`Photocopying permitted by license only
`1:>1987 Harwood Academic Publishers GmbH
`Printed in the United States of America
`
`Perspectives on Antigenicity
`and Idiotypy
`
`THOMAS KIEBER-EMMONS,t ELIZABETH GETZOFF,*
`and HEINZ KOHLER..,
`tDepartment of Molecular Immunology. Roswell Park Memorial Institute, 666
`Elm St., Buffalo. NY 14263
`*Departments of Molecular Biology and Immunology, Research Institute of
`Scripps Clinic, La Jolla, CA 92037
`
`INTRODUCTION
`
`Over ten years have passed since the concept of using synthetic peptides to
`probe antigenicity was first developed [l, 2]. Since then, prominent among
`the applications of synthetic immunogen technology in biology and medi(cid:173)
`cine (3, 4] is the utilization of synthetic peptides derived from the antigen for
`vaccine development [5]. The network hypothesis of Jerne [6] offers still
`another elegant concept for vaccine development. The anti-idiotype concept
`[7] provides an approach whereby an antigen can be substituted by an
`antibody possessing characteristics of that antigen. This can be demon(cid:173)
`strated by using an anti-idiotypic antibody (Ab2) as a surrogate antigen that
`can stimulate an antigen-specific immune response [7]. This avenue
`provides an alternative in cases where the production of antigen based upon
`molecular biological approaches may not be feasible.
`The development of idiotope (Id) derived vaccines rests on the principle of
`molecular mimicry. An understanding of the structural basis of molecular
`mimicry could improve the production of idiotype vaccines, moving it from
`
`339
`
`1 of 18
`
`Celltrion, Inc., Exhibit 1054
`
`
`
`340
`
`T. KlEBER-EMMONS, E GETZOFF, and H. KOHLER
`
`an experimental state to a rational approach [8]. The importance of molecu(cid:173)
`lar mimicry by monoclonal anti-idiotypes (anti-Id) is the ability to make T
`ceU-independent antigens T cell-dependent. For example, anti-idiotypes
`would allow the presentation of carbohydrate antigens as mimicked by the
`structure and conformation of the protein surrogate [9]. The mimicking
`abilities of such "internal image" antibodies also sets the stage for the
`possibility of producing fully synthetic idiotope vaccines using essential
`sequence information obtained from idiotope hybridoma antigens. The use
`of such designed idiotope-derived synthetic peptides would thwart problems
`associated with the administering of mouse hybridomas to humans (10) .
`Of fundamental importance in designing new peptide antigens is the
`faithfulness or fidelity of the molecular mimicry. Antibody-antibody interac(cid:173)
`tions are modulated by their large surface areas, so the complete description
`for an Id may entail contact points which are close in space but remote in
`sequence. Depending on the idiotope, there may be two components which
`contribute to the degree of mimicking fidelity: essential mimicking residues,
`and contact residues whose complementary interactions lend to the overall
`association constant for a particular complex formation. This latter compo(cid:173)
`nent may also play an important secondary role in helping to stabilize a
`particular structural environment required for full antigenic mimicry. To
`disentangle these possible effects and ultimately achieve the successful
`development of a functional antibody or peptide vaccine, it is imperative to
`fully understand the structure of the antibody molecule, the basis of idiotypic
`expression in three dimensions and the mechanisms by which large surface
`areas on proteins modulate protein-protein interactions. These points are
`addressed in this volume. Here, we present an overview of the salient
`features of these topics.
`
`THE ANTIGENIC NATURE OF IMMUNOGWBULINS
`
`ldiotopes represent a particular category of antigenic determinants which
`can activate clones bearing complementary paratopes through a self-recogni(cid:173)
`tion process. This behavior implies that idiotopes are auto-antigens: self(cid:173)
`antigens recognized by the immune system . One model proposes that the
`response to such self-proteins is directed against sequence regions that
`exhibit the highest evolutionary variability [II] . Therefore, according to this
`model, sequence-variable regions are antigenic and evolutionarily conserved
`regions induce tolerance. From a structural perspective, variable, and there-
`
`2 of 18
`
`Celltrion, Inc., Exhibit 1054
`
`
`
`PERSPECTIVES ON ANTIGENICITY AND ID!OTYPY
`
`341
`
`fore antigenic, regions can tolerate local changes in conformation and should
`correlate with sequence regions which are Oexible and surface exposed (12).
