering of fibrillar components into a highly crystalline
pattern. The biochemical and mechanical sequences that operate
during the budding process, and which bring about the circular
orientation of fibrils in an otherwise highly disordeied fibrillar net-
work, have been discussed in detail elsewhere ( Nickerson and Fal-
cone, 1959; Falcone and Nickerson, 1959).

Evidence for the existence of a crystalline ringlike region in the
living cell at the juncture of bud and mother cell can be seen ( Fig.
7) from the work of Currier (1957) with aniline blue fluorescence

' ^1

m ^ 1

i "

^■■Mii^>'''''''^

1/4

Fig. 8. Electron micrograph of isolated cell wall of bakers' yeast. Note
difFerence in appearance between surface of outer layer and granular inner
surface of wall.

218

MACROMOLECULAR COMPLEXES

microscopy. In mechanically isolated wall fragments, the scar
region (Figs. 8 to 10) gives evidence of morphological differences
between outer and inner boundaries of the wall. The scar in Fig. 9,
seen from "inside" the cell, appears to be composed of small spheres
with dense centers (as seen in GMP-I in Fig. 5), and the plug in

Fig. 9. Electron micrograph of isolated cell wall of bakers' yeast. Note bud
scar as seen from inside cell; mark in center of scar appears as a small papilla.
Scale as in Fig. 8.

the center of the scar appears as a papilla. In Fig. 10, the scar is
seen from "outside" the cell, and the mark in the center of the scar
appears as a depression.

On the matter of the existence of different layers in the cell wall
of yeasts, Houwink and Kreger ( 1953 ) commented that none of their
micrographs provided evidence that the cell wall consists of an inner
and an outer layer. However, it should be pointed out that all except
their "untreated" specimens were boiled in alkali, and they state,
"Boiling with 3% NaOH for three hours, on the assumption that it
would prove useful to remove any remaining protein, leaves a

POLYSACCHARIDE-PROTEIN COMPLEXES 219

thinner cell wall." Photographs of Northcote and Home (1952)
clearly show two layers in wall material that had been extracted
only with lipid solvents, but their photographs reveal only one laver
in wall material that had been treated with 3 per cent aqueous
NaOH to remove mannan and protein.

Fig. 10. Electron micrograph of isolated cell wall of bakers' yeast. Bud
scar, as seen from outside of cell, has central mark that appears as a depres-
sion.

The initial ultracentrifugal studies on the glucomannan-protein
complex in the primaiy alkaline extract of isolated walls had shown
(Fig. 3) the presence of a major component that sedimented as a
single peak, but the height of the boundary to the left of the peak
indicated the presence of material of low molecular weight. On
subsequent fractionation of the primary extract with ammonium
sulfate, the components GMP-I and GMP-II were obtained. Ultra-
centrifugal studies on these components (Figs. 11, 12) show GMP-I
to be a slowly sedimenting substance of low molecular weight,

220

MACROMOLECULAR COMPLEXES

whereas GMP-II is responsible for the major peak observed .in
earher sedimentation diagrams. On prolonged centrifugation, the
GVIP-II diagram reveals the presence of two similar components
(Fig. 12).

Fig. 11. Sedimentation pattern of glucomannan-protein I (GMP-I) in analyti-
cal ultracentrifuge. Material is evidently of low molecular weight.

Fig. 12. Sedimentation of GMP-II in ultracentrifuge. Note correspondence
to pattern of peak shown in Fig. 3. On prolonged centrifugation, evidence for
two components is seen.

Analysis of Polysaccharide-Protein Complexes. Lipid-free wall
preparations, subjected to the fractionation procedure shown in Fig.
4, yielded recoveries of the individual complexes from the various
yeasts as listed in Table 1. The glucan-protein fraction may account

TABLE 1

Polysaccharide-Protein Complexes of Yeast Cell Walls
in Per Cent *

Organism Glucan-Protein GMP-I GMP-II

Bakers' yeast 41.6 13.6 34.7

S. cerevisiae 18.29 28.3 55.8 11.9

C. albicans 582 47.4 3.0 27.2

C. albicans me 45.0 2.1 34.3

" Data expressed as percentages of lipid-free wall recovered when fractionated

according to scheme shown in Fig. 4; based on data of Kessler and Nickerson ( 1959).

POLYSACCHARIDE-PROTEIN COMPLEXES 221

for approximately 40 per cent of the wall material, whereas the
relative proportions of the two glucomannan-protein fractions varv
from one organism to another.

On acid hvdrolysis of the complexes and analysis for monosac-
charide constituents, glucose and mannose were found in the com-
plexes in the percentages given in Table 2. The data indicate the
presence of a glucan complex and of two glucomannan complexes
in the cell wall. Closely similar carbohydrate composition was found
in the three complexes from preparations of bakers' yeast, as well
as from preparations of C. albicans.

TABLE 2

Relative Content of Glucose and Mannose in Cell Wall Components
in Per Cent *

Glucan-Protein GMP-I GMP-II

Organism Glucose Mannose Glucose Mannose Glucose Mannose

Bakers' yeast 95 5 47 53 35 65

S. cerevisiae 18.29 95 5 45 55 33 67

C. albicans 806 95 5 48 52 37 63

C. albicans RM806 97 3 52 48 29 71

C. albicans 582 98 2 54 46 36 64

* Data from Kessler and Nickerson ( 1959).

In all of the glucan preparations that have been isolated, a small
amount of mannose has been detected. This is probably due to the
fact that the mild alkaline treatment which is used for the fraction-
ation of the wall preparations is not able to dissociate the glucoman-
nans completely from the insoluble glucan skeleton. A prolonged
exposure of the glucan to alkali can remove these traces of gluco-
mannan, but at the same time this treatment causes a degradation or
even a destruction of the protein that is associated with the carbo-
hydrate material. Equal amounts of glucose and mannose are found
in the material that can be precipitated from alkaline solution with
ammonium sulfate, whereas mannose accounts for approximately
two-thirds of the total sugar in the hydrolysate of the glucomannan
complex II.

Chromatographic analysis revealed that the glucomannan-pro-
tein II of bakers' yeast is composed of 13 amino acids, the glucan-
protein of 15 amino acids, and the glucomannan-protein I of 17

222

MACROMOLECULAR COMPLEXES

amino acids. These findings hold for both the low-protein bakers'
yeast and the high-protein strain {Saccharomijces cerevisiae, 18.29).
Quantitative determinations on the amino acid composition of the
protein components have been carried out on the three fractions;
values for the recoveries of the principal amino acids are shovi^n in
Table 3. These figures reveal that the proteins associated with the

TABLE 3
Amino Acid Analysis of Cell Wall Components *

Amino Acid

Glucan-Proteint

GMP-II

GMP-IIt

Glutamic acid 10.9 17.8 9.2

Aspartic acid 29.8 13.1 31.1

Cvsteic acid 1.5 0.2 8.4

Tyrosine 2.2 5.0 2.7 —

Serine 4.1 4.1 4.6

Threonine 6.0 5.0 5.9

Glycine 3.3 3.5 4.6

Alanine 6.4 6.9 6.5

Valine 6.5 4.6 5.4

Proline 7.8 2.2

Methionine 1.6

Isoleucine 4.5 5.1 4,9

Leucine 4.6 9.1 7.0

Lysine 5.1 8.1 6.2

Histidine 2.7

Arginine 3.7 7.2 3.5

Phenylalanine 3.6 3.8

' Values expressed as percentages of total recovery from acid hydrolysates of each

component; from data of Kessler and Nickerson (1959).

t Glucan-protein isolated from bakers' yeast.

\ Glucomannan-proteins I and II isolated from strain 18.29 of Soccharoimjces
cerevisiae.

three fractions are highly acidic; glutamic and aspartic acids, to-
gether, account for 31 per cent (GMP-I), 35 per cent (GP), and 39
per cent ( GMP-II ) of the total recovery. Although total recoveries
of approximately 75 per cent of the amino acid components were
obtained for GP and GMP-I, it must be borne in mind that some
destruction and alteration of amino acid residues could not be
avoided under the conditions used for the protein hvdrolvsis; trypto-
phan, if present, would have been completelv destroyed.

