PROGRESS IN
`
`Nucleic Acid Research
`and Molecular Biology
`
`edited by
`
`KlVlE MOLDAVE
`
`Department ofMolecular Biology and Biochemistry
`University of California, Irvine
`Irvine, California
`
`Volume 60
`
`/HD ACADEMIC PRESS
`
`San Diego London Boston New York
`Sydney Tokyo Toronto
`
`IMMUNOGEN 2167, pg. 1
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2167, pg. 1
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

` hat; 1’ dawn-301“”?! 1.;hrmffi
`
`,
`.mffiadtamz
`
`.241) M..
`..-:.CC."Z;.MVQ
`Madison, Wl araaaaaaa
`
`This book is printed on acid—free paper.
`
`Copyright © 1998 by ACADEMIC PRESS
`
`All Rights Reserved.
`No part of this publication may be reproduced or transmitted in any form or by any
`means, electronic or mechanical, including photocopy, recording, or any information
`storage and retrieval system, without permission in writing from the Publisher.
`The appearance of the code at the bottom of the first page of a chapter in this book
`indicates the Publisher’s consent that copies of the chapter may be made for
`personal or internal use of specific clients. This consent is given on the condition,
`however, that the copier pay the stated per copy fee through the Copyright Clearance
`Center, Inc. (222 Rosewood Drive, Danvers, Massachusetts 01923), for copying
`beyond that permitted by Sections 107 or 108 of the US. Copyright Law. This consent
`does not extend to other kinds of copying, such as copying for general distribution, for
`advertising or promotional purposes, for creating new collective works, or for resale.
`Copy fees for pie-1998 chapters are as shown on the title pages. If no fee code
`appears on the title page, the copy fee is the same as for current chapters.
`0079-6603/98 $25.00
`
`Academic Press
`a division of Harcourt Brace & Company
`525 B Street, Suite 1900, San Diego, California 92101-4495, USA
`http://www.apnet.com
`
`Academic Press Limited
`24-28 Oval Road, London NW1 7DX, UK
`http://www.hbuk.co.uk/ap/
`
`International Standard Book Number: 0-12—540060—8
`
`PRINTED lNTHEUNITED STATES OFAMERICA
`989900010203BB987654321
`
`IMMUNOGEN 2167, pg. 2
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2167, pg. 2
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Development of More
`Efficocious Antibodies
`
`for Medical Therapy
`and Diagnosis
`
`AMEURFINA D. SANTOSl AND
`EDUARDO A. PADLAN
`
`Laboratory of Molecular Biology
`National Institute ofDiabetes and
`Digestive and Kidney Diseases
`National Institutes of Health
`Bethesda, Maryland 20892
`
`I. Introduction ...................................................
`
`11. Procedure for Reducing the Immunogenicity of a Nonhuman Antibody
`to a Minimum .................................................
`
`A. Sample Design of a “Humanization” Protocol by SDR Transfer .....
`III. Generation of Multivalent/Multispecific Antibodies ..................
`A. Sample Construction of a Bispecific Molecule Using Two Single-Chain
`Fvs Dimerized via Electrostatically Complementary Fc Regions .
`.
`.
`.
`B. Sample Construction of a Tetraspecific Molecule from Two Bispecific
`Single-Chain Diabodies Dimerized via Electl'ostatically
`Complementaly Fc Regions ...................................
`IV. Conclusion ....................................................
`References ....................................................
`
`169
`
`172
`
`182
`183
`
`188
`
`190
`190
`192
`
`Two procedures for improving the efficacy of medically important antibodies
`are described. The first procedure is designed to reduce the immunogenicity of
`nonhuman antibodies to the barest minimum—the “humanization” is accom-
`
`plished by transplanting only the specificity-determining residues of the nonhu-
`man antibody onto a human antibody template. The second procedure is designed
`to permit the easy production of multispecific/multivalent antibodies via heterodi-
`mer formation of electrostatically complementary Fc regions.
`© 1998 Academic Press
`
`I. Introduction
`
`Antibodies represent a major factor in our defense against invading
`pathogens and noxious substances. Antibodies are generated to bind specif—
`ically to the foreign substance (antigen) and to neutralize it and facilitate its
`
`1Permanent address: National Institute of Chemistry, University of the Philippines, Diliman,
`Quezon City 1101, Philippines.
`
`Progress in Nucleic Acid Research
`and Molecular Biology, Vol. 60
`
`l 69
`
`0079-6603/98 $25.00
`
`IMMUNOGEN 2167, pg. 3
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2167, pg. 3
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`170
`
`AMEURFINA D. SANTOS AND EDUARDO A. PADLAN
`
`elimination by normal biological processes. Antibodies are multivalent (at
`least bivalent) molecules, thus they can cross-link antigens, thereby immobi-
`lizing them. The binding of antibodies to antigens also may cause the re-
`cruitment of other molecules or of certain cells—additional components of
`the immune system—which would then act to dispose of the invading sub
`stance or organism.
