Ann. Rev. Biophys. Biophys. Chem. 1987. 16: 139-59
`
`reserved Copyright © 1987 by Annual Reviews Inc. All rights
`
`
`THE STRUCTURAL BASIS
`OF ANTIGEN-ANTIBODY
`RECOGNITION
`
`R. A. Mariuzza
`
`
`
`Departement d'Immunologie, Institut Pasteur, 75724 Paris Cedex 15,
`
`
`
`
`
`France
`
`S. E. V. Phillips
`
`Astbury Department of Biophysics, University of Leeds, Leeds,
`
`
`
`
`
`
`United Kingdom
`
`R. J. Poljak
`
`
`
`Departement d'Immunologie, Institut Pasteur, 75724 Paris Cedex 15,
`
`
`
`
`
`France
`
`CONTENTS
`PERSPECTIVES AND OVERVIEW ....................................................................................... 139
`THE THREE-DIMENSIONAL STRUCTURE OF ANTIGEN AND ANTIBODY IN THE COMPLEX ........ 1 4 1
`141
`
`
`
`
`Overall Conformation o f the Fab-Lysozyme Complex...............................................
`
`
`
`
`
`The Antigen-Antibody Interface and the Nature of the Antibody Combining Site
`...... 142
`of Antigen and Antibody.......... 146
`
`
`Effect o!Complex Formation on the Coriformations
`
`ANTIGEN VARIABILITY AND ANTIBODY SPECIFICITY ....................................................... 148
`
`
`STRUCTURE OF ANTIGENIC DETERMINANTS .................................................................... 150
`
`LOCK AND KEY VERSUS INDUCED-FIT MODELS OF ANTIGEN-ANTIBODY RECOGNITION......... 153
`154
`STRUCTURE, SPECIFICITY, AND GENETIC CONTROL OF ANTIBODIES....................................
`155
`SUMMARY .....................................................................................................................
`
`PERSPECTIVES AND OVERVIEW
`
`Antibody molecules, acting as specific antigen receptors at lymphocyte
`
`
`
`
`
`
`
`
`
`surfaces or in solution in serum and other biological fluids, are a central
`139
`
`
`
`0883-9182/87/0610-0139$02.00
`
`Annu. Rev. Biophys. Biophys. Chem. 1987.16:139-159. Downloaded from www.annualreviews.org
`
` Access provided by Reprints Desk, Inc. on 07/20/19. For personal use only.
`
`Lassen - Exhibit 1052, p. 1
`
`

`

`140 MARIUZZA, PIDLLIPS & POLlAK
`
`part of the immune defense systems of higher organisms. The immune
`
`
`system can normally discriminate between self and nonself antigens and
`
`
`
`be,tween closely related structural forms of chemical molecules with a high
`
`
`specificity, which is based on the complementarity of close molecular
`
`
`
`
`interactions. Biochemical, immunochemical, and physicochemical studies
`in references 4, 7, 16, 28)
`
`
`of immunoglobulins and antibodies (reviewed
`
`
`have shown that the structural bases of antigen recognition reside in the
`
`Fab part of the antibody molecule. The complementarity-determining
`
`regions (CDRs) of the heavy (H) and light (L) polypeptide chains of
`
`
`immunoglobulins determine, by their hypervariable sequences, the anti­
`
`gen-binding specificity of antibody molecules.
`
`
`X-ray crystallographic studies of ligand-antibody (Fab) complexes (5,
`54; reviewed in 4, 16) provided the most detailed structural
`pictures of
`
`specific binding reactions at the combining sites of antibody molecules.
`
`
`The crystalline material for these X-ray diffraction studies was obtained
`
`from human and murine myeloma immunoglobulins. However, these stud­
`
`ies were not sufficient to allow characterization of the combining site of
`
`
`antibodies because the ligands, vitamin KjOH (5) and phosphorylcholine
`(54),
`
`are relatively small and do not contact all the combining site residues.
`
`Thus, for example, the fact that no conformational changes were observed
`
`
`
`in the complexed Fabs (5,54) could be attributed to the insufficient number
`of contacts made by the ligands with the combining site, or to the fact that
`
`
`
`the antibodies were not highly specific or did not have a sufficiently high
`(KA '" 1 X 105 mol-I) for those ligands.
`affinity constant
`Equally import­
`
`
`
`ant questions remained about the nature of antigenic determinants (or
`
`
`
`epitopes) recognized by antibody molecules. For example, are the epitopes
`
`
`
`
`n:cognized by antibodies made of continuous residues of the sequence of
`
`
`
`
`the antigen, or are they discontinuous, topographical features assembled
`
`
`
`by the three-dimensional folding of the antigen molecule? In addition, are
`
`
`there conformational changes in the antigen following complex formation
`
`
`of with the antibody? How can antigen variations influence the specificity
`
`
`antibodies? These and other questions could only be answered by the study
`
`of an antigen-antibody complex.
`The advent of cellular hybridization techniques for producing cell lines
`
`
`
`
`
`
`secreting antibodies of predefined specificity (33) provided an experimental
`
`approach to these questions. We chose lysozyme as an antigen model
`is known to high resolution (10),
`
`because its three-dimensional structure
`and because it has been used as a model antigen in different laboratories
`in 7). We expected a crystalline
`(reviewed
`complex between the antigen,
`
`
`hen egg-white lysozyme (HEL), and the Fab fragment of a specific mono­
`
`clonal anti-HEL antibody (acting as a monovalent antibody) to provide a
`
`
`
`structural model for the interpretation of an antigen-antibody interaction.
`
`Annu. Rev. Biophys. Biophys. Chem. 1987.16:139-159. Downloaded from www.annualreviews.org
`
` Access provided by Reprints Desk, Inc. on 07/20/19. For personal use only.
`
`Lassen - Exhibit 1052, p. 2
`
`

