`
`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
`
`Immunochemical characterization of antibodies against a(l -+ 6) dextran has given
`insights into the size and shape of the antibody-combining site and the nature of
`the interaction between antibodies and antigen. We are now attempting to correlate
`the immunochemical properties of the antidextran antibodies with their primary
`structure. In the course of these studies cDNAs from three monoclonal anti-a(l-+ 6)
`dextran hybridoma cell lines, 14.6b.1, 5.54 1 and 19.22.1 (1, 2), were cloned, and the
`nucleotide sequences of their V H and V L regions were determined (3) (Table I). All
`synthesize an identical K light chain with the V K-OXl germline gene ( 4) rearranged
`to the J K2 segment; the heavy chains differ by only one or two amino acids in their
`complementarity-determining regions (CDRs)2. When compared with 14.6b.1, 5.54
`and 19.22.1 have an identical Thr-+ Asn amino acid change at position 60 in VH;
`5.54 has an additional change (Ser-+ Gly) at position 31 in CDRl. The changes
`in heavy chain sequence result in 5.54 and 19.22.1 having a 10-fold or greater reduc(cid:173)
`tion in their binding constants for both polymeric dextran and isomaltoheptaose
`(IM7) when compared with 14.6b.1 (Table I).
`The Thr-+ Asn change in 5.54 and 19.22.1 leads to the loss of a potential N-linked
`glycosylation site (Asnss-Tyr59-Thr60) present in 14.6b.1. The purpose of this study
`was to determine whether this potential N-linked glycosylation site is used and if
`so, whether the addition of carbohydrate (CHO) to CDR2 affects the binding con(cid:173)
`stant for dextran. It is difficult to demonstrate glycosylation of VH in the original
`hybridoma antibodies since both IgA and lgM isotypes are glycosylated within their
`CHl domains and CHO present in Fd could be linked to either VH or CH. There(cid:173)
`fore, we have transferred the three V H regions to the human IgG4 constant region,
`which is devoid of CHO in its CHl domain. In this report we demonstrate the pres(cid:173)
`ence of carbohydrate within the V H of 14.6b.1. Comparison of the association con(cid:173)
`stants for aglycosylated tunicamycin (Tm)-treated and -untreated antibodies shows
`
`This work was supported in part by grants AI-19042 , CA-16858, CA-22736, and CA-13696 (to the Cancer
`Center) from the N ational Institutes of Health, and by grant DBM-860-0778 from the National Science
`Foundation. Address correspondence to Dr. Sherie L. Morrison, Department of Microbiology, 540 Mo(cid:173)
`lecular Biology Institute, University of California at Los Angeles, Los Angeles, CA 90024.
`1 The 5.54 mAb was designated as 5.54.4.24.l by Newman and Kabat (2).
`2 Abbreviations used in this paper: CDR, complementarity-determining region; CHO, ca rbohydrate;
`IM7, isomalto heptaose; Staph A, Staphylococcus aureus protein A; Tm, tunicamycin .
`
`J. ExP. MED. ©The Rockefeller University Press · 0022-1007/88/09/1099/11 $2 .00
`Volume 168 September 1988 1099-1109
`
`1099
`
`1 of 11
`
`Celltrion, Inc., Exhibit 1131
`
`
`
`(cid:52)(cid:89)(cid:70)(cid:80)(cid:77)(cid:87)(cid:76)(cid:73)(cid:72)(cid:4)(cid:55)(cid:73)(cid:84)(cid:88)(cid:73)(cid:81)(cid:70)(cid:73)(cid:86)(cid:4)(cid:21)(cid:16)(cid:4)(cid:21)(cid:29)(cid:28)(cid:28)
`
`-0
`
`0
`
`>E
`:;ti
`~ r
`
`tTl
`:;ti
`tTl
`C'J
`
`0 z
`C'J
`~
`0
`0
`[/} -<
`r
`~
`0 z
`> 'Tl
`'Tl
`tTl
`0 ...,
`> 'Tl
`~ z
`::i
`-<
`
`[/}
`
`TABLE I
`Immunochemical Properties of Hybridoma Antibodies Specific for Dextran B512
`
`Hybridoma
`
`Mouse strain
`
`I so type
`
`Site size
`
`K)
`
`Kia (IM7)l.s
`
`CDRl
`
`CDR2
`
`CDR3
`
`14.6b.1,
`5.54**
`19.22.1,
`
`BALB/c
`C57BL/6
`BALB/c
`
`IgA,k
`IgA,k
`IgM,k
`
`6
`6
`7
`
`mllg
`4.43 x 105
`1. 78 x 104
`8.87 x 103
`
`liter/mole
`5.76 x 104
`3.02 x 103
`6.46 x 10 3
`
`31 Ser __,. Gly
`-
`
`60 Thr __,. Asn
`60 Thr __,. Asn
`
`JH
`
`3
`3
`3
`
`Heavy Chain amino acid changes
`vs. 14.6b.1 prototypell
`
`• Maximum number of a(I _. 6)-linked glucose residues that fit the antibody combining site.
