COLD SPRING HARBOR SYMPOSIA

ON QUANTITATIVE BIOLOGY

VOLUME XLI

Origins of Lymphocyte Diversity

COLD SPRING HARBOR LABORATORY

1977

1 of 13 Celltrion, Inc., Exhibit 1116

COLD SPRING HARBOR SYMPOSIA ON QUANTITATIVE IUOL05\r VOLUME XLI

© 1977 by The Cold Spring Harbor Laboratory

International Standard Book Number 0-87969-040-2 !clothbound)

Library of Congress Catalog Card Number 34-8174

Printed in the United States of America

All rights reserved

COLD SPRJN(; HARBOR SYMPOSIA ON QUANTITATIVE B/OLO<;Y

Founded in 1931 by

RE<;(NALD C. HARRIS

JJirertor of the Biologiral Laboratory I 924 to I '116

l'rn1iu/I.\ Syrtt/Josia Volumes

I 11933) Surface Phenomena II 11934) Aspects of Growth III (1935) Photochemical Reactions IV (1936) Excitation Phenomena V (1937) Internal Secretions VI 11938) Protein Chemistry VII 0939) Biological Oxidations VIII 0940) Permeability and the Nature of Cell Membranes IX 0941) Genes and Chromosomes: Structure and Organization X ( 1942) The Relation of Hormones to Development XI (1946) Heredity and Variation in Microorganisms XII (1947) Nucleic Acids and Nucleoproteins XIII 0948) Biological Applications of Tracer Elements XIV 0949) Amino Acids and Proteins XV (1950) Origin and Evolution of Man XVI (1951) Genes and Mutations XVII (1952) The Neuron XVIII (1953) Viruses XIX 0954) The Mammalian Fetus: Physiological Aspects of De­

velopment XX 0955) Population Genetics: The Nature and Causes of Genetic

Variability in Population

XXI 0956) Genetic Mechanisms: Structure and Function XXII 0957) Population Studies: Animal Ecology and Demography XXIII (1958) Exchange of Genetic Material: Mechanism and Con-

sequences XXIV 0959) Genetics and Twentieth Century Darwinism XXV 0960) Biological Clocks XXVI 0961) Cellular Regulatory Mechanisms XXVII (1962) Basic Mechanisms in Animal Virus Biology XXVIII 11963) Synthesis and Structure of Macromolecules XXIX 0964) Human Genetics XXX 0965) Sensory Receptors XXXI 0966) The Genetic Code XXXII 0967) Antibodies XXXIII 11968) Replication of DNA in Microorganisms XXXIV I 1969) The Mechanism of Protein Synthesis XXXV I 1970) Transcription of Genetic Material XXXVI 11971) Structure and Function of Proteins at the Three-

dimensional Level XXXVII 0972) The Mechanism of Muscle Contraction XXXVIII (1973) Chromosome Structure and Function XXXIX (1974) Tumor Viruses XL (1975) The Synapse

The Symposium Volumes are published by the Cold Spring Harbor Laboratory, Cold Spring Harbor, New York 11724, and may be purchased directly from the Laboratory or through booksellers. Price of Volume XLI-2-part set $60.00 line. postage). May be purchased only as a complete set. Price subject to change without notice.

2 of 13 Celltrion, Inc., Exhibit 1116

Model-building Studies of Antigen-binding Sites: The Hapten-binding Site of MOPC-315

E. A. PADLAN, D.R. DAVIES, I. PECHT,* D. GrvoL* AND C. WRIGHTt

Laboratory of Molecular Bwlogy, Natwnal Institute nf Arthritis, Metabolism and Digestive Diseases, National Institutes of Health, Bethesda, Maryland 20014; *Department of Chemical Immunology, The Weizmann Institute of Science, Rehovot.

Israel; t Laboratory of Molecular Biophysics, Department of Zoology, Oxford University, Oxford, England

The molecular basis for the structural diversity of antibody combining sites has become apparent through the recent X-ray diffraction studies on several immunoglobulin (lg) fragments (see Davies et al. 1975a,b for a review). These structures reveal that the combining site is formed by bringing to­gether the three hypervariable regions (Wu and Kabat 1970) of V1, and of V H to form a continuous complementarity-providing surface. A quantitative comparison of the tertiary structures of a number of variable domains from both light and heavy chains has demonstrated that their nonhypervari­able or framework regions are very similar, with the principal differences occurring in the hypervariable loops (Padlan and Davies 1975).

In view of this structural invariance, the im­munoglobulin variable region can be regarded as consisting of a rigid framework to which are at­tached the hypervariable loops. These loops are not large, generally consisting of at most 17 resi­dues; in addition, the configurations of a number of different hypervariable loop regions are already known from X-ray diffraction. Thus it might be possible to construct by model building new loop regions (and hence the lg binding sites) from a knowledge of the sequences alone. Since it is clearly impractical to determine by X-ray diffraction the structures of all the interesting antibodies, this model-building method offers an attractive alter­native approach when sequence data are available. We have examined the potential usefulness of this possibility by attempting to build the V regions of the rabbit anti-type-III polysaccharide antibody (Davies and Padlan 1976) and of protein MOPC-315, a mouse immunoglobulin with dinitrophenyl (DNPl-binding specificity (Eisen et al. 1968), based on the known three-dimensional structure of McPC-603 Fab' (Padlan et al. 1973, 1974; Segal et al. 1974a,b) together with the amino acid se­quences of these immunoglobulin Fab's.

In this paper, we describe the results of two independent attempts to construct the model of MOPC-315, one at the National Institutes of Health and the other at the Weizmann Institute. The two models were ultimately merged to give a final model that is presented and is discussed in terms of the known binding specificity, the kinetic mapping of the binding site (Pecht et al. 1972a,b;

627

Haselkorn et al. 1974), and the results of affinity labeling with various reagents (Goetz! and Metzger 1970a,b; Givol et al. 1971; Haimovich et al. 1972). We discuss some of the limitations and the potential of this method of model building, the assumptions that have to be made, and some of the additional information that will be needed to put this tentative approach on a firmer footing.

MATERIALS AND METHODS

Sequence Alignment

The amino acid sequences of the V 11 domains of proteins McPC-603 (Rudikoff and Potter 1974) and MOPC-315 (Francis et al. 1974; L. Hood, M. Margolies, D. Givol and R. Zakut, unpubl.) were aligned for maximum sequence homology and structural analogy, as presented in Table la. The sequence of New V H (Poljak et al. 1974) is included for comparison. A similar alignment of the 603 (Segal et al. 1974a) and 315 V1, (Dugan et al. 1973) sequences is given in Table lb. However, only the sequence of the first 49 residues is known for 603, and the residues given for the rest of the sequence are those which most frequently occur in mouse kappa chains (McKean et al. 1973). Included in Table lb are the sequences of the V L domains of REI (Epp et al. 1974), Meg (Fett and Deutsch 1974), and New (Chen and Poljak 1974) proteins.

The sequence numbering schemes used in these tables are those obtained from the original refer­ences. Whenever a residue position is referred to in the text, however, the quoted number corresponds to the 315 sequence. The hypervariable regions of the light chain are: Ll, residues 23-36; 12, residues 52-58; and L3, residues 91-99; and those of the heavy chain are: Hl, residues 31-36; H2, residues 49-66; and H3, residues 99-104.

