`
`The Galactan-Binding Immunoglobulin Fab J539:
`An X-Ray Diffraction Study at 2.6-A Resolution
`
`Se Won Suh, 1 T.N. Bhat,1 Manuel A. Navia, 1 Gerson H. Cohen, 1 D.N. Rao, 2 Stuart Rudikoff,2 and David R. Davies1
`1 Laboratory of Molecular Biology, National Institute of Diabetes, Digestive and Kidney Diseases and 2 Laboratory of
`Genetics, National Cancer Institute, Bethesda, Maryland 20892
`
`search procedure based on the known structure of
`McPC603 Fab. This analysis confirmed the overall
`similarity of these two Fab structures and provided a
`basis for the higher-resolution analysis reported here.
`
`ABSTRACT The crystal structure of the Fab of
`the galactan-binding immunoglobulin J539 (a mouse
`lgA,K) has been determined at a resolution of approx(cid:173)
`imately 2.6 A by X-ray diffraction. The starting
`model was that obtained from the real space search
`described previously (Navia, M.A., Segal, D.M., Pad(cid:173)
`Ian, E.A., Davies, D.R., Rao, D.N., Rudikoff, S. and
`Potter, M. "Crystal structure of galactan-binding
`mouse immunoglobulin J539 Fab at 4.5 A resolu(cid:173)
`tion." Proc. Nat. Acad. Sci. USA, 76:4071-4074, 1979).
`This Fab structure has now been refined by re(cid:173)
`strained least-squares procedures to an R-value of
`19% for the 11,690 unique reflections between 8.0 A
`and 2.6 A. Therms deviation from ideal bond lengths
`is 0.025 A. The overall structure differs from
`McPC603 Fab, another mouse lgA,K antibody, in that
`the elbow bend, relating the variable and constant
`parts of the molecule, is 145° vs. 133° for McPC603.
`The region of the molecule expected to be the anti(cid:173)
`gen binding site contains a large cavity with two
`clefts leading away from it. This has been fitted with
`a model of an oligo-galactan.
`
`Key words: antibody, crystal structure, anti-galac(cid:173)
`tan, J539
`
`INTRODUCTION
`
`Current knowledge of the three-dimensional struc(cid:173)
`ture of antibodies is based mostly on the X-ray dif(cid:173)
`fraction investigations of fragments. 1·2 The structures
`of three Fahs have been determined: Kol 3 and New,4
`two human myeloma proteins, and McPC603,5 •6 a
`mouse lgA,K protein. Of these, New and McPC603
`have known binding specificities, while the specific(cid:173)
`ity of Kol is not known. There is therefore only lim(cid:173)
`ited information available concerning the interaction
`of antibodies with their antigenic determinants. In
`this paper we describe the crystal structure determi(cid:173)
`nation of J539, a mouse immunoglobulin Fab with
`binding specificity for /'.l(l-6)-D-galactan. J539 is a
`member of a group of several antigalactans whose
`binding properties have been extensively studied. It
`is the only carbohydrate-binding antibody to have
`been so far studied by X-ray diffraction.
`Previously the structure of J539 Fab had been ana(cid:173)
`lyzed at low resolution. 7 There, a poorly defined 4.5 -
`A electron density map was examined by using a
`
`PUBLISHED 1986 BY ALAN R. LISS, INC.
`
`•
`
`0
`
`0
`
`MATERIALS AND METHODS
`Crystallization and Data Collection
`J539 Fab was prepared as described previously,8
`except that the purified material was further sub(cid:173)
`jected to preparative isoelectric focusing in order to
`produce material that would ultimately yield large
`crystals suitable for the production of high-resolution
`X-ray diffraction data.9 The crystallizations were car(cid:173)
`ried out by the hanging drop method at room temper(cid:173)
`ature by using the vapor diffusion technique to
`equilibrate the protein droplet with a solution con(cid:173)
`taining approximately 35% saturated ammonium
`sulfate, 0.07 M imidazole, 0.03 M zinc sulfate, pH 6.8.
`The crystals were orthorhombic, space group P21212i.
`0 7
`with a = 54.1 A, b = 74.2 A, c = 130.8A.
