J. Mol. Biol. (1991) 217, 133-151
`
`Structure, Function and Properties of
`Antibody Binding Sites
`
`I. Saira Mian1t, Arthur R. Bradwell2 and Arthur J. Olson1j
`
`1 Department of Molecular Biology
`Research Institute of Scripps Clinic
`10666 North Torrey Pines Road, La Jolla, CA 92037, U.S.A.
`2 Department of Immunology
`Medical School, Birmingham University, B15 2TJ, U.K.
`
`(Received 16 May 1990; accepted 19 July 1990)
`
`Do antibody combining sites possess general properties that enable them to bind different
`antigens with varying affinities and to bind novel antigens? Here, we address this question
`by examining the physical and chemical characteristics most favourable for residues
`involved in antigen accommodation and binding. Amphipathic amino acids could readily
`tolerate the change of environment from hydrophilic to hydrophobic that occurs upon
`antibody-antigen complex formation. Residues that are large and can participate in a wide
`variety of van der Waals' and electrostatic interactions would permit binding to a range of
`antigens. Amino acids with flexible side-chains could generate a structurally plastic region,
`i.e. a binding site possessing the ability to mould itself around the antigen to improve
`complementarity of the interacting surfaces. Hence, antibodies could bind to an array of
`novel antigens using a limited set of residues interspersed with more unique residues to
`which greater binding specificity can be attributed. An individual antibody molecule could
`thus be cross-reactive and have the capacity to bind structurally similar ligands. The
`accommodation of variations in antigenic structure by modest combining site flexibility
`could make an important contribution to immune defence by allowing antibody binding to
`distinct but closely related pathogens.
`Tyr and Trp most readily fulfil these catholic physicochemical requirements and thus
`would be expected to be common in combining sites on theoretical grounds. Experimental
`support for this comes from three sources, (1) the high frequency of participation by these
`amino acids in the antigen binding observed in six crystallographically determined
`antibody-antigen complexes, (2) their frequent occurrence in the putative binding regions
`of antibodies as determined from structural and sequence data and (3) the potential for
`movement of their side-chains in known antibody binding sites and model systems. The six
`bound antigens comprise two small different haptens, non-overlapping regions of the same
`large protein and a 19 amino acid residue peptide. Out of a total of 85 complementarity
`determining region positions, only 37 locations (plus 3 framework) are directly involved in
`antigen interaction. Of these, light chain residue 91 is utilized by all the complexes
`examined, whilst light chain 32, light chain 96 and heavy chain 33 are employed by five out
`of the six. The binding sites in known antibody-antigen complexes as well as the postulated
`combining sites in free Fab fragments show similar characteristics with regard to the types
`of amino acids present. The possible role of other amino acids is also assessed. Potential
`implications for the combining regions of class I major histocompatibility molecules and the
`rational design of molecules are discussed.
`
`t Present address: Sinsheimer Laboratory, Biology
`Department, University of California Santa Cruz, Santa
`Cruz, CA 95064, U.S.A.
`t _._A1uthor to "vhom reprint requests should be
`addressed.
`§Abbreviations used: CDR, complementarity
`determining region; MHC, major histocompatibility
`complex; FR, framework.
`
`0022-2836/91/010133-19 $03.00/0
`
`1. Introduction
`Antibodies are powerful recognition and binding
`molecules that the immune system employs to
`eliminate foreign molecules. Antibody binding sites
`are formed by six hypervariable loops or comple(cid:173)
`mentarity determining regions (CDRs§). The CDRs,
`133
`
`© 1991 Academic Press Limited
`
`1 of 19
`
`BI Exhibit 1114
`
`

`

`134
`
`I. S. Mian et al.
`
`three from each of the heavy and light chain vari(cid:173)
`able domains, are connected to a relatively invar(cid:173)
`iant /3-sheet framework (Alzari et al., 1988; Davies &
`Metzger, 1983; Capra & Edmundson, 1977; Wu &
`Kabat, 1970). Early analysis of a data bank of
`complete and partial sequences of 415 light and 197
`heavy chains demonstrated that CDRs are rich in
`(Kabat et al., 1977). The
`residues
`aromatic
`combining region represents only a small part of the
`antibody molecule, whose overall three-dimensional
`structure is highly conserved. Although, the pairing
`of light and heavy chains can generate some anti(cid:173)
`body diversity, most of it is generated by the
`somatic recombination of variable region gene
`segments (Yancopoulos & Alt, 1986; Wysocki &
`Gefter, 1989). Such genetic mechanisms yield anti(cid:173)
`bodies exhibiting extensive diversity in hyper(cid:173)
`variable loop sequences. This potential repertoire is
`estimated to be approximately 109 in mouse (Berek
`et al. , 1985). However, the initial repertoire that
`confronts an antigenic challenge is smaller than the
`potential repertoire, since it is restricted to the
`existing
`on
`expressed
`specificities
`antibody
`immunocompetent B cells at a point in time
`(Holmberg et al., 1986). This available repertoire
`can yield an apparently unlimited repertoire of
`antigen binding specificities and affinities.