Model Building

Model 1 was constructed at the Weizmann Insti­tute from Nicholson molecular models (Labquip, Reading, England), which have a scale of 1 cm c= 1 A, and CPK space-filling models (1.25 cm = 1 AJ. Model 2 and the final model were constructed at the National Institutes of Health from Kendrew

3 of 13 Celltrion, Inc., Exhibit 1116

628 PADLAN ET AL.

(2 cm = 1 Al molecular models (Repetition Engi­neers, Cambridge, England).

The general principle used for model building was to construct first the framework part of the variable region, based on the structure of 603 Fab'. The hypervariable loops were then constructed, leaving the structure as little changed as possible except when forced by amino acid insertions and deletions.

An attempt was made to maximize the structural stability within each loop by forming hydrogen bonds whenever possible and maintaining the phi and psi peptide angles within reasonable limits (Ramakrishnan and Ramachandran 1965 ). The interactions between loops were then maximized, leaving no large holes in the domain interior, while minimizing steric hindrance between groups.

The Ll regions in the kappa chains 603 and REI are simple loops, whereas the corresponding re­gions are helical in New (Poljak et al. 1974) and Meg (Schiffer et al. 1973), which are both lambda chains. Since the light chain of protein 315 is of the lambda type and since this region has the same number of residues as New and Meg with no gross differences in the nature of the amino acid side chains, it is most likely that this region in 315 will have a similar configuration to that found in New and Meg. Accordingly, the Ll loop in 315 was built to conform as closely as possible to the correspond­ing region in the lambda chains with the aid of the atomic coordinates of the Meg backbone kindly provided by M. Schiffer and coworkers (pers. comm.).

L2 was built by assigning to it the same back­bone conformation as in protein 603. In this region, 603 and REI are not significantly different (Padlan and Davies 1975). Moreover, kappa and lambda light chains have the same number of residues in this part of the molecule (Dayhoff 1972); an excep­tion is New (Table lb), which has a seven-residue deletion in this part of the molecule. The L2 region in Meg appears to have the same structure as 603 (Edmundson et al. 1974). Thus it is reasonable to assume that 315 L2 will have the same configura­tion as that of 603.

L3 was built essentially as in 603, the main dif­ference being the possibility of forming more hy­drogen bonds in 315. The L3 loop in 315 was built as two, fully antiparallel, segments, the loop in 603 being less regular. Although the sequence of 603 is not known in this region, the six mouse kappa chains sequenced so far (McKean et al. 1973) all have a proline at position 94. The possible occur­rence of this residue in 603 could explain the fewer hydrogen bonds formed in 603 L3.

Protein 315 has one more residue than 603 in Hl (Table la). Originally, the sequence of 315 Hl con­tained a lysine in position 35 (Francis et al. 1974). Structurally, this residue is analogous to Met 34 of 603 since both occur two positions from the struc­turally invariant Trp 37. In 603, Met 34 is buried in

the domain interior. Its replacement by a lysine side chain would cause a significant rearrangement of the Hl backbone in order to expose the charged amino group to solvent. A reexamination of the sequence of this part of 315 led to the assignment of a Trp instead of the Lys in position 35 ( L. Hood, M. Margolies, D. Givol and R. Zakut, unpubl.). A Trp side chain can easily be accommodated in the domain interior without significantly changing the configuration from that observed in 603.

The alignment shown in Table la leads us to con­clude that the additional residue in Hl of 315 can be best accommodated in the exposed loop at the beginning of Hl. The glycine at position 32 permits the construction of a sharp turn in this part of 315.

H2 has three more residues in 603 than in 315, whereas 315 and New are of the same length in this region. An initial model of 315 H2 was built by simply excising three residues from the end of the 603 H2 loop. Minor adjustments were then made to make the 315 H2 loop resemble as closely as possi­ble the corresponding region in New (Poljak et al. 1974).

Basically the same procedure was employed in building the 315 H3 region, which has the same length as New and which is two residues shorter than 603. Here, however, the C-terminal segment of 315 H3 was made to approximate the correspond­ing region in 603, rather than in New. The differ­ence between 603 and New, aside from the two­residue insertion in 603, is the configuration of the segment immediately preceding the phenylalanine at position 105, which almost always contains a large hydrophobic residue (Dayhoff 1972). In 603, the side group of Trp 104a, which is structurally analogous to Leu 103 of 315, projects into the hapten-binding cavity. On the other hand, the back­bone of this segment in New appears to be different (Poljak et al. 1974). In view of the greater structural similarity of residues 103-105 of 315 (Leu-Tyr­Phe) to those of 603 (Trp-Tyr-Phe) rather than to those of New (Gly-Cys-Ile), the configuration of 603 was followed in building this segment of 315. However, it should be kept in mind that an alterna­tive configuration might be that observed in New.

As soon as a tentative model of the .binding site was completed, a likely site for DNP-binding was located using the criteria that the nitro groups of the DNP moiety must be hydrogen-bonded to the protein and that the stacking van der Waals inter­action between the DNP ring and the side group of Trp 93 (LJ must be maximal. The model was then adjusted to accommodate the DNP-hapten while maintaining the general stability of the various loop structures. Further adjustments were then made to ensure the feasibility of labeling Tyr 33 (L) and Lys 52 (H) by specific affinity reagents (Givol et al. 1971; Haimovich et al. 1972). The adjustments made to the tentative model to accommodate the ligand were minor and involved only slight displace-

4 of 13 Celltrion, Inc., Exhibit 1116

Table I. Amino Acid Sequences of V H Domains of Proteins McPC-603 and MOPC-315

McPC-603 MOPC-315 New

(a) Alignment of V11 Sequences

0 2 0 4 0 2 0 4 0 0 1 Glu-Val -Lys - Leu-Val-Glu-Ser-Gly-Gly -Gly 1 Asp-Val -Gln-Leu-Gln-Glu -Ser -Gly-Pro -Gly 1 Pea -Val -Gln-Leu-Pro - Glu -Ser -Gly-Pro -Glu

0 4 0 0 0 0 0 4 0 4 11 Leu-Val -Gin-Pro -Gly - Gly - Ser -Leu-Arg -Leu 11 Leu-Val -Lys-Pro -Ser-Gin - Ser -Leu-Ser -Leu 11 Leu-Val -Ser-Pro -Gly- Glx - Thr-Leu-Ser -Leu

0 4 0 4 0 1 4 0 4 0 21 Ser -Cys -Ala-Thr-Ser - Gly-Phe-Thr-Phe -Ser 21 Thr -Cys -Ser - Val -Thr -Gly -Tyr -Ser -Ile -Thr 21 Thr-Cys -Thr-Gly-Ser - Thr- Val-Ser -Thr -Phe

0 4 2 4 2 4 c 3 c 31 Asp- - Phe-Tyr -Met - Glu -Trp-Val -Arg -Gin 31 Ser -Gly -Tyr-Phe-Trp-Asn-Trp-Ile -Arg -Gin 31 Ala- -Val-Tyr-Ile -Val-Trp-Val-Arg-Gln

2 0 0 0 0 c 2 c 4 4 40 Pro -Pro -Gly-Lys -Arg-Leu -Glu-Trp-Ile -Ala 40 Phe-Pro - Gly-Asn-Lys-Leu -Glu-Trp-Leu -Gly 40 Pro -Pro - Gly - Arg-Gly-Leu -Glu-Trp-Ile -Ala