`All antibodies share the same basic structural unit that consists of two
`
`identical heavy chains (MW of each, ~50,000 to ~77,000) and two identi—
`cal light chains (MW of each, ~25,000). Each light chain is usually linked to
`a heavy chain by a disulfide bond and the heavy chains are usually linked to-
`gether by one or more disulfide bonds. Each chain has variable and constant
`domains. The N—terminal domain of both light and heavy chains is variable
`and is followed by one constant domain in the light chain (CL) and by three
`or four constant domains in the heavy chain (CHI, CHZ, CH3, and CH4) de—
`pending on antibody class.
`The antibody class, or isotype, is determined by the constant domains.
`The light chain exists in two distinct isotypes called kappa (K) and lambda (A).
`The heavy chain may be (X, y, B, e, or M type, which defines the antibody class
`as IgA, IgG, IgD, IgE, or IgM, respectively. IgG is the major antibody class in
`human serum; IgE is the antibody responsible for allergic reactions. The con—
`stant domain of the heavy chain determines the effector function(s) of an an-
`tibody, e.g., complement activation, and F0 receptor binding. Different class—
`es have different biological properties.
`The variable region contains the antigen-binding site. Each variable do—
`main consists of three hypervariable segments, called the complementarity-
`determining regions (CDRs) (l), flanked by four relatively less variable
`framework regions. The antigen-binding site is built mainly with CDR
`residues, with occasional contribution from neighboring framework
`residues. The lengths and sequences of the CDRs vary from antibody to an—
`tibody, resulting in different antigen—binding specifities. The VLzVH module
`is often referred to as the FV fragment and the VLCL:VHCHl as the Fab frag—
`ment. The CHZ and CH?) domains of the two heavy chains, plus the CH4 in
`the case of IgE and IgM, constitute the Fc fragment.
`The exquisite specificity of the binding of an antibody to its antigen and
`the ability of the immune system to respond to challenge by all sorts of anti—
`gens have found many uses in medical therapy and diagnosis. The use of an-
`tivenom against snake bites, antitoxins against bacterial infections, immune
`serum globulin against certain diseases, and so forth, are some of the well—
`known uses of specific antisera (see Ref. 2 for a review). Among the more
`recent uses of antibodies in medicine is the specific targetting of cells or
`tissues, e.g., tumor cells, either for location (in viva imaging) or for destruc—
`tion (3).
`
`IMMUNOGEN 2167, pg. 4
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2167, pg. 4
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`MORE EFFICACIOUS ANTIBODIES FOR MEDICINE
`
`171
`
`With the advent of hybridoma technology (4), monoclonal antibodies of
`virtually any desired specificity can be produced. Further, the development
`of novel expression systems has permitted the generation of pure antibodies
`in large amounts. For various reasons, including ethical considerations,
`monoclonal antibodies are usually obtained from nonhuman sources. Un—
`fortunately, the human immune system will react to, and attempt to elimi—
`nate, any nonhuman (or nonself) entity. This necessitates the “humaniza—
`tion,” i.e., the reduction of the immunogenicity, of the nonhuman antibodies
`prior to their use in human patients, especially if the treatment protocol re-
`quires protracted use of such molecules. Various procedures have been de—
`vised to humanize nonhuman antibodies (5—11). Here, we present a new pro-
`cedure, currently being developed in our laboratory, that seeks to reduce
`immunogenicity t0 the barest minimum while at the same time preserving
`the antigen—binding properties of the original antibody.
`Another topic of interest is the generation of multispecific/multivalent an—
`tibodies. A molecule that can bind different ligands has many potential uses.
`For example, a molecule of a desired reactivity can be brought to close prox—
`imity to a target cell by using a bispecific antibody that can bind the mole-
`cule via one site and an antigen on the surface of the cell via the other. Like-
`wise, two different cells can be brought together with the use of a bispecific
`antibody. Further, there are instances when an antigen has only one site (epi—
`tope) to which a given antibody type can bind. Such an antigen cannot be
`cross—linked by antibodies, the binding sites of which have the same binding
`properties; at least one other antibody type, with a different and nonover—
`lapping specificity, will be required to cross-link the antigen. In nature, sev—
`eral different antibody types are elicited by a single antigen (in a normal poly—
`clonal response). If an antibody could be engineered so that its binding sites
`have different specificities, such an antibody could by itselfproduce the same
`effect as two, or more, different antibodies. Various techniques are currently
`in use for the generation of bispecific antibodies (12—2 6).
`Multivalency amplifies the affinity of antibodies for their specific anti—
`gens. Usually, antibody—antigen reactions are characterized by nanomolar
`(or better) affinity constants, when the antigen is a protein. With carbohy-
`drate antigens, the binding constants are often much lower. The binding of
`an antibody to its antigen can be improved by increasing the numb er of com—
`bining sites on the antibody (the antigen—binding sites) so that, even with a
`low intrinsic binding affinity per site, the avidity due to the presence of mul-
`tiple sites can be substantial. This is observed in nature in antibodies of the
`IgM class, which can have as many as 12 identical antigen-binding sites.
