PERIODICAL RCXJM
`LTH SCIENCES LIBRARY
`
`THE J_OURNAL OF
`EXPERIMENTAL
`MEDICINE
`
`Edited by
`Anthony Cerami Zanvil A. Cohn Maclyn McCarty Ralph Steinman
`
`Assistant Editors
`Allan R. Goldberg Carl Nathan Ellen Pure
`
`Frederick W. Alt
`K. Frank Austen
`Baruj Benacerraf
`Constantin A. Bona
`Patrick Brennan
`George Cross
`John R. David
`Frank J. Dixon
`Bernard N. Fields
`Shu Man Fu
`Emil C. Gotschlich
`Alan N. Houghton
`Thomas J. Kindt
`Seymour J. Klebanoff
`Frank Lilly
`
`Advisory Editors
`Philippa Marrack
`Lloyd F. Mayer
`Hans J. Muller-Eberhard
`E. A. McCulloch
`Ralph L. Nachman
`Robert J. North
`Victor N ussenzweig
`Michael B. A. Oldstone
`William E. Paul
`Russell Ross
`Pablo Rubinstein
`
`David H. Sachs
`Matthew D. Scharff
`Hans Schreiber
`Benjamin D. Schwartz
`Kendall A. Smith
`Jonathan Sprent
`Timothy A. Springer
`Alain Townsend
`Giorgio Trinchieri
`Zena Werb
`Alan F. Williams
`Darcy B. Wilson
`Henry J. Winn
`Samuel D. Wright
`
`Published Monthly by The Rockefeller University Press
`Volume 168, No. 3 September 1, 1988
`
`JEMEAV 168(3) 823-1210
`
`(1988)
`
`•
`
`ISSN 0022-1007
`
`PFIZER EX. 1131
`Page 1
`
`

`

`INFORMATION FOR CONTRIBUTORS Send all contributions to the
`Editors of The Journal of Experimental Medicine, The Rockefeller University,
`Box 289, 1230 York Avenue, New York, New York 10021. Each manuscript is
`acknowledged by the Editors' office, and the author should notify the office if
`he does not receive acknowledgment within a reasonable time. Papers will be
`printed as soon as feasible after acceptance and, if possible, in the order in which
`they are received.
`Regular Contributions should be limited to 20 printed pages including illus(cid:173)
`trations (approximately 40 typed pages, double-spaced). Articles that will be
`longer will be returned to the authors to be cut to the acceptable maximum.
`Brief Definitive Reports are accepted for rapid publication provided they repre(cid:173)
`sent complete, original studies that are documented by references, experimental
`procedures, and a summary. The reports must be limited to less than 6 standard
`Journal pages, or the manuscript will be returned to the authors.
`1\vo copies of the manuscript should be submitted: (1) the original typescript,
`double-spaced- including references and legends, and (2) a fully legible carbon,
`mimeographed, or Xerox copy.
`PREPARATION OF MANUSCRIPT AND ILLUSTRATIONS See inside back cover.
`PAGE CHARGES A basic charge of $30.00 is made for each page. An author's
`inability to honor page charges will not affect publication of acceptable
`articles.
`ExcEss PAGES oF PHOTOMICROGRAPHS Four pages of photmnicrographs can be
`included at the $30.00 per page charge; a surcharge of $100 will be made for
`each additional page.
`OFFPRINTS Authors may purchase offprints of their articles at a cost indicated
`on the offprint order form that accompanies galley proof.
`
`INFORMATION FOR SUBSCRIBERS The subscription price per year (two volumes, six issues
`each) is $150, postal surcharge Europe only, $40, and for a single number is $12.50. Subscriptions,
`payable in advance, start with the first issue of the calendar year. Special student and intern rates
`available.
`MICROFORM Subscriptions and back copies from Vol. 1 on, are available in microfiche and microfilm
`from University Microfilms, 300 North Zeeb Road, Ann Arbor, Michigan 48106.
`Inquiries about subscriptions, back issues, and special student and intern rates should be ad·
`dressed to the Order Service of The Rockefeller University Press, 222 East 70th Street, New York
`10021.
`Remittances should be made by check or money order payable to The Rockefeller University Press
`and addressed to Post Office Box 5108, Church Street Station, New York 10249. All checks must
`be drawn on U. S. banks.
`Claims for nonreceipt of issues will be honored when received within 3 months of mailing for U.S.A.
`subscribers, 6 months for foreign subscribers.
`
`The Journal of E~perimental Medicine (ISSN 0022·1007) is published monthly for $150 per year
`(postal surcharge Europe only, $40 net) by The Rockefeller University Press, 222 East 70th Street,
`New York, NY 10021. Second-class postage paid at New York, NY and additional mailing offices.
`POSTMASTER: Send address changes to The Journal of Experimental Medicine, Order Service,
`The Rockefeller University Press, 222 East 70th Street, New York, NY 10021.
`Copyright© 1988 by The Rockefeller University Press, 222 East 70th Street, New York, NY 10021.
`Made in the United States of America.
`The appearance of the code at the bottom of the first page of an article in this journal indicates that the Publisher gives consent
`for individual copies of that article to be made for personal or internal use. This consent is given on the condition, however, that - for
`copying beyond the limited quantities permitted under Sections 107 or 108 of the U. S. Copyright Law- the copier pay the stated
`percopy fee, for this journal $2.00 per article, through the Copyright Clearance Center, Inc., 21 Congress Street. Salem, Massachusetts
`01970, a nonprofit organization. This consent does not extend to other kinds of copying, such as copying for general distribution,
`for advertising or promotional purposes. for creating new collective works, or for resale.
`
`PFIZER EX. 1131
`Page 2
`
`

