Proc. Natl. Acad. Sci. USA
`Vol. 88, pp. 4181-4185, May 1991
`Immunology
`
`Reshaping a therapeutic CD4 antibody
`(humanized antibody/chimeric antibody/to|erance/autoimmune disease/transplantation)
`
`SCOTT D. GORMAN, MICHAEL R. CLARK, EDWARD G. ROUTLEDGE, STEPHEN P. COBBOLD,
`AND HERMAN WALDMANN
`
`Department of Pathology. University of Cambridge. Tennis Court Road, Cambridge CB2 1QP. United Kingdom
`
`Communicated by Cesar Milstein, February I 1, 1991
`
`An immunosuppressive rat antibody (Cam-
`ABSTRACT
`path-9) against human CD4 has been reshaped for use in the
`management of autoimmunity and the prevention of graft
`rejection. Two different forms of the reshaped antibody were
`produced that derive their heavy chain variable region frame-
`work sequences from the human myeloma proteins KOL or
`NEW. When compared to a chimeric form of the CD4 anti-
`body, the avidity of the KOL-based reshaped antibody was only
`slightly reduced, whereas that of the NEW-based reshaped
`antibody was very poor. The successful reshaping to the
`KOL-based framework was by a procedure involving the
`grafting of human framework sequences onto the cloned rodent
`variable region by in vitro mutagenesis.
`
`At present, unwanted immune responses in autoimmunity
`and graft rejection are managed by the long-term adminis-
`tration of immunosuppressants such as steroids, azathio-
`prine, and cyclosporine. Patients receiving such long-term
`therapy are at continuous risk of infection and unwanted side
`effects of the drugs. An ideal alternative to sustained immu-
`nosuppression would be to establish a state of immunological
`tolerance to the inciting antigens. Here the intent is to provide
`short-term treatment to modulate the host immune system
`such that antigen—responsive T cells are either deleted or
`rendered anergic. In rodents, a short course of therapy with
`CD4 and CD8 monoclonal antibodies can tolerize to antigens
`as diverse as bone marrow, skin, and heart grafts (1-4) as well
`as preventing induction of a wide range of experimental
`autoimmune diseases (4-6). Remarkably, large doses of rat
`CD4 antibodies administered to mice can induce tolerance to
`themselves, thus avoiding an antiimmunoglobulin response
`that might neutralize their biological activity. However,
`lower doses fail to self-tolerize although they can still be
`tolerogenic for other antigens given under the CD4 umbrella
`(5, 6). In humans, rodent CD4 antibodies have thus far proven
`quite immunogenic despite their immunosuppressive prop-
`erties (7, 8). This problem can be minimized by reshaping the
`rodent antibody to a human form. In this way a human
`antibody is created that contains only the six complementa-
`rity-deterrnining regions (CDRs) from the heavy and light
`chain variable (VH and VL) regions of the rodent antibody of
`interest (9). To maximize the opportunities to use CD4
`antibodies for tolerance therapy, we have converted a known
`immunosuppressive CD4 monoclonal antibody of rat origin
`into a human form. The rat form of this antibody, Campath-9
`(Wellcome Foundation), was demonstrated to be therapeu-
`tically useful
`in combination with Campath-1H (CDw52)
`antibody to achieve a long-lasting remission in a patient with
`autoimmune systemic vasculitis (10).
`Reshaping antibodies is a relatively new procedure where
`success cannot necessarily be guaranteed for any individual
`antibody. Here, we describe a further approach of reshaping
`
`The publication costs of this article were defrayed in part by page charge
`payment. This article must therefore be hereby marked ‘‘advertisement‘’
`in accordance with 18 U.S.C. §1734 solely to indicate this fact.