1 4 2 0 0 0 0 0 2 0 50 Ala -Ser -Arg-Asn-Lys-Gly -Asn-Lys-Tyr -Thr 50 Phe-Ile -Lys-Tyr -Asp-Gly - -Ser 50 Tyr -Val -Phe-Tyr -His - Gly - -Thr

2 0 3 1 0 0 4 0 0 1 58bThr-Glu -Tyr-Ser -Alii-Ser -Val-Lys -Gly -Arg 57 Asx -(Tyr , Gly) Asx - Pro - Ser -Leu- Lys -Asn -Arg 57 Ser -Asp -Thr- Asp-Thr-Pro -Leu-Arg-Ser -Arg

4 0 4 0 2 0 0 0 0 3 68 Phe -Ile - Val -Ser-Arg -Asp-Thr-Ser -Gin -Ser 68 Val -Ser -Ile -Thr-Arg-Asp-Thr-Ser -Glu -Asn 68 Val -Thr -Met-Leu-Val-Asn -Thr-Ser -Lys -Asn

0 4 1 4 0 4 0 0 4 0 78 Ile -Leu -Tyr-Leu-Gln-Met-Asn-Ala - Leu -Arg 78 Gln-Phe -Phe-Leu-Lys-Leu -Asp-Ser -Val -Thr 78 Gin -Phe -Ser-Leu-Arg-Leu-Ser -Ser -Val -Thr

0 0 2 3 4 c 4 c 4 c 88 Ala -Glu -Asp-Thr-Ala-Ile -Tyr-Tyr-Cys -Ala 88 Thr-Glx -Asx-Thr-Ala-Thr -Tyr-Tyr-Cys -Ala 88 Ala -Ala -Asp-Thr -Ala- Val -Tyr -Tyr -Cys -Ala

3 3 2 c 0 0 cc cc 98 Arg-Asn-Tyr-Tyr-Gly-Ser -Thr-Trp-Tyr -Phe 98 Gly -Asp -Asn-Asp-His - -Leu-Tyr -Phe 98 Arg-Asx -Leu-Ile -Ala- -Gly-Cys -Ile

1 4 c 4 0 4 2 0 4 0 106 Asp-Val -Trp-Gly -Ala-Gly -Thr-Thr-Val -Thr 106 Asp-Tyr -Trp-Gly -Gln-Gly -Thr-Thr-Leu -Thr 106 Asx-Val -Trp-Gly-Gln-Gly-Ser-Leu-Val -Thr

4 0 0 116 Val -Ser -Ser 116 Val -Ser -Ser 116 Val -Ser -Ser

(a ) Above each residue in the McPC-603 sequence is its structural location designated by: 0, complete ly exposed to solvent; 1, mainly exposed; 2, partly exposed, partly buried in the domain interior; 3, mainly buried; 4, completely buried; or C, in contact with the homologous domain. The numbers at the left alongside the sequences correspond to the sequence number of the first residue in each row as obtained from the origina l publica­tions.

629

5 of 13 Celltrion, Inc., Exhibit 1116

McPC-603 REI MOPC-315 Meg New

Table I. (continued)

(b) Alignment of V L sequences

0 3 0 4 0 4 1 0 0 0 1 Asp-Ile -Val-Met-Thr-Gln -Ser -Pro -Ser -Ser 1 Asp-Ile -Gln-Met-Thr-Gln -Ser -Pro -Ser -Ser 1 Pea-Ala -Val-Val-Thr-Glu -Glu- -Ser -Ala 1 Pea -Ser -Ala-Leu-Thr-Gln -Pro -Pro- -Ser 1 Pea-Ser -Val-Leu-Thr-Gln-Pro-Pro- -Ser

4 0 4 0 0 0 0 0 4 0 11 Leu-Ser -Val-Ser -Ala-Gly -Glu-Arg-Val -Thr 11 Leu-Ser -Ala-Ser -Val-Gly -Asp-Arg-Val -Thr 10 Leu-Thr -Thr-Ser -Pro -Gly -Gly-Thr-(Val, Ile) 10 Ala -Ser -Gly-Ser -Leu-Gly -Gin-Ser -Val -Thr 10 Val -Ser -Gly-Ala-Pro -Gly-Gln-Arg-Val -Thr

4 0 4 0 4 0 0 0 3 0 21 Met-Ser -Cys-Lys -Ser -Ser -Glx-Ser -Leu -Leu 21 Ile -Thr -Cys-Gln-Ala -Ser -Gln-20 Leu-Thr -Cys-Arg-Ser -Ser -Thr-Gly-Ala -Val 20 Ile -Ser - Cys-Thr - Gly -Thr - Ser -Ser -Asp - Val 20 Ile -Ser -Cys-Thr-Gly-Ser -Ser -Ser -Asn -Ile

0 0 cc c 1 c 3 4 e 27 Asx-Ser -Gly-Asx-Glx -Lys-Asx-Phe-Leu -Ala 28 -Asp-Ile -Ile -Lys -Tyr -Leu -Asn 30 -Thr-Thr-Ser -Asn-Tyr-Ala -Asn 30 -Gly-Gly -Tyr -Asn-Tyr - Val -Ser 27e -Gly -Ala -Gly-Asn-His -Val -Lys

4 c 3 c 0 0 0 cc 35 Trp -Tyr -Glx-(G]x)-Lys-Pro -Gly-Glx -Pro -Pro 35 Trp-Tyr -Gln-Gln-Thr-Pro -Gly-Lys-Ala -Pro 37 Trp -Ile -Glx-Glx -Lys -Pro -Asx-His -Leu -Phe 37 Trp -Tyr -Gin-Gin-His -Ala -Gly-Lys-Ala -Pro 34 Trp -Tyr -Gin-Gin -Leu-Pro -Gly-Thr-Ala -Pro

0 c 4 4 45 Lys -Leu -Leu-Ile 45 Lys -Leu -Leu-Ile 47 Thr-Gly -Leu-Ile 47 Lys - Val -Ile -Ile 44 Lys -Leu -Leu-Ile

c 1 1 0 1 0 -Tyr - X - X - X - X - X -Tyr -Glu -Ala -Ser -Asn -Leu -Gly -Gly -Thr-Ser -Asp -Arg -Tyr -Glu -Val -Asn-Lys -Arg -Phe-His -Asn-Asn-Ala -Arg

0 0 0 4 1 0 1 4 0 4 55 X - X -Gly-Val -Pro -Ala -Arg-Phe-Ser -Gly 55 Gin -Ala -Gly-Val -Pro -Ser -Arg-Phe-Ser -Gly 57 Ala -Pro -Gly-Val -Pro -Val -Arg-Phe-Ser -Gly 57 Pro -Ser -Gly- Val -Pro -Asp-Arg-Phe-Ser -Gly 61 -Phe-Ser -Val

0 4 0 0 2 0 4 0 4 0 65 Ser - Gly - Ser -Arg-Thr -Asp- Phe-Thr - Leu -Thr 65 Ser -Gly --Ser-Gly -Thr -Asp-Tyr-Thr-Phe -Thr 67 Ser -Leu -Ile -Gly-Asp-Lys-Ala-Ala-Leu -Thr 67 Ser - Lys -Ser - Gly -Asn -Thr -Ala - Ser - Leu -Thr 64 Ser - Lys -Ser - Gly - Ser - Ser - Ala -Thr - Leu -Ala