Configuration of Polysaccharides in Wall Components. Im-
munochemical studies on the cross-reactions of glucomannan-pro-

POLYSACCHARIDE-PROTEIN COMPLEXES 223

teins I and II with antisera to pneumococcus and Salmonella typhi
have been carried out by Heidelberger. Negative reactions with
both GMP-I and -II in type II pneumococcus antiserum indicated
the absence of a-1, 6 or a-1, 4 Hnkages, whereas stronglv positive
cross-reactions in type VIII antiserum with GMP-I indicate the
presence of yS-1, 4 glycosidic hnkages. A stronglv positive cross-
reaction in this antiserum was also obtained with lies glucomannan
(Heidelberger and Rebers, 1958), from which a cellobiose moiety
has been obtained on partial hydrolvsis (Smith and Srivastava,
1956). In contrast, GMP-II showed little, if any, cross-reaction with
the t\pe VIII antiserum. It is, therefore, of considerable interest to
note that yeast "mannan," obtained by hot alkaline extraction pro-
cedures, has not been found to cross-react with any pneumococcal
antiserum (Heidelberger, 1959). Both GMP-I and -II also gave
strongly positive cross-reactions in typhoid antiserum, as had been
found previously for yeast "mannan" (Heidelberger and Cordoba,
1956).

Additional evidence against the presence of a-pyranose deriva-
tives in glucomannan-protein I is provided by infrared absorption
spectra. Barker et al. ( 1956 ) have shown that a-pyranose residues
absorb strongly at 844 cm~^ and /3-pyranose derivatives absorb at
about 890 cm \ Examination of Fig. 13 reveals practically no ab-
sorption in the 844 cm~^ region by GMP-I, but significant absorp-
tion by GMP-II. On the other hand, GMP-I exhibits a small peak
at 890 cm"^ which is not shown by GMP-II.

Bonding Between Polysaccharide and Protein. A thoroughly
washed, clean cell-wall preparation does not leach out material on
standing in water. This insolubility of the wall is in sharp contrast
to the solubility of GMP-1 and -II in water. Glucan is the insoluble
polysaccharide component of the wall to which the soluble com-
ponents are bound. The high content of acidic amino acids in the
protein components of the wall fractions immediately suggests that
protein and polysaccharide mav be joined through ester linkage.
This possibility is being explored through studies of infrared spectra,
and of the hydroxylamine ester-cleavage reaction. It is not yet pos-
sible to assess quantitatively the relative contributions of ester link-
age (or other covalent linkage) and hydrogen-bonding in uniting
the various components of the wall together to form the complex
wall structure, but some understanding of this union may be forth-
coming in the near future.

224

MACROMOLECULAR COMPLEXES

CO

z
<

2500
2000

1500

1000 800

900

1200
FREQUENCY (cm"')

700

2500

1500

1000

800

2000

1200 900

FREQUENCY (cm'')

700

Fig. 13. (Above) Infrared absorption spectra for glucan-protein (GP).
GMP-I, and GMP-II from high-protein strain 18.29 of Saccharomyces cerevisiae.
(Below) Spectra obtained with isolated cell wall of bakers' yeast and C. albi-
cans are shown for comparison.

POLYSACCHARIDE-PROTEIN COMPLEXES 225

Operation of Wall Components in Cellular Division

The mechanism of celhilar division in yeasts has been a subject
of investigation in this laboiatorv for many vears. It is intended at
this time merelv to indicate briefly the alterations in wall compo-
nents that occur during the process of budding. From studies on
divisionless mutant strains of Candida albicans it was found ( Nick-
erson and Falcone, 1956) that an enzyme (termed "protein disulfide
reductase," PDS reductase) which exists in normal strains causes
reduction of disulfide bonds of the protein of glucomannan-protein.
In the initial stages of this study, the primary glucomannan-protein
was employed as substrate. (In an unfortunate typographical con-
fusion, the primary glucomannan-protein, when used as substrate
for PDS reductase, was referred to as GMP-I in one of our publi-
cations, Nickerson and Falcone, 1959. ) In light of our recent frac-
tionation of primary GMP, it is hoped that we will be able to identify
more precisely the sulfur-rich substrate of PDS reductase.

The consequences of reduction of disulfide bonds in a wall com-
ponent have been interpreted in terms of randomly disordered,
covalent-bonded polymer matrices in which localized structural
weakness is introduced by rupture of covalent bonds. At the weak-
ened site, an explosive blowout occurs, for which evidence has been
presented from short-interval, time-lapse photomicrographv ( Nick-
erson and Falcone, 1959). The naked protoplasmic sphere that is
blown out of the mother cell constitutes the bud-initial. It rapidly
becomes covered with wall substance bv growth of the wall sub-
stance from the mother cell at the base of the bud-initial.

The force of the outward thrust at the instant of explosive bud-
ding serves to orient the fibrils of the glucan-protein and the spher-
ical structures of GMP-I components, as shown in Figs. 6, 9, and 10
and discussed in detail elsewhere (Falcone and Nickerson, 1959).
The breaking of covalent disulfide bonds within or between compo-
nents of the wall fabric sets in motion the train of events witnessed
in the budding process: the weakened wall cannot contain the in-
ternal pressure, a bud is extruded, and the macromolecular com-
ponents, momentarily partially freed of bonding to neighboring
components, are ordered by the outward fluid flow into the toroidal
pattern that would be predicted from purely physical considerations.

226 MACROMOLECULAR COMPLEXES

References

Barker, S. A., E. J. Bourne, and D. H. Whiffen. 1956. Use of infrared analysis
in the determination of carbohydrate structure. Methods of Biochem. Anal. 3: 213-
245.

Barton, A. A. 1950. Some aspects of cell division in Saccharomyces cerevisiae. J.
Gen. Microbiol. 4: 84-86.

Bell, D. J., and D. H. Northcote. 1950. The structure of a cell-wall polysaccharide
of baker's yeast. /. Chem. Soc. Pp. 1944-1947.

Currier, H. B. 1957. Callose substance in plant cells. Am. J. Botany 44: 478-488.

Falcone, G., and W. J. Nickerson. 1956. Cell-wall mannan-protein of baker's
yeast. Science 124: 272-273.

Falcone, C, and W. J. Nickerson. 1959. Enzymatic reactions involved in cellular
division of microorganisms. In W. J. Nickerson (ed. ). Biochemistry of Morpho-
genesis. Pergamon Press, Ltd., London.

Garzuly-Janke, R. 1940. Uber das Hefemannan. /. prakt. Chem. 156: 45-54.

Harden, A., and W. J. Young. 1912. The preparation of glycogen and yeast-gum
from yeast. /. Chem. Soc. Pp. 1928-1930.

Hassid, W. Z., M. a. Joslyn, and R. M. McCready. 1941. The molecular consti-
tution of an insoluble polysaccharide from yeast, Saccharomyces cerevisiae. J. Am.
Chem. Soc. 63: 295-298.

Haworth, W. N., R. L. Heath, and S. Peat. 1941. The constitution of yeast
mannan. /. Chem. Soc. Pp. 833-842.

Haworth, W. N., E. L. Hirst, and F. A. Isherwood. 1937. Polysaccharides. Part
XXIV. Yeast mannan. /. Chan. Soc. Pp. 784-791.

Haworth, W. N., E. L. Hirst, F. Isherwood, and J. K. N. Jones. 1939. 2:3:4-
Trimethyl mannose. J. Chem. Soc. Pp. 1878-1880.

Heidelberger, M. 1959. All polysaccharides are immunologically specific. In M. L.
WoLFROM (ed.). Carbohydrate Chemistry of Substances of Biological Interest.
Pergamon Press, Ltd., London.

Heidelberger, M., and F. Cordoba. 1956. Cross-reactions of antityphoid and anti-
paratyphoid B horse sera with various polysaccharides. /. Exptl. Med. 104: 375-
382.

Heidelberger, M., and P. A. Rebers. 1958. Cross reactions of polyglucoses in
antipneumococcal sera. VI. Precipitation of type VIII and type III antisera by
/3-glucans. /. Am. Chem. Soc. 80: 116-118.

HouwiNK, A. L., and D. R. Kreger. 1953. Observations on the cell wall of yeasts.
An electron microscope and x-ray diffraction study. Antonie van Leeuwenhoek.
J. Microbiol. Serol. 19: 1-24.

HouwiNK, A. L., D. R. Kreger, and P. A. Roelofsen. 1951. Composition and
.structure of yeast cell walls. Nature 168: 693-694.

Kessler, G., and W. ]. Nickerson. 1959. Glucomannan-protein complexes from
cell walls of yeasts. /. Biol. Chem. 234: 2281-2285.

Lamanna, C, and M. F. Mallette. 1954. Use of glass beads for the mechanical
rupture of microorganisms in concentrated suspensions. /. Bacteriol. 67: 503-504.

Ling, A. R., D. R. Nanji, and F. J. Paton. 1925. Studies on glycogen. Part I. The
nature of yeast glycogen, its preparation, estimation, and its role in yeast metab-
olism. /. Inst. Brewing 13: 316-321.

