`\.:..
`NOVEMBER 1987
`VOLUME 84
`NUMBER 22 88
`
`;
`, ~
`
`L: 1.'11\1\lL
`
`NATURn
`
`~.r-.,,.,E.S
`
`'"" ...... .... J .....
`
`Proceedings
`
`OF THE
`
`National Acadellly
`of Sciences
`
`OF THE UNITED STATES OF AMERICA
`
`PFIZER EX. 1581
`Page 1
`
`
`
`Proceedings
`OF THE
`National Academy
`of Sciences
`OF THE UNITED STATES OF AMERICA
`
`Officers
`of the
`Academy
`
`Editorial Board
`of the
`Proceedings
`
`FRANK PRESS , President
`JAMES D. EBERT, Vice President
`PETER H . RAVEN , Home Secretary
`WILLIAM E. GORDON, Foreign Secretary
`ELKAN R. BLOUT, Treasurer
`
`ROBERT H. ABELES
`GORDON A. BA YM
`WILLIAM F. BRACE
`RONALD BRESLOW
`MICHAEL J. CHAMBERLIN
`
`MAXINE F. SINGER, Chairman
`MARY-DELL CHILTON
`EDWARD E. DAVID, JR.
`STUART A. KORNFELD
`DANIEL E. KOSHLAND, JR.
`PETER D. LAX
`DANIEL NATHANS
`
`HERBERT E. SCARF
`SOLOMON H. SNYDER
`HAROLD V ARM US
`THOMAS A. WALDMANN
`FRANK H. WESTHEIMER
`
`Managing Editor: FRANCES R. ZWANZIG
`Senior Associate Editor: GARY T. CocKs
`Associate Editor: CAY BUTLER
`Associate Editor: JOHN M. MALLOY
`Associate Editor: MARILYN J. MASON
`Associate Editor: JANET L. MORGAN
`Associate Editor: T. PEARSON
`Associate Editor: DOROTHY P. SMITH
`Associate Editor: COLENE WALDEN
`Assistant Managing Editor: JOANNE D'AMICO
`
`Senior Production Editor: LYNN A. WASSYNG
`Production Editors: BARBARA A. BACON, RUTH E. CROSSGROVE, ScoTT C. HERMAN,
`MARGARET M. MADELEINE, JANET L. OVERTON, KATHLEEN RtJBY,
`DoN C. TIPPMAN, DEBORAH I. WEINER
`Administrative Assistants: DELORES BANKS, CYNDY MATHEWS
`Manuscript Coordinators: EvELYN JARMON, JACQUELINE PERRY
`Circulation: JULIA LITTLE, VIRGINIA TREADWAY
`
`Editorial correspondence: PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES, 2101 Constitution Avenue, Washing(cid:173)
`ton, DC 20418.
`Business correspondence: Circulation Office of the PROCEEDINGS, National Academy of Sciences, 2101 Constitution Ave(cid:173)
`nue, Washington, DC 20418.
`Information for Contributors: See issue Number 1, January 1987.
`Copyright: The National Academy of Sciences has copyrighted this journal as a collective work and does not own copy(cid:173)
`rights for individual articles. Requests for permission to reproduce parts of individual articles or for reprints of individual
`articles should be addressed to the authors. Microforms of complete volumes are available to regular subscribers only and
`may be obtained from University Microfilms, Xerox Corporation, Ann Arbor, MI 48103.
`Subscriptions: All correspondence concerning subscriptions should be addressed to the Circulation Office of the PROCEED(cid:173)
`INGS. Subscriptions are entered on a calendar year basis only. For 1988, subscription rates are as follows-in the United
`States: student/postdoctoral, $70; personal, $200; institutional , $250; elsewhere: student/postdoctoral, $110; personal,
`$240; institutional, $290. Information regarding air mail postage rates is available from the Circulation Office. Subscribers
`are requested to notify the Circulation Office of the PROCEEDINGS 6 weeks in advance of any change of address; also the
`local postmaster. The Academy is not responsible for nonreceipt of issues because of an improper address unless a change
`of address is on file. The notice of address change should list both the old and new addresses. Claims for replacement
`copies will not be honored more than 60 days after the issue date for domestic subscribers and not more than 90 days after
`the issue date for foreign subscribers.
