This preview shows page 1. Sign up to view the full content.
Unformatted text preview: ANTIBODY STRUCTURE AND MOLECULAR
Nobel Lecture, December 12, 1972,
G E R A L D M. ED E L M A N The Rockefeller University, New York, N.Y., U.S.A. Some sciences are exciting because of their generality and some because of
their predictive power. Immunology is particularly exciting, however, because
it provokes unusual ideas, some of which are not easily come upon through
other fields of study. Indeed, many immunologists believe that for this reason,
immunology will have a great impact on other branches of biology and
medicine. On an occasion such as this in which a very great honor is being
bestowed, I feel all the more privileged to be able to talk about some of the
fundamental ideas in immunology and particularly about their relationship
to the structure of antibodies.
Work on the structure of antibodies has allied immunology to molecular
biology in much the same way as previous work on hapten antigens allied
immunology to chemistry. This structural work can be considered the first
of the projects of molecular immunology, the task of which is to interpret
the properties of the immune system in terms of molecular structures. In
this lecture, I should like to discuss some of the implications of the structural
analysis of antibodies. Rather than review the subject, which has been amply
done (1-4), I shall emphasize several ideas that have emerged from the
structural approach. Within the context of these ideas, I shall then consider
the related but less well explored subject of antibodies on the surfaces of
lymphoid cells, and describe some recently developed experimental efforts
of my colleagues and myself to understand the molecular mechanisms by
which the binding of antigens induces clonal proliferation of these cells.
Antibodies occupy a central place in the science of immunology for an
obvious reason: they are the protein molecules responsible for the recognition
of foreign molecules or antigens. It is, therefore, perhaps not a very penetrating
insight to suppose that a study of their structure would be valuable to an
understanding of immunity. But what has emerged from that study has resulted
in both surprises and conceptual reformulations.
These reformulations provided a molecular basis for the selective theories
of immunity first expounded by Niels Jerne (5) and MacFarlane Burnet (6)
and therefore helped to bring about a virtual revolution of immunological
thought. The fundamental idea of these theories is now the central dogma of
modern immunology: molecular recognition of antigens occurs by selection
among clones of cells already committed to producing the appropriate antibodies, each of different specificity (Figure 1). The results of many studies by
31 32 Physiology or Medicine 1972 Fig. 1.
A diagram illustrating the basic
features of the clonal selection theory.
The stippling and shading indicate
receptors cells have antibody of different specificities, although the specificity of all receptors
on a given cell is the same. Stimulation
by an antigen results in clonal expanmitosis and anti- sion (maturation, body production) of those cells having
receptors complementary to the antigen. cellular immunologists (see references 1 and 2) strongly suggest that each cell
makes antibodies of only one kind, that stimulation of cell division and antibody
synthesis occurs after interaction of an antigen with receptor antibodies at
the cell surface, and that the specificity of these antibodies is the same as that
of the antibodies produced by daughter cells. Several fundamental questions
are raised by these conclusions and by the theory of clonal selection. How can
a sufficient diversity of antibodies be synthesized by the lymphoid system?
What is the mechanism by which the lymphocyte is stimulated after interaction with an antigen?
In the late 1950’s, at the beginnings of the intensive work on antibody
structure, these questions were not so well defined. The classic work of Landsteiner on hapten antigens (7) had provided strong evidence that immunological specificity resulted from molecular complementarity between the
determinant groups of the antigen molecule and the antigen-combining site
of the antibody molecule. In addition, there was good evidence that most
antibodies were bivalent (8) as well as some indication that antibodies of
different classes existed (9). The physico-chemical studies of Tiselius (10)
had established that antibodies were proteins that were extraordinarily heterogeneous in charge. Moreover, a number of workers had shown the existence
of heterogeneity in the binding constants of antibodies capable of binding a
single hapten antigen (11). Despite the value of all of this information, however,
little was known of the detailed chemical structure of antibodies or of what are
now called the immunoglobulins.
T HE M ULTICHAIN S TRUCTURE OF A N T I B O D I E S : PR O B L E M S OF S IZE AND H ETEROGENEITY If the need for a structural analysis of antibodies was great, so were the
experimental difficulties: antibodies are very large proteins (mol. wt. 150,000
or greater) and they are extraordinarily heterogeneous. Two means were adopted around 1958 in an effort to avoid the first difficulty. Following the
work of Petermann (12) and others, Rodney Porter (13) applied proteolytic
enzymes, notably papain, to achieve a limited cleavage of the gamma globulin
fraction of serum into fragments. He then successfully fractionated the digest,
obtaining antigen binding (Fab) and crystallizable (Fc) fragments. Subsequently, other enzymes such as pepsin were used in a similar fashion by
Nisonoff and his colleagues (14). I took another approach, in an attempt to
cleave molecules of immunoglobulin G and immunoglobulin M into polypeptide chains by reduction of their disulfide bonds and exposure to dissociating solvents such as 6 M urea (15). This procedure resulted in a significant
drop in molecular weight, demonstrating that the immunoglobulin G molecule
was a multichain structure rather than a single chain as had been believed
before. Moreover, corresponding chains obtained from both immunoglobulins
had about the same size. The polypeptide chains (16) were of two kinds (now
called light and heavy chains) but were obviously not the same as the fragments obtained by proteolytic cleavage and therefore the results of the two
cleavage procedures complemented each other. Ultracentrifugal analyses indicated that one of the polypeptide chains had a molecular weight in the
vicinity of 20,000, a reasonable size for determination of the amino acid sequence
by the methods available in the early 1960s.
Nevertheless, the main obstruction to a direct analysis of antibody structure
was the chemical heterogeneity of antibodies and their antigen binding
fragments. Two challenging questions confronted those attempting chemical
analyses of antibody molecules at that time. First, did the observed heterogeneity of antibodies reside only in the conformation of their polypeptide
chains as was then widely assumed, or did this heterogeneity reflect differences
in the primary structures of these chains, as required implicitly by the clonal
selection theory? Second, if the heterogeneity did imply a large population of
molecules with different primary structures, how could one obtain the homogeneous material needed for carrying out a detailed structural analysis?
These challenges were met simultaneously by taking advantage of an
accident of nature rather than by direct physicochemical assault. It had been
known for some time that tumors of lymphoid cells called myelomas produced
homogeneous serum proteins that resembled the normal heterogeneous immunoglobulins. In 1961, M. D. Poulik and I showed that the homogeneity
of these proteins was reflected in the starch gel electrophoretic patterns
of their dissociated chains (16). Some patients with multiple myeloma excrete
urinary proteins which are antigenically related to immunoglobulins but
whose nature remained obscure since their first description by Henry Bence
Jones in 1847. These Bence Jones proteins were most interesting, for they
could be readily obtained from the urine in large quantities, were homogeneous, and had low molecular weights. It seemed reasonable to suggest (16)
that Bence Jones proteins represented one of the chains of the immunoglobulin
molecule that was synthesized by the myeloma tumor but not incorporated
into the homogeneous myeloma protein and therefore excreted into the urine.
