Three Dimensional Structure of Immunoglobulins

ANNUAL REVIEWS
Copyright 1975. All rights reserved
Further
Quick links to online content
THREEDIMENSIONAL STRUCTURE OF IMMUNOGLOBULINS

David R. Davies and Eduardo A. Padlan
:-:897
Annu. Rev. Biochem. 1975.44:639-667. Downloaded from www.annualreviews.org by Naresuan University on 04/08/13. For personal use only.
Laboratory of Molecular Biology, National Institute of Arthritis, Metabolism, and Digestive Diseases, National Institutes of Health, Bethesda, Maryland 20014
David M. Segal
Immunology Branch, National Cancer Institute,
Bethesda, Maryland 20014
CONTENTS
INTRODUCTION. CHEMICAL STRUCTURE.
Light Chains.. . Heavy Chains .....

J Chain....
Hypervariability ...........
OVERALL THREE-DIMENSIONAL STRUCTURE.
Physical Chemistry and Electron Microscopy. Preliminary X-Ray Studies on F e F ragments . Low Resolution X-Ray Analysis of the Intact Molecule.
HIGH RESOLUTION CRYSTAL STRUCTURES.
Crystallographic Studies on the Meg Bence-Jones Protein. Structure of the Meg Light Chain Dimer ........... . Binding Studies on Meg Crystals... Crystal Structure of Bence-Jones Protein REI Crystallographic Studies on Human Fab Fragments...
. . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . .. . . . .
. . . . . .
COMPARISON OF STRUCTURES.
Crystallographic Studies 011 Mouse Myeloma Proteins with Antigen Binding Properties. . ........ . ..
Domain Tertiary Structure.. Quaternary Structure. . .........

CONCLUSION
The Antigen Binding Site. . ... . ... . .
639 640 640 641 642 642 643 643 643 643 644 644 645 648 649 650 652 655 656 659 662 664
INTRODUCTION
The immune system of vertehrate species is characterized by the capacity to synthesize humoral antibodies in response to challenge with antigen. These antibody molecules will bind to the antigen and with multivalent antigens will cause precipitation by formation of a crosslinked lattice. In vivo, antibodies can also play a crucial role in a variety of cellular and humoral responses.
639
640
DAVIES, PADLAN & SEGAL
Except under very special conditions, the antibodies produced generally form a heterogeneo us spectrum of molecules, both in terms of amino acid composition and seq uence and with regard to the strength of binding to antigen . Because o f th is heterogeneity, most structural studies have been carried o ut on homogeneous myeloma proteins from t umor cells. There is now a considerable body o f evidence to show that some o f these proteins are imm unologically, chemically, and physically indistinguishable from natural antibodies (1-5). Since the last reviews on this s ubject (2, 6) there has been tremendous progress in o ur understanding o f the three-dimensional structure o f antibodies and the nature o f the interaction between antibody and antigen. High resolution X-ray crystallography on a n umber o f immunoglobulin fragments has shed a great deal o f light on antibody structure. In this review we shall to a large extent be concerned with these studies and with other investigations that relate either directly or indirectly to these structure determinations. We make no claim to have been comprehensive in these other areas and rather have been q uite selective. A n umber o f recent reviews have been written on physical-chemical studies (7-9) and on the amino acid sequences ( 10-12) o f imm unoglobulins.
CHEMICAL STRUCTURE
Struct ural immunology is unusual in th at long before the recent crystallographic contributions, many ideas on structure and function had been proposed based on a wealth o f chemical data. These include the concept o f the basic immuno globulin structure along with the location o f the antigen binding site and the concept of domain structure. The molecular organization of immunoglobulins has b een discussed in many previo us reviews (2, 6, 8, 9, 1 3, 14). The basic structure consists of two identical light (L) chains of molecular weight 22,500 and two heavy (H) chains of molecular weight 50,000-75,000 linked by noncovalent interactions and disulfide bridges to form a structure with twofold symmetry (Figure 1). This basic immunoglobulin structure can join with other like structures through dis ul fide bridges between the heavy rx and J1 chains to form IgA molecules (chiefly dimers) and IgM molecules (chiefly pentamers), respectively. Earlier reviews s ummarize the vario us classes o f immunoglob ulins (2, 6, 8). Porter ( 1 5) showed that limited proteolysis results in three fragments of ro ughly equal weight, two Fabs and one Fc (Figure 1). The antigen binding site was localized on the Fab part of the molec ule, whereas the Fc portion plays an important role in complement fixation and other biological functions. Between each Fab and the Fc lies a "hinge" r egion, some 25 residues long, containing the interchain disul fides and noteworthy for often having an unusually high percentage o f proline (16).
Light Chains
The light chains are on average 2 1 7 amino acids long (16). They may be divided into two groups, K and A, clearly distingui shable by their amino acid sequence.
THREE-DIMENSIONAL STRUCTURE OF IMMUNOGLOBULINS
641
The relative prop or ti ons of K and A vary considerabl y with species, being 65 to 35% in humans and 97 to 3 % in mice. Within each group there are several subgroups that ma y be identified by the degree of sequence homology within the subgroup (16). Each light chain is divided into two approximatel y equal parts, the amin o ter minal V half w hich is quite variable between different i mmun oglobulins, and the carbox yl -terminal C half w hich is constant or common in different light chains of the same group ( 17). L ig ht chains, either w hole or fragmented, have been observed for over 100 years ( 1 8) in the urine of patients with mul tiple myeloma (Bence J ones proteins).
Heavy Chains
Depending on the i mmun oglobulin class, heav y chains con tain ab out 450 or 576 amino acids. Appreciable variation arises from inserti ons and deleti ons in the variable region as well as in the regi on near the hinge residues. The original sequence determinati on of a human )II molecule revealed that the)' heavy chain has t hree C H domains with homol ogous sequence in addition to the amino- terminal VH d omain ( 19, 20). The)' chain C d omains are also homologous with the light chain C d omain. No homolog y exists between the variable and constant domains. Nevertheless, the y are of similar length and both p ossess an internal disulfide l oop of abou t 60 (C) or 6 7 (V) amin o acid residues, s tarting at abou t position 23. I t has been
Figure 1
Schematic representation of polypeptide chains in the basic immunoglobulin
structure.
64 2
DA VIES, PADLAN & SEGAL
propo sed that all do main s form similar di screte globular units, each speciali zed for carrying out a parti cular function ( 19, 20). From 1 2 to 18 residues beyond the first cysteine of t he internal di sulfide is an "invariant" tryptophan. The YH region sequences, like t ho se of YL, permit separation into subgroups, d istingui shable by very high homology within a subgroup and by having less t han 50% homology between member s of different subgroups. T he complete amino acid sequence of the 576 residues of the fl c hain of a human IgM i mmunoglobulin ( Ou) has recently been reported ( 21 ). This chain has five domains of sequence homology, two (YH and C"I ) are associated with light chain, and two (C3 and C4) are released in t he Fc fragment after limited tryptic dige stion. An intervening do main (C2) and a segment t hat probably corresponds to t he hinge region of y chain s are degraded during thi s digestion. T he intersubunit cy steine u sed to for m the pentameric IgM is located in C3 at residue 414. Preliminary result s indicate that IgE, another class of i mmunoglobulin, also has five heavy chain domain s (22).
J Chain
A t hird type of polypeptide, known as J c hain, has been found in polymeric IgA and IgM ( 23-28) from a variety of species ( 29-31). Becau se J is always found with polymeric immunoglobulin and never with the monomer, it is generally con sidered to be involved with joining [hence t he term "J" ( 23)J i mmunoglobulin monomers. J chain has been i solated by a variety of techniques ( 24 -28, 32, 33). It has a molecular weig ht of approximately 1 5,000 ( 24, 25, 34) and an axial ratio of 1 7.9 ( 32). J c hain is bound covalently to polymeric immunoglobulin via di sulfide bridges ( 23). A ssuming one mole of J bind s to one mole of IgM pentamer ( 24, 25), it appear s t hat t he J chain does not contain su fficient SH groups to bind to every fl c hain ( 32). Alpha chains bind J at t he penultimate cysteine ( 35). In one study, J chain was removed from IgM pentamer without cau sing t he polymer to convert to monomer ( 36). It t hu s appears J does not link all mono mer s together directly (32, 36).
H ypervariability
Comparison of YL sequences revealed that some regions, notably around residue s 24-34, 89-96 ( 13, 37 , 38), and 5 0--5 5 (39), showed unu sually high variability. Wu & Kabat (40) carried out a statistical analysi s of the data t hat firmly establi shed these t hree regions as hypervariable. Similar region s ( 31-35, 5 0--65, 95-10 2) were noted in t he heavy chain s by Kabat & Wu (4 1) and Capra & Kehoe ( 12, 4 2), who also ob served hypervariability at po sitions 8 1-85. In subsequent discu ssion s t he first, second, and third hypervariable regions of t he light chain will be de signated L1, L2, and L3, respectively; the homologous hypervariable region s in the heavy chain will be designated HI, H2, and H3, and the additional hypervariable region (resid ues 8 1 -85) will be called He. It was propo sed ( 38, 40) t hat these hypervariable region s would be fo und together at the antigen binding site. The amino aeid residues involved in binding were independently determined
THJ{EE-DIMENIUNAL STRUCTURE OF IMMUNOGLOBULINS
643
by the me thod of affinity lab eling (43). In this techniqu e a reagen t, s tructurally similar t o the hapten ligand, is covalently b ound to the immunoglobu lin and th e p ositi on of the amin o acid side chains with which i t r eacts is d eterm in ed. In a recent r eview Givol (44) has listed the affinity- labeled residues and it is apparent that t hey are a ll located in the hypervariable regions of b oth L and H chains. M oreover, all t he hypervariable regi ons, wi th the excepti on of that at position 8 1-85 of the H chain , were labeled in one prot ein or another.
OVERALL THREE-DIMENSIONAL STRUCTURE
Physical Chemistry and Electron Microscopy
The m odel of three r elatively rigid fragments linked by a flexible h inge region was prop osed originally by N oe lken ei al (45) on the basis of a physical chemical analysis of the fragm ents and the intact IgG molecule. Th ese au th ors als o concluded that there was relative ly lit tle C( h e lix i n the s tructure. This overall model was con firmed by the electron m icrographs (46) of rabbit anti-DNP (2, 4dinitr op henyl) antib odies lin ked by a bivalent hapten. The aggregation of the basic immun oglobu lin structure to form IgA (dimers) and IgM (mos tly p en tamers) has b een confirmed by many e legan t micrographs (47-58). In IgA the molecules join thr ough th e Fc r egi on to form end- to-end dimers. In IgM t he y aggregate again t hrough t he Fc to for m a pentameric structure with a central disc. T he studies of Feinstein and colleagues (55-58) emphasized the flexibili ty of attachment of th e individual units to the central disc.
Preliminary X-Ray Studies on Fc Fragments
Since the Fc p ortion is id en tical in a wid e variety of oth erwise heterogen eous antibodies, large amounts of pure fragment can be obtained b y digestion of normal IgG. In fact, Fc from rabbit IgG crystallizes spontan eously from the digestion mix ture ( 15). Preliminary X-ray data have b e en presented for human and rabbit Fc, but early work (59, 60) was h indered by twinning of the crysta ls. Crystals more su itable for high resolu tion X-ray studies were obtained by Humphrey (6 1 ) fro m papain digests of a human IgG myeloma protein. The s ymmetry of these cr ystals changed when they were soaked in dilute solutions of KAuCh or o-chlor om ercuriphen ol. The n ew crystal form produced by this soaking contained only half an Fc fragment per as ymmetric unit, indicating that the fragment must have a twofold symm etry axis (62, 63).
Low Resolution X-Ray Analysis of the Intact Molecule
Crysta llization of the human yGl cryog lobu lin D ob (64) provided some information about the intact m olecule. In particu lar, it emphasized the dyad axis of the mol ecul e with t he Fc par t l ocated on a twofold symmetry axis. Similar observati ons of twofold symmetry havc bcen reported for other intact human IgG molecules ( 65, 66). The work with D ob was extended (67-70) to provide a l ow r es olution (6 A) elec tron d ensity map of the protein . This exhibi ted three g lobular regions of d ensity correspond ing to the Fc and two Fabs. Because of th e crystal packing i t
644
DA VIES, PADLAN & SEGAL
was di ffic ult at this resol ution to distinguish between various ways o f linking the Fc to the two Fabs, b ut a T-shaped model was preferred. The dimensions for this model placed the two combining sites 140 A apart, at the ends of the arms o f the T. An accompanying electron microscope investigation of these crystals yielded essentially the same result (69-72). Difficulties with crystal q uality have hitherto limited the X-ray analysis to low resol ution. It is interesting to note that in Dob and in one other crystalline intact IgG (Mcg), there is a deletion in the hinge region of the molecule (73, 74).
HIGH RESOLUTION CRYSTAL STRUCTURES
All the high resolution results hitherto achieved have dealt with fragments of the intact molecule. This has been d ue pri marily to the difficulty of getting acceptable crystals o f the intact molecule, together with the obvious attractions o f having to collect and analyze fewer data for the fragments. To date, there are reports of fo ur struct ures fro m fo ur di fferent laboratories (Table 1). Two of these, the h uman light chain di mer and the h uman Fab fragment, appeared approxi mately simultaneo usly and have been closely followed by the struct ures of the h uman Bence-Jones V L di mer and the mouse myelo ma Fab with phosphorylcholine binding specificity. Already, then, it is possible to make so me co mparisons of the struct ures, although because most of the p ublished results are preliminary and, with one exception (75), do not include atomi c coordinates, s uch co mparisons will only be possible at a q ualitative level. Where specific amino acid resid ues are referred to, the sequence n umbering is taken from the original p ublications.
Crystallographic Studies on the M cg Bence-Jones Protein
Structural studies on the Mcg Bence-Jones protein are particularly interesting since crystals of the whole IgG1 (65, 76) myeloma protein fro m the same patient have been obtained. This whole molecule pres umably contains the same light chains as fo und in the Bence-Jones dimer, and its crystal structure is being determined.
Table 1
High resolution immunoglobulin structures Molecular Source composition Human L-chain dimer Fab (pepsin) Class or type
A.
Protein Mcg
Resolution 2. 3 A
Bound ligands Dnp compounds and many other small molecules Vitamin K-l O H and several others
Reference 65, 76-82,88