`Although a single antibody has a unique three(cid:173)
`dimensional structure, biophysical and biochemical
`evidence indicates that it is multispecific or cross(cid:173)
`reactive (Richards et al. , 1975). This capacity to
`combine both with its inducing antigen and with
`antigens of similar or disparate structure augments
`the genetically determined antigen-binding capabili(cid:173)
`ties of antibodies. The extent of molecular comple(cid:173)
`mentarity between determinants on the antigen
`molecule and amino acid residues in the combining
`site determines the degree of antibody specificity.
`the
`is at
`therefore,
`Increased cross-reactivity,
`expense of specificity and affinity.
`An improved understanding of both antibody
`cross-reactivity and binding can be obtained by a
`interactions at the
`study of antibody-antigen
`atomic level. The role of residues in the definition of
`interaction with
`combining site structure and
`antigen can be assessed as a function of the chemical
`and structural properties of individual amino acids.
`First, we examine those characteristics that appear
`to be of general importance in antibody- antigen
`interactions. This is followed by a detailed study of
`the binding sites in six antibody-antigen complexes
`and four free Fab fragments of known three-dimen(cid:173)
`sional structure, and the much larger database of
`antibody sequences. Padlan (1990) has performed a
`similar, though not identical , analysis of antibody
`combining sites in general, and three anti-lysozyme
`antibody-antigen complexes in particular. On the
`basis of their propensity to occur in the combining
`sites and their greater exposure relative to those in
`the framework regions, he has suggested that these
`amino acids determine specificity. Our results and
`their interpretation lead us to conclude that Tyr
`residues may play more generally important roles in
`
`and
`binding
`interactions.
`
`non-specific
`
`antibody-antigen
`
`2. Physical and Chemical Properties
`of Amino Acids
`Since antibody binding sites are formed by six
`a highly
`supported on
`loops
`hypervariable
`conserved P-sheet framework, there is likely to be a
`bias towards amino acids that are generally found in
`non-helical regions of proteins. Figure 1 shows the
`normalized frequencies of occurrence of amino . acids
`in a-helix, /3-sheet and reverse turns in 66 globular
`proteins comprising 31 different conformations
`(Levitt, 1978). In these structures, the occurrence of
`Pro,G~,T~,~,Thr,Aw,~,A~,lli~dT~
`in a-helices is less frequent than random. Leu, His,
`Trp, Thr, Tyr, Phe, Ile and Val have a greater than
`random probability of occurring in /3-sheets; the
`same .is true for Thr, Tyr, Asn, Ser, Asp, Gly and
`Pro in reverse turns. Arg appears to be equally
`tolerated in all the secondary structures elements
`considered. In general structural terms, Tyr and
`Thr seem to be the most useful non-helix forming
`residues, since they could be positioned in either the
`strand or turn regions of the hypervariabli;i loops.
`The free energy of interaction between an anti(cid:173)
`body and its antigen is a function of both enthalpy
`and entropy. Non-bonded forces between the inter(cid:173)
`acting molecules include hydrophobic, hydrogen
`bond, van der Waals' and electrostatic interactions
`(for a review, see Fersht, 1985). In general terms,
`antibody combining Rite residues need to be as
`multifaceted as possible to accommodate the varied
`the
`features of
`stereochemical and electronic
`antigen. Hence, amino acids with non-polar (for
`example Leu, Ile and Val) and charged (for example
`Asp, Glu, Lys and Arg) side-chains would be of more
`limited usefulness than, for example, His, which is
`known to be capable of cross-linking sequentially
`distant but spatially close regions of proteins (Baker
`& Hubbard, 1984; l.S.M. & A.J.O ., unpublished
`results) . Similarly, the amides Asn and Gln would be
`generally more preferable than Asp and Gin, since
`the former pair are both hydrogen bond donors and
`acceptors whereas their- charged counterparts are
`only acceptors.
