J. Mol. Biol. (1983) 167, 661-692
`
`Structure of a Novel Bence-Jones Protein (Rhe)
`Fragment at 1·6 A Resolution
`
`W. FUREY JR, B. C. WANG, C. S. Yoo AND M. SAx
`
`Biocrystallography Lab1Yratory
`Box 12055, V.A. Medical Center, Pittsburgh, Pa 15240, U.S.A.
`and Department of Crystallography
`University of Pittsburgh, Pittsburgh, Pa 15260, U.S.A.
`
`(Received 13October1982, and in revised form 3February1983)
`
`The crystal structure of Rhe, a ,\-type Bence-Jones protein fragment, has been
`solved and refined to a resolution of l ·6 A. A model fragment consisting of the
`complete variable domain and the first three residues of the constant domain
`yields a crystallographic residual Rr value of 0·149. The protein exists as a dimer
`both in solution and in the crystals. Although the "immunoglobulin fold" is
`generally preserved in the structure, there are significant differences in both the
`monomer conformation and in the mode of association of monomers into dimers,
`when compared to other known Bence-Jones proteins or Fab fragments. The
`variations in conformation within monomers are particularly significant as they
`involve non-hypervariable residues, which previously were believed to be part of a
`"structurally invariant" framework common to all immunoglobulin variable
`domains. The novel mode of dimerization is equally important, as it can result in
`combining site shapes and sizes unobtainable with the conventional mode of
`dimerization. A comparison of the structure with other variable domain dimers
`reveals further that the variations within monomers and between domains in the
`dimer are coupled. Some possible functional implications revealed by this coupling
`a.re greater variability, induced fitting of the combining site to better accommodate
`antigenic determinants, and a mechanism for relaying binding information from
`one end of the variable domain dimer to the other.
`In addition to providing the most accurate atomic parameters for an
`immunoglobulin domain yet obtained, the high resolution and extensive
`refinement resulted in identification of several tightly bound water molecules in
`key structural positions. These water molecules may be regarded as integral
`components of t he protein. Other water molecules appear to be required to
`stabilize the novel conformation.
`
`1. Introduction
`Rhe is a ,\-type Bence-Jones protein fragment consisting of a complete variable
`domain and the first three residues of a constant domain. The protein exists solely
`as a non-covalent dimer both in solution and in the crystalline state. The crystals
`are orthorhombic, with space group P212 12 and cell constants a= 54·63(2) A,
`b = 52·22(3) A and c = 42·62(3) A. The unit cell contains four monomers (114
`661
`
`0022-2836/83/190661-32 $03.00/0
`
`© 1983 Academic Press Inc. (London) Ltd.
`
`1 of 32
`
`BI Exhibit 1125
`
`

`

`W. FUREY ET AL.
`662
`residues each) and approximately 51 % solvent by volume. The crystallization
`conditions have been reported (Wang & Sax, 1974), as has a preliminary analysis
`(ex-carbon level) of the structure at 3·0 A resolution (Wang et al., 1979). In the 3 A
`analysis, several aspects of the structure were described, the most significant
`being the discovery of a mode of association of variable domains into dimers
`different from that observed previously in either Bence-Jones proteins or Fab
`fragments. This novel mode of dimeriza.tion is correlated with a change in
`conformation within monomers, and does not represent merely a rotation of one
`domain relative to the other. Furthermore, the segment within each monomer
`that changed conformation consists solely of non-hypervariable residues.
`Previously, these residues (sequence numbers 36 to 50) were believed to be part of
`a "structurally invariant" framework common to all immunoglobulins.
`In order to determine factors responsible for this novel conformation, it is
`necessary to extend the resolution beyond the 3 A level and to complete the
`chemical sequencing of the protein. Details of the phase extension and refinement
`to 1·9 A resolution were reported (Furey et al., 1979), but no attempt was made to
`interpret the structure, since this was merely an interim step toward a detailed
`analysis at l ·6 A resolution. The current high-resolution analysis confirms the 3 A
`structure interpretation and indicates that, although the conformational change
`involves non-hypervariable residues, it is very likely ca.used by interactions with
`residues in the first and second hypervariable regions. The analysis also suggests a
`role for solvent in stabilizing the novel protein conformation. An alternative
`interpretation of the data suggests a plausible trigger mechanism for relaying
`information from the combining site end of variable domain dimers to the switch
`regions connecting them to constant domains.
`
`2. Experimental Procedures
`The X-ray data. collection and reduction process has been described in detail (Furey et
`al., 1979), so only aspects reflecting the quality of the data. need be stated here. Rhe
`crysta.ls diffract X-rays extremely well, so it was possible to collect all data to l ·6 A
`resolution from l crystal with only a. 15% decrease in standard reflection intensities. In
`addition, the moderate size of the unit cell enabled collection of diffractometer data by the
`8-28 scan method commonly used in small molecule crystallography. The overall quality of
`the data. set is reflected in the fact that over 75% of all reflections to 1 ·6 A resolution are
`considered "observed", even by the rather stringent l/o(I) > 3 criterion. The percentage of
`observed reflections as a function of resolution is given in Ta.ble l. Note that the data were
`collected at room temperature with a relatively low power (0·5 kW) X-ray source, hence it
`should be possible to collect even higher resolution data if low temperature techniques are
`applied and a more powerful (1·5 kW) X-ray tube is used.
`Extending the procedure described in the 1·9 A paper (Furey et al., 1979), refinement was
`resumed after including the additional X-ray data. to l ·6 A resolution. The 1 ·9 A refined co(cid:173)
`ordinates (Rr = 0-28, Rr = (.EllF0l- IFclll/l.'1F01) served as initial parameters. Since chemical
`sequence information was not available, electron density maps (either 2F0 -F. or F 0 - F.)
`were examined frequently. Residues to be examined were always deleted from the phasing
`process before map calculation. Occasionally, errors in amino acid sequence or atomic
`positions were indicated by the maps. The errors were corrected by non-interactive
`computer graphics (Furey et al., 1979), optical compare.tor techniques (Richards, 1968) and,
`in the last stages, interactive computer graphics. In all cases after sequence changes were
`implemented, restrained reciprocal space refinement (Hendrickson & Konnert, 1978) was
`
`2 of 32
`
`BI Exhibit 1125
`
`