`

`ANTIGEN-ANTIBODY RECOGNITION 141
`
`After a systematic exploration of 27 monoclonal anti-HEL antibodies we
`
`
`
`
`
`
`obtained such a complex in crystalline form and studied it by X-ray
`
`
`
`diffraction techniques as reviewed below.
`We do not endeavor to present here an exhaustive review of antigen­
`
`
`
`
`
`
`
`
`antibody interactions or of the pertinent literature. Instead we discuss
`
`
`some of the conclusions of our recent crystal-structure determination of
`
`
`
`this antigen-antibody complex. We emphasize that all the experimental
`
`
`
`data that have been obtained by immunochemical studies with HEL and
`D1.3 in solution
`
`
`
`the specific anti-HEL monoclonal antibody
`(20) are in
`
`
`
`
`agreement with the observations and conclusions of the crystal structure
`
`
`
`
`analysis (2, 3), and thus cQnfirm the relevance of this analysis in the
`
`
`interpretation of immunochemical phenomena.
`
`THE THREE-DIMENSIONAL STRUCTURE OF
`ANTIGEN AND ANTIBODY IN THE COMPLEX
`
`The production of hybrid cell lines secreting murine monoclonal anti-HEL
`
`
`
`
`
`
`
`antibodies, and the purification and crystallization of the complex between
`Fab D1.3 and lysozyme
`(41). Crystals
`have been described
`were grown
`
`
`from 15-20% polyethylene glycol 8000 solutions at pH 6.0. The three­
`
`dimensional crystal structure was solved to a resolution of 6 A and sub­
`
`
`
`
`
`
`sequently to 2.8 A. The conventional R value after high-resolution refine­
`
`ment is currently 0.28.
`Overall Conformation of the Fab-Lysozyme Complex
`
`
`
`
`
`
`A low-resolution (6 A) electron-density map of the Fab-Iysozyme complex
`
`
`
`
`revealed the typical domain structure of a Fab fragment (45) plus an
`
`
`
`
`
`additional globular region, corresponding to lysozyme, in close contact
`
`
`
`
`with the Fab. The a-carbon backbone of the previously determined hen­
`
`
`
`lysozyme structure could be fitted to the map using only rigid-body
`
`
`
`
`
`rotations and translations. The a-carbon backbone ofthe variable domain
`(VL + VH) of Fab New! (46, 47) was similarly
`
`fitted, but the constant
`domain (CH 1 + CL) could be fitted
`
`only by rotation about an axis through
`
`
`
`
`residues 103L and 117H at the flexible switch regions. This changed the
`
`
`
`angle between the variable and constant domains from 137° in Fab New
`
`
`to about 180°, closer to the 166° angle observed in Fab Kol (23). The low­
`
`resolution model (resembling that shown in Figure 1, obtained at 2.8-A
`
`
`
`
`
`
`resolution) showed that the interactions between Fab and lysozyme extend
`over a large area of about 20 x 30 A.
`
`
`
`I Human Fabs and myeloma immunoglobulins
`
`are labeled with the first three letters of the
`
`patient's surname.
`
`The electron-density map of the complex at 2.8-A resolution confirmed
`
`Annu. Rev. Biophys. Biophys. Chem. 1987.16:139-159. Downloaded from www.annualreviews.org
`
` Access provided by Reprints Desk, Inc. on 07/20/19. For personal use only.
`
`Lassen - Exhibit 1052, p. 3
`
`