`

`GLYCOSYLATION OF A VH RESIDUE OF
`A MONOCLONAL ANTIBODY AGAINST a(l-6) DEXTRAN
`INCREASES ITS AFFINITY FOR ANTIGEN
`
`Bv SUSAN C. WALLICK, ELVIN A. KABAT, AND SHERIE L. MORRISON
`
`From the Departments of Microbiology, Genetics and Development, and Neurology, and
`the Cancer/ Institute for Cancer Research, Columbia University College of Physicians and Surgeons,
`New York, New York 10032
`
`lmmunochemical characterization of antibodies against a(1 --+ 6) dextran has given
`insights into the size and shape of th e antibody-combinin g site and the nature of
`the interaction between antibodies and antigen. We are now attempting to correlate
`the immunochem ical properties of the antidextran antibodies with their primary
`structure. In the course of these st udies cDNAs from three monoclonal anti-a(l --+ 6)
`dextran hybridoma cell lines, 14.6b.l, 5.54 1 a nd 19.22 .1 (1, 2), were cloned , and the
`nucleotide sequences of their V H and V L regions were determi ned (3) (Table 1). All
`synthesize an identical K light chain with the V K-OX1 germline gene ( 4) rearranged
`to the J K2 segment; the heavy cha ins differ by only one or two am ino ac ids in their
`complementarity-determining regions (CDRs)2
`. When com pared with 14.6b.l , 5.54
`and 19.22 .1 have an identical Thr--+ Asn am ino acid chan ge at position 60 in VH ;
`5.54 has a n additional change (Ser --+ Gly) at position 31 in CDRl. The changes
`in heavy cha in sequ ence result in 5.54 and 19.22 .1 having a 10-fold or greate r reduc(cid:173)
`tion in their binding co nsta nts for both polymeric dextran and iso m altoheptaose
`(IM7) when compared with 14.6b.l (Table 1).
`The Thr--+ Asn chan ge in 5.54 a nd 19.22 .1 leads to the loss of a potenti al N-linked
`glycosylation site (Asn5s -Tyr5g -Thr60) present in 14.6b.l. The purpose of this study
`was to determin e whether this potential N- linked glycosylation site is used and if
`so, wh ether the addition of carbohydrate (C HO) to CDR2 affects the binding con(cid:173)
`stant for dextran. It is difficult to demonstrate glycosylat ion of V H in the original
`hybridoma a ntibodies since both l gA a nd IgM isotypes are glycosylated within their
`C1-11 domains a nd CHO present in Fd could be linked to either VH or C1-1. There(cid:173)
`fore, we have transferred the three V 1-1 regions to the hum an l gG4 con stant region,
`which is devoid of CHO in its C1-1 1 domain. In this report we demonstrate the pres(cid:173)
`ence of carbohyd rate within the V 1-1 of 14.6b.l. Comparison of the associat ion co n(cid:173)
`sta nts for aglycosylated tunicamycin (Tm)-treated and -untreated antibodies shows
`
`This wo rk was supported in pa rt by grants Al -19042, CA-16858, CA-22736, a nd CA-13696 (to the Cancer
`Center) from th e Nationa l Inst itu tes of Health , a nd by gra nt DBM-860-0 778 from th e National Science
`Foundation. Address corres pondence to Dr. She ri c L. M o rri son, Department of M icrob iology, 540 M o(cid:173)
`lecul ar Biology Institute, Un ive rsity of Californi a a t Los Angeles, Los Angeles, CA 90024.
`1 The 5.54 mAb was designated as 5.54.4.24.1 by Newman and Kaba t (2).
`1 Abbreviations used in this paper. CDR, complementa rit y- determinin g regio n; C HO, carbohyd rate;
`IM7 , isomaltoheptaose; Staph A, Staphylococcus aureus protein A; Tm , tuni camycin .
`
`J. ExP. MED. ©The Rockefeller U ni versity Press · 0022-1007/88/09/1099111 $2.00
`Volume 168 September 1988 1099- 1109
`
`1099
`
`PFIZER EX. 1131
`Page 3
`
`