`

`STRUCTURE OF BENCE-JONES PROTEIN Rhe
`
`663
`
`TABLg l
`Observed reflections as a furu;tion of resolittion
`
`dmln
`(A)
`
`3·0
`2·4
`1·9
`1·6
`
`Total
`
`2667
`5091
`10,073
`16,665
`
`Observed if I > 3 u(/)
`
`No.
`observed
`
`%
`(in range)
`
`%
`(cumulative)·
`
`2582
`4781
`8975
`12,842
`
`96·8
`90·7
`84·1
`58·7
`
`96·8
`93-9
`89·1
`77·1
`
`t'esumed. The computation wa.s performed with the space group general array processor
`version (Furey et al., 1982) of the Hendrickson-Konnert program.
`The next step in the refinement process was to introduce individual isotropic thermal
`factors for each of the non-hydrogen atoms, and to include solvent molecules in the model.
`Despite the obvious presence of water molecules in some of the early maps, solvent atoms
`were not incorporated into the model until late in the refinement process (Rp = 0·23) to
`aNoid mistaking erroneously sequenced side-chain atoms for water molecules. Wate1·
`oxygen positions were obtained from difference electron density maps by scanning through
`the largest peaks and determining the shortest distance between eaeh peak and all atoms in
`the current model. If the shortest distance was between 2·3 and 3·4 A and the model atom
`involved was capable of forming hydrogen bonds, a water oxygen ·was added at the peak
`position. The new model wa.s then subjected to several cycles of least-squares refinement.
`The procedure was iterated several times and, after discarding water molecules that moved
`far from their original locations, resulted in the inclusion of 186 water oxygen atoms (102
`partially occupied). The mean electron density at the water sites is l ·26 e/A 3 . The
`estimated error in the electron density function is -0·20e/A 3 (Cruickshank, 1949), hence
`the water molecules should be reasonably well-determined. Thermal and occupancy factors
`for the water molecules ranged from 5 to 68 A 2 and 0·30 to I ·00, respectively. Hydrogen
`atom contributions for hydrogen atoms in the protein were included in all structure factor
`calculations, but their positions were recomputed every few cycles rather than refined.
`At this time, partial sequence information became available (W. Brown, personal
`communication) for comparison with the X-ray model. There were 11 discrepancies, all of
`which were resolved in the model by adding, deleting or changing the atomic type of, at
`most, 2 atoms/residue. The chemical sequence is now known except for distinctions within
`the pairs, Gin, Glu and Asn, Asp. Since these residues are isosteric and each can form
`hydrogen bonds, they cannot always be distinguished at l ·6 A resolution by X-ray data.
`Therefore acid/amide distinctions in the model were based mostly on the sequences of other
`.\-type variable domains (Ka.bat et al., 1977). In favorable cases, however, obvious
`hydrogen bonding partners and/or the chemical composition in short peptides enabled a
`choice to be made. The amino acid sequence in the final model is given in Table 2. When
`compared with the sequence deduced from the 3 A map, significant discrepancies were
`revealed, although the original interpretation was reasonable (60% correct).
`The final model contains 1019 non-hydrogen atoms, 833 from protein (all fully occupied),
`and the 186 water oxygen atoms. The model fits into the electron density extremely well,
`a.s indicated in Fig. I. The mean isotropic thermal factor is 16·5 A 2 when a.II a.toms are
`included, and 12·5 A 2 for protein atoms only. For solvent atoms, the mean Bt is 34·2 A 2
`• A
`histogram indicating the distribution of thermal factors for the protein is given in Fig. 2.
`The final crystallographic residual RF value is 0·149 for the 12,763 observed reflections with
`d spacings ranging from I 0 to l ·6 A. The R factor is plotted as a function of resolution in
`t B (isotropic thermal factor) = 8"202
`, where O is the root-mean-square amplitude of atomic
`vibration.
`
`3 of 32
`
`BI Exhibit 1125
`
`