`Thus, the combination of intrinsic factors such as mobility and accessibility,
`and extrinsic host factors such as tolerance, immune response genes, idi(cid:173)
`otype networking and structural gene repertoire appear to describe protein
`antigenic structure [ll , 13).
`Historically, complementary relationships in the recognition properties of
`immunoglobulins have been attributed to the hypervariable nature of immu(cid:173)
`noglobulin sequences; complementary determining regions (CDRs) [14).
`These regions have been typically associated with the antigen binding site
`[15] and have shown some correlation with the self-association of light and
`heavy chains [16, 17). Conversely, residues which are not classically hyper(cid:173)
`variable can be complementary in the context of idiotope recognition [18).
`
`SURFACE VARIABILITY ANALYSIS
`
`Considering that the classical views of immunoglobulin (lg) hypervariability
`and binding site complementarity may not necessarily be equivalent con(cid:173)
`cepts, it may be appropriate to re-evaluate variability from other viewpoints.
`A theoretical approach, referred to as surface variability analysis, couples
`both intrinsic and extrinsic factors of antigenicity by considering the evolu(cid:173)
`tionary variability of protein surface regions [ 19). This method characterizes
`autoantigenic loci in protein families based upon examination of the vari(cid:173)
`ability in the hydrophilic properties of evolutionarily variant protein se(cid:173)
`quences. Surface variability is measured as a function of hydrophilicity and
`evolutionary sequence variation [19). For each sequence, hydration poten(cid:173)
`tials defining the affinity of each amino acid side chain for solvent water (20,
`21) are averaged over six residues and inverted to make hydrophilic values
`positive . At each sequence position, the resulting hydrophilicity profiles are
`averaged to form a consensus value and assayed for variability according to
`the formula of Wu and Kabat [14): number of different (hydrophilicity)
`values divided by the frequency of the most common (hydrophilicity) value.
`The product of the consensus and variability of hydrophilicity values is used
`to define a surface variabilty index giving maximal values for surface(cid:173)
`exposed sequences which varied significantly during evolution, i.e. those
`likely to form antigenic determinants. Advantages of surface variability
`analysis over consideration of surface-accessibility to antibodies are 1) a
`canonical ensemble of structures is evaluated, 2) parameters associated with
`
`3 of 18
`
`Celltrion, Inc., Exhibit 1054
`
`
`
`342
`
`T. KlEBER-EMMONS. E. GETZOFF, and H. KOHLER
`
`the intrinsic factors of protein structure are related to the extrinsic biological
`factors of evolution that play a role in defining antigenicity, and 3) the
`distinction shown between strong and weak immunodominant regions corre(cid:173)
`lates with serological trends.
`Surface variability analysis of variable domain lg sequences highlights
`potential autogenic surface regions, referred to as idiotope determining
`regions (IDRs). Surface variability profiles (Fig. l) for a family of2S mouse
`and human lg variable region sequences [22] show that, although the major(cid:173)
`ity of IDRs correspond to hypervariable regions, IDRs also occur in frame(cid:173)
`work regions. It is clear from the profiles that the majority of surface variabil(cid:173)
`ity in this lg family resides in the heavy chain. In Figure 2, the antigenic
`topography of the Fv region is illustrated by mapping surface variability
`values (classified into 4 categories from Figure I: most variable, more
`variable, Jess variable, and least variable) onto the surface of the 3-dimen(cid:173)
`sional structure of MPC603 (23, 24] . IDRs are depicted by the most brightly
`colored surface regions (Fig. 2B) contributed by residues with high surface
`variability (shown with labels in Fig. 2A). Clearly, IDRs cover a continuum
`of binding sites in the variable region. The large repertoire of IDRs should
`allow many combinatorial possibilities for idiotope expression in three
`dimensions, including those formed solely by light chain residues, those
`formed solely by heavy chain residues, and those formed by residues of both
`chains. Topographic mapping of one idiotypic system has shown a linear
`idiotope map spanning from the antigen binding site to the vicinity of the
`constant region (see Greenspan and Monafo, this issue). Surface variable
`regions including framework residues may be recognized by more cross(cid:173)
`reactive anti-idiotypic antibodies, since fewer CDR residues are involved .