`l Determined by affinity gel electrophoresis according to the method described by Takeo and Kabat (17).
`I Association constants of antidextran combining sites with isomaltoheptaose (IM7).
`II According to Akolkar et al. (3).
`1 According to Sharon et al. ( 1 ).
`** According to Newman et al. (2); designated as 5.54.4.24. l by Newman ct al.
`These sequence data have been submitted to the EMBL/GenBank Data Libraries under accession number Y00809.
`
`2 of 11
`
`Celltrion, Inc., Exhibit 1131
`
`
`
`(cid:52)(cid:89)(cid:70)(cid:80)(cid:77)(cid:87)(cid:76)(cid:73)(cid:72)(cid:4)(cid:55)(cid:73)(cid:84)(cid:88)(cid:73)(cid:81)(cid:70)(cid:73)(cid:86)(cid:4)(cid:21)(cid:16)(cid:4)(cid:21)(cid:29)(cid:28)(cid:28)
`
`WALLICK ET AL.
`
`1101
`
`that the presence of CHO increases the aKa of 14.6b.l for dextran. The effect on
`binding is unique to the carbohydrate present in VH, since absence of CHO from
`CH2 does not change the aKa for dextran. Lastly, we have demonstrated that the
`CHO in V H is more exposed than in CH2.
`
`Materials and Methods
`Cell Lines.
`5.54 is a mouse hybridoma cell line synthesizing a C57BL/6 IgA, K antibody
`specific for a(l ._ 6) dextran. D3 is a spontaneous heavy chain-loss variant of 5.54 that syn(cid:173)
`thesizes only the K light chain characteristic of the antidextran hybridomas. The D3 light
`chain variant cell line was isolated by Dr. P. N. Akolkar (Columbia University, NY). Cell
`lines were grown in Iscove's Modified Dulbeccds medium (IMDM) (Gibco Laboratories,
`Grand Island, NY) supplemented with 3-5% FCS (Hyclone Laboratories, Logan, UT).
`Gene Transfection. Gene transfection was by protoplast fusion using the method of Oi et
`al. (5) and modified as described by Tan et al. (6). Transfectant culture supernatants were
`tested for antibody production and dextran binding by ELISA (7). Dextran B512 was prepared
`from Leuconostoc mesenteroides strain B512 cultures by Dr. L. Matsuuchi as described (8). Horse(cid:173)
`radish peroxidase affinity purified goat anti-human lgG antibody was purchased from Sigma
`Chemical Co. (St. Louis, MO). D3 recipient transfected cells from positive wells were sub(cid:173)
`cloned once in soft agarose (9), and clones that stained heaviest with rabbit anti-human IgG
`Fe antiserum (Cooper Biomedical, Inc., Malvern, PA) were chosen for further analysis.
`Biosynthetic R.adiolabeling and Papain Digestion. Transfectant cells were labeled in the presence
`of 15 µCi/ml of [35S]Met or 100 µCi/ml o-[ 14C]glucosamine hydrochloride as described (10).
`Secretions from the cells were digested with pa pain (Sigma Chemical Co.) at 1:100 enzyme/
`protein ratio for 4 hat 37°C. The reaction was stopped by addition ofiodoacetamide to 0.03
`M . The Fe fraction and undigested antibody protein were precipitated by incubation with
`IgG-Sorb (Enzyme Center, Malden, MA). Fab was precipitated from the supernatant using
`rabbit anti-human Fab (prepared by Letitia A. Wims, Columbia University, NY) or by in(cid:173)
`solubilized dextran (Sephadex G75). Samples were reduced with 2-ME (0.15 M) and ana(cid:173)
`lyzed using 5% SDS-PAGE (5).
`Inhibition of Glycosylation. Tm at a concentration of 8 µg/ml (Boehringer Mannheim Bio(cid:173)
`chemicals, Indianapolis, IN) was used to inhibit N-linked glycosylation. Cells were biosyn(cid:173)
`thetically labeled for 3 h with [35S]Met in the presence of Tm as described above. After
`pretreatment, secreted lg in the culture supernatant was discarded, the cells were washed
`twice with IMDM, fresh Tm and [35S]Met added, and treatment continued overnight at
`37°C. Removal of CHO from lg was verified by immunoprecipitation of the secreted anti(cid:173)
`body and analysis by SDS-PAGE.