`Back Issues: Volumes 79-83, January 1982 and thereafter, are available from the Circulation Office ·of the PROCEEDINGS.
`The price of a single issue is $13.00.
`PRINTED IN THE USA
`Second class postage paid at Washington , DC, and at additional mailing offices.
`PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA (ISSN-0027-8424) is pub(cid:173)
`lished semimonthly by THE NATIONAL ACADEMY OF SCIENCES, 2/01 Constitution Avenue, Washington, DC 20418.
`
`© 1987 by THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA.
`
`POSTMASTER: Send address changes to: PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED
`STATES OF AMERICA, 2101 Constitution Ave., Washington, DC 20418.
`
`PFIZER EX. 1581
`Page 2
`
`
`
`Proc. Nat/. A cad. S ci. USA
`Vol. 84, pp . 8075-8079, November 1987
`Immunology
`
`Three-dimensional structure of an antibody-antigen complex
`(immunoglobulins/ epitope /x-ray crystallography/ complementarity /lysozyme)
`STEVEN SHERIFF* , ENID w. SILVERTON*, EDUARDO A. PADLAN*, GERSON H. COHEN*,
`SANDRA J. SMITH-GILLt, BARRY c. FINZEL+§, AND DAVID R. DAVIES*
`'Laboratory of Molecular Biology, National Institute of Diabetes, Digestive and Kidney Diseases, Bethesda, MD 20892; t Laboratory of Genetics , National
`Cancer Institute , Bethesda, MD 20892 ; and lGenex Corporation. Gaithersburg, MD 20877
`
`Contributed by David R . Davies , July 30, /987
`
`ABSTRACT We have determined the three-dimensional
`structure of two crystal forms of an antilysozyme Fab-lysozyme
`complex by x-ray crystallography. The epitope on lysozyme
`consists of three sequentially separated subsites, including one
`long, nearly continuous, site from Gln-41 through Tyr-53 and
`one from Gly-67 through Pro-70. Antibody residues interacting
`with lysozyme occur in each of the six complementarity(cid:173)
`determining regions and also include one framework residue.
`Arg-45 and Arg-68 form a ridge on the surface of lysozyme,
`which binds in a groove on the antibody surface. Otherwise the
`surface of interaction between the two proteins is relatively flat,
`llthough it curls at the edges. The surface of interaction is
`approximately 26 x 19 A. No water molecules are found in the
`interface. The positive charge on the two arginines is comple(cid:173)
`mented by the negative charge of Glu-35 and Glu-50 from the
`heavy chain of the antibody. The backbone structure of the
`antigen, lysozyme, is mostly unperturbed, although there are
`some changes in the epitope region, most notably Pro-70. One
`side chain not in the epitope, Trp-63, undergoes a rotation
`oC"=180° about the CfJ-CY bond. The Fab elbow bends in the
`two crystal forms differ by 7°.
`
`Until recently knowledge of the structural aspects of anti(cid:173)
`body-antigen interactions has been based on the x-ray
`analysis of four Fab structures and on some complexes with
`hapten (1-5). Haptens were observed to bind in grooves or
`pockets in the combining sites of the New and McPC603
`Fabs, and these occupied a small fraction of the total
`available area of these sites . When haptens bind to these
`Fabs, no large conformational change occurs. However, one
`cannot rule out the possibility that the behavior of antibodies
`would be different when they are bound to larger antigens,
`such as proteins . For example, the interaction with a much
`greater fraction of the combining site might in itself be
`sufficient to induce conformational changes in the antibody.
`Also, the interacting surfaces might not possess the grooves
`and pockets observed for haptens, but might resemble more
`closely the kind of surface observed in other protein-protein
`interfaces, where exclusion of bound water is believed to play
`a key role. For this reason we undertook several years ago to
`investigate the crystal structures of complexes of the Fabs of
`several monoclonal antibodies to hen egg white lysozyme
`complexed with the lysozyme (6). In this paper we report the
`analysis of two different crystal forms of one of these
`complexes.
`The site on the lysozyme to which the antibody binds has
`been the subject of an extensive serological analysis (7)
`through a study of cross-reactivity with different avian
`lysozymes. The results of that analysis are in striking agree(cid:173)
`ment with the crystal structure observations and will be
`discussed.