`

`664
`
`W . FUREY ET AL.
`
`TABLE 2
`Amino acid sequence of Be:nce-J<Ynt,s protein Rhe (V1)
`
`7
`p
`
`8
`p
`
`Residue
`13
`12
`10 11
`9
`19 20
`1
`18
`17
`6
`14 15 16
`2
`No.
`3
`4
`5
`s
`v
`I
`P G Q R V T
`G T
`Q
`T
`Code E
`L
`S
`A
`S
`37 38 39 40
`33 34 35 36
`21
`30 31 32
`23 24 25 26 27 28 29
`22
`No.
`y Q Q v
`G S N
`I
`T G S A T D
`I W
`S V
`C
`Code S
`~ ~ U % ~ ~ a @ 00
`ITT
`57 58 59 60
`41
`53 54 55
`No.
`52
`56
`I Y Y
`G K A P K L L
`p
`N
`S G V S
`Code P
`D L L
`66 67 68 69 70 71
`72 73 74 75 76 77 78 79 80
`62
`61
`No.
`63 64 65
`I S G L E
`A S L A
`S K S G T S
`S A
`F
`Code D R
`93 94 95 96 97 98 99 100
`No.
`85
`92
`86 87 88 89 90 91
`81 82 83 84
`w
`N D S L D E P G
`Code S E D E
`D Y Y C A A
`A
`IOI 102 103 104 105 106 107 108 109 I 10 111 112 113 114
`No.
`Code F G G G T K L T V L G Q P K
`
`One-Jetter amino acid code: A, alanine ; C, cysteine ; D, aspa.rtate ; E , glutamate ; F , phenylalanine ;
`G, glycine ; I , isoleucine; K , lysine; L , leucine ; N, asparagine ; P . pr'Oline ; Q, glutamine ; R , &crginine; S,
`serine ; T , threonine; V. valine ; W , tryptophan ; Y, tyrosine; 0 , water.
`
`Fig. 3. Throughout the refinement, the protein model was restrained with respect to
`stereochemistry. resulting in a. _final model with root-mean-square deviations from " idea.I"
`bond distances of 0·024 A. The root-mean-square deviation from planarity for the
`appropriate atomic groups (including peptide links) is 0·020 A. The last refinement cycle
`resulted in a. root-mean-square shift in atomic positions of 0·006 A with a. corresponding
`estimated standard deviation of 0-0ll A. The latter is most certainly a.n underestimate of
`the true error in atomic positions, which we estimate to be -0·04 to 0·08 A by the method
`of Luzza.ti (1952). Mean electron densities for the main-chain a.toms, N, Ca, C and 0 a.re
`2·89, 2·24, 2·44 a.nd 3·17 e/A 3 , respectively.
`
`3. The Molecular Structure
`(a) The monomet'ic subimit
`The Rhe monomer consists of nine strands, as is the case for all immunoglobulin
`variable domains of known structure. A stereoscopic drawing of the ex-carbon
`structure is given in F igure 4. If we neglect the N-terminal residue, which is
`highly disordered and cannot be located accut'ately, the structure begins with a. P(cid:173)
`turn involving residues 2 to 5. The nine strands contain residues 6 to 12, 17 to 24,
`33 to 39, 44 to 50, 54 to 57, 60 to 67, 72 to 79, 84 to 92 and 97 to 114. All strands
`are in an extended conformation; however, there is one short helix of roughly l ·5
`turns (residues 25 to 32) connecting strands 2 and 3.
`As seen in Figure 4, the nine strands a.re connected by ten turns, resulting in a
`pair of antiparallel ti-sheets. The first, designated sheet A, contains strands 2, 6
`and 7, while the second (sheet B) contains strands 3, 4, 8 and 9. The two sheets a.re
`also connected by a disulphide bridge between Cys22 (strand 2) and Cys89
`(strand 8), and by a salt-bridge between Arg62 and Asp83. Only strands 1 and 5
`are isolated from the sheets, although segments of strands 3 and 4 eventually
`break away from sheet B. This latter feature ha.s important consequences with
`regard to the mode of domain-domain association upon dimerization. In previous
`
`4 of 32
`
`BI Exhibit 1125
`
`