`Determination of the Antibody Protein Concentration in Culture Supernatants. Antibodies in culture
`supernatants diluted into BBS (0.02 M borate-buffered 0.75% saline, pH 8.3) were bound
`to polystyrene microtiter wells (Corning Glass Works, Corning, NY) for 3 hat 37°C. After
`blocking any unreacted sites with 1 % BSA/PBS/0.05 % Tween 20 for 1 h at room tempera(cid:173)
`ture, the ELISA plates were washed with PBS/0.05% Tween 20 three times, PBS once, and
`then bound lg was quantitated by reaction with horseradish peroxidase-labeled anti-human
`IgG antibody and compared with a human IgG standard of known concentration. Assay
`results have been reproduced at least three times. Direct binding of antibody to microtiter
`plates was a more reproducible method than binding supernatants to plates sensitized with
`anti-human 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 antibody is defined
`as the reciprocal free ligand concentration necessary for occupying one half of the antibody(cid:173)
`combining sites. If a fixed amount of antibody is reacted with an increasing amount of free
`ligand on a plate coated with antigen, the reciprocal of the free ligand concentration that
`causes 50% inhibition of binding to the plate is considered to be a function of the intrinsic
`K . and is designated as the apparent affinity constant (aK.). The aKa is calculated from the
`
`3 of 11
`
`Celltrion, Inc., Exhibit 1131
`
`
`
`(cid:52)(cid:89)(cid:70)(cid:80)(cid:77)(cid:87)(cid:76)(cid:73)(cid:72)(cid:4)(cid:55)(cid:73)(cid:84)(cid:88)(cid:73)(cid:81)(cid:70)(cid:73)(cid:86)(cid:4)(cid:21)(cid:16)(cid:4)(cid:21)(cid:29)(cid:28)(cid:28)
`
`1102
`
`VARIABLE REGION GLYCOSYLATION AFFECTS AFFINITY
`
`amount ofligand necessary for 50% inhibition of binding. The following experimental con(cid:173)
`ditions were used to measure the aKa values: Corning microtiter plates were coated with 0.5
`µg/ml or 20 µg/ml dextran B512 (high-affinity and low-affinity assay conditions, respectively).
`Bound lg was quantitated using anti-human IgG labeled with horseradish peroxidase.
`
`Results
`The expressed V H regions from the three hybridoma antibodies against a(l -+
`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(cid:173)
`fectoma antibodies used in this study are presented in Table II.
`To determine if the 14.6b.1 chimeric antibody contained CHO in VH, we frac(cid:173)
`tionated the molecule into Fab and Fe by papain cleavage, reduced the molecules
`with 2-ME, and analyzed them on 5% SDS-PAGE gels. Proteins were labeled with
`[ 35S]Met, and the Fab was precipitated using specific anti-Fab antiserum (Fig. 2
`A). Transfectoma antibodies with VH derived from 5.54 and 19.22.1 cDNA 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 Kand Fe are present (data not shown). In contrast, in transfectoma anti(cid:173)
`bodies with the H chain variable region of 14.6b.1 (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 VH ofT14.6b, we labeled secreted lg with
`14C]glucosamine, prepared Fab and Fe fractions, and analyzed the products by
`[
`SDS-PAGE (Fig. 2 B). As anticipated, the 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 lgG Fe fragment that contains N(cid:173)
`linked CHO within its CH2 domain (12). However, the Fab from only T14.6b, with
`its Fd containing the 14.6b.1 VH, shows glucosamine labeling. The reduced 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(cid:173)
`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 VH 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 transfectoma
`antibodies. Although Tm is a potent inhibitor of N-linked glycosylation (15), it is
`difficult to produce proteins completely free of glycosylated species. From recon(cid:173)
`struction experiments it was apparent that even a trace contamination of high-affinity
`
`4 of 11
`
`Celltrion, Inc., Exhibit 1131
`
`
`
`\'li'ALLlCK &T Al.