`If a positive charge is required in the antibody
`combining site, Arg would be more suitable than
`Lys because of its greater functional versatility; for
`example, Arg can form a larger number of hydrogen
`bonds than Lys. As a consequence of its planar
`nature and n-electron system, the terminal guanidi(cid:173)
`nium group of Arg often exhibits pseudo-aromatic
`behaviour by participating in most of the inter(cid:173)
`tru.e
`for
`catalogued
`previously
`actions
`aromatic-aromatic interactions (l.S.M. & A.J.O.,
`unpublished results). These interactions occur at the
`intersubunit interfaces of a number of oligomeric
`including viral coat proteins and a
`proteins,
`the photosynthetic reaction
`membrane protein;
`centre of Rhodopseudornonas viridis. The ability to
`form hydrogen bonds, hydrophobic interactions and
`
`2 of 19
`
`BI Exhibit 1114
`
`

`

`Antibody Binding Sites
`
`135
`
`1-Sr---...-----.------.----~--~--~
`
`2·0 .---------..-----,-----...----~---~
`
`1·4
`
`0·6
`
`w
`
`G
`
`M
`
`A
`
`Q
`
`K
`
`c
`
`G
`
`N
`
`EK Q
`
`c
`
`T y
`
`A
`
`w
`
`M
`
`05.__ _ _ ...__ _ _ _._ _ _ - - ' - - - - ' - - - - - ' - - - - '
`1·0
`1·2
`0·6
`Prot>abiHty a-tle!ix
`
`O·O '------~---~----'----~----'
`1·0
`0·6
`1·2
`0·8
`Prababillty '3-slleet
`
`(a)
`
`(cl
`
`2·0
`
`G
`
`K Q
`
`E.
`
`c
`
`w
`
`A
`
`M
`
`O·O '-------'-----'-----'------'------'-~---'
`0·8
`0·6
`04
`1·2
`
`Probatiility a-helix
`
`(b)
`
`Figure 1. Scatter diagrams , showing the normalized
`frequencies of occurrence of amino acids in IX-helix,
`in 66 globular proteins
`turns
`P-sheet and reverse
`comprising 31 different conformations (Levitt, 1978).
`/3-sheet.
`ex-helix
`forming
`(a) Probability of
`versus
`(b) Probability of forming IX-helix versus reverse turn.
`(c) Probability of forming /3-sheet versus reverse turn. The
`values represent the ratio of the fraction of residues of
`each amino acid that occurred in the secondary structure
`element to this fraction for all residues. To eliminate a
`bias towards structures that were determined more than
`once, the values were each weighted by a factor of
`l/(number of related proteins with same conformation).
`Normalized frequencies of I indicate random occurrence,
`than
`whilst > I
`indicate more frequent occurrence
`random. The actual point is marked by the bottom left of
`the I letter amino acid code: A, Ala; R, Arg; N, Asn;
`D, Asp; C, Cys; Q, Gin; E, Glu; G, Gly; H, His; I, Ile;
`L, Leu; K, Lys; M, Met; F, Phe; S, Ser; T, Thr; W, Trp;
`Y, Tyr; and V, Val.
`
`attractive electrostatic interactions between posi(cid:173)
`tively charged groups and aromatic rings permits
`Tyr and Trp to interact with structurally diverse
`antigens. Another functional advantage in locating
`Tyr and Trp in antibody combining sites is that,
`unlike amino acids having shorter side-chains, such
`as Asn and Ser, they lack the capacity to interact
`easily with other groups on the antibody surface but
`are ideally suited to interact with another molecule.
`The accommodation of charged areas on the
`antigen need not necessitate an antibody combining
`site possessing amino acids of complementary
`charge. Analysis of Arg, Lys, Glu and Asp side(cid:173)
`chains buried at the intermolecular interfaces of
`oligomeric systems indicates that oriented dipoles
`are usually preferred over countercharges in stabi(cid:173)
`lizing these buried residues (I.S.M. & A.J.0., unpub(cid:173)
`lished results). Thus, the peptide backbone and
`polar side-chains of hypervariable loop residues
`could be deployed to stabilize both negatively and
`positively charged regions . In some instances, this
`
`may be as effective as employment of formally
`charged amino acids: in cases of charge-charge
`interaction, the steric effects of neighbouring regions
`may prevent the formation of geometrically optimal
`ion-pairs such that the potentially available energy
`is not fully realized.
`The non-covalent association between antibodv
`and antigen requires the removal of water fro~
`interacting molecules.