`
`The publication costs of this article were defrayed in part by page charge
`payment. This article must therefore be hereby marked "advertisement"
`in accordance with 18 U.S. C. §1734 solely to indicate thi s fact.
`
`8075
`
`During the course of this analysis two reports of related
`x-ray studies of Fab--antigen complexes have appeared (8, 9) ,
`one being a description of another lysozyme-antilysozyme
`complex , although to a different epitope of the lysozyme, and
`the other describing a complex with the neuraminidase of
`influenza virus. The observations and conclusions from these
`two investigations differ in important ways from one another,
`and we describe below how our results can be related to
`them.
`
`MATERIALS AND METHODS
`Monoclonal antibody (mAb) HyHEL-5 and the Fab--lyso(cid:173)
`zyme complex were prepared as described previously (6, 7).
`Crystals were grown by vapor diffusion against 20% (wtj vol)
`polyethylene glycol 3400 (Aldrich) in 0.1 M imidazole hydro(cid:173)
`chloride, pH 7.0, 10 mM spermine with an initial protein
`concentration of7 mg/ ml. The crystals grow polymorphically
`in space group P21, differing principally in the length of the
`b axis , which was observed to vary between crystals in an
`unpredictable manner between 65 A and 75 A.
`One set of data on a crystal with cell dimensions of a = 54.9
`A, b = 65.2 A, c = 78.6 A, and f3 = 102.4° was collected at
`Genex (Gaithersburg, MD) using a single detector-single axis
`Nicolet-Xentronics (Madison, WI) area detector; 20,074
`observations yielded 7565 unique reflections. Lorentz, po(cid:173)
`larization , and absorption corrections were applied (10) , and
`the different frames were scaled together giving an overall
`merging Rsym of 0.044, where Rsym = Ihk/I;I7 - 1;!/Ihk/7,,.
`Greater than 90% of the theoretical data were observed to 4.5 A
`spacings, greater than 60% from 4.5 A to 4.0 A spacings , and
`about 40% from 4.0 A to 3.0 A spacings.
`A second data set was collected on a crystal with cell
`dimensions of a = 54.8 A, b = 74.8 A, c = 79.0 A, and f3 =
`101.8° at the University of California, San Diego, using the
`Mark II multiwire detector system with two detectors (11) ;
`38 ,689 reflections were collected from one crystal , of which
`15 ,673 were unique. Lorentz, polarization, and absorption
`corrections were applied, and the different frames were
`scaled together giving an overall merging R sym of 0.044. Of
`the 15,673 unique reflections, 15,166 are within 2.66-A
`resolution (86.3% of the theoretically observable) ; there are
`an additional 507 reflections between 2.66 and 2.54 A reso(cid:173)
`lution (18.7% of the theoretically observable).
`The structure was determined by molecular replacement
`using the program package assembled by Fitzgerald (12).
`Three probes were used : (i) tetragonal lysozyme (2L YZ)
`deposited by R. Diamond in the Protein Data Bank (13); (ii)
`CL + CHI of the McPC603 Fab (4); and (iii) V L + V H of the
`
`Abbreviations : CDR, complementarity-determining region; CDRs 1,
`2, and 3 for the light chain are referred to as Ll , L2, a nd L3 , and for
`the heavy chain a s H1, H2, and H3; mAb , monoclonal antibody ; C,
`constant region ; V , variable region .
`§Present address: Central Research and Development , E . I. Dupont
`Experimental Station ES 228/ 3168 , Wilmington , DE 19898.
`
`PFIZER EX. 1581
`Page 3
`
`
`
`8076
`
`Immunology: Sheriff e t a/.