`rotation
`Intensity data were
`collected by
`photography10 on Kodak No-Screen Medical X-ray
`films by using an Elliott GX-6 rotating anode X-ray
`generator operating at 40 kV and 40 mA. A Franks
`double-bent mirror system 11 purchased from Bran(cid:173)
`deis University was used to focus the X-ray beam.
`Data were collected on an Enraf-Nonius Arndt-Won(cid:173)
`acott rotation camera with a nominal 80-mm crystal(cid:173)
`to-film distance. The c-axis of the crystal was parallel
`to the rotation axis of the camera and the rotation
`range for each film pack was 2.3° with 0.3° overlap
`between film packs. This range permitted us to dis(cid:173)
`card the partially recorded reflections. A total rota(cid:173)
`tion of91° was sufficient to record all the independent
`reflections, except for those in the cusp along the c(cid:173)
`axis. Precession camera data for the layers Okl, lkl,
`hOl, hll, h21, hkO, hkl, and hhl were also collected to
`assist in interfilm scaling and to supply some of the
`remaining unobserved reflections.
`
`Received June 3, 1986; accepted June 16, 1986.
`S.W. Suh is now at the Department of Chemistry, College of
`Natural Sciences, Seoul National University, Seoul, Korea.
`M.A. Na via is now at the Merck Sharpe and Dohme Research
`Laboratories, Rahway, NJ 07065.
`Address reprint requests to D.R. Davies, Laboratory of Mo(cid:173)
`lecular Biology, National Institute of Diabetes, Digestive and
`Kidney Diseases, Bethesda, MD 20892.
`
`1 of 7
`
`Celltrion, Inc., Exhibit 1121
`
`
`
`J539 CRYSTAL STRUCTURE
`
`75
`
`The films were scanned at 100-µm raster steps on
`an Optronics P-IOOO film scanner. The resulting digi(cid:173)
`tal data were processed on a VAX 11/780 computer
`with a rotation film program written by G. Cornick
`and M.A. Navia (unpublished work), which incorpo(cid:173)
`rated the dynamic mask procedure of Sjolin and
`Wlodawer12 to extract the integrated intensities of
`reflections. The program was modified to allow the
`use of roughly measured coordinates of eight strong,
`fully recorded reflections as supplementary fiducial
`marks in obtaining the initial [Q] matrix.13 This pro(cid:173)
`vided a starting transformation for a more accurate
`determination of the [Q] matrix by a least-squares
`procedure based on all fully recorded reflections which
`met certain criteria.
`Intensities from the three films in each pack were
`first scaled, corrected for Lorentz and polarization
`factors, and symmetry averaged. Data from individ(cid:173)
`ual packs were merged and scaled together with those
`from the set of zero- and upper-level precession films
`by using a program of Y. Satow (unpublished work).
`The current data set to 2.6-A resolution contains 8I o/o
`of the possible reflections: 90% to 3.0-A resolution
`and 64% between 3.0-A and 2.6-A resolution. Ap(cid:173)
`proximately 77% of the independent reflections were
`measured more than once. The statistics of the data
`processing are given in Table I.
`
`Generation and Refinement of the Model
`The location and orientation of the starting model7
`were confirmed using the rotation method14 and an
`R-value search. 15•16 Initial refinement of the model
`was carried out by using CORELS. 17 Six parameters
`defining the orientation and position of the molecule
`were refined by using 10-A to 7-A data. Then con(cid:173)
`stant and variable domains were refined as separate
`rigid groups by using IO-A to 5.4-A data. No notice(cid:173)
`able improvement in the R-value was obtained when
`light and heavy chains were also treated as separate
`rigid groups.
`This model was then refined with 3.5-A data by
`using PROLSQ. 18
`•19 Residues of Fv were "mutated"
`from the McPC603 original model to those appropri(cid:173)
`ate for J539 during the course of refinement. As the
`phases improved higher-resolution data were incor(cid:173)
`porated to the limit of 2.6 A and the J539 sequence
`was substituted for the McPC603 residues. The se(cid:173)
`quence used was that of Rudikoff et al.20 for VL, Rao
`et al. 21 for VH, Svasti and Milstein22 and Hamlyn et
`al.23 for CL and Auffi-ay et al.24 for CHI.
`Initial model building was done by using the inter(cid:173)
`active graphics program BILDER25 to fit the model
`to 2Fo-Fc and 6F maps. With the availability of
`the OMITMAP procedure26 the graphics program
`FROD027 was used to fit the model to an OMITMAP.