4 0 0 2 2 0 0 1 2 4 75 Ile -Asx -Pro-Val -Glx -Ala -Asx-Asp-Val -Ala 75 Ile -Ser -Ser-Leu-Gin -Pro -Glu-Asp-Ile -Ala 77 Ile -Thr -Gly-Ala -Glx -Thr-Glx -Asp-Asp -Ala 77 Val -Ser -Gly-Leu-Gln -Ala -Glu-Asp-Glu -Ala 74 Ile -Thr -Gly-Leu-Gln -Ala -Glu-Asp-Glu -Ala

e 4 c 4 c 4 c 2 0 0 85 Thr-Tyr -Phe-Cys- X - X - X - X - X - X 85 Thr-Tyr -Tyr-Cys-Gln-Gln-Tyr-Gln-Ser -Leu 87 Met-Tyr - Phe- Cys -Ala - Leu -Trp - Phe -Arg -Asx 87 Asp-Tyr -Tyr-Cys-Ser -Ser -Tyr-Glu-Gly -Ser 84 Asp-Tyr -Tyr-Cys-Gln -Ser -Tyr-Asp-Arg -Ser

0 c 3 c 4 e 0 4 0 95 - X - X - X -Phe-Gly-Gly-Gly-Thr -Lys

6 of 13 Celltrion, Inc., Exhibit 1116

MODELS OF ANTIGEN-BINDING SITES 631

Table I. r continued)

(b) Alignment of V L sequences

95 -Pro -Tyr-Thr-Phe-Gly-Gln-Gly-Thr -Lys 97 -His - Phe-Val-Phe-Gly-Gly-Gly-Thr -Lys 97 Asp-Asn - Phe-Val -Phe-Gly -Thr-Gly-Thr -Lys 94 -Leu -Arg-Val -Phe-Gly-Gly-Gly-Thr -Lys

4 0 3 0 0 104 Leu-Glu -Ile -Lys-Arg 104 Leu-Gin - Ile -Thr-Gly 106 Val -Thr - Val-Leu-Gly 107 Val-Thr - Val-Leu - Gly 105 Leu-Thr -- Val-Leu-Arg

Cb) See footnote to a. The McPC-603 sequence is known only to residue 49; for the rest of the sequence, the most frequently occurring residue in other mouse kappa chains (Mc· Kean et al. 1973) is given for each position except for the hypervariable residues, which are simply designated by X. Ala 34, Thr 85, and Gly 100 in 603 project into the V11 :V1. inter· face but are not in actual contact with V,,; this structural classification is designated by the lower case c.

ments of main-chain atoms and rotation of side groups about single bonds.

Atomic coordinates of the 315 hypervariable residues were measured from the model and ad­justed using Diamond's model-building program (Diamond 1966). In building the DNP-hapten into the model, the nitro groups were assumed to be coplanar with the phenyl ring. Although the crystal structureof2,4-dinitroaniline has not been reported, theoretical arguments (Pauling 1948) would pre­dict a planar molecule. This is consistent with the observat~ons of Trotter (Trotter 1961; Trotter and Williston 1966) on the crystals of various nitro­phenyl compounds; Trotter found small departures from coplanarity but attributed these to intermo­lecular forces in the crystal. In DNP-lysine and its derivatives, a hydrogen bond was assumed between one oxygen of the 2-nitro group and the NH of the 1:-amino group of the lysine, thus fixing the position of the C1:.

RESULTS

Sequence Comparisons and Correlation with Structure

Included in Table la,b is the structural classifi­cation of each residue, as observed in 603, according to whether it is buried in the domain interior, ex­posed to solvent, or involved in the contact with the homologous domain.

Comparison of framework residues. As expected, the residues that are completely or mainly buried in the domain interiors are more conserved than those that are completely or mainly exposed to solvent. In V H, 16 of the 29 residues that are com­pletely or mainly buried are identical in 315 and 603. Of the buried residues that are different, a Val in 315 replaces a Thr (position 24), a Tyr re­places a Phe (position 27l, and 2 Leu replace an Ile (position 48) and a Met (position 83) in 60::! . Of the

residues that are completely or mainly exposed to solvent, 18 of 44 are identical, 5 involve Ser-Thr interchanges (positions 21, 25, 28, 30, and 71), a Lys replaces an Arg (position 44), and a Glu re­places a Gin (position 76). The residues involved in beta bends (Venkatachalam 1968) are quite con­served. The bend around position 14-15 is facili­tated by the sequence Pro-Gly in 603; Pro 14 is also present in 315. The other hairpin bend around posi­tion 41-42 is due to the sequence Pro-Gly which is present in both 315 and 603.

The lack of complete sequence information for 603 precludes a detailed correlation of the V1, sequences with structure. On the basis of the tentative se­quence shown in Table lb, 17 of the 26 completely or mainly buried residues are identical in 315 and 603 V1,. Of the buried residues that are different in 315 and 603, a Thr in 315 replaces a Val in 603 at position 12 and a Leu replaces a Met at position 20. Of the completely or mainly exposed residues, 16 of 41 are identical, and 2 Thr for Ser interchanges oc­cur at positions 11 and 21. The Gly at position 16, which is involved in a beta bend, is present in both 315 and 603, the sharp bend in 315 being further facilitated by Pro 15. Another bend around positions 42-43 involves the sequence Pro-Gly in 603; Pro is also present at position 42 in 315. The additional residue at position 8 or 9 (Table lb) produces a kink in this region in kappa chains (Epp et al. 197 4; Segal et al. 1974a,b). However, the a-carbons ofresidues 8 and 10 are only about 5 A apart, so that the excision of this additional residue requires only minor struc­tural alterations. The substitutions at positions 45 and 46 result in one more potential hydrogen bond in 315 compared to 603.

The replacement of an interior residue that in­volves a change in side-group size is generally accompanied by a compensatory substitution else­where so that no large voids are created in the do­main interiors. For example, in V H• the Ile (315) for Phe (603) substitution at position 29 is balanced

7 of 13 Celltrion, Inc., Exhibit 1116

632 PADLAN ET AL.

by the Phe (315) for Leu (603) change at position 79. The cluster involving Met 83, Leu 86, and Val 114 in 603 is replaced by the grouping of Leu 83, Val 86, and Leu 114 in 315. Furthermore, the replacement of Val 107 in 603 by a Tyr in 315 is counterbal­anced by the substitution of a Gly in 315 for the Arg 98 in 603. In V1.. Leu 68 and Ala 73 of 315 re­place Gly 66 and Phe 71, respectively, of603.

Position 6 in 603 V1. is occupied by a Gin which is completely buried, whereas a Glu exists at this posi­tion in 315. The substitution is structurally feasible since the Co:-Cf3 bond of this residue is oriented such that the side group could swing in or out of the domain interior. Exposing the side group of 315 Glu 6 to solvent creates a void that could be filled by a structural rearrangement of the flexible seg­ment Gly-Gly-Gly at positions 101-103. In 315, Gly 101 is probably closer to the main body of Vi., as in VH where a Glu also occurs at position 6.