`

`STRUCTURE OF BEN CE- JONES PROTEIN Rhe
`
`665
`
`I bl
`
`( c)
`
`FIG. I .
`
`5 of 32
`
`BI Exhibit 1125
`
`

`

`666
`
`W. FUREY ET AD.
`
`(d)
`
`Fw. l. Stereoscopic projections of some typical residues superimposed on the 1 ·6 A electron density
`map illustrating the fit. The electron density contours begin at 0·6 e/A 3 and increa.se in steps of0·6 e/A 3
`.
`Note that all rings, even prolines. have " holes" in their centers. Also note the electron density for a
`water molecule near the OH group in {d). C.oefficients for the electron density synthesis were
`F 0 exp (i~0) with the illustrated atoms omitted from the phasing.
`
`descriptions of the immunoglobulin fold (Amzel & Poljak, 1979 ; Davies et al. ,
`1975), both three and four-strand structures a.re mentioned, but with definitions
`opposite to those given above. Usually, strands 1, 2, 6 and 7 a.re referred to as the
`four-strand sheet with strands 3, 8 and 9 forming the three-strand sheet. As will
`be shown la.ter, the hydrogen-bonding pattern in Rhe suggests that strand 1 is
`much more closely related to sheet B , and that it should not be considered a
`member of the "four-strand" sheet (A). Surprisingly, the corresponding hydrogen-
`
`"
`
`40 • ' •• .. •• .. .. .. •• ..
`,\ 0 u •• •• .. .. .. .. .. .. ••
`~·\) .. ..
`
`ft I
`tt I
`tft
`ttf
`t • t
`ttt
`t t l I
`flt I
`
`,,, ••• .....
`
`tt•• •• ttt
`
`I ft tt
`f t
`t f f fl , •• _.,
`t f t t
`f t t l I
`l tt"ff l t f l f
`
`••ttttt lt ttttttt t t t t
`
`. ............... £•••• I ··•••11•
`
`' t • • •
`
`ta• t
`l t t ft t
`I l t t •t
`l
`
`' 1 ttttt
`
`1
`I
`
`'
`
`tt t l t t ttt C 1
`
`t fttttt t-tttt"lttt[ttt t '' • ttttttt 111 • ttt • tt t tcttttt • tttttt(tt( tttttttttttt
`
`tt•-.t(tttt( ttt ' t tttttCt
`
`••• • t 1 t•• r.••" ttttt•••••••••C.•••••• • •••••••• ttt.tt'ttU •••• ••••••• • •••• ••E•• C• • ••• •••••t t I •£-.••••E••••l • tttataatuEt
`ttt • t • ttt Ctttt tttttattttttttEtttttttttt"tttttttttte tftUttttttttt • tttttt( tt( attttttttttttt(l"tttl( tn:e£t UUtt t t tt[t
`••• I
`.
`
`r~VI TC>r'f'SASG I If.OW r I SC fGSA f [lJ c.s11sv ' "YOOVJ•Gt\Af-h l l . YYHUlf..lf"SGYSOf<F'SASJl.SGT!i"Sl "I ~LCS£0E#llffYYCAAW NhSl h(f"Gf GCG IM IVLGOF-..
`•o
`
`:JO
`
`60
`
`Rt!HIJUI:.
`
`10
`
`AO
`
`100
`
`11 0
`
`Fro. 2. Histogram of mean thermal factors (B) for each residue. ( *) Main chain ; ( !) side chain : E ,
`main &nd side-chains equal.
`
`I
`o
`
`'
`I
`
`'' '
`'
`'
`'
`
`'
`"
`l
`
`l
`
`I
`I
`j
`
`I
`
`fl t
`I t t
`I
`I tt I
`I tt !
`l f t I
`I 1 I I t t
`I
`l f t l . f t t t t l l
`l f f lttt.t t tl
`
`I
`I
`
`I ttt I I
`I
`l f l l l l f t f f l
`I
`t •I 1 1
`l f f f t f I
`
`I
`I !
`
`"
`I
`
`I
`
`f I
`I
`f
`
`t t
`f *l' I
`ft'l t '
`l f t [ •I
`lf ff ( 81
`
`I
`
`a
`t
`I
`• t. t
`ll t
`St. t
`t r t
`I
`l tt[ •
`
`I
`I
`I t I
`
`""•••t•
`
`111111 1 t • ttttt•tttttt( t
`
`[ ttttttttttt •
`
`11 ttt£ tt •
`
`t •• 1 • tttEt
`
`'" .. '
`...
`...
`. ..
`. ..
`. ' '
`' ....
`... ..
`,, . ..
`' ....
`"'' •••"•••••••£• C•••••••••"
`··••£•••
`·········
`·············••E•••• I
`•••• _. • • , . , , ····················••C••£••·········· ·••••E•• .... '••••••£•
`••• •• • •••t .. •• u•••••u•••uC•••••• • •••••••••u • • tt••••••••••••• ••••••t••E••••••••••••• •c•••••E••••E•••. , .. .._.,,,r,
`I . ttf' • ••••••••••• •••••tE:•·-···· •• __. ,_. •••••••••••••••••••••••• I • • • • ••E••C• •••••••••••• •C•*-• ••C••••C•••••••••-'•£•
`u • • t • • ••C • •••• • • • ••• • .-. ••• tE ••••-••tu••••••••••••••••••••••••••••••• ••C••E• •••••• ••• ••••E•••••E••••C•••••• '-•••«£•
`• • • • •' • • •C• • • • • • • • • •" • • •• ••£•••ut•_.•, .___,, •• ••• •• •• • •• •• •••••••'-• ••••••£• •t• ••• ••• ''-' ••••Ee:-••••£ • '-'-•E•••••• ._._ .• , • E •
`.
`.
`.
`.
`.
`.
`.
`
`6 of 32
`
`BI Exhibit 1125
`
`