`
`
`4181
`
`that grafts onto a rodent variable (V) region the framework
`sequences from a human V region that is most homologous
`to that of the rodent. We compare the effectiveness of two
`reshaped versions of the rat Campath-9 antibody, where one
`has derived its human VH region framework sequence from
`the myeloma protein KOL and the other from NEW.*
`
`MATERIALS AND METHODS
`
`cDNA Cloning. cDNA encoding the VL and V“ regions
`were generated by a polymerase chain reaction (PCR)-based
`method (11) except that the primer 5 ’—TGA GGA GAC GGT
`GAC CGT GGT CCC TTG GCC-3' was substituted for
`VHIFOR and 37°C and 50°C PCR annealing temperatures
`were used for VL and V“ region cDNA amplifications,
`respectively. V._ and VH cDNA regions were cloned into the
`M13-based vectors M13-VKPCR1 and M13-VHPCR1, re-
`spectively, as described (11). VL region clones were screened
`by hybridization with a 32P-labeled probe (5’-GT1" TCA TAA
`TAT TGG AGA CA-3’) specific for CDR 3 of the Y3-Ag1.2.3
`VL region cDNA (12); clones not hybridizing to this probe,
`and V“ region clones, were sequenced by the dideoxy
`method (13).
`Construction of Genes for Chimeric Antibodies. The plas-
`mid pVHrat/C51 encoding the chimeric heavy chain consists
`of the following adjacently ligated fragments: the 6.6-kilobase
`(kb) vector pHBAPr-1 (14) linearized at its cloning site with
`HindIII and BamHI containing the [3-actin promoter, xan—
`thine-guanine phosphoribosyltransferase, and ampicillin-
`resistance genes; a 39-base—pair (bp) HindIII—Nco I synthetic
`linker fragment (5'-AAG CTT TAC AGT TAC TGA GCA
`CAC AGG ACC TCA CCA TGG-3’); a 698-bp Nco I—BamHI
`fragment from an M13-VHPCR1 clone containing the Cam-
`path-9 antibody V3 region cDNA; a 2.3-kb BamHI—Sph I
`fragment containing the human G1 constant (Cm) region gene
`(15); and a 20-bp Sph I—Bgl II synthetic linker fragment
`(5’-GCA TGC GCG GCC GCA GAT CT-3’).
`The plasmid pV._rat/C,, encoding the chimeric light chain
`consists of the following adjacently ligated DNA fragments:
`a 7.5-kb BamHI—HindIII fragment from the plasmid pLD9
`(16) containing the B-actin promoter, dihydrofolate reduc-
`tase, and ampicillin-resistance genes; the 39-bp HindIII—Nco
`I synthetic linker; a 587-bp Nco I—BamHI fragment from an
`M13-VKPCR1 clone containing the Campath-9 antibody VL
`region cDNA; and a 4.9-kb BamHI fragment containing the
`human K constant (CK) region gene (17).
`Construction of Genes for Reshaped Antibodies. The plas-
`mid pVHNEW/CG, encoding the NEW-based reshaped
`heavy chain is identical to the plasmid pNH316 (16) except
`that the three CDRS of the Campath-9 antibody VH region
`
`Abbreviations: V, variable; VH, heavy chain variable; VL, light chain
`variable; CG1, G1 constant; C,, K constant; CDR, complementarity-
`determining region; PCR, polymerase chain reaction; ADCC, anti-
`bodydependent cell-mediated cytotoxicity.
`*The sequences reported in this paper have been deposited in the
`GenBank data base (accession nos. M61884 and M61885).
`Mylan v. Genentech
`Mylan v. Genentech
`IPR2016-01694
`IPR2016-01694
`Genentech Exhibit 2023
`Genentech Exhibit 2023
`
`

`
`4182
`
`Immunology: Gorman et al.
`
`Proc. Natl. Acad. Sci. USA 88 (I991)
`
`were CDR-grafted (9) into the 1.5-kb HindIII fragment from
`pNH316 encoding the Campath-1H antibody heavy chain by
`in vitro mutagenesis with three oligonucleotides.
`The KOL-based reshaped VH region was created by in
`vitro mutagenesis of a Campath-9 antibody VH region cDNA
`clone in M13-VHPCR1 with five oligonucleotides that were
`designed to mutate the Campath-9 V5 region framework
`residues into the corresponding residues of the KOL anti-
`body (18). These five mutagenic oligonucleotides were si-
`multaneously introduced in a single mutagenesis reaction.