`the
`surfaces buried by
`Antibody regions involved in this process Rhoukl he
`capable of tolerating both the polar and non-polar
`environments that exist before complex formation
`and upon antigen binding, respectively. Individual
`residues exposed on the surface of the free antibody
`can become completely or partially buried in the
`complex. In addition to residue amphipathicity,
`residue size might be a factor. There is a good
`correlation between the surface area of amino acids
`and their free energies of transfer from water to an
`organic phase (Chothia, 1974, 1975; Gelles &
`Klapper , 1978). A value of 1 A2 (I A= O·l nm) of
`
`3 of 19
`
`BI Exhibit 1114
`
`

`

`136
`
`I. S. Mian et al.
`
`G
`
`c
`
`M
`
`· 10
`
`N
`
`a
`E H
`
`w
`
`R
`
`·20 '-----'------'-----~--__,R..._._ _ _,
`150
`200
`70
`250
`100
`Surface aree. (A2}
`
`Figure 2. Comparison of the size of amino acids and the
`affinities of their side-chains for water. The surface area
`(Rose et al. , 1985) comprises the mean solvent access(cid:173)
`ibility for amino acid X in an ensemble of Gly-X-Gly
`tripeptides. The hydration potential (Wolfenden et al. ,
`1981) is the effective free energy of transfer from the
`vapour phase to dilute aqueous solution buffer at pH 7 of
`molecules having the structure R-H, where R is the side(cid:173)
`chain of each amino acid; for P (Pro), only the surface
`area is
`indicated, since no hydration potential was
`evaluated. Side-chains were modelled by the following
`compounds: A, methane; R , methylguanidine; N, aceta(cid:173)
`mide; D, acetic acid; C, methanethiol; Q, propionamide;
`E, propionic acid; G, H 2 ; H , 4-methylimidazole; I , isobu(cid:173)
`tane; L, butane; K, n-butylamine; M, ethylmethyl
`sulphide; F,
`toluene; S, methanol; T, ethanol; W,
`3-methylindole; Y, p-cresol; and V, propau.,. As a result of
`technical difficulties (Wolfenden et al., 1981), methyl(cid:173)
`guanidine (shorter than the side-chain of Arg by 2 methy(cid:173)
`lene groups) was employed to estimate the value for
`propylguanidine; this probably leads to the hydrophilic
`and hydrophobic nature of Arg being over- and under(cid:173)
`estimated, respectively.
`
`surface area gives a hydrophobic energy of
`25 cal/mol (1 cal = 4· 184 J: Chothia, 1974). Whilst
`these van der Waals' energies may be small
`compared to a hydrogen bond, when summed over
`the entire combining site they may be important in
`stabilizing the complex. Figure 2 compares the
`affinities of amino acid side-chains for water
`(Wolfenden et al., 1981) with the surface area of the
`entire amino acid (Rose et al., 1985). The classical
`groupings into small, large, hydrophobic and hydro(cid:173)
`philic amino acids are evident. With respect to
`amphipathicity, Ser, Thr, Tyr and Trp seem desir(cid:173)
`able residues to locate in antibody binding sites,
`since their side-chains are in the midrange of hydro(cid:173)
`gen potential values. The aromatic residues Tyr and
`Trp are also two of the largest and are capable of
`contributing significantly to the total intemction
`energy (Fig. 2).
`
`Figure 3. A diagram illustrating specificity and cross(cid:173)
`reactivity for a given antibody. The specific binding of the
`antibodies A and B to antigens A and B is a function of
`the high degree of complementarity between
`their
`molecular surfaces in terms of shape, size and functiona(cid:173)
`lity. (a) Cross-reactivity may arise as a result of structural
`similarity of epitopes between antigens A and A'
`(Richards et al. , 1975). A poor fit in one region may be
`compensated for by a good fit elsewhere. This could result
`in a sufficient number of short-range interactions to
`produce a stable antibody-antigen complex. Another
`cross-reacting antigen, A", may
`fit
`the antibody
`combining site in a slightly different way. (b) The anti(cid:173)
`body B may accommodate the related antigens B and B'
`if it is able to vary the stereochemical features of the
`combining site, i.e. if it is intrinsically pliable.