Meicen, W., and A. Spreng. 1908. Uber die Kohlehydrate der Hefe. Hoppe-Seyler's
Z. physiol. Chem. 55: 48-73.

Nabel, K. 1939. iJber die Membran niederer Pilze, besonders von Rhizidiomyces
bivellatus nov. spez. Arch. Mikrobiol. 10: 515-541.

Nickerson, W. J., and G. Falcone. 1956. Identification of protein disulfide re-
ductase as a cellular division enzyme in yeasts. Science 124: 722-723.

Nickerson, W. J., and G. Falcone. 1959. Function of protein disulfide reductase
in cellular division of yeasts. In R. Benesch (ed.). Sulfur in Proteins. Academic
Press, Inc., New York.

POLYSACCHARIDE-PROTEIN COMPLEXES 227

NoRTHCOTE, D. H., and R. W. Horne. 1952. The chemical composition and struc-
ture of the yeast cell wall. Biochem. J. 51: 232-236.

RoELOFSEN, p. A. 1953. Yeast mannan, a cell wall constituent of baker's yeast.
Biochem. et Biophtjs. Acta 10: 477-478.

RoELOFSEN, p. A., and I. Hoette. 1951. Chitin in the cell wall of yeasts. Antonie
van Leeiiwenhoek. J. Microbiol. Serol. 17: 297-313.

S.\LKOwsKi, E. 1894a. Ueber die Kohlehydrate der Hefe. Ber. dent. chem. Ges. 27:
497-502.

S.^LKOWSKI, E. 1894b. Ueber die Kohlehydrate der Hefe. II. Die Hefecellulose.
Bcr. (lent. chem. Ges. 27:3325-3329.

Schmidt, M. 1936. Makrochemische Untersuchungen iiber das Vorkommen von
Chitin bei Mikroorganismen. Arch. Mikrobiol. 7: 241-260.

Smith, P., and H. C. Siuvastava. 1956. Acetolvsis of the glucomannan of Ile.s
mannan. /. Am. Chetn. Soc. 78: 1404-1408.

VON EuLER, H., and A. Fodor. 1911. Zur Kenntnis des Hefengummis. Iloppe-
SeijlersZ. physiol. Chem. 72: 339-346.

Zechmeister, L., and G. Toth. 1934. Uber die Polyose der Hefemembran. Bio-
chem. Z. 270: 309-316.

DISCUSSION
/. W. Green, W. J. Nickerson, H. Ducoff, R. D. Preston, ]. D. Bergeron

Dr. Green (Rutgers University): During the rupture of the cell wall and
explosive budding, is the plasma membrane ruptured? In osmotic hemolysis,
the rupture of the red cell can be shown bv rapid photography to occur, not
explosively, but by slow swelling.

Dr. Nickerson; Judging from the appearance of the bud-initial, as seen
in the dark-field time-lapse photomicrographs, its resemblance to a spherical
protoplast— albeit of natural origin— leads me to suspect that the plasma mem-
brane is intact.

Dr. Ducoff (University of Illinois): How rapid is the change from the
filamentous to the non-filamentous form? Is there a "breakup" comparable to
that seen in induced bacterial filaments?

Dr. Nickerson: I have never observed a "breakup" in the filamentous
veast comparable to that which we have studied in bacterial filaments. Under
appropriate conditions, an inoculum of filamentous cells of a divisionless strain
is "diluted out" by budding, to give a population of largely normal yeast cells.
These revert to their filamentous phenotype when the condition promoting
cellular division is removed.

Dr. Preston (University of Leeds): How sure are you that the hole asso-
ciated with a bud scar is actually blown out? In the algae we are studying,
the production of zoospores is followed by the production of a hole in the wall
of the parent cell. Stripping through the wall, and electron microscope exami-
nation, suggest that these holes are made by progressive dissolution of the wall;
only the last few lamellae are blown out by. turgor pressure.

Dr. Nickerson: Our use of the term "explosive blowout" to describe the
budding of yeast is based on two independent observations. First, the lapse
of time between consecutive frames showing an intact wall in one and a cell
with an extruded bud-initial in the second, is 30 seconds. Second, the orienta-
tion of fibrils in the wall components at the site of budding can only be

228 MACROMOLECULAR COMPLEXES

achieved by outward thrust of Hquid at considerable velocity. We are cur-
rently attempting to calculate a minimum velocity for the thrust to result m
the observed crystalline orientation.

Dr Bergeron (Brookhaven National Laboratory): Is the cross-lmkmg m
the bud scar irreversible with regard to the later formation of overlappmg

"" Dr Nickerson: Apparently so. The photographs of Barton (1950), who
first observed bud scars, show no overlapping, even on cells with numerous
scars. Our electron micrographs also show no overlapping of bud scars.

Cellulose-Protein Complexes in Plant Cell Walls

R. D. Preston ^

Introduction

The walls of living plant cells consist in the main of a mixture
of polysaccharides and their derivatives organized in such a way as
to be chemically stable while allowing the mutual displacements
necessary in the dimensional changes associated with growth. Ex-
cept in the fungi, which will not be dealt with here, the structural
element of these walls is the polysaccharide, cellulose, whose con-
stitution will be examined below. This substance takes the form of
long threads known as microfibrils, some 100-200 A wide (depend-
ing on the plant species— see refs. in Preston, 1952) and about half
as thick (Preston, 1951) (Fig. 1), which are often oriented, either
very roughly or with precision ( Fig. 2 ) . Both the synthesis and the
orientation of these microfibrils must involve proteins, and it is the
aim of this article to inquire how far present evidence will allow us
to go in defining the cellulose-protein complex which may be in-
volved. With few exceptions, those cell walls which have been
examined have been found to contain proteins (Thimann and
Bonner, 1933; Allsopp and Misra, 1940; Myers et al, 1956; Bishop
et al., 1958; Northcote et al., 1958) and, though in some preparations
much of the protein is derived from cytoplasmic debris, there is
no doubt that some part of it occurs in the wall itself. Occasional
reference has been made to the inclusion in cell-wall preparations
of electron microscopically visible particles which appear to be
cytoplasmic ( Myers et al., 1956 ) .

It seems probable that the protein content of the wall sensu
strictii is low, perhaps not more than a few per cent. A cellulose-
protein complex is to be expected only during cellulose deposition,
after which most of the protein will be resorbed. Even this small
percentage remainder in the wall is therefore significant.

' University of Leeds, Leeds, England.

229

230

MACROMOLECULAR COMPLEXES

Fiq. 1. Isolated microfibrils of Valonia cellulose. Shadowed Pd-Au. X45,000.

CELLULOSE-PROTEIN COMPLEXES

23]

1

-^^

>i

Fig. 2. Three wall lamellae in Valonia ventricosa. Note the two abundant
sets of microfibrils lying almost at 90° to each other and a third, less abundant,
set lying obliquely. The lamellae are slightly torn. Shadowed Pd-Au. X22,500.

232 MACROMOLECULAR COMPLEXES

Polysaccharide-protein complexes are beginning to acquire sig-
nificance both in plant and in animal physiology; whether the struc-
tural component is dominantlv protein or dominantly polysaccharide,
it may well be that the same principles are involved. It is proposed
here, therefore, to examine evidence from both fields.

Although the principal structural components of animal tissues
are undoubtedly protein, there are many recorded observations
which show that polysaccharides of several kinds are involved, some-
times in considerable abundance. An outstanding case is the de-
velopment of tunicin, indistinguishable from plant cellulose, in the
tunica of sea squirts, but many cases of polvsaccharide association
with proteins have been reported. These are often acid polysac-
charides, the acidic group being represented by hvaluronic acid,
chondroitin sulfuric acid, and some partially sulfated variants of
these two. Some, however, are devoid of hexosamine or uronic acids.
Grassman and Schleich (1935) identified glucose and galactose in
hydrolysates of skin extracts, an observation confirmed by Gross
et al. (1952), who added mannose as a constituent sugar. More
recently, other polysaccharides rich in glucose have been reported
from mammalian sources. Dische and Borenfreund (1954) claim to
have demonstrated the presence in lens capsule of a polysaccharide
containing galactose and glucose in the ratio 3:2, and Dische and
Zelemenis ( 1955 ) found a similar polysaccharide in the vitreous
body. Again, polysaccharide material containing galactose, man-
nose, and glucose in the ratio 6:1:2 has been isolated from kidney
( Dohlman and Balazs, 1955 ) . It may be, however, that these repre-
sent mixtures of polysaccharides, since Dische et al. ( 1958 ) have
found the composition of similar polysaccharides in the bones of
cattle to vary with the age of the animal. Indeed, a polysaccharide
fraction in abdominal subcutaneous tissue has been shown to con-
sist of pure galactans and pure mannans (Consden and Bird, 1954).
In all these cases, the polysaccharide is associated with a protein
which is, in every case, collagen. There is no direct evidence that
glucans alone exist in any of these tissues, but the hexasaccharide
isolated from umbilical cord by Akiya and Tomoda (1955) and
considered by them to consist of /3-D-glucose residues joined bv 1-4
links, might be a precursor, or a degradation product, of cellulose.