Comparison of residues involved in the V L:V11 con­tact. It is remarkable that the V H residues in­volved in the contact with V1. are virtually invari­ant. Of the eight contacting residues, seven are identical and the other involves an Ile for Val inter­change at position 38. Of the eight V1. residues in contact with V11 , three are the same in 315 and 603. Substitutions at the other contact positions gen­erally complement each other so that the relative disposition of the two variable domains is prob­ably the same in 603 and 315. For example, in the light chain, the replacements of 603 Tyr 36, Leu 46, and Tyr 49 by the smaller 315 Ile 38, Gly 48, and Gly 51, respectively, could cause VL to move closer to V H in 315. This tendency is counteracted, on the other hand, by the replacement of 603 Pro 43 and Pro 44 by 315 Leu 45 and Phe 46, respectively, as well as the replacement of the hypervariable residue Ala 34 in 603 by Asn 36 in 315. Furthermore, the substitution of 603 Thr 85 by 315 Met 87 in V L com­plements the change from 603 Lys 43 to 315 Asn 43 in VH.

Changes involving structural hypervariable resi­dues. Certain changes in framework residues ap­pear to be necessary to accommodate substitutions at hypervariable positions. For example, the re­placement in V L of 603 Ile 2 by 315 Ala helps to ac­commodate the bulky side group of Phe 94 from L3. Also in Vi., the substitution of 603 Met 4 by 315 Val helps accommodate the buried Leu 92. In addition, this Val-for-Met substitution at position 4, as well as the Ala-for-Phe replacement at position 73 and the Ala-for-Leu interchange at the hypervariable position 35, all facilitate the formation of a helical Ll region in 315. In V H, the replacement of the con­tact residue Val 37 by Ile in 315 complements the Asn for Glu substitution at the hypervariable posi­tion 35 in Hl.

The Antigen-binding Site of MOPC-315

As soon as a tentative model was completed, even before any adjustments were made to fit hapten, several features of the binding site became ap­parent. The middle of the hypervariable surface was dominated by a rather pronounced cavity. Com­pared to 603 , the 315 cavity contained substantially more aromatic residues. It seemed most likely that the DNP moiety would be bound within this cavity.

A schematic drawing of the hypervariable loops of 315 is shown in Figure lA. For comparison, the hypervariable loops of 603 are shown in Figure lB. The disposition of the hypervariable residues of 315, including the side-chain positions obtained from the model , is shown in Figure 2.

The hapten-binding site is bounded by Ll on one side and by Hl and H2 on the other. The floor is formed by L3 , and the roof is formed primarily by H3. The hypervariable residues that project into the hapten-binding cavity include Phe 34, Asn 36, Asp 99, and Leu 103 of the heavy chain and Tyr 34, Asn 36, Trp 93 , and Phe 98 of the light chain. Ser 32 and Asn 96 of the light chain and Phe 50, Lys 52, and Asp 101 of the heavy chain ring the opening of the pocket. As in 603, L2 is screened from the hapten-binding site by Ll and H3, and no residues from L2 project into the binding cavity.

The distorted fl-helix in Ll <Poljak et al. 1974) of 315 involves residues 26-31 , with a hydrogen bond between the carbonyl oxygen of Thr 26 and the amide nitrogen of Thr 31. The helix is initiated by a sharp bend at position 26-27, with a hydrogen bond between the carbonyl oxygen of Ser 25 and the amide nitrogen of Ala 28. The helical configuration places Ser 24, Ala 28, Val 29, and Ala 35 in the domain interior, with Ala 28 juxtaposed against Phe 94 of L3 . Aside from Tyr 34 and Asn 36, which project into the cavity, the other residues in Ll are either completely or mainly exposed to solvent. A pair of hydrogen bonds can be formed between main­chain atoms of Asn 36 and Ala 91 ofL3. All the resi­dues in L2 are exposed to solvent, and a beta bend is formed with residues in positions 51-54. In L3, Leu 92, Phe 94, and Val 99 are buried, Ala 91 pro­jects into the V1.:V 11 interface, and Arg 95, Asn 96, and His 97 are exposed. The L3 loop has a beta bend involving residues 94-97. Phe 98 of L3 and the non­hypervariable Phe 105 of the heavy chain delimit the maximum depth for hapten-binding. Trp 93 lines the floor of the binding cavity and probably serves to orient the planar hapten.

The restructuring of Hl leaves Tyr 33 of 315 in the domain interior, in an analogous position to 603 Phe 32. Phe 34, Asn 36, and, to a lesser extent, Phe 50 and the nonhypervariable Trp 47 form one side of the binding pocket. A beta bend is formed with residues 30-33, and hydrogen bonding is possible between the carbonyl oxygen of Gly 32 and the

8 of 13 Celltrion, Inc., Exhibit 1116

MODELS OF ANTIGEN-BINDING SITES 633

A

d~ '""~i31H Hl 99H

36 /36L ~'~ 2JLSJ L3

B

d''"

2 SQ

L 1

Figure I. Stereo drawings of the hypervariable loops of ( AJ 315 and (BJ 603 in approximately the same orientation. The numbers indicate the first and last hypervariable positions in each loop. These and other figures were drawn using the OR TEP program of Carroll K. Johnson (ORNL-3794).

amide nitrogen of Asp IOI, as well as between the amide nitrogen of Phe 34 and the carbonyl oxygen of Asp 99. Phe 34 from HI introduces a rather bulky hydrophobic side group into the binding cavity. Beta bends are formed with residues 52-55 and 6I-64 in H2. Leu 103 from H3 contributes to the hydrophobic nature of the hapten-binding site of 3I5 and with LI residues Tyr 34 and Asn 36 forms

the other side of the cavity. The roof of the cavity is formed mainly by the backbone of H3, the side group of Asp 99, and partly by Phe 34 of HI and Leu I03 of H3. A beta bend involves positions 100-I03 in H3. As in 603, H3 is in contact with LI.

The residues that would appear to interact most with the hapten are Phe 34, Asn 36, and Leu 103 from the heavy chain and Asn 36 and Trp 93 from

Figure 2. Stereo drawing of the 315 antigen-binding site showing all nonhydrogen atoms. The orientation and scale are the same as in Fig. lA. This figure is rotated through 90° in relation to the frontispiece.

9 of 13 Celltrion, Inc., Exhibit 1116

634 PADLAN ET AL.

ASP 99 LEU\03 I PHE 34

ASH 36 :< Y /r~ --\~-~ '

1TA-;". ~) ~

TAP 93PttE 98

~ L TS 52

~E 50

Q

ASP 99

LEU\03 1 PHE ~4

RSH3K y Ar -\.~ .t

1TA~~·~P~ 1RP SJ PHE 98

~ LTS 52

I/HE 50

Figure 3. Stereo drawing of the hypervariable residues projecting into the 315 hapten-binding cavity. A drawing of BADE (without the bromine atom) is included in the figure to show the hapten-protein interactions and the possibility of affinity-labeling Tyr 34 (L). The figure is rotated 15° relative to Fig. IA.

the light chain. The nitro groups of the DNP are then at suitable distances for hydrogen bonding to the amide of the L-chain Asn 36 side group and to the amide of the H-chain Asn 36. The aromatic side group of Trp 93 is essentially parallel to the planar DNP moiety and is 3.5-4 A from it. There is considerable overlap of the two conjugated ring systems. The hapten-protein interactions are illus­trated in Figure 3. In our model of DNP binding to 315, the specific affinity reagents used by Haimo­vich et al. (1972) can react with Lys 52 of the heavy chain and Tyr 34 of the light chain, but with no other.