This hypothesis was corroborated one exciting afternoon when my student 34 Physiology or Medicine 1972 Fig. 2.
Comparisons of light chains isolated from serum IgG myeloma proteins with urinary
Bence Jones proteins from the same patient. (a) Starch gel electrophoresis in urea. 1) serum
myeloma globulin, 2) urinary Bence Jones protein, 3) Bence Jones protein reduced and
alkylated, 4) myeloma protein reduced and alkylated. L -- light chain; H -- heavy
chain. (b) Two-dimensional high voltage electrophoresis of tryptic hydrolysates. Pattern
on left is of urinary Bence Jones protein; that on right is of light chain isolated from the serum myeloma protein of the same patient. Joseph Gally and I (17) heated solutions of light chains isolated from our
own serum immunoglobulins in the classical test for Bence Jones proteinuria.
They behaved as Bence Jones proteins, the solution first becoming turbid,
then clearing upon further heating. A comparison of light chains of myeloma
proteins with Bence Jones proteins by starch gel electrophoresis in urea (17)
and by peptide mapping (18) confirmed the hypothesis (Figure 2). Indeed,
Berggård and I later found (19) that in normal urine there were counterparts
to Bence Jones proteins that shared their properties but were chemically heterogeneous.
No physical means was known at the time that was capable of fractionating
antibodies to yield homogeneous proteins. It was possible, however, to prepare
specifically reactive antibodies by using the antigen to form antigen-antibody
aggregates and then dissociating the complex with free hapten. Although
we knew that these specifically prepared antibodies were still heterogeneous
in their electrophoretic properties, it seemed possible that antibodies to
different haptens might show differences in their polypeptide chains. Baruj
Benacerraf had prepared a collection of these antibodies, and together with
our colleagues (20) we decided to compare their chains, using the same methods
that we had used for Bence Jones proteins. The results were striking: purified
antibodies showed from 3 to 5 sharp bands in the Bence Jones or light chain
region and antibodies of different specificities showed different patterns. In
sharp contrast, normal immunoglobulin showed a diffuse zone extending over
the entire range of mobilities of these bands. These experiments showed not
only that antibodies of different specificities were structurally different but
also that their heterogeneity was limited. Antibody Structure and Molecular Immunology 35 The results of these experiments on Bence Jones proteins and purified
antibodies had a number of significant implications. Because different Bence
Jones proteins had different amino acid compositions, it was clear that immunoglobulins must vary in their primary structures. This deduction, confirmed
later by Koshland (21) for specifically purified antibodies, lent strong support
to selective theories of antibody formation. Moreover, it opened the possibility
of beginning a direct analysis of the primary structure of an immunoglobulin
molecule, for not only were the Bence Jones proteins composed of homogeneous
light chains, but their subunit molecular weight was only 23,000. The first
report by Hilschmann and Craig (22) on partial sequences of several different
Bence Jones proteins indicated that the structural heterogeneity of the light
chains was confined to the amino terminal (variable) region, whereas the
carboxyl terminal half of the chain (the constant region) was the same in
all chains of the same type. This finding was soon extended by studies of other
Bence Jones proteins (23).
Although some work had also been done on the heavy chains of immunoglobulins, there was much less information on their structure. For instance,
it was suspected but not known that they also had variable regions resembling
those of light chains. Comparisons of heavy chains and light chains even at
this early stage did, however, clarify the nature of another source of antibody
heterogeneity: the existence of immunoglobulin classes (24).
Antibodies within a particular class have similar molecular weights, carbohydrate content, amino acid compositions and physiological functions (Table
1) but still possess heterogeneity in their net charge and antigen binding
affinities. Studies of classes in various animal species indicated that both the
multichain structure and the heterogeneity are ubiquitous properties of
immunoglobulins. The different classes apparently emerged during evolution
(25) to carry out various physiologically important activities that have been
named effector functions in order to distinguish them from the antigenbinding or recognition function. The various manifestations of humoral
immune responses as well as their prophylactic, therapeutic and pathological
consequences can now be generally explained in terms of the properties of
the particular class of antibody mediating that response. As a result of comparing their chain structure, it became clear that although immunoglobulins
of all classes contain similar kinds of light chains (Table l), the distinctive
class character (24) is conferred by structural differences in the heavy chains,
specifically in their constant regions, as I shall discuss later.
With the clarification of the nature of the heterogeneity of immunoglobulin
chains and classes, attention could be turned to the problem of relating the
structure and evolution of antibodies within a given class to their antigenbinding and effector functions. We chose to concentrate on immunoglobulin
G, for this was the most prevalent class in mammals and the work on chain
structure suggested that it would be sufficiently representative. T HE C OMPLETE C OVALENT STRUCTURE AND THE D OMAIN H YPOTHESIS An understanding of the chain structure and its relation to the proteolytic
fragments (26, 27) made feasible an attempt to determine the complete
structure of an immunoglobulin G molecule. My colleagues and I started this
project in 1965, and before it was completed in 1969 (28) seven of’ us had
spent a good portion of our waking hours on the technical details. One of our
main objectives was to provide a complete and definitive reference structure
against which partial structures of other immunoglobulins could be compared.
In particular, we wished to compare the detailed structure of a heavy chain
and a light chain from the same molecule.
Another objective was to examine in detail the regional differentiation of
the structure that had been evolved to carry out different physiological
functions in the immune response. The work of Porter (13) had shown that
the so-called Fab fragment of immunoglobulin G was univalent and bound
antigens whereas the Fc fragment did not. This provided an early hint that
immunoglobulin molecules were organized into separate regions, each mediating different functions. In accord with selective theories of immunity, it
was logical to suppose that V regions from both the light and the heavy
chains mediated the antigen binding functions. Early evidence that some of
the C regions were concerned with physiologically significant effector functions
was obtained by showing that Fc fragments would bind components of the
complement system (29), a complex group of proteins responsible for immunologically induced cell lysis. A more detailed assignment of structure to function
required a knowledge of the total structure. Fig. 3.
Overall arrangement of chains and disulfide bonds of the human y(;, i mmunoglobulin EU.