`Judicious engineering can generate multispecific and multivalent anti—
`bodies. Here, we present a procedure that we are developing for the easy gen—
`eration of multispecific/multivalent molecules.
`
`IMMUNOGEN 2167, pg. 5
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2167, pg. 5
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`172
`
`AMEURFINA D. SANTOS AND EDUARDO A. PADLAN
`
`II. Procedure for Reducing the Immunogenicity
`of a Nonhuman Antibody to a Minimum
`
`The goal of humanization is to produce a molecule that has the same anti-
`gen—binding properties as the original antibody, but that is nonimmunogenic
`in humans. This can be accomplished by transplanting the structures that de—
`termine binding properties from the nonhuman antibody to a human scaf-
`fold. The feasibility of transplanting a combining site from one antibody to
`another was recognized (2 7) as soon as the first antibody structures had been
`determined.
`
`Various physical and chemical techniques have shown that antigen bind—
`ing occurs via the variable domains of the light and heavy chains of the anti—
`body. Thus, the first attempt at reducing the immunogenicity of a nonhuman
`antibody was accomplished by producing a chimera, the variable domains of
`which were from the nonhuman molecule whereas the constant domains
`
`were from a human antibody (28, 29). Such chimerae do possess the same
`antigen—binding properties as the original molecules, but the nonhuman vari—
`able domains usually elicit adverse immune reactions by the host. A major
`improvement in reducing the immunogenicity of chimeric molecules is
`achieved by transplanting only the CDRs (5).
`Three—dimensional structures obtained by X~ray crystallography are now
`available for the antigen-binding regions of many different antibodies from
`different species and with a variety of ligand—binding specificities, many in
`complex with specific ligand (30). These structures allow us to make some
`generalizations: (1) VH and also VL domains have very similar three—dimen—
`sional structures regardless of species origin, (2) the framework regions are
`essentially superposable so that the variable domains might differ in struc—
`ture only in their CDRS, (3) CDRs, which have the same number of residues
`and possess certain critical residues, usually have the same loop conforma—
`tions (31), and (4) the combining site of an antibody is mainly formed by CDR
`residues, with the occasional involvement of one or two framework residues.
`These results strongly suggest that transplantation of only the CDRs and
`possibly of some framework residues also can achieve a successful transfer
`of antigen—binding properties. The implications for humanization by CDR
`grafting (5) are obvious, especially because the CDRs are frequently shared
`among antibodies from different species (2 7, 32) and, thus, can be expected
`to be the least immunogenic of the various parts of an antibody molecule.
`Indeed, the successful humanization of numerous nonhuman (usually ro-
`dent) antibodies has been accomplished by GDR grafting, although the reten—
`tion of some nonhuman framework residues also is often required (6—9, 11);
`those framework residues presumably influence the structure of the CDR
`loops.
`_
`As more structural data become available, however, it is becoming obvi-
`
`IMMUNOGEN 2167, pg. 6
`Phigenix v. Immunogen
`|PR2014—00676
`
`
`
`IMMUNOGEN 2167, pg. 6
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`MORE EFFICACIOUS ANTIBODIES FOR MEDICINE
`
`173
`
`ous that not all of the CDR residues are involved in the interaction with the
`
`ligand. The antibody residues that have been observed to be in actual con—
`tact with antigen, the specificity-determining residues (SDRs), are presented
`in Table I. It is found that, even‘ in the most extensive interaction with anti—
`gen [in the case of antibody NC4 1 binding to influenza virus neuraminidase
`(33)], only 22 residues from the antibody participate in the binding. The
`SDRS, therefore, represent only a small fraction of the (IDES, which could
`be composed of anywhere from 46 to 100 amino acids [from the authors’ sur—
`vey of human antibody sequences (data not shown)]. The immunogenicity of
`a humanized antibody can be further reduced by transplanting only those
`parts of the CDRS that contain the SDRs, and those “abbreviated” CDRs
`have been defined (34). But, even better, immunogenicity can be reduced to
`the barest minimum if only the SDRS are transferred to a human scaffold
`(34).
`With SDR transfer, it may be necessary to transplant a few critical frame-
`work residues also, but those would be fewer than with CDR grafting. Dur-
`ing CDR grafting, all the CDR residues are transplanted, including those that
`are inward-pointing and thus cannot directly contact the antigen Therefore,
`in CDR grafting, the framework residues that contact those inward~pointing
`CDR residues in the nonhuman antibody may need to be preserved also in
`the humanized molecule in order to produce the same combining—site struc-
`ture. In contrast, SDR transfer involves transplantation of residues that are
`in the main outward-pointing and exposed to solvent. Because there is preser-
`vation of interior volume in the variable domains (35, 36), there is then no
`need to preserve the CDR-contacting framework residues. Nonetheless,
`some of the CDR loops are deformable and their disposition may be deter~
`mined by contacts with the framework. Indeed, there are certain framework
`residues, e.g., residue 71 in the light chain (37) and residue 71 also in the
`heavy chain (38), that make important contacts with the CDRs and that
`should probably be preserved in the humanized molecule even when trans—
`planting only the SDRS.