`
`1103
`
`MPC 11
`
`, .. , .. .,
`
`Liq-ate
`
`l
`
`l l
`
`'
`
`l
`
`cONA
`•n
`"'"'
`
`'
`
`.... ,,,,
`
`'
`
`•
`
`E Coa- 0111
`
`~·
`
`•• IOI.lit& 1. Sut.iu11ion u( the g'1'110inUC 'I/ H rtgioo wtrb v M c.ONA •nd borype switdl. A acmrnk
`E<o RJ ftll$mC'nt wic•il'ling tlw: ).;IPC11 H chllin promoccr, leader 1eq1,WnCe, ~,,_,~ V ~
`ginn. a nd lg enh.anc:cr (2i) ... as do111Cd liltt> 11\e f.co R I U1col a pflR'22 dcrivari"'I! £rom .. ·hich
`the acqu.-nca lt'il'lg bctweoen ihc Hil)d; Ill tiK- (nutlcoul.k 29} a.rid eke PY\I It die ( n.\ldco1idc:
`2.066) Ii.ad bctn dtltted. Us.in1 cONA produced from 1he anti-.( I - 6)dt.xuu hybr\do111N(3),
`dw V repon of dMi t-i PCU -. ~p)•oed by the a•1idt:11.tN.A V usion by inwning the- P\-u ll·P11
`I ' ONA fragmcn1 into Pvu ll-J>s1 J~kk\,.d ~t PCll . The: fine (ou.r V11 am int) acid1 •re dttivcd
`(rolJI MPCll. but 11~ idimti«J 10 tho..- Mund in the 1httt cDN'A• (24). The Eco RI ("'61111:n1
`coota:ning tM dcltlran V H w:a1 jolr1cd to a hum•n lirC, conMaru tqion within the pSV'l· gpc
`exf>":Sllion we1or (2.5, S). Th.- coctmg ~enc.:. ol the M.PCIJ •nd cl).'IA ~es •tt 11hown u
`llObd and hatdlcd lina, rnpettl\tly. The Cl'()lt:h~ched boxes rtp:tefern the- ~'<>ding ~11elt(U qi
`chc ham.an lgG1 eorut u t region, 1111: n1apt IHI'! 001 dr-n. to Killk!. Rttlrinion cn:ii:~ ha....(cid:173)
`bocn ~·i.ucd .u follows: Eco R I, Jll; Pvu II, PU; 1'11 I. Pl: &i1n H I, !%Sal I, S; andSm1 I, S~.
`
`5 of 11
`
`Celltrion, Inc., Exhibit 1131
`
`
`
`l 104
`
`VARIABl.J: Rf..CION Cl.\'COSVt.A'l'ION Art1::c;1'S AFFINl1'Y
`
`TAllLI!. II
`/t.~rl4lDrr fe Tumsf«UW. • o"d lifl'irlorM A •tiW'tt
`I,~ name or ltybridlltll;a/
`1rantlcc10m.a antibodit~
`14 6b.I
`19.22. 1
`5.54.•.24. I
`S.54
`l'XCJ.?
`·r14 6b
`1'M\13.ll
`T tt.11
`TJC8.S
`T$!i 4
`' An l111111CJ111di«\l 11111itlud:y .. rntht1i:d.1tg ( d i lilll' produc:td by &"tM r.raMl'l)C·
`•ion ""d e.:pn'HiOft tt'(h11oiqucs.
`: Tlw nll!Ms ol ehc cnndttioma ~ l!n.tt hav<; b«fl 1h1>11!'.nrd rw call:' and
`ob.d1r !n tt:odi"& 1~ int 1'ht 9"ttr ·r dcrwtff traosk<toma: d~ nvmbrr fol·
`IU¥oini Ii.so t.ht oticin ol v .. -CR'-1! 14.(ib It <ltr+o.~d O<oc11 mAb 14 Qi I;
`lt .22 fl.on, 19.22. l: !I~· from !i 54.4 24 I A dash i•i<-att• U.11 d•<' n1unt
`bu !IQ( ba!n ~ ·
`
`antibody could dram.atica.lly increa11e the .appuent binding constant for dextran of
`the l°"'-affinity at11ibody (data not thown). ·to avoid this we used Con A, which binds
`high·mannose and biantennary complex olipaccharidcs (16), to separau: un·
`glycosylated from glyoosylated lg. Adtorpcion experime1lts shO\lo-ed that the CHO
`i n Tl9.2'2 antibody was not otCCC$$-ibJe by binding to Con ;\.o&pharoi;e (Fig, 3 A,
`lanes :J and 5). In contrast, the •rt4.6b antibody was adsorbed to Con A~pharosc
`( Fig. 3 A. lann 2 iind 4), thus the additional CHO present in V11 . unlike the CHO
`buried bef....·een CH2, mu!R be accessible to binding by Con A. The rc9idual 'rJ4.6b
`0tntibody seen in lhc Con A supernatant (Fig. 3 A. lane 2) may re-Beet our inabiJity
`to separa1e the Con A s-Jurl')' completely from the culture 8uid.
`Trvt treatment ofbtxh Tl4'.6b andTl9.22 a.ntibodies rt"Whc:d in om c lectrophoretii.:
`mobility change consistent with the lou of CHO f.rom the H chain (Fig. 3 8, lanes
`1-4). H ch11.ins tha1 contain tw~ one, 11.nd xcro N·linled CHO moieties ( f"ig. 3 B,
`
`A
`
`7
`7
`~
`
`~
`
`t.