Half-cystinyl residues are I-XI; I-V designates corresponding half-cystinyl residues in light and heavy chains. PCA, pyrrolidonecarboxylic acid; CHO, carbohydrate. Fab(t) and
Fc(t) refer to fragments produced by trypsin, which cleaves the heavy chain as i ndicated
by dashed lines above half-cystinyl residues VI. Variable regions, V H a nd V L , are homologous. The constant region of the heavy chain (C H ) is divided into three regions, C H 1 ,
C H 2 and C H 3, that are homologous to each other and to the C region of the light chain.
The variable regions carry out antigen-binding functions and the constant regions the
effector functions of the molecule. 38 Physiology or Medicine 1972 Amino acid sequence analysis of the Fc region of normal rabbit y chains
by Hill and his colleagues (30) demonstrated that the carboxyl terminal
portion of heavy chains was homogeneous. On the basis of internal homologies
in this region, Hill (30) and Singer and Doolittle (31) proposed the hypothesis
that the genes for immunoglobulin chains evolved by duplication of a gene
of sufficient size to specify a precursor protein of about 100 amino acids in
length. Although direct confirmation of this hypothesis is obviously not
possible, it was strongly supported by the results of our analysis (28) of the
complete amino acid sequence and arrangement of the disulfide bonds of an
entire 1gG myeloma protein.
Comparisons of the amino acid sequences of the heavy chain of this protein
with others studied in Porter’s laboratory (32) and by Bruce Cunningham and
his colleagues in our laboratory (33) s h owed that heavy chains had variable
(V H ) regions, i.e., regions that differed from one another in the sequences of
the 110-120 residues beginning with the amino terminus (Figure 3).
Examination of the amino acid sequences (Figures 4 and 5) allowed us to
draw the following additional conclusions:
1) The variable (V) regions of light and heavy chains are homologous to
each other, but they are not obviously homologous to the constant regions
of these chains. V regions from the same molecule appear to be no more
closely related than V regions from different molecules.
2) The constant (C) region of y chains consists of three homology regions, THR Fig. 4.
Comparison of the amino acid sequences of the V H a nd V L r egions of protein Eu. Identical
residues are shaded. Deletions indicated by dashes are introduced to maximize the homology. F ig. 5.
C o m p a r i s o n o f t h e a m i n o a c i d s e q u e n c e s o f C L , C H 1, C L 2 and C L 3 regions. Deletions,
indicated by dashes, have been introduced to maximize homologies. Identical residues are
darkly shaded; both light and dark shadings are used to indicate identities which occur in
pairs in the same position. C H1, CH2 and CH3, each of which is closely homologous to the others and to
the constant regions of the light chains.
3) Each variable region and each constant homology region contains one
disulfide bond, with the result that the intrachain disulfide bonds are linearly
and periodically distributed in the structure.
4) The region containing all of the interchain disulfide bonds is at the
center of the linear sequence of the heavy chain and has no homologous
counterpart in other portions of the heavy or light chains. 40 Physiology or Medicine 1972
The domain hypothesis. Diagramarrangement of domains molecule. The arrow refers to a
dyad axis of symmetry. Homology
regions (see Figures 3, 4 and 5)
which constitute each domain are
i n d i c a t e d : VL , VH -- domains made up of variable homology
r e g i o n s ; C L , CH l , CH 2 , a n d CH 3 domains made up of constant
homology regions. Within each of these groups, domains are assumed to have similar threedimensional structures and each is assumed to contributed to an active site. The V domain
sites contribute to antigen recognition functions and the C domain sites to effector functions. These conclusions prompted us to suggest that the molecule is folded in
a congeries of compact domains (28,33) each formed by separate V homology
regions or C homology regions (Figure 6). In such an arrangement, each
domain is stabilized by a single intrachain disulfide bond and is linked to
neighboring domains by less tightly folded stretches of the polypeptide chains.
A twofold pseudosymmetry axis relates the V L C L to the VHC H1 domains and
a true dyad axis through the disulfide bonds connecting the heavy chains
relates the C H 2 - CH 3 domains. The tertiary structure within each of the
homologous domains is assumed to be quite similar. Moreover, each domain
is assumed to contribute to at least one active site mediating a function of the
This last supposition is nicely demonstrated by the interaction of V region
domains. The reconstitution of active antibody molecules by recombining their
isolated heavy and light chains (34, 35, 36) as well as affinity labelling experiments (31) confirmed our early hypothesis that the V regions of both heavy
and light chains contributed to the antigen-combining sites. Moreover, the
experiments of Haber (37) provided the first indication that Fab fragments of
specific antibodies could be unfolded after reduction of their disulfide bonds
and refolded in the absence of antigen to regain most of their antigen binding
activity. This clearly indicated that the information for the combining site
was contained entirely in the amino acid sequences of the chains. That this
information is contained completely in the variable regions is strikingly shown
by the recent isolation of antigen-binding fragments consisting only of V L
and V H ( 38). The chain recombination experiments suggested an hypothesis
to account in part for antibody diversity: the various combinations of different
heavy and light chains expressed in different lymphocytes allow the formation
of a large number of different antigen-combining sites from a relatively small
number of V regions.
One of the remaining structural tasks of molecular immunology is to obtain
a direct picture of antigen-binding sites by X-ray crystallography of V domains
at atomic resolution. Although crystals of the appropriate molecule or fragment
yielding diffraction patterns that extend beyond Bragg spacings of 3.0 Å Antibody Structure and Molecular Immunology 41 have not yet been obtained, it is likely that continued searching will provide
them. The details of a particular antigen-antibody interaction revealed by
such a study will be of enormous interest. For example, certain sequence positions of V regions are hypervariable (39) and are very good candidates for
direct contribution to the site. It will be particularly important to understand
how the basic three-dimensional structure can accomodate so many amino
acid substitutions. X-ray crystallographic work may also show in detail how
the disulfide bonds in each of the V domains provide essential stability to the
site (28, 33, 40).
The proposed similarities in tertiary structures among C domains have not
been established nor have the functions of the various C domains been fully
determined. There is a suggestion that CH2 may play a role in complement
fixation (41). A good candidate for binding to the lymphocyte cell membrane
is CH 3, the function of which may be concerned with the mechanism of
lymphocyte triggering following the binding of antigen by V domains. The
C H 3 domain has already been shown to bind to macrophage membranes
(42) and there is now some evidence that lymphocytes can synthesize isolated
domains (43, 44, 45) similar to C H3 as separate molecules.