`Particularly critical are the framework residues involved in the associa—
`tion of the VL and VI_I domains, because the quaternary structure of the anti-
`gen—binding region of the antibody can modulate antigen—binding properties
`(39). Further, the N termini of both light and heavy chains of an antibody are
`in the periphery of the combining site (40), so that one or both could be in-
`volved in antigen binding; both N—terminal residues should be preserved in
`the humanized version.
`
`If the three-dimensional structure of the antibody—antigen complex is
`available, the SDRs as well as the critical framework residues can be readily
`identified. Such structures are available for only a few antibodies. Neverthe—
`less, an attempt can be made to identify the SDRs from the results for other
`complexes (Tables I and II) and from the analysis of sequences. Indeed, the
`
`IMMUNOGEN 2167, pg. 7
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2167, pg. 7
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`H@13qu
`
`
`
`SWMHVHUDMHM.H<Zn¥mzm«&Hflwmmmwflfi7<$OvflmmfivMHNMAM<AOMVnHZéMflJTJfiHOmHEA<,HZHmDea.ZHRSA<UHQHMmHHmEHPVHO<HZnflUZHWMDhEmMMHJw/
`
`
`
`onUmom
`
`H|UASH
`
`Ambfl
`
`HIHmWH
`
`ahmm
`
`mh>H
`
`Zhflm
`
`«UZH
`
`mEZH
`
`m>DH
`
`HdHH
`
`ZmZH
`
`HMbH
`
`HIHOUH
`
`hMHH
`
`WUdH
`
`Emmfl
`
`EMBH
`
`meN
`
`mMHH
`
`UMWH
`
`HMDU
`
`omwaUAMFN
`
`oa-amm>m
`
`muqmmhm
`
`m.Ho
`
`mH.HHn
`
`H.¢vo
`
`h.MH.mm
`
`m.m.mow
`
`mvflmn
`
`oaz
`
`kuz
`
`oauz
`
`m.mm
`
`Halmwvmvm
`
`H.0m
`
`H.mm
`
`mU
`
`mmmfi
`
`NHMHm
`
`mth
`
`m\mm
`
`174
`
`IMMUNOGEN 2167, pg. 8
`Phigenix v. Immunogen
`|PR2014-00676
`
`mu
`
`.
`
`a
`
`‘III
`
`"bumrvmm
`
`IMMUNOGEN 2167, pg. 8
`Phigenix v. Immunogen
`IPR2014-00676
`
`
`

`

`HIOZMH
`
`mmmfl
`
`mme
`
`meH
`
`qua
`
`mosN
`
`m¢mfl
`
`meH
`
`mmnfl
`
`HlwaH
`
`MUUN
`
`MHMH
`
`QZHH
`
`AGZH
`
`HImDNH
`
`
`
`«Immflmm
`
`mmmm
`
`Hou¢o>m
`
`MOHamb
`
`wombzu
`
`wmhfl
`
`bwww
`
`wmmm
`
`hmH
`
`momumoz
`
`om|o¢
`
`NOZ¢
`
`mmn
`
`OHImN
`
`w.woz
`
`omlwlw
`
`mmwdmo
`
`mez
`
`NH\wam
`
`
`
`
`
`
`
`
`
`
`
`no53%E0:535389BE..wozgfiPamacawafloémmafine28?EU$3»me3.8533mowooHwfiofioao5859.?Pauofinoo2w3839,585¢meEva0%SEE
`
`
`
` 2.?"Mambo05we“a:0505;2%E3302089852%$9:mw8065g.Ewfi2%nonm>wwmmBough—mm5RowemoOgumflammvAandMadamEND53ono5Maw50H0%
`
`U.3§prwMmonowcgnoo06mgomomWEQQEEHBE.0703gumMGMhowrwdsmugm58@152po”~me0%HowohmExam£33
`
`
`
`.mmammmN3@88me8dmmwmoofloSWomA
`
`m
`
`175
`
`IMMUNOGEN 2167, pg. 9
`Phigenix v. Immunogen
`|PR2014-00676
`
`IMMUNOGEN 2167, pg. 9
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`I76
`
`AMEURFINA D. SANTOS AND EDUARDO A. PADLAN
`
`VH RESIDUES IN CONTACT WITH THE LIGAND IN MURINE ANTIBODY—LIGAND COMPLEXES OF
`KNOWN THREE—DIMENSIONAL STRUCTURE”
`
`TABLE II
`
`10
`
`20
`
`CDRl
`35ab
`
`30
`
`40
`
`52abc
`
`CDRZ
`
`60
`
`............................. T SDY..—- .............. Y.S---YS.S.Y .......
`
`
`
`HyHEL-lO
`D1.3
`HyHEL-S
`D44.1
`D11.15
`Jel42
`NC41
`F9.13.7
`NClO
`E5.2
`409 . 5 .3
`
`TE33
`Bl3I2
`17/9
`26/9
`C3
`R45-45—11
`50.1
`59.1
`59155-4
`ER96
`
`Je1103
`BV04-01
`48G7
`1F7
`17E8
`CNJ206
`28B4
`MCPC603
`
`40-50
`CHA255
`NC6.8
`88C6/12
`DB3
`26—10
`N1G9
`4-4—20
`
`..Y.—-gN.DtS .......