There is some evidence that these polvsaccharides pla\' an im-
portant role in the stabilization of the associated collagen. Jackson

CELLULOSE-PROTEIN COMPLEXES 233

(1953) and Wood (1954) have shown that a polysaccharide frac-
tion resembhng chondroitin sulfate is essential for the maintenance
of some of the physical properties of collagen fibers. Similarly, Hall
et al. (1952) and Hall (1955) have demonstrated that the removal
of polysaccharides from elastic fibers leads to a marked change in
the resistance of the fiber to chemical and enzymic attack. More-
over, for these fibers, particularly in the aorta, it has been claimed
(Saxl, 1957) that pathological degradation is bound up with the
topographical concentration of polysaccharides. Excessive resist-
ance of connective tissue fibers to chemical attack has been ascribed
to variations in the content of associated polysaccharide (Hall and
Reed, 1957; Keech, 1958). Stabilization of this kind is regarded by
Yu and Blumenthal (1958) as due to the existence of complexes
between proteins and pol\ saccharides.

The more recent evidence in favor of a close relationship between
cellulose and proteins in the cell walls of plants comes from four
main sources. These will be dealt with separately below.

Cellulose-Protein Complexes in Mammalian Tissue

The recent demonstrations of the presence of cellulose in mam-
malian tissue (Hall and Reed, 1957) and of its association with pro-
tein in healthy and pathological skin (Hall et al, 1960; Hall and
Saxl, 1961 ) gives a specific lead. The fibers concerned were first
observed as highly birefringent inclusions in the alkaline degrada-
tion products of collagen (Burton et al, 1955; Hall et al, 1960).
They were reported to be rich in polysaccharide and highly resistant
to chemical attack, and the polysaccharide has been identified as
cellulose for the following reasons : ( 1 ) The x-ray diagram is indis-
tinguishable from that of cotton cellulose. (2) The fibrils of the
polysaccharide appear to be uniaxial with refractive indices of 1.596
and 1.525, very close to those of ramie fibers. (3) Paper-partition
chromatography of hydrolysates reveals the presence of only traces
of sugars other than glucose, and in particular no traces were found
of glucosamine, galactosamine, glucuronic acid, or galacturonic acid.
(4) The polysaccharide is soluble in Schweizer's reagent but not in
hot water, alkali, or boiling dilute sulfuric acid. (5) The electron
microscope appearance and the electron-diffraction diagram closely
resemble those of cellulose.

234 MACROMOLECULAR COMPLEXES

The fibers consist of a core of protein, sometimes completely in-
cased in polysaccharide. When the coating is not complete (Fig. 3),
however, it can be seen to consist of fibrils running spirally round
the fiber. The inclination of the fibrils to fiber length varies broadly
around 72° (cf., the value of 75° in elongated, growing plant cells-
Preston, 1947; Preston and Wardrop, 1949; Middlebrook and Pres-
ton, 1952).

Fig. 3. Antisotropic fiber from bovine epidermis with incomplete polysaccha-
ride sheath, as seen between crossed nicols. The polysaccharide appears
bright, the protein core dark. X520.

The composition of the protein core difiers from that in collagen
and elastin. Like elastin, the protein is rich in glycine, proline, and
valine; it contains some hydroxyproline but less than does collagen.
On the other hand, appreciable quantities of acidic and basic amino
acids are present, a feature more characteristic of collagen.

It is not, of course, certain that the cellulose is svnthesized in situ;
the conclusion does, however, seem inescapable that the protein, a
degenerative product of collagen, is involved in at least the orienta-
tion of the cellulose.

CELLULOSE-PROTEIN COMPLEXES

235

The Cellulose of Plant Cell Walls

The structure of cellulose itself is not without significance in this
respect. Classically, cellulose is regarded as a polvmer of /3-D-glucose
residues joined in long chains by 1-4 links. Over certain parts of
their length these chains lie parallel to each other and regularly
spaced in a lattice dimensioned as in Fig. 4, which is now known
to be only approximately correct (Honjo and Watanabe, 1958; Pres-
ton, 1958). The lattice is such as to yield a more or less well-defined
diffraction diagram when a beam of X-radiation or electrons is

Fig. 4. A unit cell of cellulose after Meyer and Mark. Dimensions are
given in Angstrom units. The black circles represent oxygen atoms.

passed through it. From such diagrams the spacing between promi-
nent planes in the lattice may be calculated, and particular note may
be taken of the spacings at 3.9, 5.4, and 6.1 A between planes lying
parallel to chain direction, and at 2.56 A from planes lying normal
to chain direction (Figs. 4, 5). The lattice occurs, of course, inside
the microfibril, which therefore contains a number of molecular
chains (of the order of 200) lying parallel to its length.

In general, celluloses isolated from plant cell walls include poly-
saccharides containing sugars other than glucose, and it is now
known, that these are not always contaminants from the associated
amorphous non-microfibrillar comj^onents. In a series of analyses
controlled by electron microscope observation of wall extracts, Cron-

236 MACROMOLECULAR COMPLEXES

Fig. 5. Electron diffraction diagram of Valonia cellulose; orientation correct
for comparison with Fig. 4.

shaw et al. (1958) have shown that the microfibrils themselves
usually contain sugars other than glucose, and some of the results
of these analyses are presented in Table 1. These fully confirm the
opinion, which has been gaining ground for a number of years, that
cellulose is a mixed polysaccharide. With two species (Valonia and
Cladophora) and possibly a third (Chaetomorpha) (Table 1),
sugars other than glucose have not been detected. The cellulose
involved appears to be a "true" substance in the strict chemical and
crystallographic sense, and it has been suggested that it should be
distinguished as eucellidose (Myers and Preston, 1959). The struc-
ture proposed for cellulose microfibrils is illustrated in Fig. 6. This
represents diagrammatically a transverse section of a microfibril
(Preston and Cronshaw, 1958; Preston, 1958). The short oblique
lines represent the projection of chain molecules in this plane. The
central core, represented as a lattice, is crystalline and is surrounded
by a cortex in which the arrangement of the chains is less regular.
Within this cortex occur chains of glucose residues ( solid lines ) and
of other sugars ( broken hues ) . These may be aggregated together
to form larger groups, within which the microfibrils are still dis-
tinguishable. This has led Frey-Wyssling (1959) to refer to the
single structures as elementary fibrils and to the aggregates as micro-

CELLULOSE-PROTEIN COMPLEXES

237

fibrils. It seems unnecessary to coin a new terminology in order to
include the chance aggregates of variable size, and the term "micro-
fibrils" will be used here as referring to the unit of structure only.

TABLE 1
Composition of Celluloses from Various Sources

Source

Composition

Valonia ventricosa

S

Glucose

Cladophora rupestris

S

Glucose

Chaetomorpha melagot

litim

S
W

Glucose
Arabinose

Enteromorpha sp.

S
M
M

Glucose

Xylose

Rhamnose

Ulva lactuca

S
S

Glucose
Xylose

Halidrys siliquosa and

all Fucales

s

w

w

Glucose

Xylose

Fucose

Laminaria digitata and L. saccharina

s

M

Glucose
Uronic acid

Ptilota plumosa

s

w

w

Glucose

Galactose

Xylose

Gri-ffithsia -flosctdosa

s

w

w

Glucose
Galactose

Xylose

Rhodymenia palmata

s
s

Glucose
Xylose

Porphyra sp.

s

Mannose

S. M., W refer to the intensity of the
chromatogram. S = strong; M = medium

spot on the
W = weak.

Any protein which complexes with this structure must be able to
associate with a variety of sugars just as it appears to do in animals.
Such a complex will be possible presumably only if the surfaces
of the microfibril and the protein have structural features in com-
mon, and there is some crystallographic evidence that this may be so.