Many hypervariable residues in one chain are in contact with hypervariable residues in the other chain. These include Asn 36, Phe 50, and Leu 103 of the heavy chain and Tyr 34, Asn 36, Asx 96, and Phe 98 of the light chain.

DISCUSSION

Correlation of the Model with Known Chemical Data

A number of chemical studies of the affinity label­ing (Metzger and Potter 1968; Goetz] and Metz­ger 1970a,b; Haimovich et al. 1970, 1972; and Givol et al. 1971) and kinetic mapping of the combining site ( Haselkorn et al. 1974) of protein 315 have been reported. The results of these studies are for the most part consistent with our model. The affinity labeling with a m-nitrobenzene diazonium com­pound (Goetz! and Metzger 1970a,b) and with BADE ( Haimovich et al. 1970, 1972; Givol et al. 1971J both indicated that Tyr 34 (Ll was close to the binding site. In the model of 315, examination of the positions of the reactive groups when the re­agents are bound through the nitrophenyl groups indicates that they are in a suitable position for them to react with this tyrosine. Similarly, the re­action of BADE (Haimovich et al. 1970, 1972; Givol et al. 1971) with Lys 52 (H) may be accounted for by its position in the model. The reaction of a bifunc­tional reagent (Givol et al. 1971) with protein 315 to cross-link the light and heavy chains is in accord with the positions of the Tyr 34 (L) and Lys 52 (H ).

It should be noted that although these results are

quite consistent, to a large extent they derive from coarse rather than fine detail in the model. The Ca-to-Ca separation of 20.9 A for Tyr 34 ( L) and Lys 52 (HJ is principally due to the fact that they occur in the first and second hypervariable regions of the light and heavy chains, respectively. Since these loops occur on opposite sides of the binding cavity, they will produce a significant separation of the residues regardless of the fine detail in the structure. However , if the orientation of Tyr 34 ( L) had been similar to that of the corresponding resi­due in 603 <Phe 33), then no reaction with these reagents would have been possible, since here the group points away from the binding pocket . The difference in orientation of Tyr 34 in 315 and Phe 33 in 603 lies in the difference in the configuration of the Ll loops of these two proteins. In thi s model, the phenolic oxygen of Tyr 34 (L) and the epsilon nitrogen of Lys 52 (H) are 15.l A apart. However, it should be remembered that the lysine side chain is quite flexible and can effect considerable varia­tion in this distance. This can account for the ob­servation of Givol et al. (1971 ) that affinity reagents that are slightly shorter than BADE could also label this lysine.

In our model of hapten binding to 315, the DNP moiety interacts extensively with the side group of Trp 93 (LJ. This accounts for the observed red shift in the absorption spectrum of DNP-lysine (Eisen et al. 1968) and the changes in its circular dichroism <Glaser and Singer 1971; Rockey et al. 1972; Inbar et al. 1973) on binding to 315. The asymmetry of the site is in accord with the induced circular polari­zation of fluorescence data of Pecht (1975).

The close proximity of the DNP ring to Trp 93 ( L ), the hydrogen bonding of the nitro groups within the cavity, and the van der Waals contacts with Leu 103 ( H) apparently represent the fundamental interactions responsible for the strong binding of DNP-haptens to 315. These interactions would re­sult in rigid binding for the DNP moiety, which is in agreement with the results of affinity labeling and with electron spin resonance results (Wong et al. 1974; Dwek et al. 1975).

The apparent low affinity of 315 for 2,6-dinitro and p-nitrophenyl compounds IHaselkorn et al. 1974) i!; gatisfactorily explained by our model. The

10 of 13 Celltrion, Inc., Exhibit 1116

MODELS OF ANTIGEN-BINDING SITES 635

binding of the p-nitro derivatives lacks the ad­ditional stabilizing force of one hydrogen bond (presumably to Asn 36 [L]l. The complex with 2,6-dinitrophenyl compounds would permit the forma­tion of only one hydrogen bond (presumably to Asn 36 fL]), since an attempt to form hydrogen bonds to both Asn 36 (L) and Asn 36 <HJ would pro­duce an impossible contact between the substituent at the 1-position and the side chain of Phe 34 (H).

Substitutions at the 1-position of the 2,4-dinitro­benzene moiety add to the interaction with the ring of Trp 93 (L), and longer substituents permit addi­tional interactions with Tyr 34 (L), Leu 103 (HJ, main-chain atoms in the H3 loop, and protein side groups farther away. Thus alkyl derivatives of 2,4-dinitroaniline as well as derivatives with branched cyclic or aromatic side chains are gen­erally found to bind better than the parent com­pound (Haselkorn et al. 1974). Branched side chains that are too bulky, however, apparently interfere with binding. Thus, though the 1-N-dimethyl derivative of 2,4-dinitroaniline is a stronger ligand, the diethyl derivative is a weaker binder than the parent (Haselkorn et al. 1974). In our model, it is clear that whereas the methyl groups in the former compound would add favorably to the hapten­protein interaction by increased van der Waals contact with Trp 93 (L), Leu 103 (H), and the main chain of H3, the ethyl groups in the latter derivative would cause too close contacts with Trp 93 ( L l and H3, leading to the instability of the complex. Simi­lar close contacts with Trp 93 (L) would occur in the binding of dialanine derivatives of DNP in which the first alanine residue is in the L-configuration, which explains the lower affinity of 315 for these compounds (Haselkorn et al. 1974). This interfer­ence would occur only when another amino acid residue is linked to the first alanine, if the peptide dihedral angles are kept within reasonable limits.

For DNP derivatives with side chains that have a hydrogen-bonding capacity, there are a number of polar amino acids nearby with which they may interact. These include Lys 52 (H), Asp 101 (HJ, His 102 (HJ, Ser 32 (L), Tyr 34 (L), and Asn 96 (Ll. This additional potential hydrogen bonding between hapten and protein could explain the observed high affinity of 315 for DNP derivatives with hydroxyl and carboxyl groups (Haselkorn et al. 1974). In addition, Lys 52 (H) is located where it could ac­count for the positive subsite observed by Hasel­korn et al. (1974).

Although our model of the binding site of 315 can account, at least qualitatively, for most of the avail­able chemical data on the binding ofhapten to 315, we cannot rule out the possibility that other hapten­binding configurations may exist. The other con­figurations that seem plausible to us are not grossly different from our present model. They retain the present location of the DNP-hapten while reorien­tating it, as well as the Trp 93 (L) side group, rela-

tive to the other groups in the cavity. Here again, only minor adjustments in the binding-site struc­ture would be required.

The rationale for the presence of Asp 99 ( H) in the binding cavity is not apparent. This residue is deep in the pocket and is strategically positioned so that it may play an important role in determin­ing the hapten-binding specificity of the protein. Its presence here raises the possibility that another, more tightly bound hapten might be found or de­signed for 315. Such a hapten would probably be positively charged to interact with Asp 99 (H), with an aromatic ring to interact maximally with Trp 93 (L l and the other hydrophobic residues in the cavity, and with side groups that could utilize the hydrogen-bonding potential of either or both Asn 36 (L) and Asn 36 (H). A similar case is found in protein New (Amzel et al. 1974l, where a Lys and an Arg are found deep in the binding site but are apparently not utilized in the binding of menadione haptens.