`..D.--AN.N.Q .......
`..R.NS .............
`Y.T---YS.T .........
`
`..............................
`..............................
`..............................
`..............................
`..............................
`
`..............................
`..............................
`
`N10
`
`AN02
`
`
`"See footnote to Table I.
`
`SDRS are found in the main to coincide with positions that display high se—
`quence and structural variability (34). The residues that are probably in—
`volved in the VLzVH association can be guessed from the results for other an—
`tibody structures (Tables III and IV).
`The most suitable human variable domains onto which the SDRS are
`
`transplanted should be those having as many of the critical framework
`residues as possible. For obvious reasons, it is probably best to use germ—line
`
`IMMUNOGEN 2167, pg. 10
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2167, pg. 10
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`MORE EFFICACIOUS ANTIBODIES FOR MEDICINE
`
`177
`
`TABLE II
`CONTINUED
`
`
`
`CDR3
`
`100abcdefghi
`
`
`110
`
`PDB Code
`3HFM
`1VFB
`BHFL
`1MLC_1
`lJHL
`lJEL
`INCA
`1FBI_1
`1NMB
`lDVF
`lIAI
`1NSN
`
`1TE’I‘
`2IGF
`1IFH
`1FRG
`1FP'I‘
`1IKF
`1GGI_1
`1ACY
`lMFA
`1CLZ
`
`LIGAND
`Lysozyme
`Lysozyme
`Lysozyme
`Lysozyme
`Lysozyme
`HPr
`Neuraminidase
`Lysozyme
`Neuraminidase
`D1 . 3 Fv
`730.1.4 Fab
`Staph. nuclease
`
`toxin peptide
`Ch.
`Mhr peptide
`Hemagg. peptide
`Hemagg. peptide
`Poliovims peptide
`Cyclosporin A
`HIV-1 peptide
`HIV~1 peptide
`trisaccharide
`Lewis Y sacch.
`
`inosine diphosph.
`lMRD
`trinucleotide
`1CBV
`TSA
`lGAF
`TSA
`1FIG
`TSA
`1EAP
`TSA
`1KNO__1
`TSA
`1KEL
`phosphocholine
`2MCP
`ouabain
`IIBG
`Indium—EMA
`IIND
`NC174 sweetener
`2CGR
`nitrophenyl hapten
`1YUH_1
`progesterone
`IDBE
`................................
`digoxin
`1IGJ__1
`................................
`nitrophenyl hapten
`lNGP
`..
`.... ..
`fluorescein
`lFLR
`.
`. ...........
`.
`................................ .
`spin—labelhapten
`lEAF
`..
`.
`
`
`
`
`.
`
`.
`
`sequences (V and J). Fortunately, the human heavy and light chain loci have
`now been completely mapped (41—43) so that the functional human germ-
`line antibody genes are known.
`The case of CDRS-H (the third complementarity—determining region of
`the heavy chain) merits special consideration because it is the result ofV—D—J
`recombination and cannot be in any way predicted from germ-line gene seg—
`ments. However, CDRS—H sequences from a large numb er of rearranged VH
`domains are available and can be used as templates.
`
`IMMUNOGEN 2167, pg. 11
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2167, pg. 11
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`z,
`
`178
`
`AIAEKJRFTPDKID.SAIJTY)S ADM) EIDLUURIXD A.I¥UDIJUN
`
`VL RESIDUES INVOLVED IN THE VLzVH CONTACT IN MURINE ANTIGEN—BINDING REGIONS OF
`KNOWN THREE—DIMENSIONAL STRUCTURE“
`
`TABLE III
`
`CDRl
`27abcdef 30
`
`....p
`
`”ununnunnunnnnuFF.Iv””FUN.“”UH””H””“HUHHHHHHHHHHHHHNmHNN.
`
`
`”A.H”L.H”H.333Humn”HF.”v””vnumn.“ARE.H.FFFFFFFFFFFFFFFH”F..Fm...A..
`
`..I........................................SPDJPP.
`
`
`
`
`
`
`
`
`
`
`
`9.3.HHHHHnn“FF.mViv“..qw{Farm”L.L.”K...K.K.K..K.K.”K....w.w.a._.a...9.a..a..
`
`101517.431324._5l6.51.2.0O35501.5LAL191.303412105022.0280136/55
`
`
`
`
`M91283W10..79.21_.m2.720.1154I341_.15674399C9659
`
`2112M8N8SCHymwomLymmnmgnamfiommmammmommammwméfi.59sax/amamamm9.63685761HJlMlDHNNDR3YBWFRDMO74EG4155N4JBTBD24NJCRLM
`
`
`
`$533.585.
`
`5555
`
`.......g.ggKggg...........qq
`
`
`
`RHFLHLHLHL..L.LL..LLFKLLLLLLFLFFLFV..LLLLLLmeLLLLLHFKVLLL999m.m
`
`
`
`
`PPPPPPPPVWINPIWNPPPPIWNPPPPPPPPPPPPPPPPPPPPPPPPPPFFFFF
`PPPPSSSSSESSSSflSSSfiSPPPSH..R.H.Hmnn.