`This improved the location of a number of residues
`and the overall constrast in the maps increased along
`with the improvement in the phases. During the final
`
`TABLE I. Statistics from Processing of the
`Film Data
`
`Total number of reflections measured
`37,222
`No. of independent reflections
`12,670
`Interfilm scaling R-value*
`0.077
`*R = 2:2: I Ihi - Ih I I 2:2.:Ih, , where Ih is the average intensity
`hi
`hi
`for a given reflection and Ihi isone of the measurements which
`were averaged to yield ih·
`
`passes of model building, 304 water molecules were
`added.
`The result of the final refinement cycle is summa(cid:173)
`rized in Table IL The rms error in positional parame(cid:173)
`ters was estimated to be 0.3 A by the method of
`Luzzati.28 The coordinates and other relevant data
`have been deposited in the Brookhaven Protein Data
`Bank.29 Table III shows the relationship of the serial
`numbering system which we have used in this work
`to the system used by Kabat. 30
`
`RESULTS
`In Figure I is drawn the carbon alpha skeleton of
`the J539 Fab. In overall appearance it resembles the
`structure of other Fahs, in particular McPC603. The
`pairs of domains, CHI and CL, are quite similar,
`having 88 carbon alphas match to give an rms differ(cid:173)
`ence of I.8 A. The relation between CHI and CL is
`that of a screw axis with a rotation of I 7 4 ° and a
`translation of 1.9 A. Similarly, with VH and VL, 99
`pairs of carbon alphas match with an rms deviation
`of 1.6 A to give a rotation of 169.4° and 0.04-A
`translation.
`As expected from their sequence identity, the con(cid:173)
`stant domains of J539 resemble quite closely the two
`corresponding domains of McPC603. The rms devia(cid:173)
`tion of the backbone atoms is 1.6 A (2.1 A for all
`atoms). The relative disposition of CL and CHI in
`J539 and McPC603 is not, however, exactly the same.
`For McPC603 the screw relation between CL and
`CHl corresponds to a I69° rotation with a 2.6-A
`translation. In the region of residues I3I-139 (J539
`numbering) there appears to be a major difference in
`the folding of the two CHI domains. This region in(cid:173)
`cludes a proline residue that has been interpreted to
`be a cis-proline in McPC603 but trans- in J539. We do
`not know whether these differences are the result of
`an inadequate analysis of weak electron density or
`represent real differences in structure. We are under(cid:173)
`taking a higher resolution study of J539 in order to
`attempt to clarify this situation.
`The angle between the constant and variable pairs
`of domains differs between J539 (I45°) and McPC603
`(133°). This angle, the "elbow bend" of the Fab,31•32
`has been found to be quite variable in different Fahs,
`going from about 133° in New and McPC603 to as
`much as I70° in the Fab Kol,3 probably signify-
`
`2 of 7
`
`Celltrion, Inc., Exhibit 1121
`
`
`
`76
`
`S. W. SUH ET AL.
`
`TABLE II. Refinement Data
`
`Actual rms
`deviationt
`
`Target
`a
`
`Average .::lF
`R-factor*
`No. of structure factors
`rms deviations from ideal distance (A)
`Bond distance
`Angle distance
`Planar 1-4 distance
`rms deviation from planarity (A)
`rms deviation from ideal chirality cA 3)
`rms deviation from permitted contact distances (A)
`Single torsion contacts
`Multiple torsion contacts
`Possible hydrogen bond
`rms deviation from ideal torsion angles (0
`For planar group (0 or 180)
`For staggered group (±60 or 180)
`For orthonormal group (±90)
`*R = I I IF o I - IF, I I /I IF o I.
`h
`h
`trms = root mean square.
`:j:The weight chosen for the structure factor refinement, the "target a" of ~F, was modeled by the function w = (liu)2 with a= 47 -
`230 x (sin (O)/'/.. - 116).
`
`0.030
`0.040
`0.030
`0.025
`0.150
`
`0.500
`0.500
`0.500
`
`3.0
`15.0
`15.0
`
`92.6
`0.19
`11,690
`
`0.025
`0.052
`0.029
`0.016
`0.199
`
`0.219
`0.289
`0.269
`
`2.5
`26.0
`22.3
`
`)
`
`ing flexibility in the polypeptide chains between V
`andC.