83,94103
New
Human
}c,yl
1(1
2.8 A
2.8
REI
Human
VL dimer Fab(pepsin)
75,89, 92, 93 Phosphorylcholine

105-110
McPC 603 Mouse
1(,G(
3. 1 A
645
The Mcg Bence-Jones protein dimer is made up of two lambda-type light chains joined together by a disul fide bond between the penultimate cysteine residues.
Structure of the M cg Light Chain Dimer
The crystal structure of the Mcg dimer (77) has been analyzed s uccessively at 6 (78), 3.5 (79), and 2.3 (80) A resol utions. Even at 6 A resolution, it was clear that the di mer was represented by two globular mod ules differing in size and structure. The electron density map at 3.5 A resol ution ( 78) was s ufficiently clear to trace the course of the polypeptide chains. The large and s mall mod ules previously observed at low resolution were identified as the V and C regions, respectively, on the basis of the known sequence of th e C-terminal half and on the location of the mercury atom insert ed into th e interchain disulfide bridge (8 1 ). Th e domains were clearly delineated and the two segments of electron density bridging the two regions were identified as the switch regions between the V and C parts of the amino acid sequence. The analysis at 2.3 A resolution (80), aided by knowledge of the co mplete amino acid sequence (82), has led to a complete molecular model. The i mproved resolution more fully revealed the details regarding the structure and interactions of the vario us domains. The relative disposition of the two monomers and the fo ur domains in the Mcg di mer is shown in Figure 2. Each domain is approximately 40 A long and the dimer is about 77 A long. The two chemically identical monomers are not equivalent in their three-di mensional structures although approximate twofold axes relate the paired do mains in both V and C regions. These axes are not collinear, rath er they intersect at an angle of 120. The spatial relationships b etween the V and C domains of the same chain arc vcry di ffcrcnt in the two monomers. In monomer 1, the distance (center to center) between the intradomain disul fide bonds is 25 A and the angle between the long axes of the V and C do mains is abo ut 70. In monomer 2, the measured val ues are 43 A and about 1 10, respectively. These angles are similar to those measured in the h uman Fab, New (83), with the do mains of mono mer 1 disposed as in the heavy chain and those of monomer 2 as in the light chain of the Fab. This, and the observation that the additional disulfide bridge fo und in rabbit light chains (84) could be for med
Schematic representation illustrating steric relationship of the four domains in Mcg light chain dimer (79). The appropriate twofold axes relating VI: V2 and CI: C2 are indicated. They intersect at an angle of about 120'. Reprinted with permission from Schiffer et al 1973. Biochemistry 1 2 : 4620. Copyright by the American Chemical Society.
Figure 2
646
between the correspo nding residues (82, 174 i n Mcg) i n mo no mer 2 b ut not i n mo nomer 1 , led the authors to s uggest that mo no mer 2 i s a c loser homolog to light chains and that mo no mer 1 fi lls the struct ural role of the heavy chai n in Fab (79, 80). Differe nces in the spatial relationships lead to di fferenc es in the longitudina l i nteractions between the V a nd, C domai ns o f the two Mcg mo nomers. Thus, i n mo no mer 1 , the ring o f Pro 8 a nd the mai n chain between A la 147 a nd Val 148 are very close, as are Ser 9 a nd Ser 11 and the side chain of Glu 202 (80). I n mo no mer 2, the correspondi ng residues are widely separated. The V a nd C domai ns were fo und to have very simi lar tertiary structur es. The basic do mai n str ucture co nsists of two layers of antiparallel segments (Figure 3). There are fo ur segme nts in o ne layer and three i n the other. The i ntradomai n disulfide bond is located i n the center o f each domai n, connecti ng the middle segme nt of the three chain layer with a parallel strand i n the other layer. The i nvariant tryptophan [residues 37 i n V and 1 52 i n C (82)] lies close to the disulfide a nd the i nterior of the domain is fi lled with hydrophobic side gro ups. In both layers, the for mation of sev eral hydroge n bonds is possibl e between adjacent segments. The commo nly observed "twisted sheet " o f fJ struct ure is seen i n the Mcg dimer, particularly in the C domai ns. The arrangeme nt of the segments in the C do main is mor e regular, with the a ntiparallel chains for mi ng two distinct layers. I n the V do mai n, the N-termi nal segme nt lies in the fo ur chain layer but is also close a nd parallel to the C-termi nal segme nt which is i n the three chain layer. I n addition to the more irreg ular arrangement of segme nts and layers i n the V than in the C domai n ( Fig ure 3), a n
Figure 3 Schematic representation o f monomer 2 o f the Mcg dimer (79). Arrows indicate direction of polypeptide chain. In both domains, white arrows represent stretches of extended chain lying in one plane, hatched arrows represent stretches lying in the other. Disulfide bonds are indicated as black bars connecting the two planes in each domain. Reprinted with permission from Schiffer et al 1973. Biochemistry 12 : 4620. Copyright by the American Chemical Society.
647
irregular loop is fo und in the former whi ch has no equivalent in the latter. This additional loop contains the second h ypervariable region. Two short right-handed helices were fo und in the C domains involving residues 1 2 5- 1 32 and 1 8 5-192. Although the two V domains are related by an approximate twofold axis, distinct differen ces between them were observed. The first h ypervariable loops, for example, were fo und to be signi ficantl y di fferent in the two domains. In addition, the struct ure of the V domain of monomer 1 appears to be generall y less ordered with apparently less h ydrogen bonding between segments. The C domains are more nearl y eq uivalent. The domains in the M cg A. chain are clearly structurally homologous, yet their association in pairs was fou nd to be very different in the V and C modules. Th us the difference in size of the two regions is d ue to the greater separation of the V do mai ns, reflected in the di stance between the i ntradomain disulfide bonds, which is 24 A in the V module b ut onl y 18 A in the C. The lateral interactions between the paired domains appear to be stronger in the C region. The fo ur chain layers of the C domains form the interface in this half of the dimer. Th ese layers form con cave and h ighl y complemen tary s urfaces with h ydrophohi c residues i nterdigitating across the boundary between domains. In marked contrast, it is fo und that in the V region, the fo ur chain layers are on the o utside. The three chain layers face each other with relatively few interactions between domains (Table 2). Among the few points of contact between th e V domai ns are those between the Tyr 51 and Asp 9 7 side gro ups, and between the GIn 40 si de chai ns from the two monomers. The si de group of T yr 89 is situated close to Pro 46 and to the main chain between Lys 44 and Ala 4 5 of the opposite monomer. The wider separation of the V dom ai ns i n the M cg dimer results i n the presence of a "solvent chann el" between the three chain layers (80). As revealed in the 2. 3 A stud y, the so lven t channel begins at the tip of the molecule as a large conical cavity approximatel y 1 5 A across at the entran ce and 1 7 A deep. At the base of the cavit y is a 5-6 A opening leading to an ellipsoidal po cket with dimensions of 8 x 8 x 10 A (80). The main cavity is lined by 2 1 side chains, most of whi ch emanate from the three h ypervariable regions (residues 2 3-36, 50--56, and 9 1-100) of each monomer. Tyr 34, 5 1 , and 9 3 and Glu 52 from both monomers as well as Asp 9 7 from monomer 2 line the rim o f the main cavity. The walls are formed b y Ser 36 and Ser 91, Val 48 and Phe 99 from both mo nomers. Tyr 38 and Phe 101 form the floor o f the cavity and the roof of the inner po cket. The walls of this pocket consist of T yr 89 , Pro 46, and G in 40 from both monomers, whi ch are not h ypervariable. The positions of various substitutions and antigenic markers were located in the M cg dimer. Position 1 4 7, at which Val was substitu ted for Ala i n protein Mz (8 5), was fo und in the close contact between V and C domains of monomer 1 and on the surface in m onomer 2. The second M z s ubstit ution, L ys for Asn at position 174, was located on an exposed loop in both monomers. The Kern [position 1 56 (86)J and Oz [position 1 9 3 (8 7)J markers were located on the surface on adjacent loops.
648
Table 2 dimers'
Residues in contact across the V region interface in the REI and Mcg Bence-Iones
REI (75) Tyr 36(OH) .. . . . . . ... .. .. . ... . . . . . . .. .. . . . . .. . . . . . . . GIn 89+ (O=C NH2) Tyr 36 . ... .. . . . . ... ... . ... . . .. . .. .. . .. . . . . . . .. ... Phe 98 Pro 44 . . . . . . . . Tyr 87 Pro 44 . . . . . . Phe 98 Ala 43 . . .. . . . . Tyr 87 Lys 42(C=O) . . . . , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tyr 87(OH) GIn 38(O=CNH2)" . . . . GIn 38(O=CHN2) Tyr 49(OH) '" . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tyr 96+ (OH) Tyr 49 . . . . . . . . Leu 94+ GIn 55+ . .. . .. . . Leu 94 + GIn 55+ . . . ... ..... . . . .... . . ...... . ..... . . . . ..... . . .... .Pro 95+
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Mcg (79) Tyr 51 . . Asp 97+ GIn 40 . .... .. . ... ... . . . .... . . ... .. . ..... . ... .. . .... . . ..GIn 40 ... . .. . Tyr 89 backbone between Lys 44 and Ala 45
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
a Each interaction occurs twice because of the twofold axis between domains; exact in REI, approximate in Mcg. The exception is the contact between Tyr 5 1 (monomer 1 only) and Asp 97 (monomer 2 only) in Mcg. The hypervariable residues are indicated by +.
Binding Studies on M cg Crystals
Binding studies in the crystal (88) revealed that dinitrophenyl (DNP) amino acids, purine and pyrimidine derivatives such as caffeine and 5-acetyluracil, aromatic compounds like menadione (vitamin K-3) and colchicine, and lipid components like triacetin bound in the solvent channel of the Mcg dimer. There were two distinct binding sites, A and B, in the main cavity. Site A, which is more spacious, is situated near the entrance but closer to monomer 2 than monomer 1 . The site related t o A b y the approximate twofold axis i s blocked b y another molecule in the crystal. Site B is located near the approximate twofold axis farther down the cavity. Sites A and B act as discrete sites in the binding of small molecules. Large molecules like colchicine, on the other hand, bind in sites A, B, and other parts of the cavity. The most probable contact residues for site A are Tyr 34, Tyr 93, and Glu 52 from monomer 2, with Tyr 5 1 in close proximity. The potential contributors of site B are Phe 99 and 101, Ser 36 and 91, and Tyr 38 from both monomers. An additional binding site was found in the inner pocket. The center of this third sitc is bctween thc two Pro 46 residues, with Tyr 38 and 89 and Phe 101 fro m b oth chains being accessible for binding. The b inding o f some of these compounds produces changes in the protein structure around the b inding site. 5-Iodo-DNP, which was covalently bound to Tyr 34 of
649
mo no mer 2, produced a displaceme nt of the phenolic ring of abo ut 3 A. I n addition, the polypeptide backbo ne from Tyr 34 to Ser 36 was shifted approximately 2.5 A. Large co mpounds like colchici ne were more disruptive. Menadio ne (vitami n K-3) binds in the i nner pocket q uite a vidly, cracks the crystals within 2 hr , a nd causes a displacement of the polypeptide chai n between Lys 44 and Ala 45 in mo no mer 2. The ease with which co mpo unds like menadi one co uld disrupt the crystal structure was attrib uted by the a uthors to the flexibility i n the protein "which permits a n o versized e ntering gro up with the proper affinity t o i nd uce a fi t " i n the binding channel (88).
Crystal Structure of Bence-Jones Protein REI
Crystals s uitable for X-ray analysis have been obtained fro m di mcrs of the variable halves of the K-type Bence-Jo nes protei ns REI (89), L EN (90), and A u (91). The structure of the REI dimer has been determi ned to 2.8 A resolution (75). The REI dimer crystallizes with the two VL fragments related by a local (non
crystallographic) twofold axis (75, 92, 93).
model of one domain was constructed
fro m an electron density map that had been averaged by rotation abo ut this axis. At 2.8 A resolution, the polypeptide chai n is fo und to fold i nto two sheets co vering the hydrophobic i nterior of the domai n as found i n the Mcg dimer. One sheet is described as being for med by fi ve strands and the other by three strands, with both sheets showing the commo nly observed left-handed twist (Figure 9). The hydrogen bo ndi ng between strands is fo und to be characteristic of a ntiparallel pleated sheets with about 50 % of the residues co ntrib uti ng to the fJ-sheet structure. As i n Mcg, the N-termi nal strand adds to the three strand layer i n an antiparallel fashion a nd to the other i n a parallel fashion. One t ur n of a distorted a-II helix for med by resid ues G In 79 to lie 83 was the o nly helical segment found i n the struct ure. Three hairpin t ur ns i nvol vi ng conserved glycine residues (positions 1 6, 4 1 , 57) were observed. Altho ugh the REI YL domai n was observed to have a close struct ural resemblance to other light chain Y do mai ns, differences existed i n the foldi ng of the first hyper variable regio n. This is an extended chai n i n YRE1 a nd is folded i nto a helical structure i n Fab New a nd i n the Mcg light chain, presumably to accommodate the three extra residues i nserted i n this regio n i n the latter protei ns. The REI di mer co ntact is found to i nclude hydrophobic as well as hydrogen bonding i nteractions across the YL: YL i nterface (Table 2). Some contact residues are hypervariable (Figure 12). Findi ng that the YL: YL co ntact i n REI is virtually identical with the YL: YL contact of the Mcg dimer a nd apparently similar to the correspo nding co ntact i n Fab New, the a uthors co nclude that the mode of associatio n o f the Y domai ns does not depend o n the presence o f the C domai ns. The mai n chai n atoms of the hypervariable regions were observed to for m a cavity around the diad axis. Side chai ns, predo mi na ntly fro m the three hyper variable regions of both chai ns, protrude i nto the cavity. T yr 49, 9 1 , and 96 fo rm a "slit pocket" aro und the diad, while Tyr 36 a nd G In 89 form the bottom of the pocket. The walls of the pocket i nclude the polar residues G In 89 a nd Asn 34. The potential bindi ng s urface i nvolves side chai ns and/or mai n chains
650
of the hypervariable segments Tyr 49 to Asn 53, Lys 3 1 to Le u 33, and in particular G in 90 to Thr 97. It was predicted that the arrangement of the residues in the cavity can nicely accommodate a hapten consisting of an aromatic ring with polar s ubstit uents, with the hapten being oriented in the pocket by the six tyrosine side chain s, b ut no binding st udies have yet been reported.
Crystallographic Studies on Human Fab Fragments