`
`Proc. Nat/. A cad. Sci. USA 84 ( 1987)
`
`McPC603 Fab with the following residues removed from the
`model: VL-27C-31 and 91-95 ; and VH-30-31, 52B-54,
`61-64, 96-1001 , and 101. Residue numbering in Fabs through(cid:173)
`out this paper follows Kabat e t a/. (14); C and V represent
`constant and variable region s, respectively, and L and H
`refer to light and heavy chain, respectively . We oriented
`McPC603 Fab so that the axis of the elbow was parallel to the
`z axi s, which allowed us to observe most of the difference in
`elbow bend directly in the y rotation function ang,le (15) . The
`fast-rotation function (16) was used with 10- to 4-A resolution
`data and a radiu s of integration of 24 A. The rotation function
`of Lattman and Love (17) was used to refine the position of
`the peak for each probe. The Crowther-Biow (18) translation
`function was then used with 10- to 4-A resolution data and a
`step size of 0.02 unit cell lengths in a and c to determine the
`x ,z translations. We also used the program BRUTE, written
`by M. Fujinaga and R. Read , University of Alberta, with 5-
`to 4-A resolution data and 1-A step size to determine x ,z
`translations. Where the two methods agreed , we used
`BRUTE to hold one or two probes stationary and search for
`the translation of the second or third probe to solve the
`problem of relative origins in space group P2 1. The resulting
`model from this analysis was examined with the program
`FRO DO (19) on an Evans and Sutherland (Salt Lake City,
`Utah) PS300 picture system to ensure that the three probes
`were assembled in a plausible manner and that the crystal
`contacts were reasonable.
`The positions of the Fab and lysozyme were refined using
`the stereochemically restrained least-squares refinement
`package PRECOR/ CORELS (20). We first refined with three
`"domains ," lysozyme, VL + VH , and CL + CH1 , starting
`with 10- to 8-A resolution data and then extending the
`refinement to 7-A spacings and finally to 6-A spacings. The
`crystallographic R = ~hklllFol - IFcll/~llk/I Fol for the long
`b-axis form was 0.44 and for the short b-axis form was 0.49.
`We then divided the complex into five domains: lysozyme,
`VL, VH , CL, and CHl. At this point we replaced the VL
`domain of McPC603 with that of antibody 1539 (5), because
`it shares 75% sequence identity with the V L domain of mAb
`HyHEL-5 (21) . We also replaced the CH1 domain of
`McPC603, which is a murine IgA, with the CH1 domain of
`antibody KOL (3 , 13), which is a human IgG1 and shares 60%
`sequence identity with the murine IgG1 CH1 domain of mAb
`HyHEL-5 . At this stage of refinement for both crystal forms
`the R was 0.42 for 10- to 6-A resolution data.
`At this point we concentrated on the long b-axis form
`because the data set extended to higher resolution. We
`removed from the model the side chains of amino acids that
`were not identical to the mAb HyHEL-5 sequence (ref. 21
`and A. B. Hartmann, C. P. Mallett, and S.J.S.-G. , unpub(cid:173)
`lished work) and also omitted entire residues of the following
`complementarity-determining regions (CDRs) : L3 (residues
`91-95) , H2 (residues 52B-54) , and H3 (residues 97-100); (see
`abbreviations footnote). We refined the model against data
`from 10- to 2.5-A resolution using the stereochemically
`restrained least-squares refinement package PROTIN / PROL(cid:173)
`SQ (22) . We first refined only the positional parameters using
`an overall isotropic B until the R = 0.344. We then used
`FRODO to rebuild the model. In the electron-density map at
`this stage there was excellent density for the deleted L3 and
`moderately good density for the deleted H2 and H3. Also, in
`most cases , electron density was apparent for side chains that
`had been omitted from the model. Following this , we refined
`the model adding individual isotropic B factors until the R =
`0.270, and then we examined the model with molecular
`graphics and rebuilt parts of the model , especially H3. We
`once again refined until the R = 0.249. At this stage we
`included an overall ani sotropic D.B (23) , which had values of
`Mu = - 5.89, M 13 = 2.15 , M 22 = 8.44, and M 33 = -2.55
`and which correlates with the variable b-axis length , suggest-
`
`ing the lattice contacts along the b axis are weak. We then
`omitted all of the interacting residues in lysozyme , refi ned the
`model, calculated an electron-density map, and examined it
`to determine whether our interpretation of the position of
`these residues was correct. We did the same thing with the
`interacting residues on the Fab . After making minor changes
`to the structure we refined again, yielding R = 0. 245 with an
`rms deviation from ideal bond lengths of 0.012 A. We have
`deposited the coordinates from this stage of the refinement in
`the Protein Data Bank (13).