`

`142 MARIUZZA, PHILLIPS & POLJAK
`
`Figure 1 Stereo diagram
`of the IX-carbon skeleton of the Fab-Iysozyme complex. Fab is
`
`
`
`
`shown in upper right; the heavy and light chains are shown with thick and thin bonds,
`
`nlspectively. The lysozyme active site is the cleft containing the label HEL. Antibody-antigen
`
`
`interactions are most numerous between the lysozyme and the heavy chain CDR loops.
`(Reproduced from Reference 3, with permission.)
`
`the results of the 6-;\ resolution study, and also clearly showed the detailed
`
`
`
`
`
`
`
`
`
`conformation of the amino-acid side chains, allowing a complete descrip­
`tion of the structure.
`
`The Antigen-Antibody Interface and the Nature of
`
`
`
`the Antibody Combining Site
`The site recognized by D 1.3 is made up of two stretches of the polypeptide
`
`
`
`
`
`
`chain of lysozyme, residues 18-27 and 116-129, which are distant in the
`
`
`amino-acid sequence but adjacent on the protein surface (Table 1).
`
`
`
`
`
`
`
`
`The antibody combining site appears as an irregular, relatively fiat
`
`
`
`
`surface with protuberances and depressions formed by the amino-acid side
`
`chains of the CDRs of VH and VL. In addition, there is a small cleft
`
`
`
`between the third CDRs ofVH and VL, corresponding to the binding site
`
`
`
`
`eharacterized in hapten-antibody complexes (5, 54). Although the cleft is
`
`
`
`
`not the geometrical center of the antigen-antibody interface, it binds the
`
`side chain of Gln121 of lysozyme (see Figure 2), AU six CDRs interact
`
`
`
`with the antigen; 16 antigen residues make close contacts with 17 antibody
`
`
`
`
`
`residues (Table 2). Hydrogen exchange experiments with polyclonal anti-
`
`Annu. Rev. Biophys. Biophys. Chem. 1987.16:139-159. Downloaded from www.annualreviews.org
`
` Access provided by Reprints Desk, Inc. on 07/20/19. For personal use only.
`
`Lassen - Exhibit 1052, p. 4
`
`

`

`ANTIGEN-ANTIBODY RECOGNITION 143
`
`Table 1 Lysozyme residues
`in contact with antibody"
`
`
`Lysozyme residues
`
`Number of antibody residues in contact
`
`AsplS
`Asnl9
`Arg21
`Gly22
`Tyr23
`Ser24
`Leu25
`Asn27
`Lys1l6
`Glyll7
`Thrll8
`Asp 119
`Vall 20
`Glnl21
`Ile124
`Leul29
`
`1 L chain
`2H,L
`IH
`4 H(3), L
`2H
`IH
`l L
`I H
`3 H
`6H
`2 H
`2 H
`I H
`5 H(l), L(4)
`2L
`I L
`
`
`
`a From Reference 3.
`
`R
`
`Figure 2 Stereo diagram of the antibody-antigen
`interface in an orientation similar to that
`
`
`in Figure I. All atoms are shown for residues
`
`involved in the interaction. Heavy and light
`by thick and thin bonds, and hydrogen bonds are indicated
`main chains are indicated
`by
`dotted lines.
`the diagonal from top left to lower right
`
`Most lysozyme residues lie below
`of
`from Reference 3, with permission.)
`the diagram. (Reproduced
`
`(39) indicated
`
`
`
`(Glu,Ala,Tyr)n sheep and rabbit antibodies 16-19 apparently
`
`
`
`site-associated amide hydrogens that did not exchange out from liganded
`
`
`
`Fab, and an additional 6-7 amide hydrogens whose exchange was retarded
`
`
`
`
`upon ligand binding. These numbers are in agreement with the number of
`
`
`
`
`
`antibody-contacting residues in the lysozyme-Fab complex described here.
`
`Annu. Rev. Biophys. Biophys. Chem. 1987.16:139-159. Downloaded from www.annualreviews.org
`
` Access provided by Reprints Desk, Inc. on 07/20/19. For personal use only.
`
`Lassen - Exhibit 1052, p. 5
`
`