`
`In an antibody- antigen complex, the stabiliza(cid:173)
`tion energy gained from the various intermolecular
`forces must more than offset losses due to conforma(cid:173)
`tional entropy and conformational strain. The free
`energy of complex formation could therefore be
`maximized by minimizing the loss of conformational
`entropy upon association. It is known that a single
`antibody is able to combine with a spectrum of
`different antigens (Richards et al., 1975). Although
`such cross-reaction may occur either because the
`antigens share epitopes, or because the epitopes are
`sufficiently similar in shape to bind the same anti (cid:173)
`body (Richards et al., 1975), it could arise also if the
`topography of the combining site could be modu(cid:173)
`lated (Fig. 3) . Thus, antibodies could utilize amino
`acids whose side-chains were sufficiently structur(cid:173)
`ally and functionally flexible to permit them to alter
`the stereochemical features of the combining site
`with minimal loss of entropy. The potential im por(cid:173)
`tance of side-chain mot ion has been further high·
`lighted by a recent comparative study of known
`antibody structures and sequences (Chothia et al,,
`1989). It has been suggested that the number of
`main-chain conformations of at least five of the six
`loops appears to be limited. The adoption of a
`specific backbone conformation is believed to be a
`reflection of only a few key conserved residues in the
`loop or framework of the antibody (Chothia et al .,
`1989). This small repertoire of canonic&! structures
`would represent a reduction in the ispeotrum of
`specificity and affinity potentially ;wail11ble to the
`antibody binding site were the number of confor!'l'Ht ·
`tions proportion11l to the number of f!equences th a.t
`coi1ld be produced genetically,
`
`4 of 19
`
`BI Exhibit 1114
`
`

`

`Antibody Binding Sites
`
`137
`
`Table 1
`Preferred conformations of amino acid side-chains as described by their torsion parameters (Cody, 1985)
`
`Residue
`
`A
`R
`
`N
`D
`c
`
`Q
`
`E
`
`G
`H
`
`L
`
`K
`M
`
`F
`
`p
`s
`T
`w
`y
`
`v
`
`x11
`
`60
`-60
`60
`-60
`-60
`60
`60
`-60
`180
`-60
`60
`
`-60
`60
`60
`180
`-60
`180
`-60
`60
`-60
`180
`-60
`60
`180
`-60
`60
`±35
`60
`-60
`60
`-60
`60
`-60
`180
`60
`-60
`
`5; -19
`-15
`
`11
`-7; ±35
`-30
`
`-20
`
`O; ±35
`
`O; 25
`-25
`
`-15; -45
`
`-18; ±36
`
`-20
`±20
`
`±20
`
`±10
`5
`-25
`
`-10
`
`±20
`
`-11; -35
`
`Torsion parameters (°)
`
`x12
`
`x2'
`
`x22
`
`-120
`-120
`
`180
`180
`180
`
`180
`180
`0
`±5
`
`180
`180
`180
`180
`180
`
`60
`60
`180
`180
`180
`-60
`-60
`60
`180
`180
`180
`180
`90
`90
`90
`±35
`
`90
`90
`90
`90
`90
`
`180
`-60
`180
`
`180
`-60
`
`180
`
`x4
`
`±10
`±10
`
`180
`
`x3
`
`180
`180
`
`-15
`25
`±15
`±15
`±15
`
`180
`-60
`180
`180
`
`±25
`
`±20
`
`±10
`
`These were derived from crystallographic studies of amino acids and their derivatives. All atoms are numbered using the Greek
`;:12 = (N-C11-C"-Ci'2 ),
`x. 21 = (Ca-CP-C1-cb 1 ),
`;J; = (0---C-Ca-~~), ;: 11 = (~~-Ca-CY1-CY 1 )i
`the C1
`letter designations starting with
`3 = (CP-C'-C6-C'), X4 = (C1-C6-C'-C'). To account for the ring pucker in Pro, x
`22 = (C"-CP-C1-C62 ), x
`4 = (C'-C6-N-C") and
`x
`xs = (C6-N-C"-CP). The most frequently observed values are given first.
`
`The number of degrees of freedom of each amino
`acid side-chain can be approximated by examining
`the range and distribution of the observed confor(cid:173)
`mations (Table 1). Residues with short side-chains
`such as Ser and Thr lose little entropy when fixed
`upon antigen association, since they have only one
`and two variable side-chain torsion angles, respect(cid:173)
`ively. However, they project only a short distance
`from the surface of the antibody and so could not
`effect substantial changes in binding site topo(cid:173)
`graphy. Large residues can elicit great changes in
`the surface contours of the combining site, since
`they can sweep out large volumes of space. Residues
`with the largest surface area (Fig. 2) are Trp, Arg,
`Tyr and Phe-and of these, Arg has twice the number
`of variable torsion angles of the other three.
`Additionally, Arg is less suitable than the other
`aromatic residues because it is charged and there-
`
`fore requires a more restrictive interaction at the
`interface.