`
`
`
`
`
`vuvwvuvuflyxvuvuyvuyvdv”.vuwuxyxyxvyyy,v_.vuXvuyyyxyxyvuvuvuvu.1yxxyfiqug
`
`
`
`"See footnote to Table I.
`
`IMMUNOGEN 2167, pg. 12
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2167, pg. 12
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`MORE EFFICACIOUS ANTIBODIES FOR MEDICINE
`
`179
`
`TABLEIH
`CONTINUED
`
`CWDE........YF.a.v.YY..S.Yv.gs.
`
`
`.................Wm...............RRV..R.N.........................NnDMINIMUMH”HUI”:n””Hun“”HUI““Huununuuunuunuuuuunnn
`
`ywxxnkxkvuvfivuyymexvyyxxyyy.1”XYYYYVEYYYXVHVHY.”v””xxxymmgaa.
`.AuumuuflflojnfiuhRviv..H.H.A
`
`.P..I...................................
`
`
`
`4.......................................................
`o.......................................................
`
`
`
`5355.555...S..SSS..S.PPPPSSSSSSSSSSSSSSPPPS.....
`
`
`
`PPPPPPPPVWIVPT.VPPPPIVPPPPPPPPPPPPPPPPPPPPPPPPPFFFFF
`RflmeflLuLflnmHL.L.L..LLL.R.L.L.L.LL.LR.LL.ELL.V.“LLnLLLhmeLLanumMLLumacawH
`
`
`”.mwnnuna.nygxggg..nuggnu”..1an..”4..qnunqnuunnamfifia.
`
`anagflaooafianoaoxoaaoooooo”noooaaooqanoooaxangnaonaxjixgmyyxyyxxxxxxxynxxxXXJLyyxxxyyyuyxxxyxxyyxxxuymxxxxywwyy
`
`6.......................................
`
`ouuuuunuhu”nuH”H“HHHHHHHHHHHHHHHHHHHUHHH
`
`
`
`ounuuuuununnuunnuuuH”Hun””HUI””H””UUHHHUUHHHHHHUHHHHHHH7.......................................................
`
`..........s....
`
`31w___—.___.____—__—_——__.____—____._.___.__—____.___._.——
`R__.P_—______.u...._________n.______—________.________._
`
`
`
`wsmwxxmmwemxmmLTsmiwxwNLLJYDDJJV.IXNV.V.NV.V.V...N.£1X....
`
`
`5_LP._PPPPPPPPPPPPPPPmPPPPPPPPPPPPPPPPPPPPPPPPPPPWWWMW
`
`www.u.muamaa..mm.sgmxymnwuuu.sggussm..Hsssag.ummp.mmwww
`
`
` .FFF.........FFFW.QQ.AAA e.......................................................
`aLLLR.mmwnwmmnmmmamvii.R.w.v“vflukv“L.mnmnknwwKLLRRmKKLLLwawww.
`
`lFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
`
`mmflvhéuunugéuwaiffiflw"Mfg”.Afiégfi””usuanuugnnnnnn
`..g.........................g.....9...................
`
`
`
`IMMUNOGEN 2167, pg. 13
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2167, pg. 13
`Phigenix v. Immunogen
`IPR2014-00676
`
`
`
`

`

`
`
`180
`
`AMEURFINA D. SANTOS AND EDUARDO A. PADLAN
`
`VH RESIDUES INVOLVED IN THE VL:VH CONTACT IN MURINE ANTIGEN—BINDING REGIONS OF
`KNOWN THREE—DIMENSIONAL STRUCTURE“
`
`
`TABLE IV
`
`..............................
`..............................
`.............................
`..............................
`..............................
`..............................
`..............................
`..............................
`..............................
`..............................
`..............................
`
`
`
`CDRl
`40
`35ab
`..... -— .I.K..
`..
`..N~-
`.V.Q..
`..
`.H--
`.F‘.Q
`..
`.H—-
`.V.Q..
`..
`..... ~~ .V.Q..
`..
`...H--...Q..
`..
`..Y..-- .V.Q..
`..
`.
`..... —-
`.V.Q..
`.W..
`.E--
`.V.Q .....
`.W..
`.E--
`...E .....
`.W..
`.E—-— .V.Q..
`.W..
`.V.Q .....
`.W..
`.V.Q .....
`.W..
`.I.Q .....
`.W..
`.V.Q..
`.
`.W..
`.V.Q
`.W..
`.V.Q..
`.W..
`.I.Q
`.W..
`.V.Q..”
`.W..
`.V.Q...
`.V.Q...K.L.w..
`
`CDRZ
`
`60
`
`52abc
`..
`
`.
`.
`.
`.
`.