`

`144 MARIUZZA, PIDLLIPS & POLJAK
`
`The overall shape of antibody combining sites is variable and is largely
`
`
`
`
`
`
`determined by the lengths of the CDR loops; short loops tend to form
`
`
`
`shallower sites. The phosphorylcholine-binding myeloma immunoglobulin
`
`
`
`McPC603 (54) has a significantly more concave combining site than D1.3
`
`
`because its VL CDR1, VH CDR2, and VH CDR3 are 6, 2, and 3 residues
`
`
`longer, respectively, than those ofD1.3.
`The interacting antigen and antibody surfaces are complementary, with
`
`
`
`
`
`
`
`
`protruding side chains of one surface lying in depressions of the other
`
`
`
`(Figure 2), as in other known protein-protein interactions (13) . Many van
`
`
`
`der Waals interactions are interspersed with hydrogen bonds, as in GIn 121,
`
`
`
`which makes close contacts with three aromatic side chains, VL Tyr32, VL
`
`
`Trp92, and VH Tyrl O l (see Figures 2, 4, and 5). The buried amide nitrogen
`
`
`ofGln121 forms a strong hydrogen bond to the main chain oxygen ofVL
`
`
`
`Phe91 (Table 3; see Figure 4). A symmetrical situation occurs in the
`
`
`
`adjacent VH TyrlOl, which extends to the surface of lysozyme, forming
`
`
`hydrogen bonds to the main chain nitrogens of Va1120 and Gln121 and
`
`to 001 of Asp1l 9 with its hydroxyl group. The antibody combining site
`
`
`
`shows a high concentration of aromatic side chains, especially in the
`
`
`
`
`
`Table 2 Antigen-antibody contacting residues'
`
`Antibody residuesb
`
`
`Lysozyme residues in contact
`
`Light chain
`CDRl His30
`Tyr32
`FR2
`Tyr49
`CDR2 Tyr50
`CDR3 Phe91
`Trp92
`Ser93
`
`Leul29
`Leu25, Glnl21, Ilel24
`Gly22
`Aspl8, Asnl9, Leu25
`Glnl2l
`Glnl21, Ilel24
`Glnl2l
`
`Heavy chain
`Lys1l6, Glyll7
`FRt
`Thr30
`Lys116, Gly117
`CDRl Gly3l
`Lys116, Gly1l7
`Tyr32
`Gly1I7, Thrll8, Asp1I9
`CDR2 Trp52
`Gly1I7
`Gly53
`Gly1l7
`Asp54
`CDR3 Arg99
`Arg2l, Gly22, Tyr23
`Gly22, Tyr23, Ser24, Asn27
`AsplOO
`Thr1l8, Asp1l9, Vall20, Glnl2l
`TyrlOl
`Asnl9, Gly22
`Arg\o2
`
`a From Reference 3.
`b Residue
`
`positions are numbered sequentially.
`
`Annu. Rev. Biophys. Biophys. Chem. 1987.16:139-159. Downloaded from www.annualreviews.org
`
` Access provided by Reprints Desk, Inc. on 07/20/19. For personal use only.
`
`Lassen - Exhibit 1052, p. 6
`
`

`

`ANTIGEN-ANTIBODY RECOGNITION 145
`
`Table 3 Hydrogen-bonded interactions between
`
`
`
`antigen and antibody'
`
`Antibody residueb Lysozyme residue
`
`Light chain
`His30 0
`Ne2
`Tyr50 Obi
`Ort
`0
`Phe91 Ne2
`Heavy chain
`Thr30 0
`Oyl
`Gly31 0
`N
`Gly53 0
`N
`Arg99 0
`Nrtl
`Obi
`Asp 100 Nb2
`Ob2
`Asp 100 Oy
`TyrlOl N
`Ort
`TyrlOI N
`Of/
`TyrJOI Obi
`Of/
`
`Leul29
`Aspl8
`Gln121'
`
`Lys II 6'
`Lys1l6
`Glyll7'
`Gly22
`Asn27
`Ser24'
`Val120
`Glnl21
`Asp 119
`
`a From Reference 3.
`b Residue
`
`positions are numbered sequentially.
`
`
`'Denotes the dosest interactions (distances ::;2.5 A).
`
`environment of Glnl2l. High concentration of aromatic side chains was
`
`
`
`
`
`
`also observed in the phosphorylcholine-McPC603 complex. The phos­
`
`
`
`
`phorylcholine lies in a similar pocket, surrounded by V L TyrlOO, V H Trp33,
`
`
`and VH Tyrl 07 (54). Many other hydrogen bonds occur between the side
`
`
`
`chains of the antigen and the main polypeptide chain of the antibody, and
`
`
`
`vice versa (Table 3). Another type of hydrogen bond, similar to those in
`
`
`
`
`p-sheet structures, occurs between Lys1 16 oflysozyme and VH Gly31 and
`between Gly117 and VH Gly53, where the lack of side chains allows
`
`
`
`
`
`closeness of antigen and antibody main polypeptide chains. There are
`
`
`many close side chain-side chain contacts which, with the contacts
`
`
`
`
`described above, form a tightly packed interface from which solvent is
`mostly excluded.
`Although the antigen-antibody interface involves all six CDRs of the
`
`
`
`
`
`
`Fab, VH CDRs, and VH CDR3 in particular, have more interactions than
`
`
`
`
`VL CDRs (Tables 1-3). The geometrical center of the interface lies near
`
`
`VH CDR3, and is occupied by the side chain ofVH AsplOO, which forms
`
`
`hydrogen bonds to the side chains of Ser24 and Asn27 of lysozyme. Of
`
`
`
`the antibody hypervariable regions, V L CDR2, known to be the least
`
`
`
`
`variable, contributes the least to antigen binding. Many of the antibody
`
`
`
`side chains in the interface (9 of 15 if we exclude the two Gly residues) are
`
`
`
`
`aromatic, and thus present large areas of hydrophobic surface to the
`
`
`
`antigen; in addition, some of the side chains, such as V L Tyr50 and V H
`
`Annu. Rev. Biophys. Biophys. Chem. 1987.16:139-159. Downloaded from www.annualreviews.org
`
` Access provided by Reprints Desk, Inc. on 07/20/19. For personal use only.
`
`Lassen - Exhibit 1052, p. 7
`
`