`

`TABLE I
`lmmunochemical Properties of Hybridoma Antibodies Specific for Dextran B 512
`
`H ybridoma
`
`Mouse strain
`
`Jso type
`
`Site size
`
`K .l
`
`CDR!
`
`C DR2
`
`CDR3
`
`14. 6b . l,
`5.54**
`19.22. 1,
`
`BALB/c
`C 57BL/6
`BALB/c
`
`lgA,k
`IgA,k
`lgM ,k
`
`6
`6
`
`mllg
`4.43 X
`I . 78 X
`8 .87 X
`
`105
`104
`103
`
`liter/mole
`5. 76 X 104
`3. 02 X J0 3
`6.46 X 103
`
`3 1 Ser --+ Gl y
`
`60 Thr --+ Asn
`60 Thr --+ Asn
`
`JH
`
`3
`3
`3
`
`H eavy C hain amin o acid chan ges
`vs. J4.6b. 1 prototypell
`
`• M ax imum number of a ( I -+ 6)-linked glu cose residues th at fit th e a ntibod y co mbinin g site.
`Determin ed by affinit y ge l electrophores is accordin g to the method described by T a keo a nd Kabat ( I 7).
`Associati on co nsta nts of a ntidex tran combining sites with iso maltohe ptaose (1M 7).
`According to Akolka r et al. (3).
`According to Sha ron et al. ( I).
`•• Acco rdin g to New ma n e t al. (2); designated as 5.54. 4. 24. 1 by Newm a n ct al.
`These sequ ence d ata have been submitted to the EMBL/GenBa nk Da ta Libra ri es unde r access ion number Y00809.
`
`......
`......
`0
`0
`
`~
`to
`~ l'
`trl
`to
`trl
`()
`0 z
`()
`~
`()
`0
`<fJ -<
`l'
`2;
`0 z
`
`;J>
`'Tj
`'Tj
`trl
`()
`-l
`<fJ
`;J>
`'Tj
`
`~ z
`=i
`-<
`
`PFIZER EX. 1131
`Page 4
`
`