The first Fab crystals were obtained from h uman myeloma proteins by Rossi & Nisonoff (94) using papain digestion and later, pepsin (95). Both digests gave isomorpho us crystals, b ut the pepsin crystals were larger (95-97) and were used in s ubsequent investigations. Crystals were obtained from three (o ut of six) myeloma proteins, two of which, New and Hil, were suitable for high resolution studies (96, 97). Since New crystals have one molecule in the asymmetric unit, whereas Hil has two , Fab ' New has been investigated first. Heavy atom derivatives have been obtained and an early 6 A electron density map (98-1 00) was sufficiently detailed at thi s resolution to permit a clean delineation of the molecular boundaries. The Fab ' struct ure consists of two weakly connected, globular struct ural s ub units, each approximately 30 x 40 x 50 A in size, whereas the entire Fab ' is appro ximately 40 x 50 x 80 A. The authors s uggest that each globular portion of the Fab ' mol ecul e represents a region consisting of two domains, one from the L chain and one from the H. Th us one subunit wo uld be the V region (VL: VH) and the other the C region (: 1). This 6 A map was consistent with the domain model for immuno globulin str uct ure originally s uggested by Edelman and co-workers (6, 19, 20). The overall Fab struct ure co uld be described as a distorted tetrahedron, with the domains of homology forming the vertices. An interesting feature of the Fab ' ' New 6 A electron density map was the existence of a cavity-like space at one end of the molecule. This cavity was compatible with the dimensional requirements of the antigen binding site, and on this basis the authors s uggested that the s ub unit containing this feat ure was the V region. This work has been extended to yield a 2.8 A electron density map of Fab' New ( 83). The VL amino acid seq uence has al so been determined (101). The Cr. and 1 seq uences were known, b ut VH was s urmised by homology wi th other h uman VH regions and from the electron density map. Using the known part of the seq uence and the electron density map, a model of the Fab New str ucture was constructed (83). At this resol ution it became clear that the subunit previo usly identified as the V region was in fact the C region . All fo ur domains are similar in their tertiary str uct ures to the domain str uctures described above. Fab' New is atypical in that VL contains a seven resid ue deletion in the second hypervariable loop. Thus the VL domain in New is more similar in struct ure to t he constant domains. In spitc of the fact that the Fab contains two chains of di ffering amino acid sequence, within each region , V or C, the paired domains are related by an appro ximate local twofold rotational axis. The ma jor axes of domains within each chain are also not collinear, the angle between the axes being 100-1 10 in the
651
Lchai n and 80-85 in the Fd. The CL and 1 domai ns i nteract over a wider area than the VL and VH d omai ns. As expecte d, the serol ogic A chai n m arkers, Kern (86) and Oz (87) , lie on the outside of the m olecule, e ach accessible to appropriate typi ng antisera. Disulfide bridges present i n other immu noglobulins, notably in human y2, y3, y4, and Jl ch ains, r abbit y, m ouse y2a and y2b, guinea pig y2a, another human yl, and r abbit K chai ns, could be formed in the New s tructure wi thout otherwise dis torting the molecule. This is taken as evidence th at all of these other classes of immu noglobulin are similar t o New in conform ation. I n b oth L and R chai ns the hypervariable sequences (40) occur at one e nd of the m olecule and are fully exposed to the solvent. The hypervariable regions occur at "adj acent bends of tightly p acked, li near polypeptide ch ai ns. " They form a sh allow groove 1 5 x 6 A i n area and 6 A i n depth. Residues 55-65 a nd 30-33 from the heavy ch ain form the l ower area of the site, while 27a-30 and residues cl ose to 50 of the L chain form the upper bou nd ary. The sides are comprised of residues 90-95 (L ch ai n) on the left and 102-107 (R chain) on the right. The heavy chain contributes more residues t o this region th an the l ight. Recently, it h as been observed th at Fab' New bi nds several apolar small molecules in solu ti on and i n the crystal (102, 103). B ou nd in solution wi th relatively l ow affini ty (103 Vmol) were orcei ne, dichlor ophenolind ophenol, coenzyme Q50, uridi ne, foli c acid, the N-termi nal hex apeptide of MSR- l and menadi one. I n contrast, onc molecule of IgG New (wh ole protein) binds two m olecules of vitamin K- l OR i n s oluti on with an affinity of 1.7 x 105 Vmol. B oth the phytyl tail and the 2-methylnaph thoquinone m oieties of the vi tamin K-l OR contribu te to the total bi nding e nergy. Difference Fourier m aps were computed at 6 A resolution on F ab' New crystals treated with dichlor ophe nolind ophenol, orcei ne, and menadione, and at 3.5 A for the vitamin K-l OR complex. All compounds showed one major peak at the same p osi ti on and several minor peaks. In the vitamin K-l OR map at 3.5 A resoluti on the minor peaks were barely above b ackgrou nd, and the m ajor peak could be fitted with the menadione ring and the phytyl t ail of the vitamin K-l OR. The l ocati on of this m ajor bindi ng si te is on the relatively flat e nd of the m olecule adj acent to the hypervari able regions of b oth chains (see Figure 13). The menadione m oiety lies i n the shallow groove between heavy and light ch ains described previously. The methylnaphth oqui none ri ng of vitami n K-l OR makes close contact with the ri ng of Tyr 90 of the L ch ain at the b ottom of the crevice, wi th the b ackbone and side ch ain of residue 104 (R chain), and wi th the b ackb one of L ch ain resi dues 29 and 30. Starting from the 2 -methyl -I, 4 naphthoqui none m oiety, the phytyl tail m akes contact with Gly 2 9 and Asn 30 of the L ch ain. I t the n appr oaches the b ackbone of L chai n residues 93 , 94, and residue 104 of the H ch ai n, and termi nates at the side ch ains of residues 54, 57, and 63 of the heavy chai n. A t least 10-12 residues from b oth VL and VH make contact with vitamin K-l OR. N o conform ational ch ange i n the protein structure w as observed subsequent t o vitami n K-l OR bindi ng.
652
Crystallographic Studies on Mouse Myeloma Proteins with Antigen Binding Properties
The availability of large amounts o f homogeneous proteins from transplantable tumors in the mouse ( 14), many with known antigen bin ding properties, has provide d a rich source of material for crystallographic investigation. Stimulated by the observation (104) that small crystals coul d be obtained from M OPC 3 1 5 Fab (pepsin) fragments, a systematic attempt was made to obtain crystals of Fab (pepsin) fragments of mouse myeloma proteins with known binding properties (105). Five proteins with phosphorylcholine bin ding activity, T EPC 1 5, HOPC 8, M OPC 1 67 , McPC 603, an d M OPC 5 1 1 ( 1 , 14), were investigated. These proteins precipitate with phosphorylcholine-containing natural antigens from a variety of organisms including Pneumococcus, Lactobacillus, Trichoderma, Aspergillus, Proteus morganii, an d Ascaris ( 1 , 14). Crystals were obtained from M OPC 167 an d McPC 603 . Of these, one crystal form from McPC 603 was suitable for X-ray diffraction analysis. It was demonstrated by chemical methods that these crys ta ls bound 1 mol ph ospho rylcholine per mole Fab, with a bin ding constant of 4 x 103 l/mo !. This bin ding constant is about 45-fold lower than that in free solution, a result assumed to be due to the ammonium sulfate of crysta llization.
Figure 4 Stereo drawing of a-carbon skeleton of McPC 603 Fab. The V region is at the top, and the light chain on the right.
653
A subsequen t study of an e lectron den sity map at 4. 5 A reso lution was reported together with the location of the bin ding site for phosphorykholine ( 106). At thi s reso lution the molecule, 40 x 50 x 80 A in overall dimen sion s, c learly consi ste d of fo ur di stinct globular region s, which were interpreted to be the fo ur domains. The domain s c luster in pairs to form two regions separate d by a hole an d linked by two thin ribbons of den sity. At this reso lution it was not possible to dist inguish H from L , but it was c lear that the str ucture c lo sely resemb led that fo und at 6 A reso lution for Fab New (99). In or der to locate the phosphorylcholine bin ding site (106), a difference Fourier synthesis was calculate d using data from crystals soaked in so lutions contain ing 2-(5'-acetoxymercury-2'-thienyl)-ethylphosphorylcholine (AMTEPC). The difference map showed two sites of incorporation , one on the surface of the molecule in what was calle d the C region , an d the other at the tip of the mo lec ule in the crevice between the two V region domains. When crystals that had been soaked originally in AM TEPC were sub sequently soaked in so lutions containing AMTEPC plus an excess of p-nitrophenylphosphorylcholine, the peak in the V r egion di sappeared whereas the C region peak persisted. Since the studies in so lution an d in the crystal showed specific bin ding of pho sphorylcholin e to on ly on e site per Fab ( 105), the displace d peak was taken to represent the hapten binding site. This conclusion was con firmed by the observation of a peak in this po sition in difference maps between native crystals and tho se soake d in a sat urating amo unt of pho sphorylcho linc. The binding studies with AMTEPC i llustrate one of the problems of working with crystals, namely that the AMTEPC difference map at the antigen bin ding site showed two peaks separate d by a region of low den sit y where there was a peak in the native map. This was interpreted as being due t o the binding in the cry stal of a sulfate ion from the ammonium sulfate of crystallization. The observation of no significant peaks other than those repre sent ing hapten b in ding was interpreted to mean that no signi ficant conformational change s had occurred on ligan d bin di ng. However, the po ssibility could not be ruled o ut that the sulfate bin ding m ight have caused the Fab fragment to adopt a ligan de d conforma tion before crystalli zation. The sub sequent extension of thi s work to 3.1 A resolution ( 107-1 10) together with complete seq uence data for VH an d partia lly for VL (1 10) led to an interpretation of the above result s in molecular terms. The overall appearance of the mo lecule (Figure 4)1 was very sim ilar to that previo usly reporte d for the h uman IgG (A) Fab New (83) an d the Mcg Bence-Jones dimer (79). The VH: VL an d CH 1 : CL domains are relate d in pairs to one another by approximate twofold axes of symmetry. These two axes are not collinear , b ut make an angle of approximately 1350 with each other . The two domain s of each chain ar e orient ed approx imately at r ight angles to on e another. The long axes of the L domains make an angle of approximately
1 The authors suggest that stereo viewers be used with figures 4-6, 8, 9, and 12. Stereo viewers may be obtained from Hubbard Scientific Company, P.O. Box 105, Northbrook, Illinois 60062, Abrams Instrument Corporation, 606 East Shiwassee Street, Lansing, Michigan 48901, or other sources.
654
DAVIES, PADLAN
&
SEGAL
1000 with each other, whereas the correspond ing angle for the H chain is approx imately 800 This results in a center-to-center distance of about 40 A between the VL and CL intradomain disulfides vs a d istance of about 30 A between V H and CH 1 . A possible explanation of this distortion of symmetry might come from the more extensive interaction between the C and V domains of the H chain vs the L chain. In the VH : CH I contact region residues 8-10, Gly-Gly- Gly, and residues 1 1 1-1 1 3, Gly-Thr-Thr of VH are close to the CHI domain . It appears that these small s ide chains facil itate the close approach of the two heavy domains. Within each domain the same sandwich-l ike structure is observed, illustrated in Figure 5. Ad jacent segments within each layer are anti parallel to one another and fre quently assume a p-pleated sheet con figuration. The interior of the sandwich contains principally hydrophobic residues. The extended segments are l inked by bends with varying degrees of sharpness, and the electron density in many of the t ight bends can be fitted w ith p-bend configurations ( 1 1 1 ). The quaternary structure of McPC 603 Fab (Figure 4) is similar to those in Fab New ( 83) and the Mcg dimer (79). The interaction between the C domains occurs principally between the four chain layers. The interactions between the V domains involve the three chain layers as well as a major portion of the additional loops found in the V domains. The C egion is more compact, the interacting layers being about 10 A apart. The distance between the intradomain d isulfide bonds in the C region is 18 A. In the V region, th e interacting segments are 1 2 -1 5 A apart and the distance between the d isulfide bonds is about 2 5 A. The residues involved in the V interface span the complete range of variabil ity. Some are quite conserved while others are hypervariable. For example, the large loop containing the first l ight chain hypervariable region is in intimate contact with most of the residues in the third heavy chain hypervariable region. The tip of the Fab molecule contai ns a large wedge-shaped cavity, approximately 12 A deep, 15 A wide at the mouth, and 20 A long, whose walls are lined exclusively with hypervariable residues (109, 110). Hapten binds in this cavity where it is located asymmetrically, being closer to the H chains than to the L chain (Figure 6). The cholin e end is bo und in the interior of the cavity with the
Stereo drawing of a-carbon skeleton of McPC 603 CL domain. This view depicts the bilayer nature of the domain. The N terminus is on the right, with the three chain layer on top and the four chain layer on the bottom. A disulfide bond in the center of the diagram connects the two la yers. Note the concavity of the four chain la yer.
Figure 5
655
Figure 6 The hapten binding site of Mcpe 603. Phosphorylcholinc is shown in black, liganded to the protein. The complete hypervariable surface is shown. Ll, L2, and L3 refer to light chain hypervariable regions, whereas HI, H2, and H3 refer to those of the heavy chain. Numbers of the first and last residues are indicated for each hypervariable region.
phosphate group more towards the exterior. The phosphate appears to be exclusIVely bound by t he H chain an d forms specific interactions with Tyr 33 (H) and Arg 52 (H). Thesc s idc chains are located so that they can form hydrogen bonds with thc phosphate, and the positively charged guan idinium group presu mab ly also acts to neutralize partly the negative charge on t he phosphate. In the case of a mono e sterified phosp hate, as in p ho sp horylcholine, the phosphate is doubly c harge d, but in t he "true" antigenic determinant which presu mably contain s p ho sphorylcholine in a doubly esterified form ( 1 12), the phosp hate will have only a single charge. The chol ine group appears to interact w ith both the L an d H chains, an d the acidic side c hain of Glu 35 (H) is only about 5 A. away from t he positively charged nitrogen of t he c holine. The c holine also for ms van der Waals interaction s with t he main c hain atoms of resi dues 102-103 of the H chain an d 91-94 of t he L c hain. T he entire hapten i s in close contact with the ring atoms o f Tyr 33 (H). The magnitu de of the hypervariable cavity in McPC 603 is to a large extent a reflection of the insertions t hat occur in regions Ll, H2, an d H3, causing t hese hypervariable loops to extend farther out, t hus increasing the cavity depth. The region L2 does not form a part of the cavity, being screened from it by t he large loop contain ing Ll.
COMPARISON OF STRUCTURES
concept of antibo dy structure developed t hr ough chemical stu dies. The region s of sequence homology fol d into compact domains w hich act as buil ding blocks in t he formation of t he whole i mmunoglobulin molecule ( 1 9, 20), an d all 1 1 chemically di stinct do mains described ab ove have t he same basic tertiary structure. Yet, w hereas t he do mains bear a strong familial resemblance, it is clear that structural changes have occurred
The cr ystallographic studies discussed above confirm the general
656
DAVIES, PADLAN
&
SEGAL
during the course of evolution resulting in specialized functions for the various domains of the antibody molecule. In this section w make a comparison of these domain tertiary structures and the manner in which they aggregate into functional units.
Domain Tertiary Structure