`
`
`
`m-
`
`m
`
`HyHEL—lO
`Dl.3
`40—50
`HC19
`CHA255
`B72.3
`48G7
`JellOB
`D44.1
`NC6.8
`HyHEL—S
`TE33
`mAb735
`88C6/12
`Je142
`D11.15
`J539
`5e155—4
`8F5
`17 -IA
`31312
`DB3
`26—10
`1F?
`R6.5
`17E8
`BR96
`N1G9
`C3
`17/9
`26/9
`36-71
`NC41
`730.1.4
`NClO
`E5.2
`F9.13.7
`L5MK16
`R19.9
`OPGZ
`MOPC2 1
`R45—45—1l
`4—4—20
`YsT9-1
`409.5.3
`BV04—01
`CNJ206
`McPC603
`N10
`ANOZ
`50.1
`59.1
`GH1002
`2884
`
`1583
`
`.V.Q...K.L.Y..
`.I ...... SL.W..
`.V.E...k.L.w..
`.V.Q...QgL.W..
`.V.Q...KRL.W..
`.V.Q..g..L.w..
`.I.Q....gL.W..
`.V.Q...K.LeW..
`.V.H...KRL.W..
`.V.Q...q.L.W..
`.V.Q ..... LkW..
`.V.Q ..... L.w..
`.V.Q ..... L.W..
`...Q..
`.gL.W..
`.V Q..
`.gL.W..
`.V.Q..
`.gLeW..
`.V.Q..
`.gLew..
`.V.Q.. K.L.W..
`...Q..EKgL.w..
`..... -~ ...Q...kRL.W..
`W.N~— .V.Q...K.L.W..
`..... -- .V.Q....AL.W..
`W S—-
`.V.Q ..... LeW..
`..... -- .V.Q...K.L.W..
`.H--
`.V.Q....gLeW..
`Y.E—— ....Q.....RLW
`...... -
`.I.Q...N.L.W..
`...... -
`.I.Q...NKL.W..
`....... .I.Q....gL.W..
`....... .I.Q.
`..gL.W..
`..... -- ...Q.....L.W..
`..... -- .A.Q....ALeW..
`.H--
`...Q....gL.W..
`
`
`“See footnote to Table I.
`
`IMMUNOGEN 2167, pg. 14
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2167, pg. 14
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`MORE EFFICACIOUS ANTIBODIES FOR MEDICINE
`
`181
`
`TABLE IV
`CONTINUED
`
`70
`
`82abc
`
`9O
`
`CDR3
`looabcdefghi
`
`110
`
`
`
`.
`.......FYYAM---..W.... ..... .
`.
`.
`.
`.
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`PDB Code
`3HFM
`1FDL
`lIBG
`1GIG
`1IND
`1BBJ
`1GAF
`lMRC
`1MLC_1
`ZCGR
`ZHFL
`lTE'I‘
`lPLG
`lYUH_1
`lJEL
`.
`.
`.D.YgAM—-—--
`lJHL
`--D. Wg .......
`
`..... . ...... ......Y...L.ngYN————
`ZFBJ
`--—a. WgQ.
`.
`. .....
`
`..HgYY -—~- ——DY W.........1MFA
`.......................... V.Y...
`........ YYDm—-—-— ——D.Wg.........
`lBBD
`
` ...YpYA.—-— ——.. Wg.........
`............................ F...
`1FOR
`......................... ......Y ...D.FYF-—--
`--D. Wg.........
`21G}?
`
`....NWYF—-— —-D. W.a........
`............................ F...
`1DBB
`
`Y ...NKWAM—-— ——D.Wg......
`lIGJ_1
`................................ ...gNYgF——----—T. w..........
`1FIG
`...LLLSF-—-----DY Wg. .
`............................ Y. ..
`.
`.
`.
`.
`. ..
`1RMF
`.........
`.Y.gSsY. ———————D. W lEAP
`............................ Y. ..
`.L. DgAWF———-—~~AY qu. .
`.
`.
`.
`. ..
`1CLZ
`..................... Y... Y....SSYF——————d.w ..........
`1NGP
`......................... ....F.
`.....DVgF———-——DY Wg.........
`1FPT
`..... .......................Y...R...dENGF——————..Wg.........
`11FH
`......... . ..... .....Y...R...dEKgF——————A. Wg... .....
`1FRG
`............................ F.
`.Y..gSyKF-—--—D W.........
`6FAB
`.
`F...
`.FgSLS——————DY Wg. . .. .....
`1NCA
`F...D. .YE.YYAM—————dYqu........
`IIAI
`.. .......................... Y. ..
`. .YDggF----—DY W.. ........
`1NMB
`Y
`.....quRgAM—~—~D.W.Q ..... ..
`1DVF
`Y
`............................
`.sYgVL---—D. W...Q......
`lFBI___1
`....... YWYVL-m-WD Wg......
`1LMK_1
`..... .
`.
`...............F ...S..lAVYYF-~D qu........
`2F19
`.......Y ........~YYAMD. qu........
`lOPG
`...... YWPYYAM .DYWg.........
`lIGC
`.
`.Yd’I‘.
`.Y VWFaD W ..........
`11KF
`.. S. ng——————————DY Wg... ......