Although many details are still lacking, the gross structural aspects of the
domain hypothesis have received direct support from X-ray crystallographic
analyses of Fab fragments (46) and whole molecules (47) in which separate
domains were clearly discerned. Indirect support for the hypothesis has also
come from experiments (38, 48) on proteolytic cleavage of regions between
It is not completely obvious why the domain structure was so strictly
preserved during evolution. One reasonable hypothesis is that although there
was a functional need for association of V and C domains in the same molecule,
there was also a need to prevent allosteric interactions among these domains.
Whatever the selective advantages of this arrangement, it is clear that immunoglobulin evolution by gene duplication permitted the possibility of modular
alteration of immunological function by addition or deletion of domains.
T R A N S L O C O N S : PR O P O S E D U NITS OF E VOLUTION AND G ENETIC F UNCTION The evolution by gene duplication of both the domain structure and the
immunoglobulin classes raises several questions about the number and arrangement of the structural genes specifying immunoglobulins. Although time does
not permit me to discuss this complex subject in detail, I should like to suggest
how structural work has sharpened these questions.
According to the theory of clonal selection, it is necessary that there preexist in each individual a large number of different antibodies with the
capacity to bind different antigens. One of the most satisfying conclusions that
emerged from structural analysis is that the diversity of the V regions of antibody chains is sufficient to satisfy this requirement. This diversity arises at
three levels of structural or genetic organization, two of which are now
reasonably well understood:
1) V regions from both heavy and light chains contribute to the antigen- 42 Physiology or Medicine 1972 binding site and therefore the number of possible antibodies may be as great
as the product of the number of different VL a nd V H r egions.
2) Analyses of the amino acid sequences of V regions of light chains by
Hood (49) and Milstein (50) and later of heavy chains from myeloma proteins
(32,33) indicated that V regions fall into subgroups of sequences which must
be specified by separate genes or groups of genes. Within a subgroup, the
amino acid replacements at a particular position are of a conservative type
consistent with single base changes in codons of the structural genes. Variable
regions of different subgroups differ much more from each other than do
variable regions within a subgroup.
Although different V region subgroups are specified by a number of nonallelic genes (50), the analysis of genetic or allotypic markers suggests that
C regions of a given immunoglobulin class are specified by no more than one
or two genes. These allotypic markers, first described by Grubb (51) and
Oudin (52) provide a means in addition to sequence analysis for understanding
the genetic basis of immunoglobulin synthesis (4). V regions specified by a
number of different genes can occur in chains each of which may have the
same C region specified by a single gene. It therefore appears that each
immunoglobulin chain is specified by two genes, a V gene and a C gene
(4, 49, 50).
Work in a number of laboratories (reviewed in reference 4) has shown
that the genetic markers on the two types of light chains are not linked to
those of the heavy chains or to each other. These findings and the conclusion
that there are separate V and C genes led Gally and me to suggest (4) that
immunoglobulins are specified by three unlinked gene clusters (Figure 7).
The clusters have been named translocons (4) to emphasize the fact that some
mechanism must be provided to combine genetic information from V region
loci with information from C region loci to make complete V-C structural
genes. According to this hypothesis, the translocon is the basic unit of immuno- Antibody Structure and Molecular Immunology 43 globulin evolution, different groups of immunoglobulin chains having arisen
by duplication and various chromosomal rearrangements of a precursor gene
cluster. Presumably, gene duplication during evolution also led to the appearance of V region subgroups within each translocon.
The key problem of the generation of immunoglobulin diversity has been
converted by the work on chains and subgroups to the problem of the origin
of sequence variations within each V region subgroup. It is still not known
whether there is a germ line gene for each V region within a subgroup or
whether each subgroup contams only a few genes (see Figure 7) and intrasubgroup variation arises by somatic genetic rearrangements of translocons
within precursors of antibody forming cells. At this time, therefore, we can
conclude that only the basis but not the origin of diversity has been adequately
explained by the work on structure. Although structural analysis of various
immunoglobulin classes will continue to be important, it does not in itself
seem likely to lead to an explanation of the origin of antibody diversity. What
will probably be required are imaginative experiments on DNA, RNA and
their associated enzymes obtained from lymphoid cells at the proper stage of
In this abbreviated and necessarily incomplete account, I have attempted
to show how structural work on immunoglobulins has provided a molecular
basis for a number of central features of the theory of clonal selection. The
work on humoral antibodies is just a beginning, however, for two great
problems of molecular and cellular immunology remain to be solved. The first
problem, the origin of intrasubgroup diversity, will undoubtedly receive great
attention in the next few years. The second problem is concerned with the
triggering of the clonal expansion of lymphocytes after combination of their
receptor antibodies with antigens and the quantitative description of the
population dynamics of the responding cells. An adequate solution to this
problem must also account for the phenomenon of specific immune tolerance as
described by the original work of Medawar and his associates (53).
For the remainder of this lecture, I shall turn my attention to some recent
attempts that my colleagues and I have made to see whether these problems
can be profitably studied using molecular approaches. L Y M P H O C Y T E S TIMULATION BY M EANS OF LECTINS The mechanisms of the cellular events underlying immune responses and
immune tolerance remain a major challenge to theoretical and practical
immunology (53,60). H ow does a given antigen induce clonal proliferation or
immune tolerance in certain subpopulations of cells?
Cells reactive to a given antigen constitute a very small portion of the
lymphocyte population and are difficult to study directly. Two means have
been used to circumvent this difficulty: the application of molecules that can
stimulate lymphocytes independent of their antigen binding specificity, and
fractionation of specific antigen binding lymphocytes for studies of stimulation
by antigens of known structure. Although the problem of lymphocyte stimula- 44 Physiology or Medicine 1972 tion is far from being solved, both of these approaches are valuable
particularly when used together.
Antigens are not the only means by which lymphocytes may be stimulated.
It has been found that certain plant proteins called lectins can bind to glycoprotein receptors on the lymphocyte surface and induce blast transformation,
mitosis and immunoglobulin production (see reference 54 for a review).
Different lectins have different specificities for cell surface glycoproteins and
different molecular structures although their mitogenic properties can be
quite similar. In addition, they have a variety of effects on cell metabolism and
transport. Such effects are independent of the antigen binding specificity of
the cell and they may therefore be studied prior to specific cell fractionation.
The fact that antigens and lectins of different specificity and structure may
stimulate lymphocytes suggests that the induction of mitosis is a property of
membrane-associated structures that can respond to a variety of receptors.
Triggering appears to be independent of the specificity of these receptors for
their various ligands. To understand mitogenesis, it is therefore necessary to
solve two problems. The first is to determine in molecular detail how the
lectin binds to the cell surface and to compare it to the binding of antigens.