`The interdomain contacts are summarized in Ta(cid:173)
`bles IV and V. These tables show, for each residue,
`the number of atom pairs in contact with residues of
`the other chain. A contact is defined as a distance
`which is less than the sum of the atomic van der
`Waals radii33 plus LOA. Similar tables have been
`published for the McPC603 Fab6 and we note that
`while the contacts between CL and CHl are rela(cid:173)
`tively preserved, the contacts between VL and VH
`
`TABLE III. Correspondence Between the Numbering
`Scheme Used Here and That of Kabat et al. 30
`
`Heavy chain
`Light chain
`This paper Kabat et al. This paper Kabat et al.
`
`1-27
`28-213
`
`1-27
`29-214
`
`1-52
`53
`54-83
`84-86
`87-104
`105
`106-135
`136-138
`139-159
`160-167
`168-180
`181-196
`197-201
`202-212
`213-218
`
`1-52
`52a
`53-82
`82a-82c
`83-100
`lOOa
`101-130
`133-135
`137-157
`162-169
`171-183
`185-200
`202-206
`208-218
`220-225
`
`contain significant differences, perhaps because of dif(cid:173)
`ferences in the hypervariable regions. In J539 some
`32 out of 86 interactions (37%) involve hypervariable
`residues only, compared with 46 out of 105 (44%) in
`McPC603. However, many of the same conserved res(cid:173)
`idues found by Novotny and Haber34 are also found
`here, making contacts across the interface. These in(cid:173)
`clude Y35, Q37, Y86, and F97 of the light chain and
`Q39, L45, W47, Y95, and W108 of the heavy chain.
`
`The Galactan Binding Site
`The binding of galactan in J539 has been exten(cid:173)
`sively studied in solution by Glaudemans and his co(cid:173)
`workers. The antibody combining site was shown to
`accommodate four sequential /3(1-6)-galactopyranosy 1
`residues.35- 37 Recently, Glaudemans et al.38 have
`concluded from a binding study of a number of deoxy(cid:173)
`fluorogalactosides that the binding involves both of
`the two solvent-exposed tryptophans in the combin(cid:173)
`ing site (residues W92L and W33H). They also pro(cid:173)
`posed a rather specific model for the orientation of
`the galactan on the surface of the antibody molecule.
`In this crystal form of J539 the presumed binding
`site for galactan is in close proximity to the constant
`region of a neighboring molecule (Table VI). Probably
`because of this close contact our attempts to diffuse
`galactan into these crystals were unsuccessful. Thus
`we have no direct crystallographic evidence to define
`the binding pocket of the Fab. Nevertheless, the three(cid:173)
`dimensional structure of the variable module does
`provide some clues. Figure 2 is a skeletal model of
`J539 showing just the complementarity determining
`
`3 of 7
`
`Celltrion, Inc., Exhibit 1121
`
`
`
`J539 CRYSTAL STRUCTURE
`
`77
`
`Fig. 1. A stereo drawing of the alpha carbon skeleton of J539 Fab. Solid bonds indicate the heavy chain.
`Filled circles show the complementarity determining residues. 30
`
`residues. A large pocket formed mainly by the CDRs
`of the light and heavy chain can be seen clearly, and
`connected to it there are two grooves. The bottom of
`the pocket is formed mainly by residues l95L, E50H,
`and L99H. Residues W33H, H52H, W90L, and Y92L
`line the side of the pocket. Residues YlOlH, Y104H,
`and Y92L form the outer surface of the pocket. Resi(cid:173)
`dues Y92L and P93L partition the large pocket into
`one central deep pocket and two clefts on either side.
`Residues S30L, T91L, and Y92L form one of these
`clefts and W33H, D54H, Y92L, and H52H form the
`other cleft.
`
`Binding experiments39 demonstrate that J539 binds
`to the middle of the galactan polymer, rather than at
`the ends. Accordingly, we have examined models con(cid:173)
`structed by placing several galactan residues partly
`in the central cavity with the chain extending away
`from the cavity in both directions along the two
`grooves.