Each domain consists of segments of extended polypeptide chains connected by bends of varying lengths and shapes (75, 79, 83, 1 10). The basic domain structure is shown schematically in Figure 7 for the V (right) and C (left) domains. The extended segments are designated SI, S2 . . . S9, and the bends connecting the
segments as B 1 2, B23 ' "
B89. In Figure 7 the segments lying within one layer
are designated by heavy lines, and those within the other layer by thin lines. The numbering of the stretches and bends in Figure 7 has been chosen to preserve homology between C and V domains. The intradomain disulfide bond joins S2 ano S8 in all the domains ( 1 09, 1 10). In Figure 8 a comparison of the tertiary structures of CL and V H of McPC 603 ( 109, 1 10) is made, and in Figure 9 th e tertiary structures of V L REI, V L and V H of McPC 603 (75, 109, 1 1 0) are compared. In general, whereas the domains have similar tertiary structures, larger differences exist between the V and C domains than between the six V domains or between the five C d omains thus far reported. The C and V domains bear essentially no homology in amino acid sequence ( 1 6) and it is therefore not surprising that the greatest structural differences exist between these domains.
A major difference between C and V domains lies in the existence of an

additional loop in the V which has no equivalent in the C domains (79, 83, 1 10). This loop can be seen in Figure 8 and is designated as S4, B45, S5 in Figure 7. This additional loop does not lie in either layer and thus the sandwich nature of the V domains is less apparent. The V and C domains (Figure 8) differ also in the length, shape, and chemical
109 140
171 I 1 68
51
52 56
128
57
1 59
122 I BI2
51
182
B23
141
889
'j9
202
53
1 836
151
B7B
Figure 7 Schematic representation o f V (right) and C (left) domain structures. Extended segments are labeled S t , S2, . . . S9, and the bends connecting them are B 1 2, B23, . . . B89. Dark lines indicate stretches of one layer, while stretches indicated by thin lines lie in the other. Hypervariable regions are indicated by short perpendicular lines in the V domain. Numbering refers to McPC 603 VH (right) and C. (left).
59
58
190 211
B 89
'" j
1 06
Z5
77
52
I BI2 15
57
I 85 67 1 61
10
58 I 845 I 5I I
1
B 67
75 I 1
S6 , 55
S4
i'656 I B34
93 118
34 I
S3
58 59
43 41
878
l ! 'l
657
Figure 8
Stereo drawing of (X-carbon backbone for McPC 603 CL (top) and VH (bottom) domains. Both domains are in approximately the same orientation. The loop that is present in the V domains, but missing from C, is indicated by unjoined bonds and lies on the bottom left part of the domain.
nature of the common stretches and bends. The straight segments in the C domains are apparently longer. Many of the V domains contain large, convoluted bends at the N-terminal end of the domain not found in the constant region. In the C domains, the four chain layer forms a concave surface (Figure 5) ; in the V domains this layer is curved in the opposite direction. Residues on the external surface of the four chain layers are mainly hydrophobic in the C domains and are involved in the strong interaction between the homologous domains across the C interface. The homologous residues in the V domains are exposed to solvent and are in general hydrophilic. All the V domains are quite similar to each other in their three-dimensional structure (Figure 9). The most remarkable similarity is between the two VL domains of McPC 603 and REI. The difference between the two is principally in the greater extension of the loop B23 in McPC 603 VL (in the first hypervariable region) where McPC 603 has a six residue insertion compared to REI. The VL REI domain has been superimposed by least square methods on VL McPC 603. For 94 residues the root-mean-square distance between homologous Q(-carbon atoms is 1 .4 A ( 109, 1 10). It is apparent from Figure 9 that VL and VH are also quite similar in conformation, but not to the same extent as the two VL domains. The major differences between >- McPC 603 VL and VH lie in the loop B23, where VL is larger, and in B45, where VH is more extensive. Comparative amino acid sequence analyses (16) indicate that the VH domain around B45 usually contains about five more residues than its
658
Stereo drawing of a-carbon backbone of three variable region domains, all in similar orientations. (Top) REI light chain domain. (Middle) McPC 603 light chain domain. (Bottom) McPC 603 heavy chain domain. H ypervariable residues are indicated by large circles.
Figure 9
light chain counterpart. When VH and VL from McPC 603 are superimposed, the root-mean-square displacement for 74 a-cltrbons is 1 .9 A. This has been represented in Figure 1 0 as a plot of displacement vs position in the two chains (109). The displacements observed in this plot can be compared with the variability plot of Kabat & Wu (4 \ ) for the amino acid sequence (Figure 1 1). It can be seen that by and large the sites of greatest displacement correspond to the hypervariable regions. The principal exception occurs in the extra hypervariable region (residues 82-89) of the H chain (42). This analysis lends strong support to the concept that immunoglobulins can incorporate large amounts of variation, i.e. substitutions, -::: insertions, and deletions, within a localized area of the V domain without significantly altering the structure of the nonhypervariable "framework" residues.
659
Quaternary Structure
Changes in the chemical nature of surface side chains have resulted in a completely differen t quaternary structure in V regions compared to C (79, 83, 110). In the C region, the four chain layers face each other and form a large surface of interaction between domains. In striking contrast, the four chain layers are on the outside in the V region and the interface involves the three chain layers (Figures 3 and 4). This mode of association of the V domains brings the three hypervariable regions together in space to form a continuous hypervariable surface. In addition, the V region interaction causes hypervariable residues to come into contact with each other across the interface (Figure 12). This could be an important factor in determining the size as well as the shape of the binding cavity. Since a detailed picture of the Fc has not been obtained, we can only speculate
L 12 22
-
Chain Residue Number 40 60 61 72 82 92 103
10
r--u--1
(3)
Iu-1
(5)
(4)
!l t: c
(5)
. 5
4 3
(1)
(I)
IT
10
21
31 H
41
-
51
66
78
89
100
110
Chain Residue Number
Differences in O(-carbon positions (Aj after McPC 603 VH is superimposed on McPC 603 VL. Regions of sequence hypervariability are shown at top and bottom ; Lt, L2, and L3 for the L chain and H I , H2, HE, and H3 for the heavy. HE is the extra hyper variable region residues 8 1 -85 (42). Arrows with numbers refer to the following : (1) One residue in L, not in H ; (2) One residue in H, not in L ; (3) Four residues in L, not in H ; (4) Six residues in H, not in L ; and (5) Two residues in H, not in L. Dotted portion represents region of uncertain structure in VL.
Figure 10
660
on the tertiary structures of the Fc domains as well as -on the nature of the interactions in this region of the molecule. The homology of the sequences of the Fc domains to those of CL and CH I implies that the Fc domains will resemble the C domains more closely than the V domains of the Fab. A closer examination of the sequences, especially around the intradomain disulfide, suggests that the quaternary structure of the CH2, and probably also of the CH 3 region, will be similar to that of the C region in the Fab. The high resolution structure analyses of the two Bence-Jones dimers (75, 79) and of the human (83) and murine ( 1 10) Fabs have provided some interesting details regarding the domain interactions in the C and V regions of these fragments. It was found, for example (Figure 5), that the four chain layers in the CL and CH 1 domains in McPC 603 Fab form slightly concave surfaces, cross each other at
HUMAN , MOU S E , RABBIT, HORSE , SHARK HEAVY CHAINS 80
G AP G AP
GAP
GAP
25
50
75
100
HUMAN H E AVY C H A I N S
25
50
POSITION
75
100
Figure 1 1
Amino acid variability among hcavy chains as a function of position within the sequence (41). Variability is defined as the number of different residues divided by the frequency of the most common residue at a position. Reprinted with permission from Kabat, E. A., Wu, T. T. 1971. Ann. N Y A cad. Sci. 190 : 386.
THREE-DIMENSIONAL STRUCTURE OF IMMUNOGLOBULiNS
661
Figure 1 2
Skeletal representation of the McPC 603 V region (top) and REI VL dimer (bottom). Both structures are in approximately the same orientation. The hypervariable surfaces are at the top, and the C termini at the bottom. HypervariabJe residues are indicated with large circles.