The second is to determine how the binding induces metabolic changes
necessary for the initiation of cell division. These changes are likely to include
the production or release of a messenger which is a final common pathway for
the stimulation of the cell by a particular lectin or antigen.
One of the important requirements for solving these problems is to know
the complete structure of several different mitogenic lectins. This structural
information is particularly useful in trying to understand the molecular
transformation at the lymphocyte surface required for stimulation. With
the knowledge of the three-dimensional structure of a lectin, various amino
acid side chains at the surface of the molecule may be modified by group
reagents which also may be used to change the valence of the molecule. The
activities of the modified lectin derivatives may then be observed in various
assays of their effects on cell surfaces and cell functions.
My colleagues and I (55) have recently determined both the amino acid
sequence and three-dimensional structure of the lectin, concanavalin A (Con
A) (Figure 8). This lectin has specificity for glucopyranosides, mannopyranosides and fructofuranosides and binds to glycoproteins and possibly glycolipids
at a variety of cell surfaces. The purpose of our studies was to know the exact
size and shape of the molecule, its valence and the structure and distribution
of its binding sites.
With this knowledge in hand, we have been attempting to modify the
structure and determine the effects of that modification on various biological
activities of the lymphocyte. So far, there are several findings suggesting that
such alterations of the structure have distinct effects. Con A in free solution
stimulates thymus-derived lymphocytes (T cells) but not bone marrowderived lymphocytes (B cells), leading to increased uptake of thymidine and
blast transformation. The curve of stimulation of T cells by native Con A
shows a rising limb representing stimulation and a falling limb (Figure 9) a Fig. 8.
‘Three-dimensional structure of concanavalin A, a lectin mitogenic for lymphocytes. (a)
Schematic representation of the tetrameric structure of Con A viewed down the z axis.
The proposed binding sites for transition metals, calcium, and saccharides arc indicated by
Mn, Ca and C, respectively. The monomers on top (solid lines) are related by a twofold
axis, as are those below. The two dimers are paired across an axis of D 2 s ymmetry to form the
tetramer. (b) Wire model of the polypeptide backbone of the concanavalin A monomer
oriented approximately to correspond to the monomer on the upper right of the diagram
in (a). The two balls at the top represent the Ca and Mn atoms and the ball in the center
is the position of an iodine atom in the sugar derivative, b-iodophenylglucoside, which is
bound to the active site. Four such monomers are joined to form the tetramer as shown in (a).
(c) A view of the Kendrew model of the Con A monomer rotated to show the deep pocket
formed by the carbohydrate binding site. (White ball at the bottom of the figure is at the
position of the iodine of b-iodophenylglucoside). The two white balls at the top represent
the metal atoms. 46 Physiology or Medicine 1972 probably the result of cell death. The fact that the mitogenic effect and inhibition
effect are dose dependent suggests an analogy to stimulation and tolerance
induction by antigens. When Con A is succinylated, it dissociates from a
tetramer to a dimer without alteration of its carbohydrate binding specificity.
Although succinylated Con A is just as mitogenic as native Con A, the falling
limb is not seen until much higher doses are reached.
Succinylation of Con A also alters another property of the lectin. It has
been shown that, at certain concentrations, the binding of Con A to the cell
surface restricts the movement of immunoglobulin receptors (56, 57). This
suggests that it somehow changes the fluidity of the cell membrane resulting
in reduction of the relative mobility of these receptors. In contrast, succinylated
Con A has no such effect although it binds to lymphocytes to the same extent
as the native molecule. Both the abolition of the killing effect in mitogenic
assays and the failure to alter immunoglobulin receptor mobility in B cells
after succinylation of Con A may be the result of change in valence or o
alteration in the surface charge of the molecule. Examination of other derivatives and localization of the substituted side chains in the three-dimensional
structure will help to establish which is the major factor. Recent experiments
suggest that the valence is probably the major factor, for addition of divalent
antibodies against Con A to cells that had bound succinylated Con A resulted
again in restriction of immunoglobulin receptor mobility.
Con A may also be modified by cross-linking several molecules. A very
striking effect is seen if the surface density of the Con A molecules presented
to the lymphocyte is increased by cross-linking it at solid surfaces (58). Con A
in free solution stimulates mouse T cells to an increased incorporation of
radioactive thymidine but has no effect on B cells. When cross-linked at a solid
surface, however, it stimulates mainly mouse B cells, although both T and B
cells have approximately the same number of Con A receptors (58). Similar
results have been obtained with other lectins (59). A reasonable interpretation
of these phenomena (although not the only one) is that the lectin acts at the
cell surface rather than inside the cell, that the presence of a high surface
density of the mitogen is an important variable in exceeding the threshold for
the lymphocyte stimulation, and that the threshold differs in the two kinds of
Alteration of the structure and function of various lectins appears to be a
promising means of analyzing the mechanism of lymphocyte stimulation. One
intriguing hypothesis is that cross-linkage of the proper subsets of glycoprotein
receptors by lectins is essentially equivalent in inducing cell transformation
to cross-linkage of immunoglobulin receptors in the lymphocyte membrane
by multivalent antigens. The central effector function of receptor antibodies,
triggering of clonal proliferation, may turn out to be specifically related to the
mode of anchorage of the antibody molecule to the cell membrane. The mode
of attachment of antibody and lectin receptors to membrane-associated
structures and their perturbation by crosslinkage at the cell surface may be
similar and have similar effects despite the difference in their specificities and
molecular structures. Antibody Structure and Molecular Immunology 47 A N T I B O D I E S O N T H E S U R F A C E S O F A N T I G E N - BI N D I N G C E L L S The most direct attack on the problem of lymphocyte stimulation is to explore
the effects of antigens of known molecular geometry on specifically purified
populations oflymphocytes. For this and other reasons, it is necessary to develop
methods for the specific fractionation of antigenbinding cells.
In carrying out this task it is important both theoretically and operationally
to discriminate between antigen-binding and antigen-reactive cells. In clonal
selection, the phenotypic expression of the immunoglobulin genes is mediated
in the animal by somatic division of precommitted cells (Figure 10). The
pioneering work of Nossal and Mäkelä and later of Ada and Nossal (see
reference 60) clearly showed that each cell makes antibodies of a single
specificity and that there are different populations of specific antigen-binding
cells. An animal is capable of responding specifically to an enormous number
of antigens to which it is usually never exposed, and it therefore must contain
genetic information for synthesizing a much larger number of different
immunoglobulin molecules on cells than actually appear in detectable amounts
in the bloodstream. In other words, the immunoglobulin molecules whose
properties we can examine may represent only a minor fraction of those for
which genetic information is available.