`Tyr represents a balance between the many
`different, though sometimes conflicting, desirable
`aspects we believe to be of general importance in
`antibody-antigen interactions. Thus, it would be
`expected to be the most common residue in anti(cid:173)
`body combining sites. Experimental evidence from
`antibody sequences and structures appears to verify
`these assumptions.
`
`3. Combining Sites in Known
`Antibody-Antigen Complexes
`Table 2 lists the antibody residues that bind
`antigen in six complexes whose structures have been
`determined by X-ray
`crystallography. The
`numbering scheme used follows the convention of
`
`5 of 19
`
`BI Exhibit 1114
`
`

`

`138
`
`I. S. Mian et al.
`
`Table 2
`Antigen binding residues observed in six crystallographically determined antibody-antigen complexes
`
`Residue position
`
`McPC603 HyHEL-5 HyHEL-IO Fab Dl.3
`
`Fab 4-4-20
`
`BI3I2
`
`Percentage composition
`
`Antibody- antigen complex
`
`FR2
`CDR2
`
`CDR3
`
`A. Light chain
`CDRI
`27D
`28
`30t
`3It
`32t
`34
`49
`50t
`53
`9lt
`92t
`93t
`94t
`95t
`96t
`B. Heavy chain
`FRI
`30t
`CDRl
`'ll+
`~· 1
`32t
`33
`35
`47
`50
`51
`52
`52At
`53t
`54t
`55t
`56
`57
`58
`95
`96
`97
`98
`99
`100
`lOOA
`lOOB
`100!
`
`FR2
`CDR2
`
`CDR3
`
`G (30)
`N (3I)
`N (32)
`
`y (50)
`Q (53)
`8 (91)
`N (92)
`
`y (96)
`
`T (30)
`Q {Ql \
`l.J \U..l./
`D (32 )
`y (33)
`
`y (50)
`s (52)
`
`y (53)
`s (54)
`s (56)
`
`y (58)
`W(98)
`
`N (30)
`y (31)
`y (33)
`
`D (49)
`
`W(90)
`G (9I)
`R (92)
`
`p (94t)
`
`w (33)
`E (35)
`w (47)
`E (50)
`
`s (55)
`s (57)
`T (58)
`N (59)
`G (99)
`
`y (101)
`
`H
`
`y
`
`y
`y
`
`F
`w
`8
`
`T
`G
`y
`
`w
`
`G
`D
`
`R
`D
`y
`R
`
`D (97)
`
`y (100)
`
`L (102)
`
`y (33)
`E (35)
`
`R (52)
`
`N (101)
`
`w (107)
`
`H
`
`y
`R
`
`8
`
`w
`
`w
`
`y
`y
`
`D
`D
`
`y
`
`G
`
`v
`
`p
`
`R
`
`A
`
`I
`s
`s
`G
`s
`y
`
`F
`y
`
`p
`F
`
`824, H20, Yl6, Gl4
`Dl7, N16, 8I6, Y9
`S25, N23, VI4, LIO
`S33, N28, TIS, H5
`Y68, N6, S4, W3
`A30, N24, H20, S9
`Y83, G9, F 3, 82
`Gl9, Dl5, Kll , RIO
`S38 , Kl 8, TI4, 8 ll
`W27, Y22, 816, Gl2
`N20, Yl8, 8I5, Dl2
`S36, EI4, HlO, Y9
`Sl 7, Nl6, Vl2, Ll2
`P63, HlO, LS, S6
`Wl9, Yl8, LIS, RIO
`
`S48, T42, K5, G 1
`S39, D34, :N7 , R6
`Y60, Fl 7, S6, T6
`Y40, G24, Wl6, A5
`H26, N21, Sl7, El5
`W94, Y3, L<l, F<l
`Yl9, El2, Al2, Rl2
`184, 89, Fl, Rl, Vl
`N27, Rl6, Sl3, Dl2
`P54, Nl9, S9, L4
`N29, G21, Sll, AlO
`S27, N26, G23, Dl6
`G56, Yl9, Sl5, D3
`825, T24, Gl6, Y lO
`T74, 112, K3, 82
`N20, Y l 9, L ll 5, El 3
`D28, Gl8, SlO, YlO
`Y26, Gl2, R9, D6
`Y32, Gl4, D6, L6
`Y32, Gl9, V7, S6
`G26, Y21, Sl6, E5
`S26, Yl2, Fl2, R7
`S23, Fl2, GlO, Y8
`Y29, Fl2, Dll, Sll
`W35, Y34, AlO, M6
`
`McPC603 (Satow et al., 1986), HyHEL-5 (Sheriff et al., 1987), HyHEL-10 (Padlan et al., 1989), Fab Dl.3 (Amit et al., 1986), Fab 4-4-
`20 (Herron et al., 1989) and Bl3I2 (Stanfield et al., 1990). Residue positions, framework (FR) and complementarity determining regions
`(CDR) are from Kaba.t et al. (1987). The sequential residue numbers of structures in the Protein Data Bank (Bernstein et al., 1977):
`McPC603 (File 2MCP), HyHEL-5 (lHFL) and HyHEL-10 (3HFM), are indicated in parentheses. At each of the positions known to bind
`antigen, the column headed Percentage composition indicates the 4 most common amino acids and their frequencies as calculated from
`the summary tables in Kabat et al. (1987). For example, one of the fluorescein binding residues in Fab 4-4-20 is His L27D (located in
`CDRl). Examination of the sequence database at this location indicates that Ser, His, Tyr and Gly are found in 74% of all sequences
`(24%, 20%, 16% and 14%, respectively) with the remaining 26% comprising the other 16 amino acids (each is < 14% of the total).