`
`I
`
`•
`~~f;'f;'l'];' ..
`" " ~ ;; ..
`'.i
`~-".i.--
`-
`..
`•
`" " .. "!
`• ~
`1vi!i::.:
`.. .. 'l ..
`• - • • - • • • • • --
`.. .. !! !! :! :! ::!
`.:::"" -~ "!
`H-
`· -H _,..
`·-·
`• -· -F·d,fc
`fd,• - -
`•
`
`- -H
`
`-~·
`
`F1ou11a 2. SOS-PACE 1111Uy.t. ol l1111t:111nopteelpil•1ff oblaiDed .after p11pai11o di!f\'uion o( {A)
`(»$1).iet or {B) 1uClglutONmil1f'olal)(-kid tttftttd lg. IS&Sl~tet·labtkd Md rcdU(lrd lg• were
`uwd u mukcra. (DA 1hc Tl4.6b ('F'ab) and TI4,6t,i 1am~ W\'rt at1aly~ ~•~~ne e SOS.
`PAGEgd.
`
`6 of 11
`
`Celltrion, Inc., Exhibit 1131
`
`
`
`A
`
`1 2 3 4 5 6
`
`-
`- IH
`- - - - L
`
`\\'Alli CK i:r AL.
`
`1105
`
`B
`
`12345678
`IH
`
`-
`
`L
`
`f.' 1ouu 3. 12.5" llis·tl)'C:ilw SOS.PA.OP. »1aJym of J~JMel•latxkd 1nu:uht01n.a cWtutt su·
`pem.alallll, lmm\ll'IOf>""t'clpitaled with rabbi: aflti- h\lm.an lg ~ antil«'Um after Con A adsotp·
`'°"' SD5-PAOE. Tbt JIOlliciC111$ of 1he H ~ L ch8in. a.r.i indica.t~ (,A) C.0. A-&pb•~ lldtorpQon
`11011 (u indiated) Mdlor tunie&m)'cin t n'iumc;m. Samples l¥ef'C ttdl!ced w\1h 0.1$ M 2·ME be·
`or ~med traiuk«om• lg. (Lanes I a11d 6) VitUUtcd TJ4.6b and T t9.22 tecnoted 15, rupee·
`tl,,dy (1.-nn 2 and 4) T l4.6b unbouiM., and ·n-t.6b bound 111d eluted from Con A-8ept..l'QK.
`(l.anl!ll J 1111d $) 1')~.22 urbouiid, tnd bound and cliu«d from Con A•Sephuoic, (8) Tunitwn)'cin·
`lff'.Jtlftl ( 1eU tu~f\*1.11nu wi11v:)....t Or wlth Con A~lll"Ote ad.orpt~. (L.IJlct I and 1) TI4.6b
`be!orc! 11nd aftn- ru.nJeamydn 1tntmm1; (Iii.net J and 4) 1'rt.22 bd"o« and 1/le:r (unleam}"titl
`u·u~n•, (la.rin 5 and 6) 111n~m~i""'"1~1 «1 ·r1•.6b Co" A •uP":m•uit and duatc, twpcc·
`1i-dy; (la.11\:6 1 and 4), tunicv.[t1)'Cin·uu1cd Tl9.22 Coo A supcrn11tan1 and ()11.a1c. rapeoiiYd)t
`
`lanes I, 3, and 2, ot I. respe.:tiYC.I>·) can be resol\-«I. The Ji c.hain bands or the un(cid:173)
`treated sampln(Pig. 3 8. 1and3) appear homogc.neous, sugge1ting that aJJ H chajns
`are uniformly glycosylated. From the lac.k of visible gi)'COS)'latcd H chain bands in
`l<1nc:1 2 and I, we Gtimate t hat TM t reatment results in >97% deglycosyladon of
`t he lg. l .. anes S-8 show the reauhs obtained from Con A adsorption orTm-1reated
`lg, Both 1bc 1' J4.6b and ~rl9.22 agl)'CO!lyla1ed a ntibodies were not boond by Con
`A (fig. 3 B • .5 and 7). Tht faint bands that represent CJ-10- t-f chains in lane 6
`probably rcftect nonspecific 1rapping in 1he Con A- sepharose slurry.
`Having tSCablished th.at Con A 11dsorpcion could rtmO\'C gl)'t'.mylatcd contaminants
`from Tl4.6b Tm·trc-ated preparations, we used Con A-adsorbed ni.aterial for do:·
`tran binding scudics. The result• fron1 one t ypical experiment a rc graphically illus(cid:173)
`trated in Fig. f . For the native 'tl 4.6b antibody !>0% inhibition of binding 10 ELISA
`plates coated with 0.5 or 20 ~g/ntl dcxttan Yt;U obtained whe n 1.2 s.ag/ ml of dextran
`
`lnhibt<IOfl by ~olublc
`F1ov11t 4.