One may distinguish two levels of expression in the synthesis of immunoglobulins that I have termed for convenience the p rimotype and the c lonotype
(4). The primotype consists of the sum total of structurally different immunoglobulin molecules or receptor antibodies generated within an organism Fig. 10.
A model of the somatic differentiation of antibody-producing cells according to the clonal
selection theory. The number of immunoglobulin genes may increase during somatic growth
so that, in the immunologically mature animal, different lymphoid cells are formed each
committed to the synthesis of a structurally distinct receptor antibody (indicated by an
arabic number). A small proportion of these cells proliferate upon antigenic stimulation
to form different clones of cells, each clone producing a different antibody. This model
represents bone marrow-derived (B) cells but with minor modifications it is also applicable
to thymus-derived (T) cells. d uring its lifetime. The number of different molecules in the primotype is
probably orders of magnitude greater than the number of different effective
antigenic determinants to which the animal is ever exposed (Figure 10). The
clonotype consists of those different immunoglobulin molecules synthesized
as a result of antigenic stimulation and clonal expansion. These molecules
can be detected and classified according to antigen-binding specificity, class,
antigenic determinants, primary structure, allotype, or a variety of other
experimentally measurable molecular properties. As a class, the clonotype is
smaller than the primotype and is wholly contained within it (Figure 10).
Although a view of the clonotype is afforded by the analysis of humoral
antibodies, we know very little about the primotype. It is therefore important Fig. 11.
Lymphoid cells from mouse spleen bound by their antigen-specific receptors to a nylon fiber
to which dinitrophenyl bovine serum albumin has been coupled. Treatment of bound cells in
(a) with antiserum to the T cell surface antigen θ a nd with serum complement destroys the T cells leaving B cells still viable and attached (b). See Table 2. Magnification: X235. Antibody Structure and Molecular Immunology 49 to attempt to fractionate the cells of the immune system according to the
specificity of their antigen-binding receptors (61). We have been attempting
to approach this problem of the specific fractionation of lymphocytes using
nylon fibers to which antigens have been covalently coupled (62,63). The
derivatized fibers are strung tautly in a tissue culture dish so that cells in
suspension may be shaken in such a way as to collide with them. Some of the
cells colliding with the fibers are specifically bound to the covalently coupled
antigens by means of their surface receptors. Bound cells may be counted
microscopically in situ by focusing on the edge of the fiber (Figure 11). After
washing away unbound cells, the specifically bound cells may be removed by
plucking the fibers and shearing the cells quantitatively from their sites of
attachment. The removed cells retain their viability provided that the tissue
culture medium contains serum.
Derivatized nylon fibers have the ability to bind both thymus-derived
lymphocytes (T cells) and bone marrow-derived (B cells) (64) according to
the specificity of their receptors for a given antigen (65) (Figure 11, Table 2).
Characterization of mouse lymphoid cells fractionated according to their antigen-binding
specificities. Nylon fibers were derivatized with hapten conjugates of bovine serum albumin
and mice were immunized with each of the designated haptens coupled to hemocyanin.
Inhibition of binding was achieved by addition of hapten-protein conjugates ( 250 µ g / m l )
or rabbit anti-mouse immunoglobulin (Ig) (250 µg/ml) to the cell suspension. High avidity
cells are defined as those which are prevented from binding by concentrations of Dnpbovine serum a lbumin of less than 4 µg/ml in the cell suspensions. Cells inhibited by higher
concentrations are defined as low avidity cells. Virtually complete inhibition occurs at
levels of homologous hapten greater than 100 µg/ml.
Antigen on Fiber Immunization none Cells Bound to Fiber
% (per cm)
Inhibition of Binding by:
Cells (per cm)
Cells (per cm)
% T Cells
% B Cells none 50 Physiology or Medicine 1972 About 60 % of spleen cells specifically isolated are B cells and the remainder
are T cells. By the use of appropriate antisera to cell surface receptors (Table
2), the cells of each type can be counted on the fibers and most of the cells
of one type or the other may then be destroyed by the subsequent addition
of serum complement. In this way, one can obtain populations of either T or
B cells that are highly enriched in their capacity to bind a given antigen
(Figure 11) .
Cells of either kind may be further fractionated according to the relative
affinity of their receptors. This can be accomplished by prior addition of a
chosen concentration of free antigen, which serves to inhibit specific attachment
of subpopulations of cells to the antigen-derivatized fibers by binding to their
receptors. As defined by this technique, cells capable of binding specifically
to a particular antigen constitute as much as 1 % of a mouse spleen cell
population. Very few of these original antigen-binding cells appear to increase
in number after immunization, however, and the cells that do respond are
those having receptors of higher relative affinities (62) (Table 2).
Whether these populations correspond to the primotype and clonotype
remains to be determined. It is significant, however, that fiber-binding cells
do not include plaque forming (66) cells, and it is therefore possible to fractionate antigen-binding cells from cells that are already actively secreting antibodies. Recent experiments indicate that the antigen-binding cells isolated
by this method may be transferred to irradiated animals to reconstitute a
response to the antigen used to isolate them. This suggests that the antigenspecific population of cells removed from the fibers contains precursors of
We have been rather encouraged by these findings, for the various methods
of cell fractionation appear to have promise not only in determining the specificity and range of T and B cell receptors for antigens but also in analyzing
the population dynamics of T and B cells in both adult and developing
animals. Now that fractionated populations of lymphocytes specific for
particular antigens are available, it should be possible to determine the
connection between lectin-induced and antigen-induced changes by comparing
responses to both agents on the same cells.
Although many experiments remain to be done in this area of the molecular
immunology of the cell surface, continued analysis of the mitogenic mechanism
should undoubtedly clarify the problems of immune induction and tolerance.
The results obtained using lymphocytes may also have general significance,
however, and bear upon the nature of cell division in normal and tumor cells
as well as upon growth control and cell-cell interactions in developmental
biology. Immunology can be expected to play a double role in these area of
study, for it will be a tool as well as a model system of central importance.
C ONCLUSION Immunology has been and is a curiously reflexive science, generating its
own tools for understanding, such as antibodies to antibody molecules themselves. While this approach is a powerful one, a fundamental understanding Antibody Structure and Molecular Immunology 51 of immunological problems requires chemical analysis. The determination
of the molecular structure of antibodies is a persuasive example and its virtual
completion has allied immunology to molecular biology in a very satisfying
1) The heterogeneity of antibodies and complexity of immunoglobulin
classes have been rationalized in a fashion consistent with selective theories
2) The structural basis for differentiation of the biological activity of
antibodies into antigen-binding and effector functions has been made clear.