`Of the framework residues in the light chain, only 49 and 85 (numbered
`sequentially according to MPC603) exhibit high surface variability (Fig. 2).
`Both are isolated sequentially and spatially from the CDRs. In the heavy
`chain, framework residues with high sequence variability include those
`adjacent in sequence to CDRs (30, 49, and 99-100), those conformationally
`adjacent to CDRs on the surface (76- 79), and those distant from the CDRs
`(84, 86, 88). Two clusters of heavy chain framework residues (76-79 and 84,
`86, 88) form likely IDRs; residues 76- 79 form a protruding beta bend made
`up of most variable residues, while residues 84, 86, and 88 (outwardly facing
`residues along a beta strand) form a relatively small flat surface patch of more
`variable residues. These surface topography and surface variability charac(cid:173)
`teristics suggest that the region emcompassing residues 76- 79 forms the
`most likely IDR outside of the antibody-combining site.
`
`4 of 18
`
`Celltrion, Inc., Exhibit 1054
`
`
`
`PERSPECTIVES ON ANTIGENICITY AND IDlaTYPY
`
`343
`
`Llghl Choin
`
`0
`
`20
`
`40
`
`60
`
`80
`
`100
`
`120
`
`A
`
`3000
`
`2500
`
`~ 2000
`:E 1500
`Cl ·c
`c > 1000
`500
`
`0
`
`:}()()()
`
`2500 Heovy Chein
`
`.... 2000
`::
`:E 1500
`Cl ;:
`c > 1000
`500
`
`0
`
`0
`
`40
`
`80
`60
`Sequence DUJnber
`
`100
`
`120
`
`B
`
`FIGURE I Plots of the linear relarionship of surface variability to sequence position in the
`variable regions of the light (A) and heavy (B) chains from 25 mouse and human immu(cid:173)
`noglobulins [22]. Residues are numbered sequentially to match the sequence of MPC603.
`Variability is averaged over six residues and plotted a1 1he third position to allow appropriate
`mapping onto the three-dimensional structure (see Fig. 2A). Complementarity-determining
`regions are shown by horizomal bars: light chain CORI (residues 24-40). CDR2 (56-62). and
`CDR3 (95-103) and heavy chain CORI (31-35), CDR2 (50-68), and CDR3 (IOI- Ill). Long
`dashed horizontal lines separate the four categories of surface variability used to color code
`Figure 2B. The residues included in each category for the light chain: most variable (residues
`25-26, 28, 29, 36, 38. 96-100). more variable (27, 37, 39, 49, 58-60, 85. 95, IOI), less
`variable (22-24, 40, 42, 45- 48, 50, 56-57, 61-63. 84, 86-89, 102-103, 108, 112), and least
`variable (1-21. 30- 35. 41, 43-44, 51-55. 64- 83, 90-94, 104-107 , 109-lll, 113-115); aod for
`the heavy chain: most variable (32-33 , 51-55. 58-64, 66. 77-79, 99-105), more variable (30.
`35, 49- 50, 65, 76, 84, 86, 88. 106-!09), less variable (4-5. 7, 18, 28-29, 31, 34, 36-37, 45,
`47- 48, 68-69, 71-75, 80-83, 85, 87. 89-91. 98. 110-111), and least variable (1- 3, 6. 8-17,
`19- 27, 38-44, 46, 56-57, 67, 70, 92-97, 112-122).