`We have estimated the rms positional error to be 0.40 A by
`the method of Luzzati (24) . In describing the results, hydro(cid:173)
`gen bonds and salt links were limited to pairs of appropriate
`atoms with an interatomic distance of < 3.4 A. Maximum van
`der Waals contact distances were defined as in Sheriff eta/.
`(25). Contacting surface area was c~lculated with program
`MS (26) using a probe radius of 1.7 A and standard van der
`Waals radii (27) .
`
`RESULTS
`Fig. 1a shows the C" skeleton of the HyHEL-5 Fab(cid:173)
`lysozyme complex. V H and V L• and CH1 and CL adopt the
`canonical relationships observed in other Fabs. The main
`difference between the two crystal forms is the elbow bend
`of the HyHEL-5 Fab , which is 161° in the long b-axis form
`and 154° in the short b-axis form.
`The contact between the antibody-combining site and the
`lysozyme epitope is extensive and involves many residues.
`The calculated buried surface (solvent inaccessible) area is
`about 750 A2 on the surface of both the Fab and lysozyme
`(= 14% of the surface). The interaction between the two
`proteins is very tight, and there are no water molecules
`between the combining site and the epitope. The current
`model contains three salt links and ten hydrogen bonds.
`Glu-50 (H2) forms salt links to both Arg-45 and Arg-68 of
`lysozyme , and Glu-35 (H1) forms a salt link to Arg-68. There
`are 74 van der Waals contacts .
`The epitope on lysozyme consists of three oligopeptide
`segments (Fig. 1b). The first consists of Gln-41, Thr-43 ,
`Asn-44, Arg-45, Asn-46, Thr-47, Asp-48 , Gly-49, and Tyr-53,
`which are in contact, and Thr-51 and Asp-52 , which are partly
`buried by the interaction . This segment is essentially one long
`continuous subsite, which consists of two {3-strands connect(cid:173)
`ed by a bend involving residues 46 through 49 (28). The
`second segment consists of the contacting residues Gly-67,
`Arg-68 , Thr-69, and Pro-70; and the partly buried residues
`Asn-65 , Asp-66, Gly-71 , and Ser-72. This second segment has
`the form of a rambling loop and contains part of an ex.te.n(cid:173)
`sively studied , disulfide-linked antigenic peptide that ehc1ts
`antibodies that could cross-react with native lysozyme (29).
`Arg-68 in this subsite has been identified as a "critical"
`residue to the epitope (7) . The third segment consists of the
`directly interacting Leu-84 and the partly buried residues
`Pro-79, Ser-81 , and Ser-85. Arg-61 , which is in none of.the
`segments , is also partly buried upon complex formatwn.
`The surface of the epitope is extensive (23 A between the
`most distant C" atoms) and relatively flat except for a
`protruding ridge made up of the side chains of Arg-45 and
`Arg-68, and curling back at the edges (Fig. 1d).
`.
`The antibody-combining site involves residues from all SIX
`CDRs (30). Each of these CDRs contributes at least one
`residue to the interaction with lysozyme , and most contribut~
`several residues (Fig. 1c). Trp-47, which is considered part 0
`the heavy chain framework, also interacts with lysozym~·
`The other interacting residues are Asn-31 , Tyr-32, Asp-4 d
`Trp-91 , Gly-92 , Arg-93 , and Pro-95 from the light cham; a;
`8
`Trp-33 , Glu-35 , Glu-50, Ser-54, Ser-56, Thr-57, As.n-
`'
`Gly-95 , and Tyr-97 from the heavy chain . Additional res1due~
`that are at least partly buried by the interaction but do n(?
`directly contact are Ser-29, Val-30, Tyr-34, and Arg-46 10
`
`PFIZER EX. 1581
`Page 4
`
`
`
`Immunology: Sheriff et a/.
`
`Proc. Nat/. A cad. Sci. USA 84 ( 1987)
`
`8077
`
`(a) Stereo diagram of ca trace
`FIG. 1.