3) The detailed analysis of antibody primary structure has provided a
basis for studying the molecular genetics of the immune response, particularly
the origin of diversity and the commitment of each cell to the synthesis of one
kind of antibody.
4) A general framework has been provided for studying antibodies at the
cell surface, opening several molecular approaches for analyzing stimulation
and cell triggering.
5) Finally, it is perhaps not too extravagant to suggest that the extensions
of the ideas and methods of molecular immunology to fields such as developmental biology has been facilitated. In this sense, immunology provides an
essential tool as well as a model with distinct advantages: dissociable cells
with unique gene products of known structure; the capacity to induce specific
cloned cell lines for in vitro analysis; the means to fractionate cells according
to their state of differentiation and binding specificity, allowing quantitative
studies of their selection, interaction and population dynamics.
Whether or not the immune response turns out to be a uniquely useful
model, we can expect that continued work by molecular and cellular immunologists will solve the major problems of the origin of diversity and the induction
of antibody synthesis and tolerance. In view of the intimate connection of
these problems with problems of gene expression and cellular regulation, their
solution should bring valuable insights to other important areas of eukaryotic
biology and again transform immunology both as a discipline and as an
increasingly important branch of medicine.
A CKNOWLEDGEMENTS By its very nature, science is a communal enterprise. I am deeply aware of
the essential contributions to this work made by my many colleagues and
friends throughout the last fifteen years. This occasion recalls the daily life
we have shared with warmth and affection as well as the personal debt of
gratitude that I owe them. I am equally cognizant of the fact that the knowledge of antibody structure was developed by many laboratories and researchers
throughout the world. Not all of this work has been cited, for specific recognition here runs the risk of an unintentional omission; reference may be made
to the reviews cited in the bibliography.
In addition to the fundamental support of the Rockefeller University, the
work of my colleagues and myself was supported by grants from the National
Institutes of Health and the National Science Foundation. 52 Physiology or Medicine 1972 R EFERENCES
1. Cold Spring Harbor Symposia on Quantitative Biology, “Antibodies,“ 32, 1967.
2. Nobel Symposium, 3, Gamma Globulins, Structure and Control of Biosynthesis (Killander, J. editor), Almqvist and Wiksell, Stockholm, 1967.
3. Edelman, G. M. and Gall, W. E., Ann. Rev. Biochem., 38, 415, 1969.
4. Gaily, J. A. and Edelman, G. M., Ann. Rev. Genet., S, 1, 1972.
5. Jerne, N. K., Proc. Natl. Acad. Sci. U.S., 4 1, 849, 1 9 5 5 .
6. Burnet, F. M., The Clonal Selection Theory of Acquired Immunity, Vanderbilt
University Press, Nashville, Tennessee, 1959.
7. Landsteiner, K., The Specificity of Serological Reactions, 2nd ed., Harvard University
Press, Cambridge, Massachusetts, 1945.
8. Marrack, J. R., The Chemistry of Antigens and Antibodies, 2nd ed., (Medical Research
Council Special Report Series, No. 230), London, His Majesty’s Stationery Office,
9. Pedersen, K. O., Ultracentrifugal Studies on Serum and Serum Fractions, Uppsala,
Almqvist and Wiksell, 1945.
10. Tiselius, A., Biochem. J., 3 1, 3 13; 1464, 1937.
11. Karush, F., Advan. Immunol., 2, 1, 1962.
12. Petermann, M. L., J. Biol. Chem., 144, 607, 1942.
13. Porter, R. R., Biochem., 73, 119, 1959.
14. Nisonoff, A., Wissler, F. C., Lipman, L. N. and Woernley, 1). L. Arch. Biochem.
Biophys., 89, 230, 1960.
15. Edelman, G. M., J. Am. Chem. Soc., 81, 3155, 1959.
16. Edelman, G. M. and Poulik, M. D., J. Exp. Med., 113, 861, 1961.
17. Edelman, G. M. and Gally, J. A., J. Exp, Med., 116, 207, 1962.
18. Schwartz, J. and Edelman, G. M., J. Exp. Med., 1 18, 4 1 , 1 9 6 3 .
19. Berggird, I. and Edelman, G. M. Proc. Natl. Acad. Sci. U.S., 49, 330, 1963.
20. Edelman, G. M., Benacerraf, B., Ovary, Z., and Poulik, M. D., Proc. Natl. Acad. Sci.
U.S., 47, 1751, 1961.
2 1. Koshland, M. E. and Englberger, F. M., Proc. Natl. Acad. Sci. U.S., 50, 61, 1963.
22. Hilschmann, N. and Craig, L. C., Proc. Natl. Acad. Sci. U.S., 53, 1403, 1965.
23. Titani, K., Whitley, E., Jr., Avogardo, L. and Putnam, F. W., Science, 149, 1090, 1965.
24. Bull. World Health Org., 30, 447, 1964.
25. Marchalonis, J. and Edelman, G. M., J. Exp. Med., 1 22, 6 01, 1965; 124, 901, 1966.
26. Fleishman, J. B., Pain, R. H., and Porter, R. R., A rch. Biochem. Biophys., Supplement
1, 174, 1962; Fleishman, J. B., Porter, R. R. and Press, E. M., Biochem. J., 88, 220,
27. Fougereau, M. and Edelman, G. M., J. Exp. Med., 1 21, 373, 1 9 6 5 .
28. Edelman, G. M., Gall, W. E., Waxdal, M. J. and Korigsberg, W. H., Biochemistry, 7,
Waxdal, M. J., Konigsberg, W. H., Henley, W. L. and Edelman, G. M., Biochemistry,
7, 1 9 5 9 , 1 9 6 8 ;
Gall, W. E., Cunningham, B. A., Waxdal, M. J., Konigsberg, W. H. and Edelman,
G. M., Biochemistry, 7, 1973, 1968;
Cunningham, B. A., Gottlieb, P. D., Konigsberg, W. H. and Edelman, G. M., Biochemistry, 7, 1983, 1968;
Edelman, G. M., Cunningham, B. A., Gall, W. E., Gottlieb, P. D., Rutishauser, U.
and Waxdal, M. J., Proc. Natl. Acad. Sci. U.S., 63, 78, 1969.