`
`5 of 18
`
`Celltrion, Inc., Exhibit 1054
`
`
`
`344
`
`T. KIEBER-EMMONS, E. GETZOFF. and H. KOHLER
`
`A
`
`FIGURE 2 Computer graphics images of immunoglobulin surface variability mapped onto
`the light and heavy chain variable domains of MPC603. Labeled alpha carbon backbones (A)
`and solid external molecular surfaces (B) are shown in the same orien1a1ion, looking down into
`the combining site from the solvent. The alpha carbon back.bone (A) of the light chain (top) is
`shown in blue with yellow CDRs and the alpha carbon backbone of the heavy chain (bottom) is
`shown in purple with green CDRs. In both the light and heavy chains, residues with high average
`surface variability are labeled (A) by number at the alpha carbon position in yellow (more
`variable) and red (most variable). The solid external molecular surface (B) is color coded by the
`surface variability (defined in texl and shown in Fig. I) using a radiating body color scale with
`increasing variability corresponding to increasingly brighter color. The highly (most and more)
`variable regions (lightest colors in B) are defined 10 be potential lDRs; the majority are also
`CDRs. The most prominent highly variable surface region outside the CDRs is formed by heavy
`chain residues 76-79 (bottom right), which are conformationally adjacent to CDR2. Other
`framework residues with high surface variability are light chain 49 (behind) and 85 (upper
`righti, heavy chain 30, 49 and 99-100 (adjacent in sequence to CORI, CDR2, and CDR3.
`
`6 of 18
`
`Celltrion, Inc., Exhibit 1054
`
`
`
`PERSPECTIVES ON ANTIGENICITY AND IDIOTYPY
`
`345
`
`respectively), and heavy chain 84, 86, and 88 (lower left). The backbone is displayed using the
`graphics programs GRAMPS [25) and GRANNY (26). The molecular surface, defined by
`mathern<ttically rolling a probe sphere (1 .4 A radius) representing a solvent water molecule over
`the van der Waals surface of the protein, is calculared and displayed using the programs AMS
`and RAMS (27, 28). Crystallographic coordinates [23, 24) are taken from the Brookhaven
`Protein .Data Base [29). (See Color Plates I and II in the color secrion of this issue.)
`
`B
`
`7 of 18
`
`Celltrion, Inc., Exhibit 1054
`
`
`
`346
`
`T. KIEBER-EMMONS. E. GETZOFF. and H. KOHLER
`
`IDIOTOPE/PARATOPE OVERLAP
`
`IDRs agree quite well with idiotope locations deduced from sequence anal(cid:173)
`ysis (Table I; see also Chen, et al., this issue). IDR areas have been shown 10
`be surface-exposed (see Novotny in this issue). Surface variability analysis
`has implied that 1) there are more IDR than CDR loci and 2) all CDR loci are
`also IDR loci. Structural and functional implications of this relationship
`between CDRs and IDRs have been discussed [22) (see also Stevens and
`Schiffer, this issue). The overlap between idiotopes and paratopes [22]
`implies that idiotopes can participate as binding sites; the dual roles of
`binding to a determinant or being bound by another antibody may be
`expressed simultaneously by the same topographic site. Consequently, sites
`on Igs may not be easily characterized functionally [22). In support of this
`
`Correspondence of Experimentally Deduced ldiotype Locations and Predicred IDRs
`
`TABLE l
`
`System
`
`Anti-Dextran
`
`Anti-Phosphocholine
`Anti-Galactan
`
`Ars·A
`
`Anti·lnulin
`
`B lymphocyte
`Anti-4-hydroxy-3-nitro-S-idiopbenyl
`IgM-RF
`
`Experimental
`Position
`
`H96- 97
`H53-54
`HFR3
`HV3
`H96-97
`H53
`L30
`H59
`HV3
`LV3
`LS3
`L56
`L53
`H94-99
`HV3
`H94-99
`L28- 31
`L53
`HV2
`LFR2
`LFR3
`
`Predicted
`Position
`
`IDRE- H
`IDRC-H
`IDRD-H
`IDRE-H
`IDRE-H
`IDRC-H
`IDRB-L
`IDRD-H
`IDRE-H
`IDRG-L
`IDRD-L
`IDRD-L
`IDRD-L
`IDRE-H
`IDRE-H
`IDRE-H
`TDRB-L
`IDRD·L
`IDRC-H
`IDRC-L
`IDRE-L
`
`References
`
`30, 31
`32, 33
`18
`34
`35
`
`36
`
`32
`
`37
`38
`39-44
`
`45
`
`Experimental positions correspond to H (heavy), L (light) chain designation with standard
`numbering scheme (93]. HV and LV designation for heavy chain variable and light chain
`variable regions I to 3, respectively (93]. The nomenclature for corresponding IDR predicted
`positions are from [22].