approximately right angles (Figure 4), and together form a rather extensive interface (109, 1 10). The segments across the C region interface are almost uniformly 10 A apart, and the C region, in comparison with the V, is more compact. Very similar results were obtained in the Mcg Bence-Jones dimer (79). In McPC 603 Fab, there are fewer aromatic side chains in the C region compared to the V region interface. In the latter, the interacting segments are more widely separated, being 12-15 A apart. Consequently, the distance between the intradomain disulfides is about 25 A (center to center) in the V region and 18 A in the C. Approximately the same distances were observed in the Mcg dimer. Moreover, the distance calculated for the REI protein, where the two VL domains are related by a twofold axis, is - 24 A. Some of the interactions observed in the McPC 603 V region (109, 1 10) were also found in the REI and Mcg dimers (Table 2), especially those near the carboxyl end of the V domains. In McPC 603 Fab, interactions betwccn many hypervariable residues have been observed. For example, almost the entire third
662
heavy chain hypervariable segment and most of the residues in the first light chain hypervariable region are in close contact. Contacts between hypervariable residues were also found in the REI dimer (Figure 12), but only one such interaction was observed in the Meg dimer. In this respect, the REI VK dimer more closely resembles the McPC 603 V region. It is interesting that the two V domains have few points of contact in the Mcg dimer. Indeed, many of the residues that are in contact in both REI and McPC 603 are found to line the walls of the pocket and the main cavity in Mcg. It seems that in Mcg, part of the interface in the V region had opened up and become accessible for binding by aromatic molecules like DNP and menadione (vitamin K-3). Comparing the known ), and K light chain sequences ( 1 6), it is not obvious why the V region interactions in the )Aype Mcg should be different from those in the K-type REI. The longitudinal interactions between the VL and CL domains and those between the VH and CH 1 domains in the human and in the murine Fab are similar to those occurring in the two monomers of the Mcg light chain dimer. In the Fabs, the domains of the heavy chain approach each other more closely at the switch region. In McPC 603, this close approach is apparently allowed by the existence of small residues at the interface.
The Antigen Binding Site
Since the binding studies with both phosphorylcholine and vitamin K-l OH have been performed with myeloma proteins, it is reasonable to question their physiological significance. Although direct evidence is not available, there is consider able indirect evidence to suggest that what is being observed is indeed a satisfactory model for antigen-antibody interaction. McPC 603 is one of a number of independently induced mouse plasma cell tumor proteins that precipitate with several natural antigens, including Pneumococcus C polysaccharide ( I , 1 13). Choline is a common constituent of all of these antigens, and Leon & Young ( 1 14, 1 15) observed that phosphorylcholine would inhibit the precipitation of Pneumococcus C polysaccharide by McPC 603. Antibodies to phosphorylcholine-containing antigens have been prepared from Balb/C mice that have been shown to be idiotypically identical with one phosphorylcholine binding myeloma, TEPC 15 (3). Although under these conditions, no idiotypic identity with McPC 603 has been observed, there is no good reason to doubt that phosphorylcholine is the major antigenic determinant in the interaction between Pneumococcus C polysaccharide and McPC 603 protein ( 1). In the case of Fab' New binding to vitamin K-l OH, evidence has been cited (102, 103) that anti idiotypic antibodies against IgG New will combine with some of the immuno globulin species from the serum of a rabbit immunized with a vitamin K-l bovine gamma-globulin complex. Secondly, the affinity constants observed in McPC 603 for phosphorylcholine and in New for vitamin K- l OH are in the right range [5 x 104-1 x 109 ljmoi ( 1 1 6)] to correspond to antigen-antibody complexes, although both are at the lower end of the range. Thirdly, the location of the binding site on the tip of the Fab in the region
663
between the light and heavy chains is reassuringly reasonable when viewed in the light of existing chemical knowledge. Tn both, the hapten is to
3.
large extent in
contact with hypervariable residues. For McPC 603 the results of affinity labeling are consistent with the interaction being predominantly with the heavy chain Although the hypervariable end of the molecule in the case of Fab New
( 1 17).
is described as being rather flat with a narrow crevice between the light and heavy
domains, whereas for McPC 603 it is more in the nature of a large cavity, both haptens bind to their respective proteins in approximately the same place (Figure 1 3). In both cases there are interactions with hypervariable regions of light and heavy chains. In McPC 603 the interactions come principally from the H chain, while in Fab New the extended phytyl chain of vitamin K- l OH ensures hapten in teraction with residues from each of the five hypervariable regions. Despite the disparity in size between vitamin K - l OH and phosphorylcholine, they both bind to their respective proteins with roughly equal affinity (102, 105). In both proteins the light chain second hypervariable region plays no role in binding-in Fab New because it has been deleted and in McPC 603 because this loop is excluded from the binding site by the extensive first L chain hypervariable region. In neither protein is there any large conformational change on binding of hapten, in agreement with a low angle X-ray diffraction study on Fab;'.. 'ltigen interactions ( 1 1 8). In a recent review, Metzger
(7) has dealt comprehensively with the data regarding
Figure 13 Binding sites of Fab' New (left) and McPC 603 (right) in similar orientations. Fab' New is shown with vitamin K- l OH (102. reproduced with authors' permission) ; McPC 603 is shown binding phosphoryicholine.
664
possible conformational changes in whole molecules as a result of antigen or hapten binding. The interesting binding studies with the Mcg L chain dimer (88) arc more difficult to assess. Three different binding sites are observed, A and B being in the hypervariable cavity in the vicinity of the binding sites observed with Fab New and McPC 603. In contrast to these proteins where the L 2 region is either absent (New) or not involved in the binding site (McPC 603), site A of Mcg involves residues 5 1 , 52 from the second hypervariable region. The third site lies within the inner pocket in a region that involves nonhypervariable residues and defines the VL : VH interface in the other proteins. These binding studies were performed in the crystal, and it would be interesting to know the binding constants involved.
CONCLUSION
The elucidation of the complete primary structure of an intact human immuno globulin revealed the existence of repeating regions of sequence homology (19). This observation led to the hypothesis that these regions would fold into compact domains with similar tertiary structures (1 8-20). The crystallographic studies described above confirm the domain hypothesis conclusively. While the domains act as the building blocks of immunoglobulin molecules, it is nevcrtheless clear from the X-ray studies, even at low resolution, that the basic structural units consist of pairs of strongly interacting, homologous domains. The immunoglobulin molecule can be viewed as a series of loosely connected structural units ; two in each Fab and two (IX, y) or three (fl, e) in the Fc. Tn the V region, the basic structural unit consisting of the two variable domains is the functional entity responsible for antigenic recognition. The distinct mode of association of the V domains produces a continuous hypervariable surface whose topology can be altered by amino acid substitutions, insertions, and deletions in the hypervariable regions of both the light and heavy chains (109, 1 10). The hypervariable surfaces present in the total immunoglobulin pool probably generate the entire immunogenic potential of the organism. The correlation of structure with function in the V region is made possible by the existence of small molecules that bind specifically to the hypervariable surface and demonstrate the antibody-hapten interaction. Similar correlations involving C region structural units require the development of- suitable markers to serve as probes for functional specialization. These probes, with the elucidation of the detailed structures of Fc and whole immunoglobulin molecules, should in the future lead to an understanding of the structural basis of effector mechanisms in the immune response.
ACKNOWLEDGMENTS
We wish to acknowledge Drs. E. Kabat, H. Metzger, M. Navia, M. Potter, and S. Rudikoff for their advice and criticisms of the manuscript. We are grateful to Dr. Enid Silverton for permitting us to use her data on comparisons in advance of publication.
665