`

`146 MARWZZA, PIDLLIPS & POLJAK
`
`Tyrl 01, participate in hydrogen bonding with the antigen via their polar
`
`
`
`
`
`
`
`atoms. About 11 % of the solvent-accessible surface (37) of lysozyme (748
`
`
`A 2) is buried upon complex formation; a similar area (690 A 2) of the
`
`
`a.ntibody surface is also buried.
`Although the contacting surfaces of antigen and antibody fit each other
`
`
`
`
`
`
`
`
`
`remarkably well, there are some unoccupied spaces between them. A water
`
`
`
`
`
`molecule probably fills the space between VH residues 52 and 100 and
`
`
`
`
`lysozyme residues 24 and 118; an electron-density peak at that location
`
`
`
`
`suggests that a water molecule is hydrogen-bonded to the nitrogen atom
`
`
`
`of lysozyme Gly117. Other unoccupied spaces appear not to be filled
`
`
`
`
`by water molecules. These observations might explain the occurrence of
`
`
`
`
`
`heteroclitic antibodies that have higher affinities for heterologous, closely
`
`
`
`
`related antigens. Such antibodies could fill the unoccupied spaces with
`
`
`
`their side chains and provide a tighter association. This might also explain
`
`
`
`
`how somatic mutations in the antibody genes could give a better fit between
`
`
`
`
`
`antigen and antibody. For example, in antibody Dl.3 an increase in affinity
`
`
`
`could be achieved by amino-acid changes that would fill the holes in the
`
`
`interface with hydrophobic side chains or permit salt links and hydrogen
`
`
`
`bonding to lysozyme residues, such as Arg21 and ThrI18, whose polar
`
`parts are unbonded in the interface.
`
`Effect of Complex Formation on the Conformations of
`
`
`
`Antigen and Antibody
`The availability of several high-resolution structures of hen lysozyme in
`
`
`
`
`
`
`
`
`
`different crystal forms allows a direct assessment of the effect of complex
`
`
`formation on antigen conformation. A least squares fit of Ox atoms of
`
`
`
`
`
`
`
`lysozyme in the complex to those of native lysozyme in tetragonal crystal
`
`
`
`form refined at 1.6 A (10) gave a root-mean-square (rms) deviation 0[0.64
`
`
`A between the two (see Figure 3). Since the error in atomic positions in
`
`the complex can be estimated (40) as approximately 0.6 A, the observed
`
`
`
`
`
`
`:rms difference is not significant. In addition, the regions most distant from
`
`
`
`antibody contacts show the largest changes (up to 1.6 A). A similar fitting
`
`
`
`
`
`
`of the tetragonal hen lysozyme form with the structure of triclinic hen
`
`
`
`
`lysozyme refined from neutron diffraction data (42) gave an rms deviation
`
`
`
`
`of 0.88 A. The rms deviation between tetragonal and orthorhombic hen
`
`
`lysozyme determined at physiological temperature (8) was 0.46 A. Somf:
`
`
`
`
`
`
`differences in side-chain conformation can be observed between complexed.
`
`
`
`
`
`and tetragonal lysozyme, but similar differences are observed betweeL
`
`
`
`
`
`different crystal structures of native hen lysozyme. Thus, it appears that
`
`
`
`
`
`complexing with antibody D 1.3 distorts the structure oflysozyme no mon:
`than crystallization.
`
`
`
`
`Additivity enzyme-linked immunosorbant assays (ELISA) and sep-
`
`Annu. Rev. Biophys. Biophys. Chem. 1987.16:139-159. Downloaded from www.annualreviews.org
`
` Access provided by Reprints Desk, Inc. on 07/20/19. For personal use only.
`
`Lassen - Exhibit 1052, p. 8
`
`