`Several conformations of galactan could be reason(cid:173)
`ably fitted to the binding pocket based on a consider(cid:173)
`ation of van der Waals contacts alone. They could be
`constructed with the carbohydrate chain running in
`either direction. One of these, shown in Figure 3, is
`
`TABLE IV. Contacts Between Residues of VL and VH*
`
`V37 Q39 G44 L45 W47 E50 H52 N59 Y60 P62 Y95 L99 Y104 N105 W108 G109
`
`1
`E
`H 33
`35
`y
`Q 37
`s 42
`43
`p
`45
`p
`48
`y
`86
`y
`w 90
`92
`y
`93
`p
`94
`L
`95
`I
`97
`F
`
`2
`
`4
`
`1
`
`1
`
`2
`
`1
`5
`
`1
`
`1
`2
`
`2
`
`3
`
`2
`
`1
`
`4
`
`8
`
`11
`4
`
`1
`
`*Residues listed in the left column are from VL; the residues listed above the columns are from VH. The numbers in the table
`correspond to the number of pairs of atoms from two residues which are within potential van der Waals contact distance. A dot(-)
`indicates no contacts. The boxed regions delineate possible hypervariable to hypervariable interactions.
`
`4 of 7
`
`Celltrion, Inc., Exhibit 1121
`
`
`
`78
`
`S.W. SUH ET AL.
`
`TABLE V. Contacts Between Residues of CL and CHI*
`1141 L145 H147 F171 P172 A174 Q186
`Y127 P128 L129 T130 L131 P132 Pl33
`
`1
`3
`
`2
`12
`1
`
`6
`
`2
`
`1
`
`4
`16
`
`2
`
`2
`
`2
`
`2
`
`1
`
`s 115
`I
`116
`117
`F
`s 120
`122
`E
`Q 123
`s 126
`s 130
`v 132
`134
`F
`N 136
`L 159
`N 160
`s 161
`w 162
`163
`T
`s 173
`M 174
`s 175
`5
`E 212
`*Residues listed in the left column are from CL; the residues listed above the columns are from CHl. The numbers in the table
`correspond to the number of pairs of atoms from two residues which are within potential van der Waals contact distance. A dot
`indicates no contacts.
`
`5
`4
`
`2
`1
`1
`
`2
`2
`
`4
`
`2
`3
`1
`3
`
`similar in general terms to the model proposed by
`Glaudemans et al.40 The preliminary and tentative
`nature of this model should be emphasized. It was
`constructed simply by fitting the galactan into the
`combining site in a manner that would avoid bad
`nonbonded contacts, but without taking into account
`possible hydrogen bonds and without minimizing the
`energy of contact. The great flexibility of the /3(1-6)
`carbohydrate linkage facilitates the fitting of these
`models but at the same time reduces the probability
`of obtaining a unique model. We are at present trying
`to grow crystals of J539 from solutions containing
`various oligogalactans; in this way we hope to ob-
`
`TABLE VI. Contacts Between Residues of Two
`Molecules Related by a Twofold Screw Axis Parallel
`to the z Unit Cell Edge*
`
`Y92 H52 D54 855 T57 HlOO YlOl Y104
`
`7
`19
`
`1
`
`2
`
`1
`
`s 137
`D 138
`v 192 10
`E 196
`1
`1
`9
`*Residues listed in the left column are from CHl of one
`molecule; the residues listed above the columns are from a
`neighboring molecule and are from VL (Y92) and VH. The
`residues Y92L, H52H, D54H, YlOlH, and Yl04H are involved
`in the putative hapten binding pocket. The numbers in the
`table correspond to the number of pairs of atoms from two
`residues which are within potential van der Waals contact
`distance. A dot indicates no contacts.
`
`serve directly, although probably in a different crys(cid:173)
`tal form, the manner in which the antibody binds to
`the galactan.