Literature Cited 1. Potter, M. 1971. Ann. N Y A cad. Sci. 190 : 306-21 2. Metzger, H. 1970. Ann. Rev. Biochem. 39 : 889-928 3. Cosenza, H., Koehler, H. 1972. Science 176 : 1027-29 4. Blomberg, B., Geckler, W. R., Weigert, M. 1972. Science 177 : 1 78-80 5. Lieberman, R., Potter, M., Mushinski, 1974. J. Exp. Med. 139 : 983-1001 6. Edelman, G. M., Gall, W. E. 1969. Ann. Rev. Biochem. 38 : 41 5-66 7. Metzger, H. 1974. Advan. Immunol. 1 8 : 169-207 8. Cathou, R. E., Dorrington, K. J. 1974. Biological Macromolecules "Subunits in Biological Systems," ed. S. N. Timasheff, G. D. Fasman, 7. New York : Dekker, 9. Gaily, 1. A. 1973. The Antigens, ed. M. Sela, 161-298. New York : Academic 10. Hood, L., Prahl, J. 197 1. Advan. Im munol. 14 : 291-351 1 1. Gaily, 1. A., Edelman, G. M. 1972. Ann. Rev. Genet. 6: 1 -46 12. Capra, 1. D., Kehoe, 1. M. 1975. Advan. Immunol. 1 9 : 1 13. Milstein, c., Pink, J. R. L. 1 970. Progr. Riophys. Mol. BioI. 2 1 : 209-63 14. Potter, M. 1972. Physiol. Rev. 52 : 631719 1 5. Porter, R. R . 1959. Biochem. J. 73 : 1 1 9-26 1 6. Dayhoff, M. 0., Ed. 1972. Atlas o f Protein /Sequence and Structure,
In press
E. B., Humphrey, W. Jr., Rudikoff, S.
chemistry 10 : 3843-50 26. O'Daly, J. A., Cebra, J. J. 1 97 1 . J. Immuno!. 107 : 436-48 27. Morrison, S. L., Koshland, M. E. 1 972. Proc. Nat. A cad. Sci. USA 69 : 1 24-28 28. Kownatzki, E. 1 97 1 . Eur. J. Immunol. 1 : 486-91 29. Weinheimer, P. F., Mestecky, J., Acton, R. T. 1971. J. Immunol. 107 : 1 2 1 1-12 30. Kehoe, 1. M., Tomasi, T . B., Ellouz, F., Capra, 1. D. 1972. J. Immuno!. 109 : 59-64 3 1. Zikan, 1. 1973. Immunochemistry 1 0 : 351-54 32. Wilde, C. E. III, Koshland, M. E. 1973. Biochemistry 1 2 : 3218-24 33. Kobayashi, K., Vaerman, 1. P., Heremans, 1. F. 1973. Biochim. Biophys. Acta 303 : 105-17 34. Schrohenloher, R . E., Mestecky, J., Stanton, T. H. 1973. Biochim. Biophys. . Acta 295 : 576-81 35. Mestecky, 1., Schrohenloher, R. E., 1974. Proc. Nat. Acad. Sci. USA 7 1 : 544-48 Tomasi, T. B. 1973. Proc. Nat. Acad. Sci. USA 70 : 3410-14 Milstein, C. 1967. Nature 2 1 6 : 330-32 Kabat, E. A. 1 970. Ann. NY A cad. Sci. 1 69 : 43-54 Franek, F. 1 969. Developmental Aspects o f Antibody Formation and Structure,
Kulhavy, R., Wright, G. P., Tomana, M.
36. 37. 38. 39.
Proc. Nat. A cad. Sci. USA 53 : 1403-9 18. Edelman, G. M., Gally, J. A. 1962. J. Exp. Med. 1 1 6 : 207-27 19. Edelman, G. M., Cunningham, B. A., Acad. Sci. USA 63 : 78-85 20. Edelman, G. M. 1970. Biochemistry 9 : 3 197-3205 2 1 . Putnam, F. W., Florent, G., Paul, c ., Shinoda, T., Shimizu, A. 1 973. Science 182 : 287-91 22. Bennich, H., von Bahr-Lindstroem, H. 1973. 9th Int. Congr. Biochem. p. 299. 23. Halpern, M. S., Koshland, M. E. 1970. Nature 228 : 1 276-78 24. Mestecky, J., Zikan, 1., Butler, W. T. 1971 . Science 1 7 1 : 1 1 63-65 25. O'Daly, 1. A., Cebra, J. J. 1971. Biou., Waxdal; M. 1. 1969. Proc. Nat.
Washington DC : Nat. Biomed. Res. Found. 1 7. Hilschmann, N., Craig, L. C. 1965.
Gall, W. E., Gottlieb, P. D., Rutishauser,
(Abstr.)
Med. 132 : 2 1 1-50 41. Kabat, E. A., Wu, T. T. 1971. Ann. N Y Acad. Sci. 190 : 382-92 42. Capra, J. D., Kehoe, J. M. 1 974. Proc. Nat. Acad. Sci. USA 7 1 : 845-48 43. Wofsy, L., Metzger, H., Singer, S. 1. 1962. Biochemistry 1 : 1031-38 44. Givol, D. 1 974. Essays Biochem. 10 : 73-103 45. Noelken, M. E., Nelson, C. A., Buckley, E. C. III, Tanford, C. 1965. J. Bioi. Chem. 240 : 2 1 8-24 46. Valentine, R. c., Green, N. M. 1967. J. Mol. Bioi. 27 : 6 1 5-17 47. Svehag, S. E., Chesebro, B., Bloth, B. 1967. Science 158 : 933-36 48. Chesebro, B., Bloth, B., Svehag, S. E. 1968. J. Exp. Med. 127 : 399-410 49. Svehag, S. E., Bloth, B. 1970. Science 168 : 847-49 50. Svehag, S. E. 1972. 3rd Int. Convoc. Immunol. Buff alo, 1 973. 8-91. Basel :
ed. 1. Sterzl, I. Riha. Prague : Czecho slovak Acad. Sci. 40. Wu, T. T., Kabat, E. A. 1 970. J. Exp.
Karger
666
DAVIES, PAD LAN & SEGAL
51. Shelton, E., McIntyre, K. R. 1970. J. Mol. Bioi. 47 : 595-97 52. Shelton, E., Smith, M. 1 970. J. Mol. Bioi. 54 : 6 1 5-17 53. Dourmashkin, R. R., Virella, G., Park house, R. M. E. 1971. J. Mol. BioI. 56 : 207-8 54. Green, N. M., Dourmashkin, R. R., Parkhouse, R. M. E. 1 971. J. M o l Bioi. 5 6 : 203-6 55. Feinstein, A., Rowe, A. 1. 1 965. Nature 205 : 147-49 56. Feinstein, A., Munn, E. A. 1 966. J. Physiol. London 186 : 64-66P 57. Feinstein, A., Munn, E. A., Munro, A. J. 1 97 1 . Nature 23 1 : 527-29 58. Feinstein, A., Munn, E. A., Richardson, N. E. 197 1 . Ann. N Y A cad. Sci. 190 : 104-2 1 59. Poljak, R. J., Dintzis, H. M. 1966. J. Mol. Bioi. 1 7 : 546-47 60. Poljak, R. 1., Dintzis, H. M., Goldstein, D. 1. 1967. J. Mol. Bioi. 24 : 351-52 61. Humphrey, R. L. 1967. J. Mol. BioI. 29 : 525-26 62. Poljak, R. J., Goldstein, D. J., Humphrey, R. L., Dintzis, H. M. 1967. Cold Spring Harbor Syrnp. Quant. Bio/. 32 : 95-98 63. Goldstein, D. J., Humphrey, R. L., Poljak, R. 1. 1968. J. Mol. BioI. 35 : 247-49 64. Terry, W. D., Matthews, B. W., Davies, D. R. 1 968. Nature 220 : 239-41 65. Edmundson, A. B., Wood, M. K., Schiffer, M., Hardman, K. D., Ainsworth, C. F., Ely, K. R., Deutsch, H. F. 1970. J. Bioi. Chern. 245 : 2763-64 66. Palm, W., Colman, P. M. 1 974. J. Mol. Bioi. 82 : 587-88 67. Sarma, V. R., Silverton, E. W., Davies, D. R., Terry, W. D. 1 9 7 1 . J. Bioi. Chern. 246 : 3753-59 68. Sarma, V. R., Davies, D. R., Labaw, L. W., Silverton, E. W., Terry, W. D. 197 1 . Cold Spring Harbor Symp. Quant. BioI. 36 : 413-19 69. Davies, D. R., Sarma, V. R., Labaw, L. W., Silverton, E. W., Segal, D. M., Terry, W. D. 1971. Ann. NY A ead. Sci. 190 : 122-29 70. Davies, D. R., Sarma, V. R., Labaw, L. W., Silverton, E. W., Terry, W. D. 197 1 . Proe. First Int. Congl'. Immunol., 25-32. New York : Academic 7 1 . Labaw, L. W., Davies, D. R. 1971. J. BioI. Chem. 247 : 3760-62 72. Labaw, L. W., Davies, D. R. 1 972. J. U Itrastruct. Res. 40 : 349-65 73. Deutsch, H. F., Suzuki, T. 1971. Ann. N Y Acad. Sci. 190 : 472-85 74. Lopes, A. D., Steiner, L. A. 1973. Fed. Proe. 32 : 1003 (Abstr.)
.
75. Epp, 0., Colman, P., Fehlhammer, H., Bode, W., Schiffer, M., Huber, R., Palm, W. 1974. Eur. J. Biochern. 45 : 5 13-24 76. Edmundson, A. B., Schiffer, M., Wood, M. K., Hardman, K. D., Ely, K. R., Ainsworth, C. F. 197 1. Cold Spring Harbor Symp. Quant. BioI. 36 : 427-32 77. Schiffer, M., Hardman, K. D., Wood, M. K., Edmundson, A. B., Hook, M. E., Ely, K. R., Deutsch, H. F. 1 970. J. Bi oi Chem. 245 : 728-30 78. Edmundson, A. B., Schiffer, M., Ely, K. R., Wood, M. K. 1972. Biochemistry 1 1 : 1822-27 79. Schiffer, M., Girling, R. L., Ely, K. R., Edmundson, A. B. 1973. Biochemistry 1 2 : 4620-31 80. Edmundson, A. B., Ely, K. R., Girling, R. L., Abola, E. E., Schiffer, M., Westholm, F. A. 1974. Progress in Immunology 1I. ed. L. Brent, J. Hol borow, 1 : 103-13. Amsterdam : North Holland 8 1 . Ely, K. R., Girling, R. L., Schiffer, M., Cunningham, D. E., Edmundson, A. B. 1973. Biochemistry 12 : 4233-37 82. Fett, 1. W., Deutsch, H. F. 1 974. Bi o chemistry 13 : 41 02-14 83. Poljak, R. J., Amzel, L. M., Avey, H. P., Chen, B. L., Phizackerly, R. P., Saul, F. 1973. Proc. Nat. Acad. Sci. USA 70 : 3305-10 84. Strosberg,A. D., Fraser, K. 1., Margolies, M. N., Haber, E. 1972. Biochemistry 1 1 : 4978-85 85. Milstein, c., Frangione, B., Pink, J. R. L. 1 967. Cold Spring Harbor Symp. Quant. B io i. 32 : 31-36 86. Hess, M., Hilschmann, N., Rivat, L., Rivat, c., Ropartz, C. 1971. Nature New Bioi. 234 : 58-60 87. Appella, E., Ein, D. 1 967. Proc. Nat. A cad. Sci. USA 57 : 1449-54 88. Edmundson; A. B., Ely, K. R., Girling, R . L.,Abola,E. E., Schiffer, M., Westholm, F. A., Fausch, M. D., Deutsch, H. F. 1974. Biochemistry 1 3 : 3816-27 89. Palm, W. H. 1 970. FEBS Lett. 1 0 : 46-48 90. Solomon, A., McLaughlin, C. L., Wei, C. H., Einstein, J. R. 1970. J. Bioi. Chem. 245 : 5289-9 1 9 1 . Schramm, H. J. 197 1. Z. Physio/. Chem. 352 : 1 1 34-38 92. Epp, 0., Palm, W., Fehlhammer, H., Ruehlmann, A., Steigemann, W., Schwager, P., Huber, R. 1 972. J. Mol. Bioi. 69 : 3 1 5-18 93. Colman, P. M., Epp, 0 . , Fehlhammer, H., Bode, W., Schiffer, M., LaUman, E. E., Jones, T. A., Palm, W. 1974. FEBS Lett. 44 : 194-99 94. Rossi, G., Nisonoff, A. 1 968. Biochem.
.
THREE-DIMENSIONAL STRUCTURE OF IMMUNOGLOBULINS Biophys. Res. Commun. 31 : 9 1 4- 1 8 95. Rossi, G., Choi, T . K., Nisonoff, A . 1 969. Nature 223 : 837-38 96. Avey, H. P., Poljak, R. J., Rossi, G., Nisonoff, A. 1 968. Nature 220 : 1248-49 97. Humphrey, R. L., Avey, H. P., Becka, L. N., Poljak, R. J., Rossi, G., Choi, T. K., Nisonoff, A. 1 969. J. Mol. Bioi. 43 : 223-26 98. Poljak, R. 1., Amzel, L. M., A vey, H. P., Becka, L. N., Goldstein, D. 1., Humphrey, R. L. 197 1 . Cold Spring Harbor Symp. Quant. Bioi. 36 : 42 1 -25 99. Poljak, R. 1., Amzel, L. M., Avey, H. P., Becka, L. N., Nisonoff, A. 1972. Nature New Bioi. 235 : 1 37-40 100. Poljak, R. J. 1973. Con temp. Top. Mol. Immunol. 2 : 1-26 101. Chen, B. L., Poljak, R. 1. 1 974. Bio chemistry 13 : 1 295-1302 102. Amzel, L. M., Poljak, R. 1., Saul, F., Varga, 1. M., Richards, F. F. 1 974. Proc. Nat. Acad. Sci. USA 7 1 : 1 427-30 103. Richards, F. F., Amzel, L. M., Konigs berg, W. H., Manjula, B. N., Poljak, R. J., Rosenstein, R. W., Saul, F., Varga, J. M. 1974. The Immune System, ed. E. E. Sercarz, A. R. Williamson, C. F. Cox, 53-67. New York : Academic 1 04. Inbar, D., Rotman, M., Givol, D. 1 97 1 . J . BioI. Chem. 246 : 6272-75 105. Rudikoff, S., Potter, M., Segal, D. M., Padlan, E. A., Davies, D. R. 1 972. Proc. Nat. A cad. Sci. USA 69 : 3689-92 1 06. Padlan, E. A., Segal, D. M., Rudikoff,
667
107. 108.
109.
1 10.
111. 1 12. 1 13. 1 14. 1 15. 1 16.