`

`STRUCTURE OF BENCE-JONES PROTEIN Rhe
`0·641
`
`667
`
`0·56i
`
`0·48
`
`(}40
`
`(): 0·32
`
`0·24
`
`0·16
`
`0-08
`
`0·00+--~--~---~--~-~
`0·10 0·20
`0·30 0·40 0·50 0·60 0·70
`2 sin en ..
`Fie. 3. Plot of R facto1· versus resolution.
`
`bonding diagrams for other immunoglobulin domains (Amzel & Poljak, 1979)
`agree with that for Rhe, yet the sheet designations are different.
`In terms of the variability of amino acid sequence, residues 23 to 35 represent
`the first hypervariable region, while the second and third hypervariable regions
`
`FIG. 4. Rhe monomer, ('(-carbon skeleton.
`
`7 of 32
`
`BI Exhibit 1125
`
`

`

`668
`
`W. FURE Y ET A/,.
`
`correspond to residues 51 to 57 and 90 to 100, respectively. All other residue
`positions usually show much less variation in sequence for ,\-type V Lt domains.
`
`(b) Main chain torsion angles a.nd tttms
`In Figure 5 the distribution of q,, ip angles is shown in the usual Ramachandran
`plot, with "allowed"
`regions marked according · to Ramakrishnan &
`Ramachandran (1965). The most striking feature of the plot is the small number
`of non-glycine residues outside allowed regions,
`indicating a very stable
`conformation. There are only three violations, with extenuating circumstances
`explaining t he occurrence of each. For example, the abnormal i/J value for Asp28 is
`readily explained when neighboring residues are also considered. From
`Figure 6(a), it is clear that any change in i/J would weaken hydrogen-bonding
`interactions with the main chain NH groups of Ser31 and Asn32. A change in i/J
`here would also affect the direction of the NH bond of Ile29. Since this group is
`already ideally oriented for hydrogen bond formation to the side-chain carboxyl
`group of Asp28, any change could only destabilize the system. There are even
`more compelling reasons for the unusual angles in Asn52. From Figure 6(b), it is
`clear that any change in q, would weaken the hydrogen-bond interactions between
`NH of Asn52 and 0 of Val34, and between main chain atoms 0 of Tyr51 and NH
`of Asp53. Likewise, any change in ip would weaken the hydrogen bond between
`main chain 0 of Asn52 and water oxygen 0118. A change in i/J would also
`weaken the NH-0 hydrogen bond already mentioned between Asp53 and Tyr5l.
`:~ ~:-~====:::;;:::;::~\-\-l:---r-----.-,,---G---------,
`...
`,,, .
`r
`:
`i
`:
`i '
`i. .. +G
`...... J G
`
`60 ::.
`
`\
`'',,)((
`
`\\.
`
`I
`
`I
`
`I
`
`. ----------·
`/
`\
`*
`\
`........ ,
`__ ....... ~. ':
`:
`................ __________ . .:,
`
`O t---~----~~~----1------..,.,...,.....--------~
`G
`52
`
`'
`•
`
`-60
`
`-120
`
`)1€028
`
`G
`
`I
`I
`
`)l(N52
`
`G
`
`,------------~
`- 120
`-60
`
`-180
`
`60
`
`120
`
`180
`
`F10 . 5. The Ra.machandran plot for Rhe. G indicates glycine residues.
`t Abbreviation used: V L• variable domain of light chain.
`
`8 of 32
`
`BI Exhibit 1125
`
`