`
`8 of 18
`
`Celltrion, Inc., Exhibit 1054
`
`
`
`PERSPECTrVES ON ANTIGENICITY AND IDrOTYPY
`
`347
`
`notion is the recent description of two antibody structures, one with a unique
`V-V packing arrangement (see Stevens and Schiffer in this issue) and the
`other emphasizing the expanse of contact area across the surface of the
`antibody [46).
`Jdiotypic and anti-idiotypic interactions have been classified according to
`the location of the idiotope on the lg surface relative to their paratope [ 6, 47-
`50). Definitions derived from such topographic relationships are arbitrary in
`that an idiotypic antibody (Ab!) that is defined for one system can be an anti(cid:173)
`idiotypic antibody (Ab2) in another system. The fact that the same antibody
`can bind epitopes while presenting itself as an antigen further emphasizes
`the correlation between paratopes and idiotopes.
`
`STRUCTURAL VERSUS TOPOGRAPHIC DETERMINANTS
`
`The phenomenon of molecular mimicry would appear to suggest a chemicaV
`structu ral equivalence between the antigen and anti-Id antibody. However,
`similar three-dimensional surface environments do not imply that the sec(cid:173)
`ondary and tertiary folding pattern of the antibody and anti-idiotype are the
`same. For protein antigens, the identification of homologous sequence
`regions between the nominal antigen and anti-Id does not ensure the identi(cid:173)
`fication of the "internal image," since different three-dimensional environ(cid:173)
`ments influence the folding patterns of the related sequences (51, 52). On the
`other hand, the framework (FR) region of an antibody represents a simple
`beta sheet scaffold onto which binding sites may be built, implying that the
`structure of CDRs is relatively independent of the FR context. The confor(cid:173)
`mation of CDR loops between beta strands depends on loop sizes and
`specific interactions between the loop and the beta sheet. Studies on the
`conformational attributes of the antibody molecule [53-57] have empha(cid:173)
`sized antigen binding. Here, we are interested in how the folding patterns if
`Igs can be related to the folding pattern of nominal antigens, in particular,
`whether certain CDRs and FR regions influence the expression of idiotopes
`in terms of sequence or backbone conformation. Historically, determinants
`that seem to be affected by single residue changes have been described as
`topographic, because the immune system must be responding to local
`changes in the surface topography of the molecule [58]. Alternatively,
`determinant that involve changes in the backbone conformation were called
`conformational determinants or structural determinants [59). Neither of
`these definitions preclude that the determinants may be discontinuous. An
`
`9 of 18
`
`Celltrion, Inc., Exhibit 1054
`
`
`
`348
`
`T. KIEBER-EMMONS, E. GETZOFF. and H. KOHLER
`
`important difference in the two determinant features is that a synthetic
`peptide may not be able to adopt a particular conformational determinant.
`The subtle difference between these determinant types may also influence
`the expression of cross-reactive idiotopes which certainly play a role as
`regulatory idiotopes [18).
`The nature of antibody-antibody interactions implies that extended areas
`of the protein surface contribute to binding, leading to a multisite interaction
`model for idiotope recognition. The required thermodynamic conditions for
`such interactions can be achieved via several sources (13] and can be quite
`precise in that the interactions can be affected by a single amino acid change
`(60-63]. Experimental observations idicate that both sequence and confor(cid:173)
`mation dictate an immune response (64, 65]. In either case, the degree of
`spatial adjustments in the molecular partners results in different free energies
`of association for any given antibody with any other given antibody or
`antigen. The overall energetics of the interaction is determined by the free
`energy cost (G) of any conformational changes experienced by either the
`antigen (anti-Id) or the antibody upon complex formation. At the molecular
`level , the relative energy cost of the spatial displacement of interacting
`groups is intrinsically associated with the relative mobility of a region. The
`concept of macromolecules as flexible entities and range of the dynamical
`nature of proteins has been extensively discussed [66- 72].