238

MACROMOLECULAR COMPLEXES

y

."-7^'-

/ ;
' I / / /

/ / / / •'

.^^^

7^^

-y^

7^

/

/

^-^

-7^

7^

7^

7^^^

J->

7^

■7^

;^

/ /

/ / /

Fig. 6. Diagrammatic representation of a cellulose microfibril in transverse
section. Oblique solid lines, projection of polyglucose chains; oblique broken
lines, projection of chains containing sugars other than glucose. The central
lattice represents the crystalline core, and the longer edges of the core are
parallel to the planes of 6.1 -A spacing.

When celluloses are immersed in solutions of salts of heavy
metals, the cation is rapidly adsorbed according to a Langmuir
isotherm (Belford and Preston, 1959; Belford et ah, 1958; Belford
et al., 1957, 1959). Parallel bundles of the microfibrils then yield
one of two electron-diffraction diagrams, of which one is illustrated
in Fig. 7. This is quite different from the diagram yielded by cellu-
lose alone (Fig. 5) and has been interpreted as arising from a two-
dimensional array of heavy metal ions (Fig. 8). The diagram, and
therefore the lattice producing it, is the same whatever metal is
used and whatever the source of the cellulose. Two txpes of array
have been calculated, defined by the parameters:

Type I p = 6.15 A, q = 7.05 A, 0 = 90°

Type II p = 7.32 A, q = 5.68 A, <f> = 87°-90° (variable)

which explain quantitatively the positions of all the arcs in the
diagrams. These parameters do not correspond to those in crystals
of the metals, in any metal salt, or in cellulose ( see spacings quoted
on p. 235). Since the only surfaces available for adsorption are those
of the microfibrils, this two-dimensional array must lie on these sur-
faces, and the mutual arrangement of the ions must be governed
by the surface. The only possible inference seems to be that the
molecular chains lying on the surface of the microfibrils are folded
in one of two wavs, bringing points of attraction into a lattice de-

CELLULOSE-PROTEIN COMPLEXES 239

fined by these parameters, in configurations demanded by surface
forces.

The significance of these observations is that precisely the same
observations have been made also with collagen fibers from the
tendon of Achilles. The surfaces of collagen and cellulose micro-
fibrils, therefore, have in common an array of active centers defined
by the parameters given above. It is not yet known whether or not
other proteins behave similarly; these form the object of further
investigations now in progress.

<7
Fig. 7 Fig. 8

Fig. 7. Electron diffraction diagram of the Cu-celluiose complex in Pseu-
dotsuga cellulose; the constituent microfibrils lie parallel to the longer edge of
the page. Contrast with Fig. 5, in which the orientation is the same.

Fig. 8. The two-dimensional array of Cu atoms presumed to lie on the
surface of cellulose microfibrils in the Cu-cellulose complex. The parameter p
lies parallel to microfibril length (for values of p and q, see text).

The Stability of Protein-Cellulose Complexes in Plants

Although little specific attention has been given to any mutual
stabilization between proteins and polysaccharides in plants, there
are indications in the hterature of such a stabilization, over and
above the frequent statements that high sugar content in a tissue
appears to prevent protein degradation. When leaves fall on the
soil, die, and are buried, then both the proteins and the polysac-
charides break down rapidly under the influence of microorganisms

240 MACROMOLECULAR COMPLEXES

and other agencies. It has been shown by Handley (1954), however,
that, under the same conditions, leaves which have died on the plant
behave very differently. Although the vascular bundles largely dis-
appear, the cells of the epidermis and mesophyll, which might have
been considered much less resistant, remain almost intact. Dealing
with mor soils, Handley remarks, "Serial sections (of recognizable
fragments of litter of Calltma vulgaris— present author) indicated
that vascular tissue disappears first, leaving a residue of leaf meso-
phyll tissue which ultimately becomes an amorphous mass contain-
ing mesophyll cell walls apparently coated with some protective
material. Similar findings were obtained for . . . beech litter, Nor-
way spruce litter and Abies pinsapo litter. . . . From serial sections
of shoots of Calluna vulgaris, having both dead and living leaves,
changes have been observed in the mesophyll tissue when the leaf
dies, while still on the living plant, which give the mesophyll tissue
the appearance it retains until it becomes amorphous material in the
mor. The material appearing . . . would seem to be that which
protects the cellulose wall. . . . Experimental evidence is put for-
ward to show that the essential part of the change occurring when
the leaf dies is a precipitation or stabilisation of cytoplasmic pro-
tein."

From this point of view, the recent findings of Steward and Pol-
lard (1956, 1957) and Pollard and Steward (1959) are not without
significance. They have shown that, when cultures of carrot-root
phloem and potato tubers are fed with C^^-labeled proline, the pro-
line can be recovered in a protein which also contains C^^ hydroxy-
proline, clearly produced from the proline. Apart from this, the
active carbon appears in no other amino acid, nor does it appear as
COo. The protein concerned is therefore stable and does not take
part in the normal protein turnover in the plant. Moreover, the
protein is not part of any of the cytoplasmic particles which Steward
and Pollard examined, and they conclude, therefore, that this pro-
tein occurs in the ground cytoplasm. Here, then, is a protein of
which at least the proline- and hydroxyproline-containing portion
is guarded from the metabolic machinery of the plant. One possibil-
ity is that this is the protein which is closely associated with the
developing cell wall. This is the only protein extracted from plants
to contain hydroxyproline, and recalls immediately the composition
of the protein in the cellulose-protein complex described by Hall,
and discussed in an earlier section of this paper.

CELLULOSE-PROTEIN COMPLEXES 241

Electron Microscope Observations
of Cell Wall-Cytoplasmic Relations

There is thus some evidence that cellulose and protein may be
associated in plants just as they are now known to be in animal skin.
We come now to the electron microscope evidence which seems
convincing.

Electron microscope evidence for the interpenetration of walls
and the proteins of cytoplasm comes from many sources. Miihle-
thaler ( 1950 ) was the first to show that in primary walls the cellu-
lose microfibrils are twisted round each other, and to state clearly
that this could not happen if the wall and cytoplasm were separated
by an interface. Similar observations can be made even with sec-
ondary walls, and the walls of the seaweeds Valonia, Cladophora,
and Chaetomorpha have proved to be particularly revealing (Frei
and Preston, 1961).

In these three species, microfibrillar orientation reaches its most
perfect state. The walls consist of many lamellae, each normally one
or two microfibrils thick, and each consisting of a compact set of
parallel microfibrils. These microfibrils lie, moreover, with the
planes of 6.1- A spacing (Fig. 2) more or less parallel to the wall.
The microfibril direction alternates from one lamella to the next
through an angle which can be as low as 60° but is usually rather
less than a right angle so that odd lamellae show one orientation and
even lamellae the other. In Valonia, in some Cladophora spp., and
in some Chaetomorpha spp. (but not all), a third orientation also
occurs, much more rarely than the other two. A model of the whole
wall of a Valonia vesicle is illustrated in Fig. 9 ( Cronshaw and Pres-
ton, 1958 ) . This regular alternation from one direction to the other,
through an angle which is not the same from point to point in a cell,
cannot be ascribed to twinning. It must reflect changes in the pro-
teins at the surface of the cytoplasm, and recalls the somewhat simi-
lar arrangement of protein fibrils found in the cuticle of earthworm
by Rudall and Reed ( see Astbury, 1949 ) . In all lamellae, however,
the microfibrils are frequently twisted around each other. Moreover,
the switch from one orientation to the other is never clean and
abrupt. Microfibrils from one lamella may pass through the next
lamella and become part of the next-but-one lamella with the same
orientation; microfibrils from one lamella may interweave with those

242

MACROMOLECULAR COMPLEXES

Fig. 9. Model of wall structure in a vesicle of Valonia ventricosa. The black
and white tapes represent the run of the two major sets of microfibrils, one in
a slow left-hand spiral and the other in a steep left-hand spiral. The gray tape,
a fairly steep right-hand spiral, corresponds to the third orientation. One "pole"
lies in the center of this figure, the other lying on the far side of the model.

of the next; and microfibrils from one lamella mav turn through 90°
and become part of the next lamella (Frei and Preston, 1961). This
seems to be clear evidence that the wall and the cytoplasm must be
intermixed during microfibril synthesis. Examples of these inter-
weavings are given in Fig. 10.