`~ ... n ofa11o11bodybindi~todu·
`tr1u1...:oa1cd F.LJSA phu"- Pt:rttnl•
`a~o( 1u1tibody bindin_g (().Tdlr1atc)
`"pklclcd 11pin11~nn in!Ubilur
`onne.;"1"*1io11 (al)&d...,), Pl.i<•
`wett ~td "'•1lh ?(} pg/mJ dei!U'IUI.
`Native ;uuibodieJ .an.cl 11n1ibMW,.
`li(l)'('OIJ)'liwnd by iunk.a11'1)'(lfl trtJll •
`nient were wio:d( lr'al..'C ~u•m i1ki u(
`d)"~lliued lg prew.n1 in runic.a·
`"l)'WI UU!cd 11f.fib~ ttrnO.'Cd
`by .ildaol'p(ion t(I Col'I A-&phMtlt'(',
`
`o>
`
`7 of 11
`
`Celltrion, Inc., Exhibit 1131
`
`
`
`(cid:52)(cid:89)(cid:70)(cid:80)(cid:77)(cid:87)(cid:76)(cid:73)(cid:72)(cid:4)(cid:55)(cid:73)(cid:84)(cid:88)(cid:73)(cid:81)(cid:70)(cid:73)(cid:86)(cid:4)(cid:21)(cid:16)(cid:4)(cid:21)(cid:29)(cid:28)(cid:28)
`
`1106
`
`VARIABLE REGION GLYCOSYLATION AFFECTS AFFINITY
`
`inhibitor was added. CHO-depleted T14.6b antibody could not bind to 0.5 µg/ml
`dextran-coated plates (data not shown). Using low-affinity binding conditions (micro(cid:173)
`titer wells coated with 20 µg/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 µg/ml
`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 µg/ml; we observed a slight affect of antibody concentra(cid:173)
`tion on apparent aKa values. The aKa values determined using the inhibition ELISA
`were, in general, slightly higher than 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.1 and 19.22.1 reported (1). In summary, it is clear that the presence of
`CHO within the antidextran V H region significantly affects its affinity for antigen,
`however, we cannot rule out an additional contribution of the altered amino acids
`to the differences in binding.
`
`Discussion
`Antibodies are glycoproteins with all heavy chains containing 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 for Dextran B512
`
`Number of
`aKa (calculated from
`aKa (tabulated
`several experiments )l
`from Fig. 4)*
`Hybridoma or transfectoma antibody
`--'-------------'-----~-~------~--~---ex-'-periments
`mllg
`2.30
`
`106
`
`± 0.1 x
`ND
`106§
`± 0.6 x
`106 ),
`± 0.3 x
`10 5
`± 0.04 x
`10411
`± 3.6 x
`104)1
`± 0.3 x
`105
`± 0.4 x
`
`1.68
`(2.10
`1.18
`8.22
`(6.5
`1.09
`
`4
`
`8
`5
`5
`10
`6
`4
`
`14.6b.1
`19.22.1
`T14.6b (without Tm)
`
`ND
`ND
`1. 7 x 106
`
`T14.6b (with Tm) Con A-adsorbed
`T19.22 (without Tm)
`
`T19.22 (with Tm) Con A-adsorbed
`
`1.1 x 105
`1.0 x 105
`
`8.3 x 104
`
`•Calculated from the reciprocal concentration of dextran B512 necessary to inhibit 50% of the maximal
`binding of antibody to dextran-coated plates. 1/(Dex]150 has been doubled to give the final aKa value be(cid:173)
`cause dextran inhibitor and antibody were added to microtiter wells at a 1: 1 molar ratio.
`t The aKa value represents an average obtained from the experiments indicated. The error for the sum total
`of all the values is represented by the first standard deviation.
`§ Antibody concentration was 0.8 µg/ml.
`II Antibody concentration was 0.3 µg/ml.
`1 Culture supernatants were not from tunicamycin experiments. Antibody concentration was 1 µg/ml.
`
`8 of 11
`
`Celltrion, Inc., Exhibit 1131
`
`
`
`(cid:52)(cid:89)(cid:70)(cid:80)(cid:77)(cid:87)(cid:76)(cid:73)(cid:72)(cid:4)(cid:55)(cid:73)(cid:84)(cid:88)(cid:73)(cid:81)(cid:70)(cid:73)(cid:86)(cid:4)(cid:21)(cid:16)(cid:4)(cid:21)(cid:29)(cid:28)(cid:28)
`
`WALLICK ET AL.