`1FLR
`..
`.YgP. ~~~~~~~~~A. W.q .....
`1MAM
`.V. R...YyaV-------D. W........Q.
`lIAI
`..
`.Q..TAWF———————AY qu. .......
`1CBV
`........ ..............YYYr ———————A.w .......
`1KNO_1
`2MCP
`lNSN
`lBAF
`1GGI__1
`1ACY
`1GHF
`lKEM
`
`
`
`.....
`
`1NLD
`
`
`
`IMMUNOGEN 2167, pg. 15
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2167, pg. 15
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`182
`
`AMEURFINA D. SANTOS AND EDUARDO A. PADLAN
`
`A. Sample Design of CI ”Humonizotion” Protocol
`by SDR Transfer
`
`Humanization by SDR transfer is illustrated here for the case of the
`murine antibody, B723, which binds to the tumor-associated antigen, TAG-
`72 (3). The three—dimensional structure of the antigen—binding region of
`B723 has been determined by X—ray crystallography (44), although only in
`the uncomplexed form. The residues involved in the VL:VI_I contact in B723
`are included in Tables III and IV.
`
`Based on the results for other antibodyzantigen complexes (Table I), the
`residues in B723 likely to be involved in antigen binding are residues 30, 32,
`and 34 in CDRI, the framework residue 49, residues 50 and 53 in CDR2,
`and residues 89, 91, 92, and 94—96 in CDR3 in the light chain; and the
`framework residue 30, the residues 3 1, 33, and 35 in CDRI, residues 50 and
`52—57 in CDR2, and residues 95—97 and 101 in CDRS of the heavy chain.
`The residues that contribute side—chain interactions to the VL:VH contact
`in B723 are 32, 34, 36, 38, 42—44, 46, 49, 87, 90, 94—96, 98, and 103 in
`the light Chain (Table III) and 35, 39, 45, 47, 50 58, 60, 61, 91, 93, 96, 97,
`103, and 105 in the heavy chain (Table IV; see also Table V).
`In addition to the putative antigen-contacting residues, those residues in—
`volved in the quarternary interaction between the variable domains, the N—
`terminal residues, and the residue at position 71 in both light and heavy
`chains should be preserved during humanization. This requirement is used
`in the identification of the human sequences that would best serve as tem—
`plates for humanization.
`A comparison with known human germ—line VK sequences reveals that
`B723 VL is most similar to VKI—ZI (45), among several, in the critical frame—
`work residues. The ]L segment of B723 is most similar to the human L2.
`The sequences are compared to Table VI. Also indicated in Table VI are the
`residues that are probably involved in the interaction with the antigen and
`the framework residues that are most probably critical to the preservation of
`the structure of the antigen—binding site. A protocol for the humanization of
`B723 VL by the method of SDB transfer is included in Table VI.
`A comparison with known human germ~line VH sequence reveals that
`B723 VH is most similar to DP—7 (41) in the critical framework residues. The
`]H segment of B723 is most similar to the human JH—l and its CDR3-H has
`the same number of residues as that of antibody JM 0—131 (46). The se—
`quences are. shown in Table VII. The protocol that we propose for the ‘hu—
`manization’ of B723 VH by the method of SDB-transfer is included in Table
`VII.
`
`The CDR residues of B723 that are transferred to the human templates
`number 16 using the method of SDR transfer. This constrasts with the 54
`
`IMMUNOGEN 2167, pg. 16
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2167, pg. 16
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`MORE EFFICACIOUS ANTIBODIES FOR MEDICINE
`
`183
`
`TABLE V
`ENVIRONMENT or SIDE CHAINS“
`,______—.________~
`
` Distance range (A) Nonpolar Polar (main chain) Polar (side chain)
`
`
`
`
`
`Number of atoms in that distance range
`
`Of Tyr-407 in IgGl Fc
`r < 3.5
`3.5 < r < 4.0
`. 4.0 <1" < 4.5
`
`5
`28
`38
`
`2
`2
`7
`
`1
`1
`1
`
`4.5 < 1‘ < 5.0
`Total
`Of Pile-506 in IgE Fe
`1‘ < 3.5
`3.5 < 1” < 4.0
`4.0 < r < 4.5
`
`fl
`113
`
`12
`23
`33
`
`_§
`19
`
`0
`2
`4
`
`_0_
`3
`
`0
`1
`1
`
`2
`_9_
`fl
`4.5 < r < 5.0
`
`
`
`115 15Total 4
`
`“These values are derived from the Ciystal structure of human IgGl Fc (PDB Entry lFCl) and from
`the model of human IgE Fc (PDB Entry 210E).
`
`that would have been transferred if the method of CDR grafting were em-
`ployed. A substantial reduction is therefore achieved in the number of non-
`human residues that will be present in the humanized molecule with a con-
`sequent reduction in its immunogenicity. This number may be reduced even
`further if the structure of the 1372.3 antibodyzantigen complex becomes avail—
`able and the actual CDR residues that make contact with antigen become
`known. We believe that humanization by SDR transfer will reduce immuno-
`genicity to the barest