The strongest evidence for the existence of a cellulose-protein
complex comes, however, from examination of the development of
a new cell wall over the surface of the spores of these plants, and
from an examination of lamellae just being deposited in adult cells
(Nicolai, 1957; Nicolai and Preston, 1959; Frei and Preston, 1961;
Preston and Kuyper, 1951; Preston et al, 1953). The first wall
lamella deposited around a zoospore, gamete, or zvgote carries

CELLULOSE-PROTEIN COMPLEXES

243

Fig. 10. Electron micrograph of lamellae stripped from the wall of Chaefo-
morpha melagonium. Note: (1) The apparent "stray" microfibrils are in situ as
deposited because they are attached to a lamella at both ends. (2) One set of
microfibrils interweaves with the other. The run of the "strays" proves that this
interweaving is real and is not due to accidental stripping away of microfibrils.
Shadowed Pt. XI 7,000.

microfibrils which are oriented at random. In the next one, deposited
to the inside of the first (Fig. 11), the microfibrils are oriented
transversely to an axis passing through the point of attachment to
the substrate. Microfibril synthesis and microfibril orientation seem,
therefore, to involve different mechanisms. The microfibrils of this
second lamella, which run in slow spirals round the sporeling, are
preceded by the development of coarse bands with similar orienta-
tion, which appear to be cytoplasmic (Fig. 11). Microfibrils de-
velop within these bands; they are first observed at a point on
what may be called the equator and appear to develop by end-
synthesis in such a way that they remain within the coarse bands.
In the third lamella, the microfibrils run from pole to pole of the

244

MACROMOLECULAR COMPLEXES

Fig. 11. Electron micrograph of the two-doy-old sporeling of Chaefomorpha
melagonium. The surface is covered with coarse bands oriented transversely
(and therefore in a slov\/ spiral opening from one "pole" at the "base" and
closing in toward the sprouting "tip"). Occasional microfibrils can already be
seen toward the lower right.

CELLULOSE-PROTEIN COMPLEXES

245

Fig. 12. Electron micrograph of innermost lamella of wall of Valonia ven-
tricosa. The microfibrils appear to radiate from a granular mass. Shadowed
Pd-Au. X45,000.

246

MACROMOLECULAR COMPLEXES

CELLULOSE-PROTEIN COMPLEXES J247

spiral thus produced (and therefore at right angles to the micro-
fibrils of this second lamella) and are again preceded by coarse
bands appropriately aligned.

Examination of new lamellae in course of deposition has proved
even more critical. The first observations were made on the giant
vesicles of Valonia (Preston and Kuvper, 1951; Preston et ah, 1953).
The innermost lamella of the wall was stripped off^ by means of
Cellotape, and the two types of critical observation made are illus-
trated in Figs. 12 and 13. There is no doubt about the latter (Fig.
13); it surely illustrates an undisturbed lamella. In the former (Fig.
12), however, the new lamella is so delicate that some distortion
might have occurred. The impression given is that microfibrils
project from a cluster of granules each about 500 A in diameter, and
the tentative explanation made at that time was that each cluster
represented an "island of synthesis." Microfibrils would then appear
to be built by end-svnthesis ( a method of synthesis later supported
by Colvin et ah, 1957), association between microfibril and enzyme
occurring only at one end. The mechanism of orientation remained
obscure, but observations of the tvpe shown in Fig. 13 suggested
that orientation occurred in the cytoplasm before the microfibrils
of the new lamella were produced.

More recent observations of Chaetomovpha, using more delicate
techniques, are more convincing on this latter point ( Frei and Pres-
ton, 1961). In these, the filaments are either fixed directly in an
osmium fixative or plasmolyzed in 0.5M sucrose before fixation, but
in either case the appearance to be described is the same. The
lamellae are then teased off in water, using fine needles under a
binocular microscope. Fig. 14 illustrates the type of observation to
which attention is now called. This is a frequent but not a constant
observation; more frequently the lamella is so densely covered by
cytoplasm that no details are visible. Occasionally the lamella is
completely free of cytoplasm. Fig. 14 shows clearly, however, ag-
gregates of dense granular bodies lying in files on the wall lamella,
the long axes of the files lying approximately at right angles to the
microfibrils, i.e., in the direction which the fibrils of a new lamella
would take. It is to be emphasized that the lamella observed here

Fig. 13. Electron micrograph of innermost lamella of wall of Valonia ventri-
cosa. Granular material, evidently of cytoplasmic origin, has taken up an
orientation which the next layer of microfibrils would have adopted. Shadowed
Pd-Au. X 22,500.

248

MACROMOLECULAR COMPLEXES

Fig. 14. Electron micrograph of the innermost layer of Chaetomorpha me-
lagonium wall. Note the linear files of aggregated granules oriented at right
angles to the microfibrils of the last lamella deposited and therefore parallel
to those of a lamella about to be deposited. These are the remnants of a
much more extensive system. Shadowed Pt. X 20,400.

is the innermost one lying next to the cytoplasm, and very few
microfibrils of the next lamella have yet been synthesized. These files
cannot therefore represent debris lying between microfibrils which
have been stripped away. When microfibrils of the new lamella do
appear, they appear to be embedded in this material and sometimes

CELLULOSE-PROTEIN COMPLEXES 249

to project from one end. One example of this, marked by an arrow,
is visible in Fig. 14.

When similar preparations are made of the innermost lamellae
of cells which have previouslv been plasmolvzed, these oriented
cytoplasmic aggregates remain attached to the lamella. They are
therefore more closely attached to the wall than is the bulk of the
cytoplasm, which retracts on plasmolysis. A new wall lamella does
not develop under these conditions, presumablv because of the with-
drawal of the supply of raw materials necessary. The new cytoplas-
mic surface does produce a new cell wall, but the microfibrils are
not oriented. When, however, filaments plasmolyzed in, say, 0.8M
sucrose are returned to sea water, they continue to grow normally.
After a few days in sea water, wall layers similar to that illustrated
in Fig. 14 can be found deep in the wall, covered by more recent
wall lamellae with the usual alternations in microfibril orientation.
A cytoplasmic surface, withdrawn from the wall and at that time
producing microfibrils oriented at random, can thus reconstitute the
whole orientation mechanism.

Discussion and Conclusions

Although, so far as the writer is aware, there is no direct bio-
chemical evidence of close association between cellulose and pro-
tein in plant cell walls, the morphological observations and struc-
tural determinations referred to above, together with the evidence
from parallel conditions in animals, however, is strongly suggestive.
It would appear to be established that microfibrils are synthesized
by what appears to be end-synthesis and that, at least during and
immediately after synthesis, the part of the microfibril concerned is
embedded in an aggregate of granules which may be enzyme com-
plexes and therefore largely protein. Microfibrils are occasionally ob-
served, on the surface of new wall lamellae, which have long taper-
ing ends running down to the limits of visibility in the electron
microscope. It would seem, therefore, possible that synthesis occurs
over long lengths of microfibril and is end-synthesis of individual
molecular chains rather than of whole microfibrils.

It would certainly be premature to attempt to give even a purely
morphological picture of the cellulose-protein complex during syn-
thesis. Appearances do suggest, however, that a cytoplasmic surface
which is to produce a wall lamella first becomes organized into

250 MACROMOLECULAR COMPLEXES

coarse bands oriented with respect to the cell axis. At some point
in the band— which presumably already contains a microfibril end-
synthesis begins. As this proceeds, the microfibril end moves for-
ward through the coarse band and new parts of the band take over
synthesis. At this time, the coarse band has a central core of cellu-
lose and an outer sheath of protein— the reverse of the condition in
animal skin. It is not clear whether the diameter of the microfibril
( which varies only within rather narrow limits ) is a function of the
size of the protein aggregates or is controlled thermodynamicallv as
suggested by the work of Fiirth ( 1955 ) .

Eventually the cellulose-protein complex must disintegrate, so
that the microfibril lies free on the wall surface.

When these protein aggregates take up a new alignment in prep-
aration for the deposition of the next lamella, it would then follow
that synthesis of the new lamella could begin only at microfibril
ends of the lamellae just completed. This would explain the manv
interweavings between microfibrils of different lamellae. This does
not in any sense "explain" the specific orientations of cellulose micro-
fibrils, but merely puts the mechanism one step further back into
the cytoplasm. In spite of earlier criticisms of orientation by proto-
plasmic streaming, it seems possible that in some plants there is
some association between streaming and orientation and that this is
achieved through orientation of proteins (Probine and Preston,
1958). It is not at the moment possible to decide whether streaming
is involved in the cases discussed in this paper.

These considerations refer specifically, of course, only to those
few algae in which microfibrillar orientation is so perfect that the
necessary critical observations have been made. It seems unlikely,
however, that the basic principles will be different in other plants.
The present situation is, in brief, that such evidence as is available
points toward the type of mechanism which mav be involved in wall
synthesis. Further progress clearly demands a thorough biochemical
study of the proteins in or near the cytoplasmic surface.