`
`Comparison With Predicted Models of J539
`The variable region of J539 has been the subject of
`numerous modeling attempts. These have ranged
`from simple backbone models to fairly complete
`models for all the atoms in the combining site
`42 (Padlan, Glaudemans, and Davies, unpub(cid:173)
`region41
`•
`lished). We have compared several of these models
`with the observed structure derived from X-ray dif(cid:173)
`fraction. Although in many of the CDR loops the
`models show quite good agreement with the X-ray
`structure, there are always some loops that show very
`poor agreement. As an example, we quote the com(cid:173)
`parison for the model proposed by Mainhart et al.,41
`whose coordinates were kindly given to us by Main(cid:173)
`hart. We observe an rms deviation for the alpha car(cid:173)
`bons of 2. 7 A for all the heavy-chain CD Rs and 1.8 A
`for the light-chain CDRs. For the individual CDRs,
`the alpha carbon rms deviations range from 1.1 A for
`Hl and L2 to 4.0 A for H3. If all the atoms are
`included, then the deviations are 4.6 A and 3.0 A for
`the heavy- and light-chain CDRs, respectively. The
`rms deviations for the individual CDRs range from
`2.0 A and 2.1 A for Ll and L2 to 3.9 A for Hl and
`L3 and 6.5 A for H3. Similar deviations are found in
`other predicted models. These large discrepancies be(cid:173)
`tween prediction and observation for key parts of the
`combining site, in spite of encouraging agreement in
`other parts, are probably due to inadequacies in the
`
`5 of 7
`
`Celltrion, Inc., Exhibit 1121
`
`
`
`J539 CRYSTAL STRUCTURE
`
`79
`
`Fig. 2. A stereo diagram of the complementarity determining residues (CDRs)30 of J539. Alternate CDRs
`are drawn with a bold line. All residues of the CDRs have been included.
`
`Fig. 3. A stereo drawing of the residues which form the surface of the suggested galactan-binding pocket
`of J539. A model of a penta-/l(l-6}-galactan which was fitted into this pocket in a manner that would avoid
`bad nonbonded contacts is shown by using filled bonds. The reducing end of the galactan is toward the left
`of the figure, consistent with the results from solution studles.40
`
`present state of modeling procedures. However, we
`cannot rule out the possibility that there are distor(cid:173)
`tions in the conformations of the hypervariable loops
`of the observed crystal structure as a result of the
`close contact with the neighboring molecule in the
`crystal.
`
`REFERENCES
`1. Davies, D.R., Metzger, H . Structural basis of antibody func(cid:173)
`tion. Ann. Rev. Immunol. 1:87-117, 1983.
`2. Amzel. L.M., Poljak, R.J. Three-dimensional structure of
`immunoglobulins. Ann. Rev. Biochem. 48:961-997, 1979.
`3. Marquart, M., Deisenhofer, J ., Huber, R., Palm, W. Crystal(cid:173)
`lographic refinement and atomic models of the intact im(cid:173)
`its antigen-binding
`munoglobulin mqlecule Kol • and
`fragment at 3.0 A and 1.9 A r esolution. J. Mo!. Biol.
`141:369-391, 1980.
`1. Saul, F.A., Amzel, L.M., Poljak, R.J. Preliminary refine(cid:173)
`ment and structural analysis of the Fab fragment from
`human immunoglobulin New at 2.0 A. J. Biol. Chem.
`253:585-597, 1978.
`5. Segal, D.M. , Padlan, E.A., Cohen, G.H., Rudikoff, S., Potter,
`M., Davies, D.R. The three-dimensional structure of a phos-
`
`phocholine-binding mouse immunoglobulin Fab and the na(cid:173)
`ture of the antigen binding site. Proc. Natl. Acad. Sci. USA
`71:4298-4302, 1974.
`6. Satow, Y., Cohen, G.H., Padlan, E.A. , Davies, D.R. The
`phosphocholine binding immunoglobulin Fab McPC603: An
`X-ray diffraction study at 2.7 A . J . Mo!. Biol. (in press)
`1986.
`7. Navia, M.A., Segal, D.M. , Padlan, E.A., Davies, D.R. , Rao,
`D.N., Rudikoff, S., Potter, M. Crystal structure of galactan(cid:173)
`binding mouse immunoglobulin J539 Fab at 4.5 A resolu(cid:173)
`tion. Proc. Natl. Acad. Sci. USA 76:4071-4074, 1979.
`8. Rudikoff, S., Potter, M., Segal, D.M., Padlan, E.A., Davies,
`D.R. Crystals of phosphorylcholine-binding Fab-fragments
`from mouse myeloma proteins: Preparation and X-ray anal(cid:173)
`ysis. Proc. Natl. Acad. Sci. USA 69:3689-3692, 1972.
`9. Bott, R., Navia, M.A., Smith, J.L. Improving the quality of
`protein crystals through purification by isoelectric focus(cid:173)
`ing. J. Biol. Chem. 257:9883-9886, 1982.