`Asn, Asp, Gin and Glu are underestimated because positions given as Asx or Glx by Kabat et al. (1987) (i.e. where there are uncertainties
`in sequence data), whilst being included in the total number of sequences, are not incorporated into totals for these 4 amino acids.
`Although positions L95 and L96 are listed separately, examination of t he crystal structures of McPC603, H yHEL-5 and HyHEL-10
`indicates that Leu L96 (102), Pro L95 (94) and Tyr L96 (96) are structurally equivalent (see also Fig. 5). In HyHEL-10, Ser L93 (93)
`a nd Trp L94 (94) are listed as being partially buried by t he antigen lysozy me (Padlan et al., 1989). In B l 3I2, Leu L27D, Cys H32, Trp
`H47, Gly H50, Ser H96, Ser H97 and Asp H98 are buried by, but not in van der Waals' contact with, the antigen (Stanfield et al., 1990).
`Also marked are residues in the hypervariable and framework regions that a comparative study of known antibody structures and
`sequences (Chothia et al., 1989) has suggested as being important (t} and mainly responsible (t} for generating the observed main-chain
`conformations of 5 of the 6 hypervariable loops (predictions for CDR3 of the heavy chain were not made). Amongst all the antigen
`binding residues in these complexes, only Pro L95 (94) of HyHEL-5 belongs to the group of key residues. Other positions that are
`mainlv responsible for producing the canonical structures occur at the followimr nositions. Lie-ht chain: 2 1114. i.e. out of a total of 4
`classe~ in ~hich this is i~portant for the observed conformation , only in canonic;;:-! ; tructure nu~ber 1 is this, ~ r~sidue that is a primary
`determinant of the loop main chain conformation), 25 (1/4), 29 (1/4), 33 (1/4), 48 (1/1), 64 (1/1), 71 (1/4), 90 ([1, 2, 3]/3), 95 ([l, 3]/3).
`Heavy chain: 26 (1/2), 27 (1/2), 29 (1/2), 34 (1/2), 52a (2/4), 55 ([l, 4]/4), 71 ([2, 3, 4]/4).
`
`6 of 19
`
`BI Exhibit 1114
`
`

`

`Antibody Binding Sites
`
`139
`
`Kabat et al. (1987) with L (light) and H (heavy)
`prefixing
`the position number
`(the sequential
`residue numbers are given in parentheses where
`available). The complexes are McPC603 (Satow et
`al., 1986: murine myeloma
`lgA-K), HyHEL-5
`(Sheriff et al., 1987: BALB/c murine monoclonal
`antibody, IgGl-K), HyHEL-10 (Padlan et al., 1989:
`BALB/c murine monoclonal antibody, IgGl-K),
`Fab Dl.3 (Amit et al., 1986: BALB/c murine mono(cid:173)
`clonal antibody, IgGl-K), Fab 4-4-20 (Herron et al.,
`1989: BALB/c murine monoclonal antibody,
`IgG2a-K) and Fab Bl312 (Stanfield et al., 1990:
`murine monoclonal antibody, IgGl-K). McP0603
`and Fab 4-4-20 bind small haptens (phosphocholine
`and fluorescein, respectively), Bl312 binds a syn(cid:173)
`thetic 19 amino acid residue peptide (a homologue
`of the 0-helix of myohaemerythrin) and
`the
`remainder bind different regions of the same large
`protein antigen (lysozyme). Because of the limited
`number of X-ray structures of antibody-antigen
`complexes, Table 2
`includes
`the frequency of
`different amino acids obtained from the more exten(cid:173)
`sive antibody sequence database (Kabat et al.,
`1987). These potential binding positions are
`obtained by projecting the information from the
`structurally determined interface residues to those
`same locations in the larger sequence database. For
`example, position L91 binds antigen in all the struc(cid:173)
`turally determined complexes considered and at this
`location in other antibody sequences, the four most
`common amino acids are Trp, Tyr, Ser and (Hy.