References

Akiya, a., and M. Tomoda. 1956. /. Pharm. Soc. Japan 76: 1092. (Cited in Hall

et al, 1960.)
Allsopp, a., and P. Misra. 1940. The constitution of the cambium, the new wood

and the mature sapwood of the common ash, the common elm and the Scotch

pine. Biochem. J. 34: 1078-1084.

CELLULOSE-PROTEIN COMPLEXES 251

AsTBURY, W. T. 1949. The structure of biological tissues as revealed by x-ray diffrac-
tion analysis and electron microscopy. Brit. }. Radiol. 22: 335-365.

Belford, D. S., a. Myers, and R. D. Preston. 1958. Electron diffraction study of
adsorbed metal ions on the surface of cellulose microfibrils. Nature 181: 1516-
1518.

Belford, D. S., and R. D. Preston. 1959. Ordered adsorption of metal ions on
cellulose. Biochim. et Biophys. Acta. (In press.)

Belford, D. S., R. D. Preston, C. D. Cook, and E. H. Nevard. 1957. Timber
preservation by copper compounds. Nature 180: 1081-1083.

Belfoiuj, D. S., R. D. Preston, C. D. Cook, and E. H. Nevard. 1959. The impreg-
nation of timber by water-borne preservatives. I. General survey. /. Appl. Chem.
(London) 9: 192-200.

Bishop, C. T., S. T. Bayley, and G. Setterfield. 1958. Chemical constitution of
the primary cell walls of Avena coleoptiles. Plant Physiol. 33: 283-289.

Burton, D., E). A. Hall, M. K. Keech, R. Reed, H. Saxl, R. E. Tunbridge, and
M. J. Wood. 1955. Apparent transformation of collagen fibrils into 'elastin.'
Nature 176: 966-969.

CoLviN, J. R., S. T. Bayley, and M. Beer. 1957. The growth of cellulose micro-
fibrils from Acetobacter xylinum. Biochim. et Biophys. Acta 23: 652-653.

CoNSDEN, R., and R. Bird. 1954. The carbohydrate of connective tissue. Nature
173: 996-997.

Cronshaw, J., A. Myers, and R. D. Preston. 1958. A chemical and physical in-
vestigation of the cell walls of some marine algae. Biochim. et Biophys. Acta
27: 89-103.

Cronshaw, J., and R. D. Preston. 1958. A re-e.xamination of the fine structure of
the walls of vesicles of the green alga Valonia. Proc. Roy. Soc. (London) B 148:
137-148.

DiscHE, A., and G. Zelemenis. 1955. A.M. A. Arch. Ophthalmol. 54: 528. (Cited in
Hall et al, 1960. )

Dische, Z., and E. Borenfreund. 1954. Am. }. Ophthalmol. 38: 165. (Cited in
Hall et al, 1960. )

Dische, Z., A. Danilczenko, and G. Zelemenis. 1958. In Chemistry and Biology
of Mucopolysaccharides. Ciba Foundation. J. & A. Churchill Ltd., London.

Dohlman, C. H., and E. A. Balazs. 1955. Arch. Biochem. Biophys. 57: 44. (Cited
in Hall et al, 1960. )

Frei, E., and R. D. Preston. 1961. Proc. Roy. Soc. (London) B. (In Press.)

Frey-Wyssling, a. 1959. Die Pftanzliche Zellwand. Springer Verlag, Berlin.

FiJRTH, R. 1955. Fundamental lengths in physics and biology. Exptl. Med. Surg.
13: 17-21.

Grassman, W., and H. Schleich. 1935. Uber den Kohlenhydratgehalt des Kollagens
II. Biochem. Z. 277: 320-328.

Gross, J., J. H. Highberger, and F. O. Schmitt. 1952. Proc. Soc. Exptl. Biol. Med.
80: 462. ( Cited in Hall et al, 1959. )

Hall, D. A. 1955. The reaction between elastase and elastic tissue. Biochem. }.
59: 459-465.

Hall, D. A., F. Happey, P. F. Lloyd, and H. Saxl. 1960. Oriented cellulose as a
component of mammalian tissue. Proc. Roy. Soc. (London) B 151: 497-516.

Hall, D. A., and R. Reed. 1957. Hydroxyproline and thermal stability of collagen.
Nature 180: 243.

Hall, D. A., R. Reed, and R. E. Tunbridge. 1952. Structure of elastic tissue.
Nature 170: 264-266.

Hall, D. A., and H. Saxl. 1961. Submitted to Proc. Roy. Soc. (London) B.

Handley, W. R. C. 1954. Mull and mor formation in relation to forest soils. For-
estry Commission Bull No. 23. H. M. Stationery Office, London.

HoNjo, G., and M. Watanabe. 1958. Examination of cellulose fibre by the low-
temperature method of electron diffraction and electron microscopy. Nature
181:326.

252 MACROMOLECULAR COMPLEXES

Jackson, D. S. 1953. The nature of collagen-chondroitin sulphate hnkages in tendon.
Biochem. J. 56: 699-703.

Keech, M. K. 1958. Ann. Rheumatic Diseases 17: 23. (Cited in Hall et al, 1960.)

MiDDLEBROOK, Mavis J., and R. D. Preston. 1952. Spiral growth and spiral struc-
ture. III. Wall structure in the growth zone of Phycomyces. Biochim. et Biophijs.
Acta 9: 32-48.

MiJHLETHALER, K. 1950. Elektronmikroskopische Untersuchungen iiber den Feinbau
und das Wachstum in Mais- und Haferkoleoptilen. Ber. schweiz. botan. Ges.
60: 614.

Myers, A., and R. D. Preston. 1960. Fine structure in the red algae. II. The struc-
ture of the cell wall of Rhodymenia palmata. Proc. Roy. Soc. (London) B. (In
press. )

Myers, A., R. D. Preston, and G. W. Ripley. 1956. Fine structure in the red algae.
I. X-ray and electron microscope investigation of Griffithsia floscidosa. Proc. Roy.
Soc. (London) B 144: 450-459.

NicoLAi, E. 1957. Wall deposition in Chaetomorpha melagonium (Cladophorales).
Nature 180: 491-492.

NicoLAi, E., and R. D. Preston. 1959. Cell wall structure in the Chlorophyceae.
III. Differences in structure and development in the Cladophoraceae. Proc. Roy.
Soc. (London) B. 151: 244-255.

NoRTHCOTE, D. H., K. W. GouLDiNG, and R. W. Horne. 1958. The chemical com-
position and structure of the cell wall of Chlorella pyrenoidosa. Biochem. ]. 70:
391-397.

Pollard, J. K., and F. C. Steward. 1959. The use of C' ^-labelled proline by grow-
ing cells; its conversion to protein and hydroxyproline. /. Exp. Botany 10: 17-32.

Preston, R. D. 1947. The fine structure of the wall of the conifer tracheid. II.
Optical properties of dissected cells in Pinus insignis. Proc. Roy. Soc. (London)
B 134: 202-218.

Preston, R. D. 1951. Fibrillar units in the structure of native cellulose. In The size
and shape factor in colloidal systems. Discussions Faraday Soc. No. 11: 165-170.

Preston, R. D. 1952. Molecular Architecture of Plant Cell Walls. Chapman and
Hall, Ltd., London.

Preston, R. D. 1958. Wall organization in plant cells. Ann. Rev. Cytol. 8: 33-60.

Preston, R. D., and J. Cronshaw. 1958. Constitution of the fibrillar and non-
fibrillar components of the walls of Valonia ventricosa. Nature 181: 248-250.

Preston, R. D., and B. Kuyper. 1951. Electron microscopic investigations of the
walls of green algae. I. A preliminary account of wall lamellation and deposition
in Valonia ventricosa. J. Exp. Botany 2: 247-255.

Preston, R. D., E. Nicolai, and B. Kuyper. 1953. Electron microscopic investiga-
tions of the walls of green algae. II. The cytoplasm-wall relationship in freeze-
dried Valonia macrophysa. }. Exp. Botany 4: 40-43.

Preston, R. D., and A. B. Wardrop. 1949. The submicroscopic organisation of the
walls of conifer cambium. Biochim. et Biophys. Acta 3: 549-559.

Probine, M. C, and R. D. Preston. 1958. Protoplasmic streaming and wall struc-
ture in Nitella. Nature 182: 1657-1658.

Saxl, H. 1957. Gerontologia. 1: 142. (Cited in Hall et al, 1960.)

Steward, F. C, and J. K. Pollard. 1956. The amino acid composition of the total
protein of growing and non-growing tissues. Plant Physiol. 31: ix.

Steward, F. C., and J. K. Pollard. 1957. Prohne and hydroxyproline in the protein
of growing cells, ibid. 32: lii-fiii.