`Twelve clones were sequenced and each clone had incorpo-
`rated the five mutagenic oligonucleotides. The plasmid
`pVHKOL/C51 encoding the KOL-based reshaped heavy
`chain consists of the following adjacently ligated fragments:
`the 9.8-kb vector pHBAPr-1-gpt (14) linearized at its cloning
`site with HindIII and BamHI containing the B-actin pro-
`moter, xanthine-guanine phosphoribosyltransferase, and am-
`picillin-resistance genes; the 39—bp HindIII—Nco I linker; a
`698-bp Nco I—BamHI fragment encoding the KOL-based
`reshaped VH region; a 2.3-kb BamHI—Sph I fragment con-
`taining a human C51 region gene (15); and the 20-bp Sph I—Bgl
`II synthetic linker.
`The plasmid pVLREI/C,‘ encoding the reshaped light chain
`is identical to the plasmid pLD9 (16) except that the three
`CDRs of the Campath-9 antibody VL region were CDR-
`grafted (9) into the 748-bp HindIII fragment from pLD9
`encoding the Campath-1H antibody light chain by in vitro
`mutagenesis with three oligonucleotides.
`Transfections and Antibody Purification. The CD4-
`expressing cell line HCD4—NB2 is a clone of the rat T-cell line
`NB2-6TG stably transfected by electroporation with the
`expression vector pSFSVneo (19) containing cDNA encod-
`ing the human CD4 antigen (20).
`Plasmids encoding antibody chains were cotransfected as
`described (21) into dihydrofolate reductase-deficient Chinese
`hamster ovary cells (106 cells per 75-cm’ flask) using 9 pg and
`1 pg of the appropriate heavy and light chain constructs,
`respectively. Transfectants were selected in medium con-
`taining 5% dialyzed fetal bovine serum for 2-3 weeks, and
`antibody-secreting clones were identified by ELISA of cul-
`ture supematants. Chimeric and reshaped antibodies were
`purified from culture supematants using protein A-Sepharose
`CL-4B (Pharmacia) column chromatography as described
`(22). Antibody concentrations were determined by absor-
`bance at 280 nm.
`Immunofluorescence and Flow Cytometry. HCD4-NB2
`cells were washed with staining medium (phosphate-buffered
`saline containing 0.1% bovine serum albumin, 1% heat-
`inactivated normal rabbit serum, and 0.1% sodium azide) and
`then incubated with either the chimeric or reshaped antibod-
`ies (105 cells per 0.1 ml) diluted in staining medium for 1 hr
`at 4°C. The cells were washed and then incubated with
`fluorescein isothiocyanate-conjugated anti-human IgG1 (y-
`chain-specific) antibodies (The Binding Site, Birmingham,
`U.K.) diluted 1:30 in staining medium for 1 hr at 4°C.
`Propidium iodide (100 ;ag/ml final concentration) was added
`during the last 10 min of incubation. Cells were thoroughly
`washed and resuspended in 0.5 ml of staining medium. Mean
`cellular fluorescence (3000 live cells per sample) was deter-
`mined with a Cytofluorograph (model 50-H Ortho Instru-
`ments). Propidiumiodide-stained dead cells were gated-out.
`Fifty percent antigen binding titers were determined by fitting
`the data to a sigmoid curve by a least squares iterative
`procedure (23).
`Antibody-Dependent Cell-Mediated Cytotoxicity (ADCC).
`Antibodies were assayed by ADCC with activated human
`peripheral blood mononuclear cells (24). Briefly, S x 10"
`HCD4-NB2 cells were labeled with “Cr and incubated for 1
`hr at room temperature with different concentrations of
`antibodies. A 75-fold excess of activated cells was added as
`
`effectors. After 4 hr at 37°C, cell death was determined by
`measuring “Cr release.
`
`RESULTS
`
`Cloning of V1, and V“ Region cDNA. cDNAs encoding the
`VL and V5 regions from the Campath-9 antibody-secreting
`clone YNB46.1.8SG2B1.19 (10) were isolated by PCR using
`primers that amplify the segment of cDNA encoding the
`amino-terminal region through the joining region (11). VL
`region clones were first screened by hybridization with a
`32P-labeled oligonucleotide probe complementary to CDR 2
`of the light chain expressed by the rat Y3-Ag1.2.3 myeloma
`cell line (12) that was used as the fusion partner to generate
`the Campath-9 antibody-secreting hybridoma. Subsequent
`nucleotide sequence analysis was restricted to clones that did
`not contain sequence complementary to this probe (about 5%
`of clones). In this manner, two cDNA clones from indepen-
`dent PCR amplifications were identified that encoded iden-
`tical VL regions. Nucleotide sequence analysis of random V5
`region clones from two independent PCR amplifications
`revealed a single species of V3 region cDNA. These cDNA
`sequences have been submitted to the GenBank data base,
`and their predicted amino acid sequences are shown (Fig. 1).