`

`ANTIGEN-ANTIBODY RECOGNITION 147
`
`Figure 3 The IX-carbon skeleton
`oflysozyme in the complex superimposed by least-squares
`
`to Fab is at the top;
`
`
`
`on that of native lysozyme in the tetragonal crystal form. The interface
`
`
`no significant conformational change is apparent in this region, except for that of the C­
`
`
`
`terminal residues, which make few contacts to Fab. Greater differences, although still not
`oCCur at the bottom
`
`significant at this resolution,
`
`
`of the molecule. (Reproduced from Ref­
`erence 3, with permission.)
`
`aration of the ternary complex Fab D1.3-HEL-other Fab in solution
`
`
`
`
`
`
`indicated that complex formation between Fab D 1.3 and lysozyme does
`
`
`
`not induce conformational changes that can be detected by other mono­
`
`
`
`
`clonal antibodies raised against native lysozyme (20). These monoclonal
`
`
`
`
`
`
`antibodies are directed against different, nonoverlapping antigenic deter­
`
`
`
`minants of HEL. One of the antibodies, D74.3, recognizes a determinant
`99, 101, 102, and 103. Another
`
`that includes HEL residues
`
`antibody that
`complex with Fab D1.3 and HEL, D44.2, recognizes
`forms a ternary
`an
`
`
`
`
`
`antigenic determinant that includes Arg68, located at the distal end of the
`
`
`HEL determinant recognized by D 1.3.
`
`
`
`Since the crystal structure of unligated Fab D1.3 has not been deter­
`
`
`
`
`mined, a direct comparison between complexed and uncomplexed forms
`
`
`
`is not possible. The similarity of complexed Fab D1.3 with other Fab
`
`
`
`
`
`
`structures suggests, however, that possible conformational changes in the
`
`
`
`
`variable domain should be small. NMR studies on unligated and hapten
`
`
`
`(dinitrophcnol)-ligated mouse myeloma protein MOPC315 support this
`
`
`
`conclusion (18). In addition, predicted structures for unligated D1.3 ( 14)
`
`
`
`based on its amino-acid sequence and on the three-dimensional structures
`
`
`of other Fabs also agree well with the determined structure of the com­
`
`
`
`
`plexed antibody in the framework p-sheet regions and in some, but not
`
`
`
`
`all, of the CDR loops. Furthermore, the unaltered relative disposition of
`
`Annu. Rev. Biophys. Biophys. Chem. 1987.16:139-159. Downloaded from www.annualreviews.org
`
` Access provided by Reprints Desk, Inc. on 07/20/19. For personal use only.
`
`Lassen - Exhibit 1052, p. 9
`
`

`

`148 MARIUZZA, PIDLLIPS & POLJAK
`
`VH and VL in Fab D1.3 indicates
`that antigen binding causes no change
`
`
`
`
`
`in quaternary structure in the variable domain.
`
`
`The axes of the variable and constant domains of Fab D1.3 make an
`angle of � 180°, i.e. an extended
`("elbow-bending")
`
`conformation. This
`
`
`
`observation is not in agreement with Huber et aI's hypothesis (23) that the
`
`
`
`antibody molecule assumes a more rigid conformation, with an
`ligated
`angle of � 120°. In fact, the angle observed
`"elbow-bending"
`in the lyso­
`
`
`
`
`zyme-Fab D1.3 complex corresponds to that postulated to occur in unli­
`
`
`
`gated Fabs. Thus, the allosteric model of antibodies (reviewed in 4, 16),
`
`
`
`
`
`according to which antigen binding induces conformational changes trans­
`
`
`
`
`mitted by close contacts across the variable and constant domains, is not
`
`
`
`consistent with the structure of this complex.
`
`
`
`
`
`established at the current resolution.
`
`No other change of conformation in the antibody or antigen can be
`
`ANTIGEN VARIABILITY AND ANTIBODY
`SPECIFICITY
`
`is of hen lysozyme The equilibrium affinity constant of Fab D 1.3 binding
`
`
`
`
`4.5 x 107 mol-I, an average
`
`
`value for monoclonal antiprotein antibodies
`
`(31). This corresponds to a free energy change of -10.3 kcal mol-I. The
`
`
`
`fine specificity of monoclonal antibody D1.3 for other avian lysozymes
`
`
`
`
`shows its ability to distinguish a single amino-acid change in the antigen,
`
`
`
`at position 121. Bobwhite quail lysozyme, which has four amino-acid­
`
`
`
`
`sequence differences from hen lysozyme (48) but none in the interface with
`
`
`
`Fab D1.3, binds with similar affinity (20). The binding of antibody D1.3
`
`
`
`
`
`to the lysozymes of partridge [3 amino-acid differences (26)], California
`
`
`
`
`
`quail [4 amino-acid differences (25)], Japanese quail [6 amino-acid differ­
`
`
`
`
`
`ences (29)], turkey [7 amino-acid differences (35)], pheasant and guinea
`(KA « 1 x 105
`
`
`fowl [10 amino-acid differences each (27») is undetectable
`
`
`
`mol-I). These lysozymes differ from hen lysozyme at position 121, which
`
`
`
`makes close contact with the antibody. In all except for Japanese quail
`
`
`
`and pheasant lysozymes, His replaces GIn at that position.
`
`
`
`Of all lysozyme residues in the interface, Gln121 loses the greatest
`
`
`solvent-accessible area upon complex formation (Table 4). Computer­
`
`
`
`
`
`graphics analysis and conformational energy calculations (3) indicate that
`
`
`
`despite the tightness of the packing around the side chain of this residue
`
`
`
`(see above), GIn can be replaced by His with only minor displacements of
`
`
`
`contacting Ab side chains; the buried hydrogen bond made by Gln121
`
`
`can also be made with good geometry by the replacing His side chain.
`
`
`
`
`Indeed, energy minimization of the complexes using the methods of Levitt
`
`
`
`(38) and Brooks et al (11) indicates that energies for the complex with His at
`
`Annu. Rev. Biophys. Biophys. Chem. 1987.16:139-159. Downloaded from www.annualreviews.org
`
` Access provided by Reprints Desk, Inc. on 07/20/19. For personal use only.
`
`Lassen - Exhibit 1052, p. 10
`
`