Thimann, K. v., and J. Bonner. 1933. The mechanism of the action of the growth
substance of plants. Proc. Roy. Soc. (London) B 113: 126-149.

Wood, G. C. 1954. Some tensile properties of elastic tissue. Biochim. et Biophys.
Acta 15: 311-324.

Yu, S. Y., and H. T. Blumenthal. 1958. /. Gerontol. 13:366. (Cited in Hall
et al, 1960. )

CELLULOSE-PROTEIN COMPLEXES 253

DISCUSSION
M. V. Edds, Jr., R. D. Preston

Dr. Edds (Brown University): How do you separate the lamellae of the
Valonia cell wall in preparation for electron microscopy?

Dr. Preston: The separation of lamellae of the Wdonia cell wall is com-
paratively easy. A square is cut out of the wall and is left to dry down on a
glass slide. A piece of Cellotape is then pressed on the wall and, on peeling
the Cellotape from the slide, some of the wall remains on the Cellotape and
some on the glass. The process can be repeated until a thin lamella is left
either on the Cellotape or on the glass. This lamella can then be separated
from the glass by flotation in water, or from the Cellotape with appropriate
solvents. A more reliable method, however, is to mount a piece of wall so
that the part of it to be examined overlaps a hole in a piece of adhesive paper.
Stripping by Cellotape then leaves the final lamella projecting over the hole
without ever coming into contact with either glass or Cellotape.

Alternatively (and better), the lamella can be teased off in water, using
fine steel needles. This requires considerable manipulative skill, but produces
a preparation with a minimum of distortion and without any contamination by
Cellotape.

Index

Aggregation, 7, 19-20, 21-24, 33-37, 53-

84
Amino acid analysis, 222; see also specific

amino acids
Antigenic patches, 164
Apatke, 54, 55, 69, 83
Arhacia, 169
Aster, 161-162, 167-168
ATP, 24, 51, 52
Autoradiography, 126
Axon, 140-141
Axoplasm, 141

Bacteriochlorophyll, 185

Barker, S. B., 83

Beck, L. v., 83

Bergeron, J. A., 179-203, 228

Bim'olecular leaflet, 139-140, 152

Bond

disulfide, 11-12, 167-169

hydrogen, 5-7, 20-21

ionic, 5-7

phosphate, 179-180
Bone, 53-84
Bud scar, 216
Budding, 217, 225

Calcification, 53-84
Candida, 225
Carbohydrates, 5
Carotenoid, 185
Cartilage, 53-84
Casein, 11
Cell

division, 161-178, 225

wall, 205, 206, 229, 241
Cellulose, 205-228, 229, 236
Cementum, 55
Centrosomes, 161
Charge distribution, 30, 32, 52, 62
Chitin, 83, 209
ChJamtjdonionas, 95, 99
Chlorophyll, 85-112, 150
Chloroplast, 85-112, 150, 180
Chloroplastin, 100-103, 199
Chondroitin sulfate, 233
Chromatium, 180
Chromatophore, 86, 179-204
Chromosome, 15

Chymotrypsin, 41

Collagen, 14, 19-52, 53-84, 232

Collagenase, 44

Complex

antigen-antibody, 164

cellulose-protein, 205-228, 229, 233

chitin-protein, 17

chlorophyll-protein, 85-112

glucomannan-protein, 205-229

lipoprotein, 115, 118

polysaccharide-protein, 205, 214, 220,
232
Conchiolin, 83 ff.
Covalent bonds, 4
Cross-linking, Ca-phosphate, 12
Cryofixation, 116, 118, 123

bromine, 138

osmium, 131, 141
Crystalline ordering, 17
Crystallization, 55-84, 105-109
Cytochrome, 86

Dentin, 55
Ducoff, H., 227

Edds, M. v., Jr., 176, 203, 253

Elastoidin, 82

Electron-spin resonance, 151

Elodea, 95

Energy, interaction, 13

Energy transfer, 150

Entropy, 13

Enzyme reactions, hydrolvtic, 13

Euglena, 90, 98, 101

Falcone, C, 205-227
Fatty acid, 17
FDNB, 69 ff., 83
Fernandez-Moran, H., 113-159
Fibril formation, 9, 11, 19-52
Fibrillar system, 161-178, 229-253
Fibrogenesis, 9, 11, 19-52
Fibrous long-spacing, 28
Forces

intermolecular, 68

London-van der Waals, 7, 17

long-range, 5, 12, 15

short-range, 5, 7
Free radicals, 116, 120

255

256

INDEX

Freeze-drying and freeze-substitution,

117
Fuller, R. C, 179-202

Gel-difFusion technique, 163
Giese, A. C, 112, 158, 177,203
Glimcher, M. J., 53-84
Glucan, 206
Glucomannan, 207
Glycoprotein, 28, 32
Green, J. W., 82, 227
Ground substance, 77
Group

carboxyl, 71

disulfide, 167

guanidino, 71

phosphorus, 12

polar, 21

sulfhydryl, 167
Group-blocking and group-substitution,
74

Helices, 5, 20, 22, 37, 40
Helium, liquid, 116, 119-122
Hexosamine, 232
Hodge, A. J., 19-52, 112
Hydroxylysine, 69, 75, 83
Hydroxyproline, 240

Ice-crystal formation, 129-131

Insulin, 9

Interaction

charge, 13

energy, 5, 7

long-range, 5, 12, 15

molecular, 3, 4, 8, 19-52, 53-84

short-range, 5, 7

sonic, 25, 37

Johnson, W. H., 51

Kessler, G., 205-227

Lamellae, 4, 8, 13, 85-112, 113-161, 241

basement, 14
Liesegang phenomena, 105
Lilium, 168

Low-temperature preparation, 115, 117
Lysine, 69, 75, 83

McElroy, W. D., 17, 51
Mannan, 207
Matrix, organic, 53-84
Mazia, D., 161-176
Membrane

double, 114 flF.

unit, 114, 142

Mercaptoethanol, 167

Metastable equilibrium, 54, 55, 62, 117

Micelle, 11

Microfibril, 229, 236, 237, 243

Mineralization, 53-84

Mitochondria, 150

Mitotic apparatus, 161-178

Molecular

fingerprint, 24

interaction, 3, 4, 8, 19-52, 53-84

organization, 150

recombination, 8
Monolayer, 91, 198
Monomolecular films, 8
Myelin, 113, 114, 126, 127, 134, 138,
150, 151

Nickerson, W. J., 17, 205-228
Nucleation, lO; 17, 19, 55, 62, 82
Nucleic acids, 5, 86

Orthogonal pattern, 14
Oyster, 83

Paracrystalline state, 108-109, 148-150

Paramyosin, 34, 51

Pattern, diffraction, 20, 22

Pepsin, 41, 44

Periodicity, 19-54, 105-109

Phase transformation, 117

Phospholipid, 185

Phosphorylation, 69

Photochemical organelle, 182, 188

Photophosphorylation, 187

Photopigment, 144, 150

Photopolymerization, 118

Photoreceptor, 85, 113, 142, 146, 150

Photosynthesis, 9, 85-112, 179-204

Photosynthetic pigments, 87, 181

Phvtol, 92, 198

Polymerization, 28, 37, 38, 43, 52, 118

Polypeptides, 5, 19, 40, 62

Polysaccharide, 186, 206, 222

Porphyrin, 92, 95, 198

Preston, R. D., 227, 229-253

Prosser, C. L., 52, 83

Protease, 41, 45

Protein, 5, 19-52, 53-84, 161-178, 185,

205-228, 229-253
Protein disulfide reductase, 225
Protofibrils, 34, 62

Retinal pigment, 146
Retinal rod, 101, 126, 142, 145
Rhodopsin, 101
Rhodospirillum, 99, 179
Richards, A. G., 17, 83

INDEX 257

Schmitt, F. O., 19-51 Trypsin, 41, 44

Schwann cell, 114, 139 Tyrosine, 47, 48
Segment long-spacing, 24

Semiconductor, 98, 151 Uronic acid, 232

Sherman, F. G., 52 ,, , ^„„ ^

Sol-gel transformation, 169 1""^"^'. P' ^"^^

Specificity, 3 ff., 62 Van Holde, K. E., 81

Spindle 161-178 y^^^^^^ 4^ g ^2, 13, 17, 151

Strongtjlocentrotus, 168 ^^^g^^ D p., 3-17, 203

Svmmetrv, 68

Went, H. A., 161-177

T w ifto Wolken, J. J., 85-112

Trillium, 168 ^ ^

Tropocollagen, 19-52 Yeast cell wall, 229-253