`10. Arndt, U.W., Wonacott, A.J. (eds.) "The Rotation Method in
`Crystallography." Amsterdam: North-Holland, 1977.
`11. Harrison, S.C. A point-focussing camera for single crystal
`diffraction. J . Appl. Cryst. 1:84-90, 1968.
`12. Sjolin, L., Wlodawer, A. Improved technique for peak inte(cid:173)
`gration for crystallographic data collected with position(cid:173)
`sensitive detectors: A dynamic mask procedure. Acta Cryst.
`Series A 37:594-604, 1981.
`
`6 of 7
`
`Celltrion, Inc., Exhibit 1121
`
`
`
`80
`
`S.W. SUH ET AL.
`
`13. Rossman, M.G. Processing oscillation diffraction data for
`very large unit cells with an automatic convolution tech(cid:173)
`nique and profile fitting. J_ Appl. Cryst. 12:225-238, 1979.
`14. Crowther, R.A. The fast rotation function. In: "The Molec(cid:173)
`ular Replacement Method." Rossmann, M.G. (ed.) New York:
`Gordon and Breach. 1972:173-178.
`15. Bott, R., Sarma, R. Crystal structure of turkey egg-white
`lysozyme: Results of the molecular replacement method at
`5 A resolution. J. Mo!. Biol. 106:1037-1046, 1976.
`16. Rae, A.D. The use of structure factors to find the origin of
`an oriented molecular fragment. Acta Cryst. Series A
`33:423-425, 1977.
`17. Sussman, J.L., Holbrook, S.R., Church, G.M., Kim, S.-H. A
`structure-factor least-squares refinement procedure for
`macromolecular structures using constrained and re(cid:173)
`strained parameters. Acta Cryst. Series A 33:800-804, 1977.
`18. Hendrickson, W.A. Stereochemically restrained refinement
`of macromolecular structures. In: "Methods in Enzymol(cid:173)
`ogy." Wyckoff, H.W., Hirs, C.H.W., Timasheff, S.N. (eds.)
`Vol 115. Orlando: Academic Press. 1985:252-270.
`19. Hendrickson, W.A., Konnert, J.H. Incorporation of stereo(cid:173)
`chemical information into crystallographic refinement. In:
`"Computing in Crystallography." Diamond, R., Rama(cid:173)
`seshan, S., Venkatesen, K. (eds.) Bangalore: Indian Acad(cid:173)
`emy of Sciences. 1980:13.01-13.23.
`20. Rudikoff, S., Rao, D.N., Glaudemans, C.P.J., Potter, M. K
`chain joining segments and structural diversity of antibody
`combining sites. Proc. Natl. Acad. Sci. USA 77:4270-4274,
`1980.
`21. Rao, D.N., Rudikoff, S., Krutzsch, H., Potter, M. Structural
`evidence for independent joining region gene in immuno(cid:173)
`glogulin heavy chains from anti-galactan myeloma pro(cid:173)
`teins and its potential role in generating diversity in com(cid:173)
`plementarity-determining regions. Proc. Natl. Acad. Sci
`USA 76:2890-2894, 1979.
`22. Svasti, J., Milstein, C. The complete amino acid sequence
`of a mouse K light chain. Biochem. J. 128:427-444, 1972.
`23. Hamlyn, P.H., Brownlee, G.G., Cheng, C.-C., Gait, M.J.,
`Milstein, C. Complete sequence of constant and 3' noncod(cid:173)
`ing regions of an immunoglobulin mRNA using the dideox(cid:173)
`ynucleotide method of RNA sequencing. Cell 15:1067-1075,
`1978.
`24. Auffray, C., Nageotte, R., Sikorav, J.-L., Heidmann, 0.,
`Rougeon, F. Mouse immunoglobulin A: Nucleotide se(cid:173)
`quence of the structural gene for the a heavy chain derived
`from cloned cDNAs. Gene 13:365-374, 1981.
`25. Diamond, R. BILDER: A computer graphics program for
`biopolymers and its application to the interpretation of th.e
`structure of tobacco mosaic virus protein disc at 2.7 A
`resolution. In: "Biomolecular Structure, Function, Confor(cid:173)
`mation and Evolution." Srinivasan, R. (ed.) Vol. 1. Oxford:
`Pergamon Press. 1981:567-590.