`

`STRUCTURE OF BEN CE-JONES PROTEIN Rhe
`
`669
`
`(o)
`
`e<Jll7
`
`.0117
`
`(b)
`
`Fm. 6. Environments around (a) Asp28 and (b) Asn52.
`
`The only other abnormal conformation occurs for Ser2. Since the torsion angles
`for Ser2 are likely to be adversely affected by the disorder in residue 1, and in any
`event a.re just slightly outside of a.n allowed region, no further comment is
`required.
`In Table 3 the ten sharp turns in Rhe a.re classified according to the scheme
`introduced by Venka.ta.chala.m (1968), and extended by Lewis et al. (1973). The
`fa.ct that all three turns of type II have Gly in position 3 is in excellent agreement
`with the findings of Venkata.cha.lam (1968) and Crawford et al. (1973). They claim
`that for a 1-4 hydrogen bond to remain intact in a type II bend, position 3 must
`be glycine. In a more recent study on the frequency of occurrence of ea.ch amino
`a cid in various positions of ~-turns in 29 proteins, Chou & Fasma.n (1977) found
`that the most frequently occurring amino acid residues a.re Pro, Gly, Asn, Asp and
`Ser. The structure of Rhe is consistent with their findings in that of the 40
`residues involved in the ten turns, 23 (57%) conform to those five amino acid
`
`9 of 32
`
`BI Exhibit 1125
`
`

`

`tl70
`
`W . FUREY E7' A/,.
`
`TABLE 3
`fJ-Tum -pat·ameters
`
`Residue no. Sequenc.-e
`
`ef.2 (.)
`
`Y,2 (. )
`
`4>3 (. )
`
`Y,3 (. )
`
`d ('(\1-('-<>4 d OJ-N4
`{A)
`(A)
`
`2-5
`13- 16
`40-43
`51- 54
`56-59
`6<HJ3
`68- 71
`80-83
`93- 96
`94-97
`
`SVLT
`TPGQ
`\'PGK
`YNDL
`PSGV
`SDRF
`SGTS
`ESED
`NDSL
`DSLD
`
`-76
`- 50
`- 55
`70
`-57
`-72
`58
`- 53
`- 65
`-81
`
`-+i
`130
`138
`-46
`142
`3
`- 117
`-37
`-25
`-36
`
`-104
`95
`79
`-143
`79
`-72
`-104
`-64
`-81
`-86
`
`146
`- 11
`II
`12
`- 6
`- 18
`15
`- 14
`- 36
`- 21
`
`6·27
`5·93
`5·71
`5·86
`6·04
`6·10
`5·60
`5·+9
`5·34
`5·00
`
`5·39
`3·13
`3·14
`4·10
`3·38
`3-19
`3·20
`2·91
`3·17
`3·72
`
`C'lass
`
`l{NIJ
`TI
`II
`IV
`II
`Ill
`II'
`III
`III
`I
`
`NI , non-ideal, i.e. one angle off l>y more than 50°.
`
`types. There is only one abnormal turn (type IV), which is the result of the
`unusual torsion angles for Asn52 discussed above. In this case. the usual 1-4
`hydrogen bond is replaced by a. 1-3 hydrogen bond. The only " non-idea.I" entry
`(residues 2 to 5) is probably still an artifact of the disordered N-termina.I residue.
`
`( c) Secondary struct11re
`The secondary structure of Rhe is almost exclusively {J-sheet, and therefore Rhe
`is classified as a {J-protein according to the scheme of Levitt & Chothia. (1976).
`Sheet A (strands 2, 6 a.nd 7), along with strands I a.nd 5, forms one half of a fJ(cid:173)
`ba.rrd. These five strands make up approximately one-half of the molecule and
`the corresponding main chain atoms a.re shown in Figure 7(a). From the Figure, it
`is obvious that the sheet is quite regular, with no fJ-bulges of the types reported
`by Richardson et al. (1978). There are few hydrogen bonds from the inner strands
`to strands I or 5. Indeed, most of the carbonyl and NH groups in strands I and 5
`are nearly perpendicular to the plane of sheet A. This is particularly the case for
`strand l, with Thr5 and Thrl3 being exceptions. The hydrogen bond to Thrl3 is a
`consequence of the {J-turn involving residues 13 to 16. In strand 5, even if the NH
`and carbonyl groups were ideally oriented, hydrogen bonds would not exist as this
`strand is considerably below the plane of sheet A. Note that strand 5 consists
`almost exclusively of members of the second hypervariable region, hence
`interactions between these residues and strand 6 a.re likely to vary considerably
`within V L domains from different sources.
`Analysis of sheet B (strands 3, 4, 8 and 9) is considerably more complicated.
`With the exception of the short helical region and a few turns linking to sheet A,
`these four strands complete the monomeric subunit of Rhe. The ma.in chain atoms
`for this ha.If of the molecule a.re shown in Figure 7(b). The most obvious features
`are the regular {J-type hydrogen bonds between the two central strands, and the
`" wide" {J-bulge (Richardson et al. , 1978) to the left involving residues Gly103,
`Glyl04 and Tyr88.
`The rightmost strand (strand 4) includes a "classic" {J-bulge involving residues
`
`10 of 32
`
`BI Exhibit 1125
`
`