Gottlieb, P. D., Cunningham, B. A., Rutishauser, U. and Edelman, G. M., Biochemistry,
9, 3155, 1970;
Cunningham, B. A., Rutishauser, U., Gall, W. E., Gottlieb, P. D., Waxdal, M. J. and
Edelman, G. M., Biochemistry, 9, 3161, 1970;
Rutishauser, U., Cunningham, B. A., Bennett, C., Konigsberg, W. H. and Edelman,
G. M., Biochemistry, 9, 3171, 1970; Antibody Structure and Molecular Immunology 53 Bennett, C., Konigsberg, W. H., and Edelman, G. M., Biochemistry, 9, 3181, 1970;
Gall, W. E., and Edelman, G. M., Biochemistry, 9, 3188, 1970;
Edelman, G. M., Biochemistry, 9, 3197, 1970.
29. Amiraian, K. and Leikhim, E. J., Proc. Soc. Exptl. Biol. Medl, 108, 454, 1 961 ;
Taranta, A. and Franklin, E. C., Science, 134 , 1 9 8 1 , 1 9 6 1 .
30. Hill, R. L., Delaney, R., Fellows, R. R., Jr., and Lebovitz, H. E., Proc. Natl. Acad.
Sci. U.S., 56, 1762, 1966.
31. Singer, S. J. and Doolittle, R. E., Science, 153, 13, 1966.
32. Press, E. M. and Hogg, N. M., Nature, 223, 807. 1969.
33. Cunningham, B. A., Gottlieb, P. D., Pflumm, M. N. and Edelman, G. N. in Progress
in Immunology, (B. Amos, ed.), Academic Press, I n c . , N . Y . , p p . 3 - 2 4 , 1 9 7 1 ; C u n ningham, B. A., Pflumm, M. N., Rutishauser, U., and Edelman, G. hf., Proc.
Natl. Acad. Sci. U.S., 64, 997, 1969.
34. Franck, F. and Nezlin, R. S., Biokhimia, 28, 193, 1963.
35. Edelman, G. M., Olins, D. E., Gally, J. A. and Zinder, N. D., Proc. Natl. Acad. Sci.
‘U.S., 50, 753, 1963.
36. Olins, D. E. and Edelman, G. M., J. Exp. Med., 119, 799, 1964.
37. Haber, E., Proc. Natl. Acad. Sci. U.S., 52, 1099, 1964.
38. Inbar, D., Hachman, J. and Givol, D. Proc. Natl. Acad. Sci. U.S., 69, 2659, 1972.
39. Wu, T. T. and Kabat, E. A., J. Exp. Med., 1 3 2 , 2 1 1 , 1 9 7 0 .
40. Edelman, G. M. Ann. N.Y. Acad. Sci., 190, 5, 1 9 7 1 .
41. Kehoe, J. M. and Fougereau, M., Nature, 224, 1212, 1970.
42. Yasmeen, D., Ellerson, J. R., Dorrington, K. J., and Pointer, R. H., J. Immunol., in press.
43. Berggård, I. and Bearn, A. G., J. Biol. Chem., 243, 4095, 1968.
44. Smithies, 0. and Poulik, M. D., Science, 175, 1 8 7 , 1 9 7 2 .
45. Peterson, P. A., Cunningham, B. A., Berggård, I. and Edelman, G. M. Proc. Natl.
Acad. Sci. U.S., 69, 1967, 1972.
46. Poljak, R. J., Amzel, L. M., Avey, H. P., Becka, I,. N. and Nisonoff, A., Nature New
Biol., 235, 137, 1972.
47. Davies, D. R., Sarma, V. R., Labaw, L. W., Silverton, E. W. and Terry, W. D., in
Progress in Immunology, ( B. Amos, e d . ) , A c a d e m i c P r e s s , N . Y . , p p . 2 5 - 3 2 , 1 9 7 1 .
$8. Gall, W. E. and D’Eustachio, P. G., Biochemistry. 11, 4 6 2 1 , 1 9 7 2 .
49. Hood, L., Gray, W. R., Sanders, B. G. and Dreyer, W. J., Cold Spring Harbor Symp.
Quant. Biol., 32, 133, 1967.
50. Milstein, C., Nature, 216, 330, 1 9 6 7 .
51. Grubb, R., T he Genetic Markers of H uman Immunoglobulins, S pringer-Verlag BerlinHeidelberg-New York, 1970; Acta Path. Microbial. Scand., 39, 195, 1956.
5 2 . O u d i n , J . , C o m p t . R e n d . A c a d . S c i . , 242, 2489; 2606, 1 956; J. Exp. Med., 112, 1 2 5 ,
5 3 . M e d a w a r , P . B . i n L e s P r i x N o b e l e n 1 9 6 0 , I mprimerie Royal, P. A. Norstedt and
54. Sharon, N. and Lis, H., Science, 177, 949, 1972.
55. Edelman, G. M., Cunningham, B. A., Reeke, G. N., Jr., Becker, J. W., Waxdal, M. J.,
and Wang, J. L., Proc. Natl. Acad. Sci. U.S., 69, 2580, 1972.
56. Yahara, I. and Edelman, G. M. Proc. Natl. Acad. Sci. U.S., 69, 608, 1972.
57. Taylor, R. B., Duffus, P. H., Raff, M. C. and DePetris, S., Nature New Biol., 233, 225,
1971; Loor, R., Forni, L. and Pernis, B., Eur. J. Immunol., 2, 203, 1972.
58. Andersson, J., Edelman, G. M., Möller, G. and Sjöberg, O., Eur. J. Immunol., 2,
233, 1 972.
59. Greaves, M. F. and Bauminger, S., Nature New Biol., 235, 67, 1972.
60. Nossal, G. J. V. and Ada, G. L., Antigens, Lymphoid Cells and the Immune Response,
Academic Press, N.Y., 197 1.
61. Wigzell, H. and Andersson, B., J. Exp. Med., 129, 23, 1 9 6 9 .
62. Edelman, G. M., Rutishauser, U. and Millette, C. F. Proc. Natl. Acad. Sci. U.S.,
68, 2153, 1971. Physiology or Medicine 1972 54 63. Rutishauser, U., Millette, C. F. and Edelman, G. M. Proc. Natl. Acad. Sci. U.S., 69,
64. Gowans, J. L., Humphrey, J. H. and Mitchison, N.A., A discussion on cooperation
between lymphocytes in the immune response. Proc. Roy. Soc. London B, 176, No.
1045 , p p . 3 6 9 - 4 8 1 , 1 9 7 1 .
65. Rutishauser, U. and Edelman, G. M. Proc. Natl. Acad. Sci. U.S., 69, 3774, 1972.
66. Jerne, N. K., Nordin, A. A. and Henry, C. in “Cell Bound Antibodies,“ (B. Amos
and H. Koprowski, eds.), Wistar Institute Press, pp. 109-125, 1963. ...
View Full Document