`Determinants are often comprised of discontinuous parts of an antigen. In
`the crystal structure of Amit and coworkers [46], the lysozyme determinant
`is made up of two stretches of polypeptide chain comprising residues 18 to 27
`and 116 to 129. Surface variability analysis of lysozyme [19] is in relatively
`good agreement, identifying the majority of residues in this epitope (resi(cid:173)
`dues 14-21 and residues 115- 126) as autogenic loci. While CDR3 of the
`heavy chain of the antilysozyme forms the principle contacts, all six CDR
`regions are involved in contacting lysozyme. Molecular modelling studies of
`Jysozyme-antilysozyme complexes [73] also suggest that the loop region
`epitope (residues 57-84) of lysozyme is described by autogenic loci (19]
`centered on positions 67- 79. In these complexes all six CDR regions are also
`involved in defining contacts.
`ldiotopes (and antigenic determfoants in general) have been localized by
`evaluating the reactivity of antibodies to the protein with synthetic peptides
`derived from different parts of a protein [33) (see Chen, et al., this issue).
`Small peptides are thought to exist in a multiplicity of transient conforma(cid:173)
`tional states in dynamic equilibrium, in contrast to the relatively stable
`structure of a protein in solution (74]. Nevertheless, small peptides can
`
`10 of 18
`
`Celltrion, Inc., Exhibit 1054
`
`
`
`PERSPECTIVES ON ANTIGENICITY AND IDIOTYPY
`
`349
`
`induce antibodies with a sequence and structural requirement for binding
`antigen comparable to antibodies raised against the native protein [3].
`Whether this is because certain peptides adopt a conformation similar to that
`found in the native protein [75] or this results from induced fit remains to be
`seen. The energetic considerations which mediate antipeptide-peptide com(cid:173)
`plexes and the relationship to antipeptide-antibody interactions with the
`parent molecule have been discussed [13, 64, 65].
`The degree of specificity of a particular complex, as well as the possible
`description of epitope regions, relies on the consorted, dynamical nature of
`an antigen-antibody interaction [12, 76, 77] . However, in some circum(cid:173)
`stances a restriction in the conformational freedom of immunizing peptides
`will result in antibodies with the same specificity as those induced by the
`proteins themselves [78, 79]. This result focuses on the supporting structural
`role of regions remote in sequence to a primary epitope center. Short
`synthetic linear peptides may not account for such environmental contribu(cid:173)
`tions [80]. The influence of a tertiary environment on the natural conforma(cid:173)
`tions of short peptide segments is clearly evident, in that identical sequence
`segments in native proteins can have different conformations [51, 52]. The
`possible fine specificity in antigenic recognition by B cells apparently
`extends to T cells, where different determinants can be formed by the same
`peptide and la molecule [81].
`
`MIMICRY OF CONTACT RESIDUES
`
`The possible ways in which anti-idiotypic antibodies can mimic antigens has
`been summarized (82) . Sequence homology with the protein antigen has
`been suggested for the mimicking capabilities of monoclonal antibodies in
`the reovirus system [83). Sequence analysis of a hybridoma that mimics the
`reovirus antigen indicates sequence homology in an expected IDR of the
`kappa light chain [83]. A linear synthetic decapeptide derived from the anti(cid:173)
`Id (LV2;hypervariable region 2 of the light chain;CDR2), which is homolo(cid:173)
`gous to reovirus in five positions and has three conservative substitutions is
`capable of inducing biological responses similar to the reovirus antigen as
`well as the anti-Id [83 and M. Greene, personal communication]. Essential
`contact residues which define this reovirus epitope appear to be retained in
`the Ab2 in a homologous, linear fashion. In a random fashion, the likelihood
`of homologous matching of five residues in this hypervariable region, as
`evidence in the reovirus system, is 7 :20s. Although such homology would be
`
`11 of 18
`
`Celltrion, Inc., Exhibit 1054
`
`
`
`350
`
`T. KIEBER-EMMONS, E. GETZOFF, and H KOHLER
`
`rare in randomly generated sequences. analysis of viral protein sequences
`suggests that rudimentary viral epitope sequences are established within the
`lg germ lines (Kieber-Emmons and Kohler, unpublished).
`In Table II, light chain sequences of the crystallographically known lg
`structures are aligned with those of the proposed reovirus epitope and its
`mimicking anti-Id. The resulting homology points out the possible structural
`similarities between the mimicking monoclonal 87. 92.6 and MPC603 in this
`region. In Table III, heavy chain-sequence comparisons with reovirus indi
`cate additional homology including the identical reovirus sequence (NSYSGS).