Hypervariable residues have been found to par­ticipate in the contact between Vi. and V tt in 603 where almost the entire H3 is in intimate contact with the large loop of Ll. In our model, the homolo­gous regions in 315 are also in contact. In the ab­sence of an actual structure determination, the effect of the interaction of hypervariable residues on the size and shape of the antigen-binding site of 315 can only be guessed. We have built the model of the combining site of 315 assuming that these hypervariable interactions will have only a modu­lating effect on the structure of the binding cavity and will not affect significantly the disposition of the pertinent side groups.

Comparison with Other Model-building Studies

Poljak et al. (1974) have briefly discussed the binding site of 315 relative to the structure of Fab New, a human myeloma protein. Their analysis is based on the unrevised sequence of Hl, with Lys at position 35, but they do not discuss the effect of this group on the conformation of Hl. Nevertheless, many of their conclusions regarding the probable nature of the amino acids lining the predicted hap­ten-binding cavity are similar to ours.

A previous attempt involved construction of a model of a homologous protein from a known three­dimensional structure for a-lactalbumin and chicken lysozyme (Browne et al. 1969). In that case, there was a striking degree of sequence homology between the two proteins, including four S-S bridges; as a result, much of the molecule could readily be constructed to the lysozyme framework.

Here, by contrast, although the degree of sequence homology between 603 and 315 is high for the framework part of the two molecules, the hyper­variable loops that form the binding surface are quite different in sequence. The invariance of the

11 of 13 Celltrion, Inc., Exhibit 1116

636 PADLAN ET AL.

framework is an essential assumption in order for the construction of the model of the 315 binding site to be feasible. The other assumption that was roughly adhered to in both studies was that loops of the same length would have essentially the same conformation. We know that this need not be pre­cisely true, and even when the same amino acid sequences occur in the loops, as in the two variable domains of the immunoglobulin light-chain dimer Meg, the loops have in places somewhat different conformations (Edmundson et al. 1974). When the amino acids are different, then this assumption is even less likely to be true. Nevertheless, this pro­vided a reasonable starting model for loops of the same size, and the biggest difficulties were encoun­tered with insertions and deletions. It is interest­ing to note that in two human V K domains, REI and Au, the entire backbone conformations, including the hypervariable regions, appear identical despite their having 16 amino acid differences (Fehlhammer et al. 1975).

Discussion of the Present Approach

This study was limited in several ways. First, it was based on a very small number of known struc­tures. Although these structures provide a reason­able basis for assuming framework invariance, they do not as yet provide enough data to enable us to make specific structural statements about the na­ture of the hypervariable surface. In particular, they leave us with no good feeling for the changes produced by insertions and deletions. To this extent, more structure determinations of immunoglobulins from different insertion and deletion subgroups are needed.

Second, the study has been restricted to the use of mechanical molecular models. At the initial stages, it might be beneficial to examine particularly large loops by methods such as those of Chou and Fasman (1974) or Wu and Kabat (1971) in order to predict the possible changes in secondary structure. In the final stages, it is clear that the use of a general energy minimization program which takes into ac­count the bad contacts and attempts to alleviate them would be of benefit. Nevertheless, until more sophisticated procedures are carried out, model­building techniques of the kind used in this study will be the only way in which an understanding of the nature of the binding sites of most immuno­globulins can be obtained.

Acknowledgment

Part of this work was supported by National Institutes of Health Grant ROI AI-11453 to D.G.

REFERENCES

AMzEL, L. M., R. J. PoWAK, F. SAUL, J. M. VARGA and F. F. RICHARDS. 1974. The three dimensional structure

of a combining region-ligand complex of immuno­globulin New at 3.5 A resolution. Proc. Nat. Acad. Sci. 71: 1427.

BROWNE, w. J., A. C. T. NORTH, D. C. PHILLIPS, K. BREW, T. C. VANAMAN and R. L. HILL. 1969. A possible three­dimensional structure of bovine a-lactalbumin based on that of hen's egg-white lysozyme. J. Mol. Biol. 42: 65.

CHEN, B. L. and R. J. PoWAK. 1974. Amino acid sequence of the (,\) light chain of a human myeloma immuno­globulin (lgG New). Biochemistry 13: 1295.

CHOU, P. Y. and G.D. FASMAN. 1974. Prediction of protein conformation. Biochemi~try 13: 222.

DAVIES, D. R. and E. A. PADLAN. 1976. Correlations be­tween antigen-binding specificity and the three-di­mensional structure of the antibody combining sites. In Proceedings of the Royal Society of Medicine Sym· posium (20-22 October 1975, Rockefeller University). Raven Press, New York. (ln press.)

DAVIES, D. R., E. A. PADLAN and D. M. SEGAL. 1975a. Three-dimensional structure of immunoglobulins. Annu. Rev. Biochem. 44: 639.

---. 1975b. Immunoglobulin structures at high reso­lution. In Contemporary topics in molecular immunol· ogy (ed. F. P. Inman), vol. 4, p. 127. Plenum Press, New York.

DAYHOFF, M. 0., ed. 1972. Atlas of amino acid sequence and structure, vol. 5. National Biomedical Research Foun­dation, Silver Spring, Maryland.

DIAMOND, R. 1966. A mathematical model-building pro­cedure for proteins. Ac:ta Cryst. 21: 253.

DUGAN, E. S., R. A. BRADSHAW, E. S. SIMMS and H. N. E1sEN. 1973. Amino acid sequence of the light chain of a mouse myeloma protein <'.'v10PC-315l. Biochemistry 12: 5400.

DwEK, R. A. , J . C. A. KNOTT, D. MARSH, A. C. McLAUGHLIN, E. M. PRESS , N. C. PRICE and A. I. WHITE. 1975. Struc­tural studies on the combining site of the myeloma pro­tein MOPC 315. Eur. J. Biochem. 53: 25.

EDMCNDSON, A. B .. K. R. ELY, R. L. GIRLING, E. E. ABOLA, M. SCHIFFER and F. A. WESTHOLM. 1974. Structure and binding properties of a ,\-type Bence-Jones dimer. In Progres.~ in immunology II: Immunochemical aspects (ed. L. Brent and J. Holborow), vol. I, p. 103. North­Holland, Amsterdam.

EISEN, H. N., E. S. SIMMS and M. POTTER. 1968. Mouse myeloma proteins with antihapten antibody activity. The protein produced by plasma cell tumor MOPC-315 . Biochemistry 7: 4126.

EPP, 0., P. COLMAN, H. FEHLHAMMER, w. BODE, M. ScHIF­FER, R. HUBER and W. PALM. 1974. Crystal and molecu­lar structure of a dimer composed of the variable por­tions of the Bence-Jones protein REI. Eur. J. Biochem. 45: 513.

FEHLHAMMER, H., M. ScHJFFER, 0. EPP, P. M. COLMAN, E. E. LATTMAN, P . SCHWAGER, W. STEIGEMANN and H. J. ScHRAMM. 1975. The structure determination of the variable portion of the Bence-Jones protein Au. Biophys. Struct. Mechanism 1: 139.