`26. Bhat, T.N., Cohen, G.H. OMITMAP: An electron density
`map suitable for the examination of errors in a macromo(cid:173)
`lecular model. J. Appl. Cryst. 17:244-248, 1984.
`27. Jones, T.A. A graphics model building and refinement sys-
`
`tern for macromolecules. J. Appl. Cryst. 11:268-272, 1978.
`28. Luzzati, V. Traitment statistique des errors dans la deter(cid:173)
`mination des structures cristallines. Acta Cryst. 5:802-810,
`1952.
`29. Bernstein, F.C., Koetzle, T.F., Williams, G.J.B., Meyer, E.F.,
`Brice, M.D., Rogers, J.R., Kennard, 0., Shimanouchi, T.,
`Tasumi, M. The protein data bank: A computer-based ar(cid:173)
`chival for macromolecular structures. J. Mo!. Biol. 112:535-
`542, 1977.
`30. Kabat, E.A., Wu, T.T., Bilofsky, H., Reid-Miller, M., Perry,
`H. (eds.) Sequences of Proteins of Immunological Interest.
`Washington: United States Department on Health and Hu(cid:173)
`man Services, 1983.
`31. Huber, R., Deisenhofer, J., Colman, P.M., Matsushima, M.,
`Palm, W. Crystallographic structure studies of an IgG mol(cid:173)
`ecule and an Fe fragment. Nature 264:415-420, 1976.
`32. Colman, P.M., Deisenhofer, J., Huber, R., Palm, W. Struc(cid:173)
`ture of the human antibody molecule Kol (Immunoglobulin
`Gl): An electron density map at 5 A resolution. J. Mo!.
`Biol. 100:257-282, 1976.
`33. Case, D.A., Karplus, M. Dynamics of ligand binding to
`heme proteins. J. Mo!. Biol. 132:343-368, 1979.
`34. Novotny, J., Haber, E. Structural invariants of antigen
`binding: Comparison of immunoglobulin VL-VH and VL(cid:173)
`VL domain dimers. Proc. Natl. Acad. Sci. USA 82:4592-
`4596, 1985.
`35. Jolley, M.E., Rudikoff, S., Potter, M., Glaudemans, C.P.J.
`Spectral changes on binding of oligosaccharides to murine
`IgA myeloma proteins. Biochemistry 12:3039-3044, 1973.
`36. Manjula, B.N., Mushinski, E.B. and Glaudemans, C.P.J.
`The formation of active hybrid immunoglobulins from the
`heavy and light chains of /3(1,6)-D-galactan binding murine
`myeloma IgA's SlO and J539. J. Immunol. 119:867-871,
`1979.
`37. Roy, A., Manjula, B.N., Glaudemans, C.P.J. The interaction
`of two polysaccharides containing /3(1,6)-linked galactopyr(cid:173)
`anosyl residues with two monoclonal antigalactan immu(cid:173)
`noglobulin Fab' fragments. Mo!. Immunol. 18:79-84, 1981.
`38. Glaudemans, C.P.J., Kovac, P., Rasmussen, K. Maping of
`the subsites in the combining area of monoclonal antigalac(cid:173)
`tan IgA J539. Biochemistry 23:6732-6736, 1984.
`39. Glaudemans, C.P.J., Bhattachargee, A.K., Manjula, B.N.
`Monoclonal anti-galactan IgA J539 binds intercatenarily
`to its polysaccharide antigen. Observations on the binding
`of antibody to a macromolecular antigen. Mo!. Immunol.
`(in press), 1986.
`40. Glaudemans, C.P.J., Kovac, P. Probing the combining site
`of monoclonal IgA J539 using deoxyfluoro- and other gal(cid:173)
`actosides as ligands. Mo!. Immunol. 22:651-653, 1985.
`41. Mainhart, C.R., Potter, M., Feldmann, R.J. A refined model
`for the variable domains (Fv) of the J539 /3(1,6)-D-galactan(cid:173)
`binding immunoglobulin. Mo!. Immunol. 21:469-478, 1984.
`42. Stanford, J.M., Wu, T.T. A predictive method for determin(cid:173)
`ing possible three-dimensional folding of immunoglobulin
`backbones around antibody combining sites. J. Theor. Biol.
`88:421-439, 1981.
`
`7 of 7
`
`Celltrion, Inc., Exhibit 1121
`
`