`
`1107
`
`and maintenance oflg conformational features that contribute to effector functions
`(19). Carbohydrate can also be found within the V region of an antibody molecule.
`15 % of human myeloma light chains have carbohydrate within their variable regions
`(20). In a study of 76 human IgG myeloma proteins, rv25 % were shown to contain
`a carbohydrate moiety on the Fab fragment (21). The carbohydrate was linked to
`either the light chains or the Fd fragments, and in a few cases to both.
`In an earlier study, Matsuuchi et al. (22) isolated and characterized a spontane(cid:173)
`ously arising mutant of the myeloma J558 [IgA, A., anti-a(l -+ 3) a(l -+ 6)dextran]
`with decreased reactivity with polymeric dextran. The mutant differed from the wild(cid:173)
`type in that it had increased amounts of sialic acid on the carbohydrate in its Fab
`region. Since the variable region of J558 does not contain the canonical carbohy(cid:173)
`drate addition sequence, the altered carbohydrate probably resides within the CH1
`domain. The change in carbohydrate content was the consequence of the altered
`availability of cellular enzymes involved in glycosylation.
`It has been well documented how variation in amino acid sequence of the variable
`region contributes to antibody diversity, and the many genetic mechanisms that can
`generate different amino acid sequences have been defined. In this report we dem(cid:173)
`onstrate directly that the presence of carbohydrate in CDR2 of V H is critical for
`the high-affinity binding of a monoclonal antibody specific for polymeric a(l -+ 6)dex(cid:173)
`tran, and we infer that the carbohydrate also contributes to the increased affinity
`for IM7. Thus not only the specific amino acid sequence of the variable region, but
`also its carbohydrate moieties can determine the specificity and magnitude of the
`antigen-antibody interaction. Variable expression of glycosyltransferases could be
`used to modulate antibody binding.
`Of great interest is the mechanism by which presence of an oligosaccharide at(cid:173)
`tached to amino acids in the combining site of the antibody 14.6b.1 leads to increased
`Ka for both polymeric dextran and IM7. X-ray crystallographic studies of unrelated
`antibodies predict that the residues to which the carbohydrate is attached in V H
`should be exposed on the hypervariable loops. Our Con A binding experiments also
`suggest that, in contrast to the carbohydrate in CH2, the V H oligosaccharide is rel(cid:173)
`atively exposed, and is positioned at the surface of the lg. Thus it is possible that
`the V H carbohydrate directly interacts with the antigen; however, it is difficult to
`see how direct interactions could occur both with polymeric dextran and a site-filling
`oligosaccharide, IM7.
`A more likely explanation for the effect of glycosylation is that the carbohydrate
`linked to amino acid 58 alters the conformation of the combining site. Such altera(cid:173)
`tions might increase the accessibility of the Thr residue at position 60 in the 14.6b.1
`VH region so that it may contact the antigen more closely. Indeed, Feldman and
`coworkers have predicted from the hypothetical space-filling model of the V region
`of the galactan-binding myeloma lgJ539 that H chain Thr residue 56 may contact
`galactan (23). The X-ray crystallographic structure of the 14.6b.1 Fab would aid in
`our understanding of how the presence of carbohydrate affects the topology of the
`combining site.
`
`Summary
`We have observed that antidextran hybridomas with potential N-linked glycosyla(cid:173)
`tion sites in V H have higher affinity for polymeric dextran and for isomaltoheptaose
`
`9 of 11
`
`Celltrion, Inc., Exhibit 1131
`
`
`
`(cid:52)(cid:89)(cid:70)(cid:80)(cid:77)(cid:87)(cid:76)(cid:73)(cid:72)(cid:4)(cid:55)(cid:73)(cid:84)(cid:88)(cid:73)(cid:81)(cid:70)(cid:73)(cid:86)(cid:4)(cid:21)(cid:16)(cid:4)(cid:21)(cid:29)(cid:28)(cid:28)
`
`1108
`
`VARIABLE REGION GLYCOSYLATION AFFECTS AFFINITY
`
`than those lacking potential glycosylation sites. In these studies we have used gene
`transfection and expression techniques to verify that the carbohydrate addition sites
`in V H were used. The carbohydrate of the V H region was accessible for binding by
`the lectin Con A. By ELISA analysis it was demonstrated that the aKa of the anti(cid:173)
`body for dextran was influenced by the presence of carbohydrate in V H, with the
`aglycosylated antibody having an aKa 15-fold lower than its untreated counterpart.
`The aKa for antigen of antibodies that contain carbohydrate only in their constant
`region was unaffected by lack of carbohydrate. Thus, not only the amino acid se(cid:173)
`quence of the variable region but also its carbohydrate moieties can determine the
`magnitude of the antigen-antibody interaction.