`

`WALLICK ET AL.
`
`1101
`
`the presence of CHO increases the aKa of 14.6b.l for dextran . The effect on
`is unique to the carbohydrate present in V H , since absence of CHO from
`H2 does not change the aKa for dextran. Lastl y, we have demonstrated that the
`HO in V H is more exposed than in CH2.
`
`Materials and Methods
`5.54 is a mouse hybridom a cell line synthesi zing a C57 BL/6 IgA, K antibody
`Cell Lines.
`·fie for a( I -+ 6) dextran. D3 is a sponta neous heavy chain-loss variant of 5.54 that syn(cid:173)
`zes only the K light chain charac teristi c of the antidextran hybridom as . The D 3 light
`n variant cell line was isolated by Dr. P. N . Akolka r (Columbia University, NY). Cell
`were grown in lscove's Modified Dulbecco's medium (IMDM) (Gibco Labo ratories,
`Island , NY) supplemented with 3- 5% FC S (H yclone Laboratories, Logan , U T ).
`Gene Transfection. G ene transfection was by protoplast fu sio n using the method of Oi et
`(5) and modified as described by Tan et al. (6). Tra nsfecta nt culture supernatants were
`for antibody production and dextran binding by ELISA (7). D extran B512 was prepared
`Leuconostoc mesenteroides strain B512 cultures by Dr. L. M atsuuchi as described (8). H orse(cid:173)
`radish peroxidase affinity purified goat anti-human IgG antibody was purchased from Sigma
`Chemical Co. (St. Loui s, MO). D3 recipient tran sfccted cells from positi ve wells were sub(cid:173)
`cloned once in soft agarose (9), and clones tha t stained heavi est with rabbit anti-hum an IgG
`Fe antiserum (Cooper Biomedical, Inc., M alvern , PA) were chosen for furth er analysis.
`Biosy nthetic Radiolabeling and Papain Digestion. Tra nsfectant cells were labeled in the presence
`of 15 ~tCi/ml of [35S]Met or 100 ~tC i/ml o-[ 14C] glucosamine hydrochloride as described (10).
`Secretions from the cell s were di gested with papain (Si gma C hemical Co.) at 1:100 enzyme/
`protein ratio for 4 h at 3rC. The reaction was stopped by addition ofi odoace tamide to 0.03
`M. The Fe fraction and undi gested antibod y protein were precipita ted by incubation with
`IgG-Sorb (Enzym e Center, M alden, MA). Fab was precipitated from the supern atant using
`rabbit a nti-huma n Fab (prepared by Letiti a A. Wim s, Columbi a Uni versit y, NY) or by in (cid:173)
`solubilized dex tra n (Sephadex G 75). Samples were reduced with 2- M E (0.1 5 M ) and ana(cid:173)
`lyzed using 5% SDS-PAGE (5).
`Inhibition of Glycosylation. Tm at a co ncentra tion of 8 ~t g/ml (Boehringer M annheim Bio(cid:173)
`chemicals, Indian a polis, IN ) was used to inhibi t N -linked glycosyla tion. Cells we re biosyn(cid:173)
`thetically labeled for 3 h with [35S]Met in the presence of Tm as described above. After
`pretreatment, secre ted Ig in the culture supernata nt was di scarded, the cells were washed
`twice with IMDM , fresh Tm and [35S]Met added , a nd treatment co ntinued overni ght at
`37°C . Removal of C HO from Ig was verified by immunoprec ipi tatio n of the secreted a nti (cid:173)
`body and analysi s by SDS-PAGE.
`Determination of the Antibody Protein Concentration in Culture Supernatants. Antibodies in cultu re
`supernatants diluted into BBS (0.02 M bora te-buffered 0.75 % saline, pH 8.3) were bound
`to polystyrene microtiter wells (Corning Glass Works, Corning, NY) for 3 h at 37 °C . After
`blockin g any unreacted sites with 1% BSA/PBS/0.05 % Tween 20 for 1 h at room tempera(cid:173)
`ture, the ELISA pla tes we re washed with PBS/0.05% Tween 20 three tim es, PBS once, and
`then bound Ig was qu antita ted by reaction with ho rse radish perox idase-labeled anti-h um an
`lgG antibody and compa red with a human IgG stand ard of known concentratio n. Assay
`results have been reproduced at least three times. Direc t binding of antibody to microtiter
`plates was a more reprodu cible method than bindin g supern ata nts to plates sensiti zed with
`anti-huma n IgG antiserum , for reasons that are not clear.
`Determination of the Apparent Association Constants of Aglycosylated Con A- adsorbed or - untreated
`Transfectoma Antibody Against Dextran B512. Apparent binding constants were determined using
`the method of Nieto et al. (11). In brief, the association constant for an antibod y is defin ed
`as the reciprocal free ligand concentratio n necessary for occupyin g one half of the antibody(cid:173)
`~ombinin g sites. If a fi xed a moun t of antibod y is reacted with an increasing amount of free
`hgand on a pla te coa ted with anti gen , the reciprocal of the free li gand concentrati on th at
`causes 50 % inhibiti on of binding to the plate is considered to be a fun ction of the intrin sic
`K. and is designa ted as the appa rent affini ty constant (aK.). The aKa is calcula ted from the
`
`PFIZER EX. 1131
`Page 5
`
`