`of lysozyme-HyHEL-5 complex . Lyso(cid:173)
`zyme (blue), lysozyme epitope (red), light
`chain (yellow), heavy chain (green), and
`CDRs (magenta) . (b) Stereo diagram of
`lysozyme highlighting epitope. Residues
`not in contact with mAb HyHEL-5 have
`only backbone atoms shown (blue) . Resi(cid:173)
`dues in epitope (red) and residues in the
`first subsite not directly in contact with the
`Fab (yellow). The first subsite (residues 41
`through 53) is toward the bottom of the
`figure , starting at Gln-41 at the lower left
`and going across to Thr-47 at lower right
`and then looping back from Asp-48 to
`Tyr-53; the second subsite (residues 67
`through 70) is at the upper right ; and the
`third subsite (lle-84) is toward the upper
`left. The side chains of Arg-45 and Arg-68
`are more or less vertical and just to the
`right of center. (c) Stereo diagram of mAb
`HyHEL-5 Fab highlighting CDRs and con(cid:173)
`tacting residues . Framework residue s
`(blue) and CDR, but not in contact (yel(cid:173)
`low) , have only backbone atoms shown.
`Contacting residues (red) are shown in
`toto . The light chain is on the left , and the
`heavy chain on the right. L1 , lower left ;
`L2, upper left ; L3 , center bottom; H1 ,
`upper right; H2, lower right ; and H3 , cen(cid:173)
`ter top. Trp-47 from the heavy chain frame(cid:173)
`work is just to the right of L3, and Glu-35
`(H1) and Glu-50 (H2) are directly above
`Trp-47 . (d) Stereo diagram of mAb Hy(cid:173)
`HEL-5 with complementary surface of ly(cid:173)
`sozyme epitope superimposed. Residues
`not in contact with lysozyme (blue). Resi(cid:173)
`dues in contact with lysozyme (red). Ly(cid:173)
`sozyme buried surface (yellow dots) . Ori(cid:173)
`entation is identical to Fig. 1c. Bright areas
`around edges illustrate curling of surface.
`(e) Stereo diagram of superposition of
`backbone atoms of lysozyme in tetragonal
`crystal form on lysozyme in complex with
`mAb HyHEL-5 . Tetragonal lysozyme
`(blue). Lysozyme in complex with mAb
`HyHEL-5 (red). White results from exact
`superposition of red and blue. Epitope is at
`the right. Pro-70 at upper right can be seen
`to differ in the two structures. Loop at
`Thr-47 and Asp-48 also shows differences
`in backbone at lower right.
`
`1ra1ne,.vm·k region 2) from the light chain and Ser-30, Asp-31,
`-32, Leu-52, Gly-55, Asn-96, and Asp-98 from the heavy
`chain. Only 30% of the residues in the CDRs are actually in
`COntact with lysozyme . If one adds the residues that are at
`least partly buried , this fraction rises to 48%. The surface of
`interaction is quite broad and extensive with the ca atoms of
`interacting combining site residues separated by as much as
`28 A. There is a groove on the surface ofthe antibody running
`from L3 to H3 and between Trp-91 of L3 and Trp-33 of Hl
`(Fig. 1d). Arg-45 and Arg-68 of lysozyme fit into this groove,
`Placing them in position to form salt links to Glu-50 (H2) and
`Glu-35 (Hl).
`We examined the lysozyme and Fab structures for indica(cid:173)
`tion at this degree of refinement of any significant confor(cid:173)
`mational change that may have occurred as a result of their
`
`assoc.at1on. The structure of hen egg lysozyme has been
`determined in four crystal forms, thus providing a database
`for comparing the structure of lysozyme in different envi(cid:173)
`ronments. The tetragonal lysozyme coordinates used in this
`comparison (D. C. Phillips, personal communication) dif(cid:173)
`fered slightly from those used for structure determination
`(rms difference = 0.32 A for 516 main-chain atoms). We
`superimposed the tetragonal lysozyme coordinates onto the
`lysozyme in our complex, using all main-chain atoms, both
`including and excluding atoms from residues that are in(cid:173)
`volved in interactions with the antibody-combining site on
`the Fab. Qualitatively we see the same behavior with both
`procedures, and we report numbers calculated when residues
`were excluded (Fig. le). For these lysozymes , the backbone
`structures are nearly identical (rms difference = 0.48 A for
`
`PFIZER EX. 1581
`Page 5
`
`
`
`8078
`
`Immunology: Sheriff et a/.