`

`ANTIGEN-ANTIBODY RECOGNITION 149
`surface area CA2) of antigen residues
`Table 4 Accessible
`
`% Loss of
`Lysozyme
`residue Uncomplexed Complexed accessibility"
`
`Lys13
`Asp18
`Asn19
`Arg21
`Gly22
`Tyr23
`Ser24
`Leu25
`Gly26
`Asn27
`Arg112
`Asn113
`Lys116
`Glyll7
`Thr118
`Asp119
`Vall 20
`Gln121
`Ile124
`Arg125
`Leu129
`
`85.37
`91.02
`49.60
`76.59
`26.16
`89.44
`148.15 142.51
`6.79
`52.82
`44.61
`20.70
`7.36
`62.99
`12.91
`26.17
`2.24
`2.66
`4.95
`14.45
`123.17
`106.86
`94.61
`96.46
`38.80
`121.75
`14.16
`77.82
`50.69
`80.70
`92.10
`56.10
`4.21
`14.55
`7.33
`143.89
`16.04
`65.11
`143.12
`180.16
`156.91
`190.66
`
`6.2
`35.2
`70.8
`3.8
`87.1
`53.6
`88.3
`50.7
`15.8
`65.7
`13.2
`1.9
`68.1
`81.8
`37.2
`39.1
`71.1
`94.9
`75.4
`20.6
`17.7
`
`'Percent loss of accessibility on complex formation. The greatest loss of
`
`
`
`is observed for Glnl21.
`accessibility
`
`position 121 are lower by approximately 2 and 10 kcal mol-I, respectively.
`
`
`
`
`
`While such methods provide unreliable estimates of absolute con­
`
`
`
`
`
`formational energy, differences due to the replacement of a single buried
`
`
`
`residue should be more dependable. Since the substitution of His for GIn
`
`
`
`
`in the interface is not stereo chemically forbidden and may even be favored,
`in affinity of D 1.3 for Iysozymes
`
`
`the explanation of the dramatic decrease
`
`
`
`
`with His at position 121 lies elsewhere. Other possibilities include differ­
`
`
`ences in hydrophobicity of the side chains of these residues and local
`
`
`
`conformational or charge differences in the antigen. Free energy of transfer
`
`
`
`
`
`experiments between water and organic solvents suggest that His is slightly
`
`more hydrophobic (0.8 kcal mol-I) than GIn, which rules out the former
`
`
`
`possibility. Since the pK of His121 is unknown, its charge at neutral pH
`
`
`
`placed cannot be assessed. The side chain of the adjacent Asp 1 19 is ideally
`
`to form a salt bridge with the N82 of His121, but the Hisl 21 side chain
`
`
`would have to be in a conformation no longer complementary to the
`
`
`antibody-combining site. Thus if there were a salt bridge, complex for­
`
`
`
`mation would require its rupture and a deprotonation and reorientation
`
`Annu. Rev. Biophys. Biophys. Chem. 1987.16:139-159. Downloaded from www.annualreviews.org
`
` Access provided by Reprints Desk, Inc. on 07/20/19. For personal use only.
`
`Lassen - Exhibit 1052, p. 11
`
`