1 1 7.
1 18.
S., Potter, M., Spande, T., Davies, D. R. 1973. Nature New BioI. 245 : 165-67 Padlan, E. A., Segal, D. M., Cohen, G. H., Davies, D. R., Rudikoff, S., Potter, M. See Ref. 103, 7-14 Padlan, E. A., Segal, D. M., Cohen, G. H., Davies, D. R., Rudikoff, S., Potter, M. 1 974. Transplant. Proc. In press Segal, D. M., Padlan, E. A., Cohen, G. H., Silverton, E. W., Davies, D. R., Rudikoff, S., Potter, M. 1 974. See Ref. 80, 93-102 Segal, D. M., Padlan, E. A., Cohen, G. H., Rudikoff, S., Potter, M., Davies, D. R. 1974. Proc. Nat. A cad. Sci. USA 7 1 : 4298-4302 Venkatachalam, C. M. 1 968. Bio polymers 6 : 1425-36 Brundish, D. E., Baddiley, 1. 1 968. Biochem. J. 1 10 : 573-82 Cohn, M. 1 967. Cold Spring Harbor Symp. Quant. Bioi. 32 : 2 1 1-21 Leon, M. A., Young, N. M. 1 970. Fed. Proc. 29 : 437 (Abslr.) Leon, M. A., Young, N. M. 1971. Biochemistry 10 : 1424-29 Karush, F. 1 962. Advan. Immunol. 2 : 1-40 Chesebro, B., Hadler, N., Metzger; H . 1 972. 3rd Int. Convoc. Immunol. Buff alo, 1973. 205-17. Basel : Karger Pilz, 1., Kratky, 0., Licht, A., Sela, M. 1973. Biochemistry 12: 4998-5005