`Comparison of the crystallographic structures of HV2 (hypervariable region
`2 of the heavy chain;CDR2) regions from NEWM and MPC603 also indi(cid:173)
`cates that the conformational properties are highly constrained in spite of
`sequence differences.
`The alignments in HV2 and LV2 provide an opportunity to compare the
`conformational properties of the sequences of the proposed reovirus epitope
`and the mimicking monoclonal. The sequences in the two tables can be
`considered as structural variations of beta bends. The structural similarities
`provide a data base [64] to help elucidate residues essential for antibody
`87.92.6 recognition and binding and residues integral to the idiotope be(cid:173)
`cause they stabilize secondary structure essential to the presentation of the
`antigenic site .
`The identification of sequence similarities between an antibody and an
`antigen is nongermane in the analysis of anti-idiotypic antibodies mimicking
`carbohydrates or haptenic antigens. For these cases, conceptual physical/
`chemical models (e.g. the proper alignment of functionally reactive groups,
`or of residues with van der Waals interaction tendencies, shape or other
`physical attributes resembling those of the antigen) must be invoked as an
`initial basis for mimicry of nonprotein antigens. Such models must also be
`examined in cases where the mimicking antibody and protein do not exhibit
`sequence homology. Thus, the degree of chemical or sequence similarities
`will certainly affect the degree of mimicry fidelity or faithfulness.
`An anti-TIS hybridoma, 4CJI, is capable of inducing biological effects
`similar to those induced by phosphorylcholine (PC) (85, 86]. Competitive
`bindings assays have shown that PC is a successful inhibitor of 4Cll-Tl5
`binding. Crystallographic analysis of the PC-binding antibody MPC603 and
`PC-binding studies provide a basis for examining potential contact residues
`in the variable domains of 4Cll that could provide chemical mimicry of PC.
`Conformational attributes of model peptides interacting in a manner similar
`to PC with MPC603 suggests that the tetrapeptide Lys-Gly-Gly-Asp is
`
`12 of 18
`
`Celltrion, Inc., Exhibit 1054
`
`
`
`Alignment of Struc1urally Known LV2 Regions with Reovirus and Anti-Id
`
`TABLED
`
`MPC603
`Reovirus
`87.91.6
`NEWM
`KOL
`
`Leu
`Ue
`Leu
`Leu
`Leu
`
`Leu
`Val
`Leu
`Leu
`Leu
`
`Ile
`Ser
`Ile
`lie
`Ile
`
`Tyr
`Tyr
`Tyr
`'fyr
`Tyr
`
`Gly
`Ser
`Ser
`Arg
`Arg
`
`Ala
`Gly
`Gly
`Asp
`Asp
`
`Ser
`Ser
`Ser
`Asp
`Ala
`
`Thr
`Gly
`Thr
`Lys
`Met
`
`Arg
`Leu
`Leu
`Arg
`Arg
`
`Gin
`Asn
`Gin
`Pro
`Pro
`
`Ser
`Trp
`Ser
`Ser
`Ser
`
`Gly
`Arg
`Gly
`Gly
`Gly
`
`The sequences for MPC603, NEWM, KOL, and reovirus were obtained from Protein Identification Resource [84]; the sequence for 87. 91.6 is from
`[83].
`
`TABLE 01
`
`Alignmenl of Selected HV2 Regions wilh Reovirus and Mimicking Anti-Id
`
`Reovirus
`36-60
`NEWM
`MPC603
`87.91.6
`
`-
`Tyr
`Trp
`Leu Asn
`Ser Gly
`Ser Gly
`Ser
`Val
`Ile
`Gly
`Ile
`Ser Met
`Gln
`Ser
`Thr Tyr
`Tyr
`Ser Gly
`Tyr
`Ser
`Tyr
`Lys
`Leu Glu His Met Gly
`Asn
`Ile
`Ser Asp Asp
`Tyr His Gly Thr
`Phe
`'fyr Val
`Lys Gly Leu Glu
`Gly
`Trp
`Ile
`Lys Arg Leu Glu Trp
`Ile
`Ala Ala
`Ser Arg Asn Lys Gly Asn Lys
`Tyr
`Thr
`Leu Glu
`Gtn Gly
`Gly Arg
`lie Asp
`Pro Ala Asn Gly Asn Thr
`Tyr
`Trp
`Ile
`
`Sequences for reovirus. 36-60, NEWM, and MPC603 are from lhe Protein Identification Resource [84]; lhe sequence for 87.91.6 is from (83].