`

`1102
`
`VARIABLE REGION GLYCOSYLATION AFFECTS AFFINITY
`
`amount of ligand necessary for 50 % inhibition of binding. The followin g experimental con(cid:173)
`dition s were used to measure the aX;, valu es: Corning microliter plates were coated with 0.5
`Jlg/ml or 20 Jlg/ml dextran B512 (high:affinity and low-affinity assay condition s, respectively).
`Bound Ig was qu a ntitated usmg a nti-human IgG labeled with horseradish peroxidase.
`
`Results
`The expressed V H regions from the three hybridoma a ntibodies against a(1 -+
`6) dextran were joined to the human IgG4 constant region gene (Fig. 1 ), and after
`transfection of D3 , a cell line producing only the hybridoma-specifc light chain (5,
`6) directed the expression of an H chain that assembled with the endogenous light
`chain and was secreted (results not shown). Nomenclature for the mAbs and trans·
`fectoma antibodies used in this study are presented in Table II.
`To determine if the 14.6b.l chimeric antibody contained CHO in V r~ , we frac(cid:173)
`tionated the molecule into Fab and Fe by papain cleavage, reduced th e molecules
`with 2-ME, and analyzed them on 5% SDS-PAGE gels. Proteins were labeled with
`
`e5S]Met, and the Fab was precipitated using specific anti-Fab antiseru m (Fig. 2
`
`A). Transfectoma antibodies with V H derived from 5.54 and 19.22.1 eD NA clones
`(T5.54 and T19.22 , respectively) show comigration of their Fd and K light chains.
`Precipitation ofFab with insolubilized dextran results in the same pattern, verifying
`that both K and Fe are present (data not shown). In contrast, in transfectoma anti(cid:173)
`bodies with the H chain variable region of 14.6b.l (T14.6b ), the Fd portion migrates
`more slowly than the L chain. The reduced mobility of the T14.6b Fd fragment is
`consistent with glycosylation of its V H.
`To confirm the presence of CHO in the V H of T14.6b, we labeled secreted Ig with
`[14C]glucosamine, prepared Fab and Fe fractions, and analyzed th e products by
`SDS-PAGE (Fig. 2 B). As anticipated, th e K light chains do not contain CHO and
`bands are absent from the position indicated by the [35S]Met-labeled K light chain.
`We find [ 14C]glucosamine labeling of the human IgG Fe fragment that contains N(cid:173)
`linked CHO within its Cr~2 domain (12). However, the Fab from only T14.6b, with
`its Fd containing the 14.6b.l Vr~ , shows glucosamine labeling. The redu ced inten(cid:173)
`sities of the Fd bands relative to the Fe is probably due to poor recovery of the Fab
`fragment rather than incomplete glycosylation (13). In SDS-PAGE gels in which we
`can resolve H chains containing no, one, or two CHO moieties (Fig. 3 B) we find
`only one heavy chain band for T14.6b.
`We have used the glycohydrolase Endo H to investigate the structure of the V H
`oligosaccharide. The di-N-acetylchitobiose linkage of high-mannose core oligosac·
`charides found on newly synthesized IgG H chains is susceptible to Endo H cleavage
`(14), while processed complex CHO are resistant to Endo H cleavage. H chains ob(cid:173)
`tained from cell cytoplasms were hydrolyzed by Endo H (data not shown). In con(cid:173)
`trast, heavy chains from the secretions of both T19.22 and T14.6b were unaltered
`by Endo H treatment. Thus the N-linked CHO present in V H does not appear to
`differ from that present in the constant region.
`To examine the role of CHO in Ag binding we determined the association con(cid:173)
`stants for Tm-treated aglycosylated and untreated native antidextran transfectorn.a
`antibodies. Although Tm is a potent inhibitor of N-linked glycosylation (15), it 15
`difficult to produce proteins completely free of glycosylated species. From recon·
`struction experiments it was apparent that even a trace contamination of hi gh-affinity
`
`PFIZER EX. 1131
`Page 6
`
`

`

`WALLIC K ET AL.
`
`1103
`
`MPCll
`
`PII
`
`PI
`
`eDNA
`
`PII
`
`PI
`
`RI
`
`PII
`
`PI
`
`RI
`
`MO<.Ae VDJ
`
`H<.mon Constant Region
`
`Amp'
`
`E. Coli gpt
`
`-----------------
`
`1 RI
`l Ligate
`
`eDNA
`
`tunon Constant Region
`
`FIG URE 1. Substitution of the genomic V H region with V H eDNA a nd isotype switch. A genomic
`Eco RI fragment conta ining the MPC ll H cha in p romoter, leader sequence, rearranged V re(cid:173)
`gion, and Ig enha ncer (2 4) was cloned into the Eco RI site of a pBR322 derivati ve from which
`the sequences lying between the Hind III site (nucleotide 29) a nd the Pvu II site (nu cleotide
`2,066) had been d eleted. Using eDNA produced from the an ti-a(! -+ 6)dextran hybridom as (3),
`the V region of the MPC ll was replaced by the a ntidextra n V regio n by insertin g the Pvu II-Pst
`I eDNA fragment into Pvu II-Pst !-cleaved MPC!l. The first four VH amino acids a re deri ved
`from MPC ll , but a re identical to those found in the three cD NAs (24). The Eco RI fragment
`conta ining the dextran VH was j oined to a human IgG 4 constant region within the pSV2-gpt
`expression vector (25, 5). T he codin g sequ ences of the MPC ll and eDNA genes a re shown as
`solid a nd hatched lines, respecti vely. The crosshatched boxes represent th e coding sequences of
`the hum an IgG4 constant region. T he maps a re not drawn to scale. Restriction enzy mes have
`been abbreviated as follows: Eco RI, RI; Pvu II, PII; Pst I, PI; Bam HI, B; Sal I, S; and Sma I, SM.
`
`PFIZER EX. 1131
`Page 7
`
`