Three Dimensional Structure of Immunoglobulins

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Three Dimensional Structure of Immunoglobulins

Uploaded by

Copyright:

Available Formats

ANNUAL REVIEWS

Copyright 1975. All rights reserved

Quick links to online content

THREEDIMENSIONAL STRUCTURE OF IMMUNOGLOBULINS

Bethesda, Maryland 20014

Light Chains.. . Heavy Chains .....

Domain Tertiary Structure.. Quaternary Structure. . .........

The Antigen Binding Site. . ... . ... . .

DAVIES, PADLAN & SEGAL

THREE-DIMENSIONAL STRUCTURE OF IMMUNOGLOBULINS

Schematic representation of polypeptide chains in the basic immunoglobulin

DA VIES, PADLAN & SEGAL

THJ{EE-DIMENIUNAL STRUCTURE OF IMMUNOGLOBULINS

DA VIES, PADLAN & SEGAL

Reference 65, 76-82,88

75,89, 92, 93 Phosphorylcholine

McPC 603 Mouse

THREE-DIMENSIONAL STRUCTURE OF IMMUNOGLOBULINS

DAVIES, PADLAN & SEGAL

THREE-DIMENSIONAL STRUCTURE OF IMMUNOGLOBULINS

DAVIES, PADLAN & SEGAL

Binding Studies on M cg Crystals

THREE-DIMENSIONAL STRUCTURE OF IMMUNOGLOBULINS

model of one domain was constructed

DAVIES, PADLAN & SEGAL

THREE-DIMENSIONAL STRUCTURE OF IMMUNOGLOBULINS

DAVIES, PADLAN & SEGAL

Crystallographic Studies on Mouse Myeloma Proteins with Antigen Binding Properties

THREE-DIMENSIONAL STRUCTURE OF IMMUNOGLOBULINS

THREE-DIMENSIONAL STRUCTURE OF IMMUNOGLOBULINS

The cr ystallographic studies discussed above confirm the general

Domain Tertiary Structure

segments as B 1 2, B23 ' "

B89. In Figure 7 the segments lying within one layer

A major difference between C and V domains lies in the existence of an

THREE-DIMENSIONAL STRUCTURE OF IMMUNOGLOBULINS

DAVIES, PADLAN & SEGAL

THREE-DIMENSIONAL STRUCTURE OF IMMUNOGLOBULINS

Chain Residue Number 40 60 61 72 82 92 103

Chain Residue Number

DAVIES, PADLAN & SEGAL

THREE-DIMENSIONAL STRUCTURE OF IMMUNOGLOBULiNS

DAVIES, PADLAN & SEGAL

THREE-DIMENSIONAL STRUCTURE OF IMMUNOGLOBULINS

(7) has dealt comprehensively with the data regarding

DAVIES, PADLAN & SEGAL

THREE-DIMENSIONAL STRUCTURE OF IMMUNOGLOBULINS

E. B., Humphrey, W. Jr., Rudikoff, S.

Kulhavy, R., Wright, G. P., Tomana, M.

36. 37. 38. 39.

Washington DC : Nat. Biomed. Res. Found. 1 7. Hilschmann, N., Craig, L. C. 1965.

Gall, W. E., Gottlieb, P. D., Rutishauser,

DAVIES, PAD LAN & SEGAL

111. 1 12. 1 13. 1 14. 1 15. 1 16.

You might also like