`
`Proc. Nat/. A cad. S ci. USA 84 ( 1987)
`
`516 atoms) . However, we do see some c~anges in the region
`of the epitope (rms difference = 0.64 A for 96 atoms). In
`particular, the C" ofPro-70 has moved about 1. 7 A, which can
`be accounted for by a hydrogen bond of the carbonyl oxygen
`and the hydroxyl of Tyr-97 (H3).
`There are also some changes in the side-chain atoms (rms
`difference = 1.22 A for 429 atoms). The side chains of
`residues not involved in the epitope (excluding Trp-63) show
`an rms difference of 1.12 A for 420 atoms. Comparing side
`chains of residues involved in the epitope gives an rms
`difference of 1.20 A for 55 atoms. The largest side-chain
`difference is Trp-63, which is not in the epitope and which
`appears to have flipped 180° around the Cf3-c-r bond.
`Although the current electron-density maps strongly favor
`movement, there is some ambiguity because there is a tail of
`electron density that points , more or less, in the direction of
`the original side-chain conformation. Preliminary refinement
`of the short b-axis form strongly suggests the movement of
`the Trp-63 side chain. Many of the other large differences are
`in residues that are part of the epitope. In particular, the side
`chains of Arg-45 and Arg-68 in the tetragonal crystal form, the
`triclinic crystal form, and the complex with the antibody are
`within hydrogen-bonding distance of one another, but in none
`of the three are the side chains in the same position relative
`to the backbone.
`To ascertain whether any changes have taken place in the
`Fab in the complex, we need crystals of the uncomplexed
`Fab , but so far we have been unable to grow such crystals
`large enough for diffraction studies. However, the fact that
`the molecular replacement technique worked so well shows
`that the positions of CL relative to CHI and of V L relative to
`V H are basically unchanged from the "canonical" structures
`found in Fab fragments from myeloma proteins. This result
`differs from the result of Colman eta/. (9), who report that V L
`and V H form a non-canonical pairing in their antineuramin(cid:173)
`idase-neuraminidase complex.
`
`DISCUSSION
`
`Our results confirm the identification of the epitope on
`lysozyme for HyHEL-5 based on serological studies (7). In
`those studies , epitope mapping was accomplished by ana(cid:173)
`lyzing the cross-reactivity of homologous proteins in the
`antibody-antigen reaction. The limited sequence variation
`among these cross-reacting proteins was utilized to pinpoint
`the residues that were important in the particular binding
`interaction . The present crystallographic results provide
`strong evidence for the validity of epitope mapping by
`serological techniques .
`The HyHEL-5-Iysozyme complex described here and the
`two other antibody-antigen complexes whose structures have
`been reported share some common features. For example, in
`all three cases , the interaction involves all of the CDRs of the
`antibody. The surface area of the combining site of the
`antibody that is buried upon complexing with antibody is 750
`A2 for HyHEL-5 , 690 A2 for the antilysozyme Dl.3 Fab (8),
`and not specified for the antineuraminidase NC41 Fab (9),
`where the third CDRs of both light and heavy chains have not
`yet been completely modeled.
`Although the areas of interaction are comparable for the
`HyHEL-5 and Dl.3 antibody, the association constant is
`greater for HyHEL-5 than for Dl.3 antibody [2.5 x 109
`(M. E. Denton and H . A. Scheraga, personal communica(cid:173)
`tion) vs. 4.5 x 107 (8)] . This indicates that the area of
`interaction alone cannot be the total determinant of the
`attraction between the antibody and antigen ; undoubtedly ,
`other factors such as electrostatic and hydrogen bonding play
`a role. In this connection, the favorable electrostatic inter(cid:173)
`actions involving Arg-45 and Arg-68 of lysozyme and Glu-35
`(Hl) and Glu-50 (H2) of the Fab, and the shape complemen-
`
`tarity between the Arg-45 , Arg-68 side-chain ridge on lyso(cid:173)
`zyme and the groove on the Fab are relevant. In contrast
`there are no electrostatic interactions between Dl. 3 antibody
`and lysozyme (8). We note that when Arg-68 becomes a
`lysine, as it does in bobwhite quail , the HyHEL-5 antibody
`binds with less avidity by a factor of 103 (7) , so that shape and
`hydrogen-bonding capacity of the charged group are also
`crucial. In antibody Dl.3 , H3 plays a dominant role in the
`antibody-antigen interaction (8). However, in HyHEL-5 , H2
`and L3 play the dominant role contributing six and four
`residues , respectively, and approximately equal surface ar(cid:173)
`eas to the interaction. CDR 1 of the light chain (Ll), Hl, and
`H3 play a lesser role by each contributing two residues and
`only 50-60% of the surface area of H2 and L3.