FEIT, J. W. and H. F. DEUTSCH. 1974. Primary structure of the Meg lambda chain. Biochemistry 13: 4102.

FRANCIS, S. H., R. G. Q. LESLIE, L. Hooo and H. N. EISEN. 1974. Amino-acid sequence of the variable region of the heavy (Alpha) chain of a mouse myeloma protein with anti-hapten activity. Proc. Nat. Acad. Sci. 71: 1123.

G!VOL, D., P. H. STRAUSBAUCH, E. HURWITZ, M. WILCHEK, J. HAIMOVICH and H . N. EISEN. 1971. Affinity labeling and cross-linking of the heavy and light chains of a myeloma protein with anti-2,4-dinitrophenyl activity. Biochemistry 10: 3461.

GLASER, M. and S. J. SINGER. 1971. Extrinsic Cotton effects

12 of 13 Celltrion, Inc., Exhibit 1116

MODELS OF ANTIGEN-BINDING SITES 637

characteristic of specific hapten-antibody interactions. Proc. Nat. Acad. Sci. 68: 2477.

GoETZL, E. J. and H. METZGER. 1970a. Affinity labeling of a mouse myeloma protein which binds nitrophenyl ligands. Kinetics of labeling and isolation of a labeled peptide. Biochemistry 9: 1267.

--. 1970b. Affinity labeling of a mouse myeloma pro­tein which binds nitrophenyl ligands. Sequence and position of a labeled tryptic peptide. Biochemistry 9: 3862.

HAIMOVICH, J., D. G1voL and H. N. EISEN. 1970. Affinity labeling of the heavy and light chains of a myeloma protein with anti-2,4-dinitrophenyl activity. Proc. Nat. Acad. Sci. 67: 1656.

HAIMOVICH, J., H. N. EISEN, E. HURWITZ and D. GIVOL. 1972. Localization of affinity-labeled residues on the heavy and light chain of two myeloma proteins with anti-hapten activity. Biochemistry 11: 2389.

HASELKORN, D., S. FRIEDMAN, D. GIVOL and I. PECHT. 1974. Kinetic mapping of the antibody combining site by chemical relaxation spectrometry. Biochemistry 13: 2210.

INBAR, D., D. G1voL and J. HOCHMAN. 1973. Circular di­chroism study of the antibody combining site. Bio­chemistry 12: 4541.

McKEAN, D., M. POTTER and L. HooD. 1973. Mouse im­munoglobulin chains. Patterns of sequence variation among K chains with limited sequence differences. Biochemistry 12: 760.

METZGER, H. and M. POTTER. 1968. Affinity site labeling of a mouse myeloma protein which binds dinitrophenyl ligands. Science 162: 1398.

PADLAN, E. A. and D.R. DAVIES. 1975. Variability ofthree­dimensional structure in immunoglobulins. Proc. Nat. Acad. Sci. 72: 819.

PADLAN, E. A., D. M. SEGAL, G. H. COHEN, D.R. DAVIES, S. RumKOFF and M. POTTER. 1974. The three-dimen­sional structure of the antigen binding site of McPC 603 protein. In The immune system (ed. E. E. Sercarz et al.), p. 7. Academic Press, New York.

PADLAN, E. A., D. M. SEGAL, s. RUDIKOFF, M. POTTE~, T. SPANDE and D. R. DAVIES. 1973. Structure at 4.5 A resolution of a phosphorylcholine-binding Fab. Nature New Biol. 245: 165.

PAULING, L. 1948. The nature of the chemical bond, 2nd Ed. Cornell University Press, Ithaca, New York.

PECHT, I. 1975. Antibody combining sites as a model for molecular recognition. In Protein-ligand interactions, p. 356. Walter de Gruyter, Berlin.

PECHT, I., D. G1voL and M. SELA. 1972a. Dynamics of hapten-antibody interaction studies on a myeloma pro­tein with anti-2,4-dinitrophenyl specificity. ,J. Mol. Biol. 68: 241.

PECHT, I., D. HASELKORN ands. FRIEDMAN. 1972b. Kinetic

mapping of antibody binding sites by chemical relaxa­tion spectroscopy. FEBS Letters 24: 331.

POLJAK, R. J., L. M. AMZEL, B. L. CHEN, R. P. PHIZACKERLEY and F. SAUL. 1974. The three-dimensional structure of the Fab' fragment of a human myeloma immu­noglobulin at 2.0 A resolution. Proc. Nat. Acad. Sci. 71: 3440.

RAMAKRISHNAN, c. and G. N. RAMACHANDRAN. 1965. Stereochemical criteria for polypeptide and protein chain conformations. II. Allowed conformations for a pair of peptide units. Biophys. J. 5: 909.

RocKEY, J. H., P. C. MONTGOMERY, B. J. UNDERDOWN and K. J. DORRINGTON. 1972. Circular dichroism studies on the interactions of haptens with MOPC-315 and MOPC-460 mouse myeloma proteins and specific anti­bodies. Biochemistry 11: 3172.

RUDIKOFF, S. and M. POTTER. 1974. Variable region se­quence of the heavy chain from a phosphorylcholine binding myeloma protein. Biochemistry 13: 4033.

SCHIFFER, M., R. L. GIRLING, K. R. ELY and A. B. EDMUND­SON. 1973. Structure of a .\-type Bence-Jones protein at 3.5 A resolution. Biochemistry 12: 4620.

SEGAL, D. M., E. A. PADLAN, G. H. COHEN, s. RUDIKOFF, M. POTTER and D. R. DAVIES. 1974a. The three-dimen­sional structure of a phosphorylcholine-binding mouse immunoglobulin Fab and the nature of the antigen binding site. Proc. Nat. Acad. Sci. 71: 4298.

SEGAL, D. M., E. A. PADLAN, G. H. COHEN,E. w. SILVERTON, D. R. DAVIES, S. RuDIKOFF and M. POTTER. 1974b. The structure of McPC 603 Fab and its hapten complex. In Progress in Immunology II: Immunochemical as­pects (ed. L. Brent and J. Holborowl, vol. 1, p. 93. North-Holland, Amsterdam.

TROTTER, J. 1961. A three-dimensional analysis of the crystal structure of m-dinitrobenzene. Acta Cryst. 14: 244.

TROTTER, J. and C. S. WILLISTON. 1966. Bond lengths and thermal vibrations in m-dinitrobenzene. Acta Cryst. 21: 285.

VENKATACHALAM, C. M. 1968. Stereochemical criteria for polypeptides and proteins. V. Conformation of a sys­tem of three linked peptide units. Biopolymers 6: 1425.

WONG, L. T. L., L. H. PIETTE, J. R. LITTLE and J.C. HsIA. 1974. Stereospecificity of murine myeloma protein-315 to enantiomeric spin labeled dinitrophenyl hapten. Immunochemistrv 11: 377.

Wu, T. T. and E. A: KABAT. 1970. An analysis of the se­quences of the variable regions of Bence-Jones proteins and myeloma light chains and their implications for antibody complementarity. J. Exp. Med. 132: 211.

---. 1971. An attempt to locate the non-helical and per­missively helical sequences of proteins: Applications to the variable regions of immunoglobulin light and heavy chains. Proc. Nat. Acad. Sci. 68: 1501.

13 of 13 Celltrion, Inc., Exhibit 1116