`

`1104
`
`VAR IABLE REGION GLYCO YLATION AFFECTS AFF I N ITY
`
`TABL E II
`No menclature fo r Transjectoma • and Hybridoma Antibodies
`
`Laboratory name of hybridoma/
`transfcctoma antibodi es
`
`arn e used in thi · stud yl
`
`14.6b. l
`19.22. 1
`5.54. 4.24. 1
`TK C3.2
`T H V8 .3
`TJ 8.5
`
`5.54
`Tl4.6b
`Tl 9.22
`T 5.54
`
`• An immortali zed antibody-sy nthesizing cell line produced by ge ne transfcc(cid:173)
`ti on and ex pression techniques .
`The names of the transfec toma cell lines have bee n shortened for ease and
`cla rit y in reading the text. The lett er T denotes transfectoma; the number fol(cid:173)
`lowin g lists the ori gin ofV 11 segment : 14.6b is deri ved from rnAb 14.6b. l ;
`19.22 from 19.22 . 1; 5.54 from 5.54.4.24. 1. A das h indicates that th e name
`has not bee n changed .
`
`an tibody could dramatically increase the apparent binding constant for dextran of
`the low-affinity ant ibody (data not shown). To avoid this we used Con A, which binds
`high-mannose and biantennary complex oligosaccharides (16), to separate un(cid:173)
`glycosylated from glycosylated Ig. Adso rption ex periments showed that the CHO
`in T 19.22 antibody was not accessible by binding to Con A--8eph arose (Fig. 3 A,
`lanes 3 and 5). In contrast, the T14.6b antibod y was adso rbed to Con A--8epharose
`(Fig. 3 A , lanes 2 and 4), thus the add itional CHO present in V H , unlike the CHO
`buried between CH2, must be accessible to binding by Con A. T he residual T14.6b
`antibody seen in the Con A supern atant (Fig. 3 A, lane 2) may refl ect our inability
`to separate the Con A slurry completely from th e culture fluid .
`TM treatmen t of both T14.6b and T19 .22 antibodies resulted in an electrophoretic
`mobility change consistent with the loss of CHO from the H chain (Fig. 3 B, lanes
`1-4). H chains that contain two, one, and zero N-linked C HO moieties (Fig. 3 B,
`
`A
`
`H-
`
`Fd ,K-
`
`'A
`Ji Ji
`g
`g
`g
`~
`~
`... ~ oO
`~
`.a
`.a
`C'4
`...
`C'4
`on .n
`0:.
`:!
`....
`....
`....
`....
`
`C'4
`C'4
`0:
`....
`
`B
`
`:a ~ :a ~
`:ii' ~
`g " g "
`g
`" ~ ~ ~ ~
`... ... N N
`14
`... " - - ' - r..........l
`.a
`~ "--'NN..A .a
`...
`oO ~
`oO
`on on ~ !?!
`.,; .,;
`:! :!
`....
`....
`....
`....
`....
`....
`....
`
`- H
`
`- Fd
`-K
`
`•
`
`FIGURE 2. SDS-PAGE analys is of immunoprecipitates obtai ned after papain di gestion of (A)
`[35S]Met or (B) [ t4C ]glucosamine- labeled secreted I g. [35S]Met-labeled a nd reduced Igs were
`used as m a rkers. I n A the T14.6b (Fab) and T14.6b sam ples were a nalyzed on a separate SDS(cid:173)
`PAGE gel.
`
`- H
`
`- Fd , Fe
`
`--K
`
`PFIZER EX. 1131
`Page 8
`
`

`

`WALLICK ET AL.
`
`1105
`
`B
`
`1234 56
`lH
`
`12345678
`-
`}H
`
`L
`
`L
`
`F IGURE 3. 12.5% Tris-glycine SDS-PAGE analysis of [35S]Met-labeled transfectoma culture su(cid:173)
`pernata nts, immunoprecipitated with rabbit anti-human Ig Fe a nti serum after Con A adso rp·
`tion (as indicated) and/o r tunicamycin treatment. Samples were reduced with 0.15 M 2- ME be(cid:173)
`fore SDS-PAGE. The positions of the H and L chains are indicated. (A) Con A-Sepharose adsorption
`of secreted transfectoma !g. (Lanes 1 and 6) Untreated Tl4.6b and Tl9.22 secreted Ig, respec(cid:173)
`tively. (La nes 2 a nd 4) Tl4.6b unbound, a nd Tl4.6b bound and eluted from Con A-Sepharose.
`(Lanes 3 a nd 5) Tl9.22 unbound, and bound a nd eluted from Con A-Sepha rose. (B) Tunicamyci n(cid:173)
`treated cell supe rn ata nts without o r with Co n A-Sepha rose adso rption. (Lanes 1 and 2) Tl4.6b
`before and after tunicamycin treat ment; (la nes 3 a nd 4) Tl 9.22 before and a fte r tunicamycin
`treatment; (la nes 5 and 6) tunicamycin-treated Tl4.6b Con A su pe rn a tant a nd eluate, res pec(cid:173)
`tively; (lanes 7 a nd 8), tuni camycin-treated Tl9.22 Con A superna ta nt and el uate, respecti vely.
`
`lanes 1, 3, and 2, or 4, respecti vely) can be resolved. The H chain bands of th e un(cid:173)
`treated samples (Fig. 3 B, 1 and 3) a ppear homogeneous, suggestin g that all H chains
`are un iformly glycosylated. From the lac k of visible glycosylated H chain bands in
`lanes 2 and 4, we estimate that TM treatment res ults in >9 7% deglycosylation of
`the I g. Lanes 5- 8 show th e results obtained from Con A adsorption of Tm-treated
`lg. Both the T14.6b and T19 .22 aglycosylated ant ibod ies were not bound by Con
`A (Fig. 3 B, 5 and 7). The faint bands that represent CHo - H chains in lane 6
`probably reflect nonspecific trapping in the Con A-seph arose slurry.
`H aving established that Con A adsorption could remove glycosylated contaminants
`from T14.6b Tm-treated preparations, we used Con A-adso rbed material for dex(cid:173)
`tran binding studies. The results from one typical ex periment are graphically illu s(cid:173)
`trated in Fig. 4. For the native T14.6b antibod y 50% inhibition of binding to ELISA
`plates coated with 0.5 or 20 11g/ml dextran was obtained when 1. 2 ~t g/ml of dex tran
`
`I Q z
`; • ::» • ;c
`
`~
`
`lE ...
`0
`~
`
`..,.....Tl4.6b!- Tm )
`HT14.6b!+Tm l
`-TI9.221- Tm l
`-n~2 ! + Tm l
`
`l nhibition by soluble
`F IGURE 4.
`dextran of antibody binding to dex(cid:173)
`tra n-coated ELI SA plates. Pe rcent(cid:173)
`age of a ntibody binding (o rdin a te)
`is plotted aga in st dextran inhibitor
`concentration
`(abscissa). Pla tes
`were coated wit h 20 ~ g/ ml dextran.
`Nat ive a ntibod ies and a ntibodies
`aglycosyla ted by tuni camyci n treat(cid:173)
`ment were used; trace quant ities of
`glycosy latcd Ig present in tun ica(cid:173)
`myc in treated Tl4.6b were removed
`by adsorp ti on to Con A-Sep ha rose.
`
`0.2
`
`PFIZER EX. 1131
`Page 9
`
`