`The lysozyme epitope for HyHEL-5 consists principally of
`two continuous segments of polypeptide chain, residues
`41-53 and 65-72. The lysozyme epitope for antibody Dl.3
`again involves two continuous segments of the protein,
`residues 18-27 and 116-129 (8). This similarity is remarkable,
`but it is probably coincidental because the epitope for NC41
`Fab on the neuraminidase involves four segments , 368-370,
`400-403, 430-434, and parts of 325-350 (9). It is interesting
`that the two continuous segments in the lysozyme epitope for
`mAb HyHEL-5 have been reported to have high mobility in
`several lysozyme structures (31).
`Conformational changes in the Fab that result from antigen
`binding cannot be quantitated because in none of the three
`cases so far reported has the structure of the uncomplexed
`Fab been analyzed. It should be noted that antibodies
`frequently have large insertions into the CDRs, and these
`sometimes project into the solvent and cannot be localized
`with certainty by x-ray diffraction. L1 in McPC603 Fab is an
`example of this (4). In any interaction involving McPC603
`Fab with a large antigen , it is likely that such a loop would
`move. However, Colman eta/. (9) have reported a difference
`in the relationship of V L to V H in NC41 Fab. We have
`duplicated the calculations in Table 1 of Colman et a/. (9) with
`results that are essentially in agreement with theirs and
`extended the table to include Fabs 1539 and HyHEL-5. We
`find that the relative disposition of the two variable domains
`in the NC41 Fab does lie at the extreme of the values
`observed. Whether this is the result of binding to antigen or
`the result of the interaction of hypervariable residues in the
`interface cannot be determined at present. We should point
`out, however, that the NC41 Fab--neuraminidase complex
`has been elucidated to only 3.0-A resolution and has been
`subjected to only preliminary refinement (R = 0. 35). These
`conclusions need to be confirmed by further refinement.
`Amit eta/. (8) have observed that no large conformational
`changes have occurred in lysozyme upon binding to the D1.3
`Fab. Over most of the molecule we find essentially the same
`results , although some changes do occur in both the back(cid:173)
`bone and the side chains. In particular, the flipp ing of the
`Trp-63 side chain, if confirmed by further refinement, to(cid:173)
`gether with movement of Pro-70 are the most notable of these
`changes . Colman eta/. (9) have also reported a few significant
`differences between the structure of the neuraminidase alone
`and that in the complex with NC41 Fab, although here, to~,
`we must await confirmation from further refinement. It 15
`apparent that some deformation of the antigen can occu;,
`especially when the epitope is in a flexible part of t e
`structure, as in HyHEL-5-Iysozyme complex. Indeed, flex(cid:173)
`
`ibility has been implicated in the antigenicity of prote_ins (3;,
`
`33) and could aid in the binding of antibody to ant1gen Y
`allowing the latter to complement more closely the structure
`of the antibody-combining site.
`d
`It has been hypothesized that changes in the elbow ben f
`may signal antigen binding (34). The different crystal for~s ~e
`the HyHEL-5-lysozyme complex are the first example o t
`same antibody-antigen complex showing different elboW-
`
`PFIZER EX. 1581
`Page 6
`
`
`
`Immunology: Sheriff et a/.
`
`Proc. Nat/. A cad. Sci. USA 84 ( 1987)
`
`8079
`
`bend angles . The values observed here lie in the middle of the
`spectrum of observed elbow bends ranging from 133o to
`zl80° (35). There appears to be no correlation between
`observed values of elbow bends for different Fabs and
`binding to hapten or antigen. It is therefore likel y that this
`range of values is simply an indication of flexibility of this part
`of the antibody molecule .
`In conclusion , the results observed for the binding of
`antibodies to protein antigens have many features in common
`with the binding of haptens . For example, it is clear that
`charge neutralization in the interface plays an important role .
`The principal difference in the case of the larger antigens is
`the much greater area of the complementary surfaces that are
`brought into contact with consequent exclusion of water
`