`

`St
`
`II
`
`la l
`
`( b)
`
`( c )
`
`FIG. 7. (a.) St1·ands 5. 6, 7. 2 and I (from left to right) comprising one-half of the Rhe monomer. (b) Strands 9. 8. 3 and 4 (from left. to right) comprising
`the other half. (c) Helical region.
`
`11 of 32
`
`BI Exhibit 1125
`
`

`

`W . FUREY ETAL.
`672
`Leu48, Ile49 a.nd Trp36. In similar regions of Bence-Jones proteins Rei a.nd
`McP0603, a. classic bulge was detected by Richardson et al. (1978), involving
`residues Leu47, Ile48 a.nd Trp35. There is another bulge further down strands 3
`and 4 a.fter they tum away from sheet B. The residues involved a.re Ala.44, Pro45
`and Gln39, but it is not clear whether this bulge should be classified as wide or
`classic. Note in Figure 7(b) that as strands 3 and 4 proceed downward, they also
`move out, ending up nearly perpendicular to sheet B. It is this feature that
`ultimately determines the mode of association of variable domain monomers into
`a dimer.
`In terms of the variability of amino aicids, nearly a.II residues shown in the {3-
`sheet portion of Figure 7 (b) a.re generally considered conserved. The loop shown in
`the upper left containing residues 90 to 100, however, is made up exclusively of
`residues from the third hypervariable region. The only other feature with
`considerable secondary structure is the short helical region (residues 25 to 32)
`shown in Figure 7(c). Note that the region is ma.de up exclusively of members of
`the first hyperva.ria.ble region. Although this segment is obviously helical, the
`distances between a-carbons i and i + 3 are considerably larger than usually found
`in helices. For the five consecutive tetra.peptides beginning with Ser25, the mean
`Ccx.-G<X1+ 3 distance
`is 5-80 A. The mea.n CcxrCa;+ 3 distances for helical
`tetra.peptides in myoglobin, lysozyme and pancreatic trypsin inhibitor are 5·15,
`5·33 and 5·40 A, respectively (Chou & Fa.sman, 1977). The extension of the helix is
`a.ppa.rently required to a.ccommoda.te the unusua.l angles for Asp28 discussed
`earlier. This is evident from an examination of the ma.in cha.in torsion angles for
`residues 26 to 31, the inner residues of the helix. If Asp28 is excluded, the mean cf>,
`i/J values for the five residues a.re -73° and -28°, respectively, which a.re close to
`the normal helical values of 4> - -60° and i/J - -30°. Therefore, if a tetra.peptide
`did not include Asp28, one would expect to observe normal helical Car-Ca;+s
`distances.
`The complete Rhe monomer is generated by placing the two large segments
`shown in Figure 7 back to back connecting them with the appropriate j}-turns.
`The short helical segment connects strand 2 to strand 3, and therefore also
`connects sheet A to sheet B. The interaction between sheets is further stabilized
`by a. disulphide bridge between Cys22 (strand 2, sheet A) and Cys89 (strand 8,
`sheet B), by hydrophobic interactions in the core of the protein, and by a salt(cid:173)
`bridge between Arg62 (strand 6, sheet A) and Asp83 (strand 8, sheet B ). Strand 1
`also interacts with strand 9 (sheet B ), forming a sea.m between the two large
`regions of j}-structure. This is evident in Figure 8, where the complete main chain
`hydrogen bonding pattern is shown. Apparently, this is the reason why strand 1
`interacts only weakly with sheet A, and also explains the carbonyl and NH
`orientations in strand 1 mentioned above. The skewed hydrogen bonds in Figure 8
`result from the fact that these two strands a.re parallel. Of the nine strands in
`Rhe, this is the only occurrence of {3-type interactions between parallel strands;
`a.U other interacting strands run in opposite directions. Note the kink in strand l
`caused by the Pro-Pro sequence for residues 7 and 8. In Bence-Jones protein Rei
`(Huber & Steigema.nn, 1974), residue 8 is also Pro, but was found to be in the cia
`conformation. All eight proline residues in Rhe a.re in the trans conformation.
`
`12 of 32
`
`BI Exhibit 1125
`
`