`

`1106
`
`VARIABLE REGION GLYCOSYLATION AFFECTS AFFINITY
`
`inhibitor was added. CHO-depleted T14.6b antibody could not bind to 0.5 1-!g/m)
`dextran-coated plates (data not shown). U sing low-affinity binding conditions (micro(cid:173)
`titer wells coated with 20 11g/ml dextran) the aglycosylated T14.6b and T19.22 anti(cid:173)
`bodies and native T19.22 antibody showed half-maximal binding when 18-24 1-!g/m)
`dextran B512 inhibitor was added.
`The apparent association constants for Tm-treated aglycosylated and untreated
`native antidextran antibodies are summarized in Table III. The binding constant
`of the CHO-depleted T14.6b was 14-15-fold lower than the native antibody. In con(cid:173)
`trast, carbohydrate removed from the Fe ofT19 .22 did not affect that antibody's ability
`to bind antigen. All experiments except those noted were performed using an anti(cid:173)
`body concentration of 1 11g/ml; we observed a slight affect of antibody concentra(cid:173)
`tion on apparent aKa values. The aKa values determined using the inh ibition ELISA
`were, in general, slightly higher th an those obtained previously using affinity gel
`electrophoresis. For a discussion of affinity gel electrophoresis see Takeo and Kabat
`(17). However, the differences in binding strength between antibodies were similar
`using the two assays. We found a 32 -fold difference in binding affinity between the
`T14.6b and T19.22 antibodies, versus the 50-fold difference between the parental
`mAbs 14.6b.l and 19.22.1 reported (1). In summary, it is clear th at the presence of
`CHO within the antidextran V H region significantly affects its affini ty for antigen,
`however, we cann ot rule out an additional contribution of the altered amino acids
`to th e differences in binding.
`
`Discussion
`Antibodies are glyco proteins with all heavy chains containin g at least one and
`frequently several N-linked carbohydrate residues (18). The role postulated for car(cid:173)
`bohydrate found on the heavy chain constant regions includes solubilization of the
`H chain , facilitation of subcellular transport and secretion , promotion of assembly,
`
`TABLE III
`Apparent Binding Constants fo r Dextran B512
`
`H ybridoma or transfectoma a ntibod y
`
`aKa (tabulated
`from Fig. 4)*
`
`aKa (calcul ated from
`several ex periments)!
`
`Nu mber of
`experiments
`
`14.6b.1
`19.22.1
`T 14.6b (w ithout Tm)
`
`T14.6b (w ith Tm) Con A-adsorbed
`T 19.22 (w ith out Tm)
`
`T 19.22 (w ith Tm) Con A-ad sorbed
`
`ND
`ND
`
`1.1 X 105
`1.0 X 105
`
`8.3 x 104
`
`nzllg
`2.30 ± 0. 1 X 106
`ND
`1.68 ± 0.6 X 1061
`(2. 10 ± 0.3 X 106)
`1. 18 ± 0.04 X 105
`8.22 ± 3.6 X 10411
`(6.5 ± 0.3 X 104)
`1.09 ± 0.4 x 105
`
`4
`
`8
`5
`5
`10
`6
`4
`
`• Calcula ted from the reciprocal concentration of dex tran B512 necessary to inhibit 50 % of the maximal
`binding of ant ibody to dextran-coated plates. 1/ [Dex]l so has been doubled to give the fin a l aKa value be(cid:173)
`cause dext ra n inhibitor and antibody were added to mi crotiter wells at a 1:1 molar ratio.
`1
`l The aKa value represents an average obtained from the expe rim ents indi cated. The error for the sum rota
`of all the valu es is represented by the first standard deviation.
`S Antibody concen trat ion was 0.8 ~t g/m l.
`II Antibody concentration was 0.3 ~t g/ml.
`C ulture supernatants were not from tuni ca m yc in ex periments. Antibody concen tra ti on was 1 11g/rnl.
`
`PFIZER EX. 1131
`Page 10
`
`