`In spite of differences in the stereochemical
`features of the antigen (the anti-lysozyme anti(cid:173)
`bodies bind to different regions of the protein), there
`appears to be an overall bias in the types of amino
`acid found in the combining sites. At both the
`known and potential binding positions, Tyr, Trp ,
`Ser and Asn are the most common residues that
`interact with antigen. As outlined in the previous
`section,
`these four, particularly Tyr and Trp,
`possess structural and functional characteristics
`that are highly desirable for antibody binding sites.
`Details of the combining region of HyHEL-5 and
`McP0603
`(Fig. 4) demonstrate
`the extensive
`binding repertoire and versatility of Tyr and Trp in
`accommodating both small and large ligands. In the
`remaining
`complexes
`(HyHEL-10, Fab Dl.3,
`Fab 4-4-20 and Fab Bl312), these residues also
`employ both the aromatic ring system and hydro(cid:173)
`gen-bonding atoms in interacting with the hapten or
`antigen (Padlan et al., 1989; Amit et al., 1986;
`Herron et al., 1989; Stanfield et al., 19!l0).
`Table 2 indicates that the concentration of Tyr is
`highest in the heavy chain. This may account for
`the possession of antigen-binding affinities by
`complete variable heavy domains (Ward et al.,
`1989). CDR3 of the heavy chain contains 19 amino
`acid positions, including a frequently inserted eight
`residue loop (Kabat et al., 1987). Tyr comprises
`25 % of the total residues; in 11 of the positions it is
`the most common residue and many individual
`sequences
`contain
`consecutive Tyr
`residues,
`including five in two sequences (human heavy chain
`
`CDR3 of proteins OU and WOL; Kabat et al., 1987).
`Chothia et al. (1989) were unable to make predic(cid:173)
`tions about the conformation of this loop. Another
`observation from the sequence data (Kabat et al.,
`1987) is that Tyr and Trp frequently alternate with
`small amino acids such as Gly, Ala and Ser. This
`pattern of residues might allow maximum mobility
`of Tyr and Trp. In HyHEL-5, the internally
`directed residue, Met L33(32), is located between
`Tyr L32(31) and Tyr L34(33) and permits the side-·
`chains of the flanking aromatics to be exposed and
`potentially mobile. This Met is one of the residues
`responsible for stabilizing canonical structure 1 of
`the light chain CDR.l (Chothia et al., 1989).
`The number of locations involved directly or in(cid:173)
`directly in antigen binding that are additional to
`those observed in the structures examined should be
`relatively
`small. Nuclear magnetic
`resonance
`spectroscopy studies of three different antibodies
`raised against the same 14 amino acid residue
`peptide antigen (Anglister & Zilber, 1990) indicate
`that the aromatic residues that interact with
`antigen are all situated at positions given in Table 2
`(His L31, Tyr L32, Phe L96, Tyr H32, Trp H50,
`Trp HlOOA). The same is true for all but one posi(cid:173)
`tion (Lys HlOOc) believed to be potentially involved
`in hapten binding in the preliminary crystal struc(cid:173)
`ture of Fab 36-71, an anti-arsonate monoclonal
`antibody (Rose et al., 1990). The other presumed
`binding positions are Phe L32 , Phe L50, Phe L96,
`Asn H35, Trp H47 , Tyr H50 and Glu H96 . Figure 5
`shows all residues observed to bind antigen in
`McPC603, HyHEL-5 and HyHEL-10, and two
`potential supplemental positions, H59 and H94.
`H59 is implicated by studies on anti-morphine
`(Miller & Glasel, 1989) and anti-common acute
`lymphocytic leukaemia antigen (Kudo et al., 1985)
`antibodies. In addition, this is the only position that
`is highly conserved amongst all six CDRs; Kabat et
`al. (1987)
`indicated that 96% of all antibody
`sequences have Tyr at H59. The importance of H94
`is supported by the observation that changes in
`affinity of variant anti-digoxin antibodies could be
`accounted for by a mutation at this position (Panka
`et al., 1988).
`As the number of crystallographically determined
`complexes with greater diversity in the nature of
`antigen increases, the types of amino acid