`

`STRUCTURE OF BENCE- JONES PROTEIN Rhe
`
`673
`
`FIG. 8. Schematic diagram showing main chain hydrogen bonds.
`
`Also, unlike in the Rei structure, there is no evidence of disorder in the disulphide
`bridge in Rhe. The S-S distance .is 2·018 A with Cp-Sy-Sy angles of 103·4° and
`100·9°. The Cp-8-S-Cp torsion angle is 87·4°.
`
`(d} Hydrogen bonding
`Ifwe temporarily restrict our attention to only one monomer, then the protein(cid:173)
`protein hydrogen-bonding interactions can be separated into three categories;
`main chain- main chain, main chain-side chain, and side chain-side chain.
`Hydrogen bonds were included in Figure 8 if the N-0 distance was less than
`3·30 A, the N-H-0 angle greater than 120°, and the C-0-H angle greater than
`90°. Although these criteria are somewhat arbitrary, they are reasonable in that
`all strong hydrogen bonds will certainly be detected, all totally unrealistic ones
`will be rejected, yet a considerable margin for error is provided for. Most
`hydrogen bonds are in fact quite normal, as indicated by mean N-0 distances and
`N-H-0 angles of 2·990 A and 160°, respectively. It should be pointed out that
`although protein stereochemistry was restrained during the refinement process, no
`restraints were included for any of the hydrogen bond interactions. Therefore, the
`hydrogen bond parameters represent unbiased structural observations. There are
`57 main chain-main chain, 21 main chain-side chain, and 11 side chain-side chain
`hydrogen bonds stabilizing the Rhe monomer. The dominance of main cha.in(cid:173)
`main chain interactions is not surprising, considering that two J3-sheets constitute
`most of the molecule.
`For the main chain-side chain hydrogen bonds, the mean donor-acceptor
`distance decreases to 2·908 A and the standard deviation rises to 0·213 A. This is
`mainly the result of including three hydrogen bonds with oxygen atoms as both
`the donor and acceptor. If these three terms are neglected, the mean distance rises
`
`13 of 32
`
`BI Exhibit 1125
`
`

`

`674
`
`W. FUREY ETAL.
`
`TABLE 4
`Side chain-side chain hyd1·ogen bonds
`
`Donor
`
`Acceptor
`
`Tl050Gl
`S770G
`S250G
`N930Dl
`D940D2
`D830DI
`D830D2
`N520Dl
`D280Dl
`N930DI
`E840E2
`
`Q6NE2
`Rl7NH2
`T270Gl
`T270Gl
`N32ND2
`R62NH1
`R62NH2
`K67NZ
`N93ND2
`S950G
`Qll2NE2
`Mean
`Standard deviation
`
`dD-A D-H- A C--0- H
`(A)
`(. )
`
`,.)
`
`3-005
`3·166
`3·209
`3·148
`3·115
`2·830
`2·994
`2·912
`2·858
`2·682
`3·254
`3·016
`0·181
`
`174
`152
`155
`164
`178
`174
`169
`172
`169
`172
`156
`167
`9
`
`114
`171
`140
`99
`102
`123
`121
`103
`11 9
`123
`96
`119
`22
`
`to 2·961 A a.nd the sta.nda.rd deviation drops to 0· 174 A. The adjusted statistics
`then a.re in rea.sona.ble agreement with the results for ma.in chain-main chain
`hydrogen bonds. The mean donor-H- a.cceptor angle for these interactions is also
`160°.
`Of the 11 side cha.in-side chain hydrogen bonds (listed in Ta.hie 4), five
`correspond to interactions exclusively between hypervariable residues, and
`therefore a.re likely to be unique to the Rhe structure. Of the remammg six
`hydrogen bonds, the most important seem to be the two involved in the salt(cid:173)
`bridge between Arg62 and Asp83 shown in Figure 9. In addition to connecting the
`two ,8-sheets, the residues in the Figure serve to insulate the non-polar core
`residues from solvent. Since these residues a.re not from hyperva.riable regions,
`this feature ma.y be common to all V L domains.
`Although the mea.n donor-acceptor distances a.nd donor--H- a.cceptor angles
`a.re similar in all three categories, the "acceptor angles" 0-0- H show a marked
`tendency to decrease as the number of main chain atoms in the hydrogen bond
`decreases (146°, 132° and 119° for main cha.in- main chain, ma.in chain-side chain
`a.nd side chain-side chain interactions, respectively). This is probably the result of
`increased flexibility in the side-cha.ins, enabling the more favorable acceptor
`
`Fm. 9. Salt-bridge between Argii2 and Asp83.
`
`14 of 32
`
`BI Exhibit 1125
`
`