`
`We are indebted to Dr. Pradip Akolkar, Columbia University, NY, who donated cDNA clones
`and L chain variant cell lines. We thank Drs. S. K. Sikder and T. Matsuda for helpful conver(cid:173)
`sations concerning carbohydrate chemistry. We appreciate Dr. John Bergman n's donation
`of Con A-Sepharose, and for informing us that biantennary complex carbohydrates could
`be adsorbed to this lectin. Again we thank Steve Brown for pointing out the potential glycosy(cid:173)
`lation site in CDR2. We thank Dr. B. F. Erlanger for letting us use his ELISA reader. We
`greatly appreciate Marian Olsen's assistance in the preparation of this manuscript.
`
`Received for publication 2 March 1988 and in revised form 26 May 1988.
`
`References
`1. Sharon, J., E. A. Kabat, and S. L. Morrison. 1982. Association constants of hybridoma
`antibodies specific for o.(1 -+ 6)-linked dextran determined by affinity electrophoresis.
`Mol. lmmunol. 19:389.
`2. Newman, B. A., and E. A. Kabat. 1985. An immunochemical study of the comining
`site specificities of C57BL16J monoclonal antibodies to o.(1 -+ 6)-linked dextran B512.
`}. lmmunol. 135:1220.
`3. Akolkar, P. N., S. K. Sikder, S. B. Bhattacharya, J. Liao, F. Gruezo, S. L. Morrison,
`and E. A. Kabat. 1987. Different V Land V H germ-line genes are used to produce similar
`combining sites with specificity for o.(1 -+ 6)dextrans.}. lmmunol. 138:44 72. Errata cor(cid:173)
`rected 198 7, 139:3911.
`4. Griffiths, G. M., C. Berek, M. Kaartinen, and C. Milstein. 1984. Somatic mutation
`and the maturation of immune response to 2-phenyloxazolone. Nature (Lond.). 312:271.
`5. Oi, V. T., S. L. Morrison, L.A. Herzenberg, and P. Berg. 1983. Immunoglobulin gene
`expression in transformed lymphoid cells. Proc. Natl. Acad. Sci. USA. 80:825.
`·
`6. Tan, L. K., V. T. Oi, and S. L. Morrison. 1985. A human-mouse chimeric immunoglob(cid:173)
`ulin gene with a human variable region is expressed in mouse myeloma cells.]. lmmunol.
`135:3564.
`7. Chen, H. T., S. D. Makover, and E. A. Kabat. 1987. Immunochemical studies on mono(cid:173)
`clonal antibodies to stearyl-isomaltotetraose from C58/J and a C57BL/10 nude mouse.
`Mol. Immunol. 24:333.
`8. Matsuuchi, L., and S. L. Morrison. 1978. Estimation of antibodies specific for dextran.
`]. Immunol. 121:962.
`9. Coffino, P., R. Laskov, and M. D. Scharff. 1970. lmmunoglobulin production: method
`for quantitatively detecting variant myeloma cells. Science (Wash. DC). 167:186.
`10. Morrison, S. L. 1979. Sequentially derived mutants of the constant region of the heavy
`chain of murine immunoglobulins.]. Immunol. 123:793.
`11. Nieto, A., A. Gaya, M. Jansa, C. Moreno, and J. Vives. 1984. Direct measurement of
`
`10 of 11
`
`Celltrion, Inc., Exhibit 1131
`
`
`
`(cid:52)(cid:89)(cid:70)(cid:80)(cid:77)(cid:87)(cid:76)(cid:73)(cid:72)(cid:4)(cid:55)(cid:73)(cid:84)(cid:88)(cid:73)(cid:81)(cid:70)(cid:73)(cid:86)(cid:4)(cid:21)(cid:16)(cid:4)(cid:21)(cid:29)(cid:28)(cid:28)
`
`WALLICK ET AL.
`
`1109
`
`antibody affinity distribution by hapten-inhibition enzyme immunoassay. Mo/. Immunol.
`21:537.
`12. Silverton, E.W., M.A. Navia, and D.R. Davis. 1977. Three-dimensional structure of
`an intact human immunoglobulin. Proc. Natl. Acad. Sci. USA. 74:5140.
`13. Anderson, D. R., P. Samaraweera, and W. J. Grimes. 1983. Incomplete glycosylation
`of Asn 563 in mouse immunoglobulin M. Biochem. Biophys. &s. Comm. 116:771.
`14. Robbins, P. W., S. C. Hubbard, S.J. Turco, and D. F. Wirth. 1977. Proposal for a common
`ol