`
`;:g
`;;o
`
`"' ~
`rn
`~
`~
`Ci
`~
`fi
`::i
`-<
`~
`8 s -<
`
`~
`
`. :D ...
`- ct
`z; ...
`• m
`r-
`
`~ -;z ,. .....
`
`v
`
`.,,,
`""
`
`13 of 18
`
`Celltrion, Inc., Exhibit 1054
`
`
`
`352
`
`T. KJEBER-EMMONS, E. GETZOFF, and H. KOHLER
`
`representative of the major antigenic determinant of 4C! l which interacts
`with Tl 5 (87] . Competitive inhibition assays have shown that the synthesized
`model tetrapeptide is capable of inhibiting 4Cll-T15 binding, though not as
`effectively as PC (87). Sequence analysis (88) and molecular modelling of
`4CI I indicates that a loose chemical mimicry of the dipolar character of PC
`localized in CD2 oftbe heavy chain may have a dominant role in determining
`the antigenic topography of the hybridoma. For a small haptenic antigen like
`PC, which is being mimicked by a comparatively large molecule like an
`antibody, mimicry may be mediated by appropriately placed contact residues
`that reside on different parts of the hybridoma. The mode of interaction of an
`antibody with a small hapten and with a large protein are different. So there
`may be multiple ways in which mimicry can be realized. Residues of anti-PC
`antibodies that interact with PC may be utilized in 4CI l binding, but other
`residues of the anti-PC antibodies that do not contact PC may now contact
`4Cl 1. Anti-PC antibodies with different sequences may not bind 4Cl I or may
`bind with very low affinity, due to the proximity of noncornplementary
`residues over a large surface area in the complexes.
`
`SUMMARY
`
`The recent crystal determination of a lysozyme-antilysozyme complex
`provides a three-dimensional prototype of the manner in which contacts in
`idiotype-anti-idiotype interactions may be realized [46]. Such interactions
`can be approximated by two complementary "flat" surfaces. Each IDR
`(autoantigenic locus) location might provide a particular recognition feature
`between two interacting partners. The combinatorial manner in which IDR
`domains are recognized by anti-idiotypic antibodies describe the repertoire
`of private and public (crossreactive) idiotopes of an antibody.
`Several interesting features emerge from consideration of the Ab contact
`residues in the crystal structure. First, framework residues are implicated in
`contacting tbe antigen: Thr 30 (FRI) of the heavy chain and Tyr 49 (FR2) of
`the kappa light chain. Both of these residues lie within predicted lDRs (22].
`Framework regions have recently been suggested to be involved in several
`anti-idiotypic systems (18, 45), although such regions have, in the past, been
`disregarded based solely upon sequence analysis. The surface variabilty
`analysis, which identifies the repertoire of complementary interacting sur(cid:173)
`faces, depicts the immunoglobulin as having more variability than generally
`
`14 of 18
`
`Celltrion, Inc., Exhibit 1054
`
`
`
`PERSPECTIVES ON ANTIGENICITY AND IDIOTYPY
`
`353
`
`thought. This variability may also extend to T cell receptors since T cell
`chains express an extensive surface variable repertoire similar to that of the
`irnmunoglobulin bght chains (Kieber-Emmons and Kohler, unpublished).
`Second , the D region plays a critical role in the generation of the anti(cid:173)
`lysozyme combining sites. Similarly, the D segment makes up the largest
`component of an IDR (22]. Third, while the CDR3 of the heavy chain
`contributes most to the antibody-lysozyme complex it is not the most
`surface-exposed (see Novotny, this issue). Nevertheless, surface variability
`analysis indicates that this region is generally immunodominant (22] which
`is also observed experimentally [33) . Together, these results indicate that
`perhaps certain JDR regions are intrinsically more antigenic.
`ldiotypic structures must be accessible for antibody recognition and
`binding. From a structural viewpoint, a single antibody molecule has a
`continuum or several d ifferent combining sites (89]. Subsequently, a single
`residue can be contained in several