`

`1 50 MARlliZZA, PHILLIPS & POLlAK
`
`of the imidazole side chain, and the free energy of association would
`
`
`
`
`
`be reduced. The presence of such a salt bridge could be confirmed by
`
`
`
`
`determination of the structure of California quail lysozyme or another
`
`
`
`lysozyme containing His at position 121.
`Knossow et al (32) used X-ray diffraction to examine another case in
`
`
`
`
`
`
`
`
`which a single amino-acid substitution in the antigen is sufficient to abolish
`
`
`
`
`
`
`antibody binding. They selected an antigenic variant of influenza virus by
`
`
`
`growing the virus in the presence of a neutralizing monoclonal antibody
`
`
`
`that recognizes the viral hemagglutinin glycoprotein. The mutant hem­
`
`
`
`
`
`agglutinin contained a single amino-acid substitution (Gly146 --+ Asp).
`
`
`
`
`Determination of its three-dimensional structure revealed only minor local
`
`
`
`
`distortions in the hemagglutinin conformation. The authors suggested two
`
`
`
`
`explanations for the decreased affinity of the monoclonal antibody for the
`
`
`
`mutant hemagglutinin: (a) The larger size and negative charge of the Asp
`
`
`
`side chain may prevent its accommodation by the antibody combining
`
`site. (b) An apparent hydrogen bond between the OD of the carboxylate
`
`
`
`
`of Asp1 46 and the Ne of Arg 1 4 1 may stabilize the local structure of the
`
`
`
`
`silte to which the antibody binds, no longer permitting the antibody to
`
`
`"induce a fit" with the antigen. Alternatively, accommodation of Asp146
`
`
`
`
`may require rupture of the hydrogen bond and repositioning of its side
`
`chain, but at an energy cost greater than the free energy of complex
`formation.
`Japanese quail lysozyme has Asn at position 1 2 1 and the additional
`
`
`
`
`
`differences of Lys and GIn replacing Asn1 9 and Arg21 , respectively.
`
`Asn121 cannot form a hydrogen bond as strong as that of Glnl 2 1 . In
`
`
`addition, replacement of Asn 1 9 by Lys causes the loss of weak interactions
`
`
`between ObI of Asnl9 and N ofVH Arg102. The positively charged Lysl9
`
`
`
`
`side chain would be repelled by VH Arg1 02 and would probably remain
`
`
`
`
`outside the interface, which would reduce packing efficiency. The side
`
`
`chain of Arg21 is external to the interface (only its main chain atoms make
`
`
`
`
`contact with Fab), so changes at this position on the antigen surface
`
`
`
`probably do not interfere with complex formation.
`
`STRUCTURE OF ANTIGENIC DETERMINANTS
`
`The antigenic determinant (defined here as the portion of a protein bound
`
`
`
`
`
`
`
`
`
`
`by a specific antibody) recognized by antibody 0 1 .3 is distinguished by
`
`
`three principal characteristics: (a) It is large, nearly 750 A2 in area, com­
`
`
`
`
`residues large number of lysozyme prising different atoms of a relatively
`
`(see Table I); (b) it is formed by two segments of amino acids (residues
`
`
`1 8-27 and 1 16- 129), which are widely separated in the primary structure
`
`
`of lysozyme but adjacent in its three-dimensional structure; and (c) it is
`
`
`
`Annu. Rev. Biophys. Biophys. Chem. 1987.16:139-159. Downloaded from www.annualreviews.org
`
` Access provided by Reprints Desk, Inc. on 07/20/19. For personal use only.
`
`Lassen - Exhibit 1052, p. 12
`
`

`

`ANTIGEN-ANTIBODY RECOGNITION 151
`
`on the surface of the antigen, fully exposed to solvent. To what extent are
`
`
`
`
`
`
`
`
`
`these likely to be general characteristics of antigenic determinants? We can
`
`
`
`
`attempt to answer this question within the context of our current under­
`
`
`
`
`standing of the three-dimensional structure of proteins and the chemical
`
`
`
`
`nature of protein-protein interactions in specific biological complexes.
`
`
`
`The size of antigenic sites is, of course, determined by the dimensions
`
`
`
`
`of the corresponding recognition site (i.e. the antibody combining site).
`
`
`
`The dimensions of the combining site are in turn set by the positions and
`
`
`
`
`size of the CDR loops, with possible limited participation by the antibody
`
`
`
`
`framework (see above). The size of an antigenic determinant therefore
`
`
`depends on which and how many of the CDR loops it contacts. In the
`
`
`
`
`
`Fab-Iysozyme complex presented here, the antigen is positioned such that
`
`
`
`
`all six CDRs participate in binding; this results in a contact area of about
`
`
`
`
`750 A2, which probably represents the average size of an antigenic site.
`
`
`
`
`One can imagine other situations; for example, the antigen may be heavily
`
`
`
`tilted toward one of the two variable domains so that not all of the CDRs
`
`
`
`of the other can make contact. (Indeed, in the D1.3-lysozyme complex
`
`there are more interactions with VH than with VL CDRs.) In such a case
`
`
`the area of interaction might be reduced by as much as half. However,
`
`
`on the basis of the three-dimensional structure of other protein-protein
`
`
`
`
`complexes (see, for example, 12, 1 3, 24, 49), it appears unlikely that two
`
`
`
`
`
`globular proteins can be juxtaposed without a contact area of at least
`
`several hundred square angstroms.
`Immunochemists have traditionally divided antigenic determinants into
`
`
`
`
`
`
`
`
`two structural categories: (a) continuous determinants, in which all the
`
`
`
`
`residues in contact with antibody are contained within a single segment of
`
`
`the amino-acid sequence of the antigen and (b) noncontinuous deter­
`
`
`
`
`
`minants, in which residues are far apart in the sequence but are brought
`
`together by the folding of the protein in its native conformation (55). What
`
`
`
`
`
`
`
`is the probability of occurrence of each of these types of determinants in
`
`
`
`
`
`native proteins? The answer depends on what proportion of the surface of
`
`
`
`a globular protein is made up oflinear arrays of residues. Barlow et al (6)
`
`
`
`
`have computed this proportion as a