`

`WALLIC K ET AL.
`
`1107
`
`maintenance oflg conformational features th at contribute to effector functions
`Carbohydra te can also be found within the V region of an antibody molecule.
`% of human myeloma light chains have carbohydrate within their variable regions
`In a study of 76 human IgG myelom a proteins, rv2 5% were shown to contain
`carbohydrate m oiety on the Fab fragm ent (21). The carbohydrate was linked to
`r the light ch ains or the Fd fragm ents, and in a few cases to both.
`In an earlier study, Matsuu chi et al. (22) isolated and characterized a spontane(cid:173)
`sly arisin g mutant of the myeloma j 558 [IgA, A. , anti-a(l - 3) a(1 - 6)dextran]
`decreased reactivity with polymeric dextran. The mutant differed from the wild(cid:173)
`in th at it had increased amounts of sialic acid on the carbohydrate in its Fab
`Since the vari able region of J 558 does not contain th e canonical carbohy-
`addition sequence, the altered carbohydra te probably resides wi thin the CH,
`· . The cha nge in carbohydrate content was the consequence of the altered
`ility of cellular enzym es involved in glycosyla tion .
`It has been well documented how vari ation in amino acid sequence of the variable
`contributes to antibod y dive rsity, and th e m any ge netic mecha nisms th a t can
`differen t amino acid sequ ences have been defin ed . In this report we dem(cid:173)
`directl y th at the presence of carbohydrate in CDR2 of V H is criti cal for
`the high-affinit y binding of a monoclonal antibody specific for polymeric a(l - 6)dex(cid:173)
`tran, and we infer th at th e carbohydrate also contributes to th e increased affinity
`for IM7. Thus not onl y the specific amino acid sequ ence of th e variable region, but
`also its carbohydrate moieti es can determin e the specificity a nd magnitude of the
`antigen-antibody interaction. Variable ex pression of glycosyltransferases could be
`used to modulate antibody binding.
`Of great interest is the mechanism by which presence of an oligosaccharide a t(cid:173)
`tached to amino acids in the combining site of the antibody 14.6b.llead s to increased
`Ka for both polym eric dextran and IM7 . X-ray crystallographic studies of unrelated
`antibodies predict th at th e residues to which the carbohydra te is attached in V H
`should be exposed on th e hypervari able loops. Our Con A binding ex periments also
`suggest tha t, in contrast to th e carbohydrate in C H 2, th e V H oligosaccha ride is rel(cid:173)
`atively exposed , a nd is positioned a t th e surface of the I g. Thus it is possible that
`the V H carbohydrate directly interacts with the anti gen; however, it is diffi cult to
`see how direct in teractions could occur both with polymeri c dextran and a site-filling
`oligosaccharide, IM 7.
`·
`A more likely ex planation for the effect of glycosylation is th at the carbohydrate
`linked to amin o acid 58 alters th e conformation of the combining site. Such altera(cid:173)
`tions might increase th e accessibility of th e Thr residue at position 60 in the 14. 6b.l
`V H region so th at it m ay contact the antigen m ore closely. Indeed , Feldman and
`coworkers have predicted from the hypotheti cal space-filling model of the V region
`of the galactan -binding myelom a I g j 539 that H chain Thr residue 56 may contact
`galactan (23). The X-ray crystallographic stru cture of the 14.6b.l Fab would aid in
`our understandin g of how the presence of carbohydrate affects the topology of the
`combining site.
`
`Summa r y
`We have observed th at an tidextran hybridom as with potential N-linked glycosyla(cid:173)
`tion sites in V H h ave higher affini ty for polym eric dextran and for isom altoheptaose
`
`PFIZER EX. 1131
`Page 11
`
`

`

`1108
`
`VARIABLE REGION G