`As no additional VL or V" region-encoding clones were
`identified, it was assumed that these sequences were derived
`from the Campath-9 antibody genes.
`Chimeric Antibody Constructs. Plasmids were constructed
`that encoded a rat/human chimeric version of the Campath-9
`antibody. The plasmid pVHrat/CG1 encodes a chimeric heavy
`chain consisting of the Campath-9 VH region (Fig. 1A) and a
`human CG; region. The plasmid pV;_rat/C, encodes a chi-
`meric light chain consisting of the Campath-9 VL region (Fig.
`1B) and a human CK. These chimeric heavy and light chains
`were coexpressed in Chinese hamster ovary cells to produce
`a chimeric antibody.
`Reshaped Antibody Heavy Chain Constructs. Possibly the
`largest unknown variable when reshaping an antibody is the
`selection of the human immunoglobulin V region from which
`the framework sequences are derived. Because the frame-
`work regions hold the CDRs in their correct spatial orienta-
`tion and can sometimes even participate in antigen binding
`(29), this selection could be important. At present, there are
`insufficient published reshaping results to generalize a “best
`framewor ” selection strategy. Reshaping experiments to
`date (9, 30-32) have not compared the effectiveness of
`different human frameworks incorporating the same rodent
`CDRs.
`To investigate the importance of framework selection and
`to maximize our chances of producing a functional reshaped
`CD4 antibody, we have designed two different versions of
`reshaped VH regions. In the first case, we designed a re-
`shaped VH region that derives its CDRs from the Campath-9
`V5 region and its framework sequences from the NEW-based
`framework that had been used previously for the reshaped
`antibody Campath-1H (9) and others (30, 31). Given the
`demonstrable antigen binding of these antibodies, it was
`reasonable to try the same framework sequences as well. A
`plasmid was thus constructed, pVHNEW/C51, that encodes
`a reshaped heavy chain consisting of an NEW-based V3
`region with Campath-9 V1; region CDRs (Fig. 1A) and a
`human CG; region.
`In the second case, we designed a reshaped V3 region that
`derives its CDRs from the Campath-9 V3 region and its
`framework sequences from the V3 region of the human
`myeloma protein KOL (18). The V3 region of KOL was
`chosen because of all known human heavy chain V regions its
`overall amino acid sequence is very homologous to the
`Campath-9 VH region (Fig. 1A) containing 72% identical
`residues (excluding gaps introduced for alignment purposes).
`
`

`
`Immunology: Gorman et al.
`
`Proc. Natl. Acad. Sci. USA 88 (1991)
`
`4183
`
`A
`
`CAHPATI-I-9
`NEW
`CAMPATH-1H
`NEW-based resh.
`KOL
`KOL-based resh.
`
`CAHPATH-9
`NEW
`CAMPATH-11-l
`NEW-based resh.
`KOL
`KOL-based resh.
`
`B
`
`CAHPATH-9
`REI
`CAHPATH-1H
`REI-based resh.
`
`CAHPATH-9
`REI
`CAMPATH-1H
`REI-based resh.
`
`70
`60
`50
`40
`30
`20
`10
`QVQLQESGGG LVQPGRSLKL SCAASGLTFS NYGHAHVRQA PTKGLEWVAT ISHDGSD--T YFRDSVKGRF
`....EQ..P.
`..R.SQ'.l'.S.
`'I'.'.l'V..S...
`.DYYT....P .GR....IGY VFYILTSDD.
`---TPLRS.V
`..R.SQT.S. T.TV..F..'l‘ DFY.N....P .GR....IGI-‘ .RDKAKGY'1'. EYNP.....V
`..R.SQ'.l'.S. r.rv..r..r . . . . . . . ..P .GR....IG.
`. . . . . ..--.
`. . .
`. . . . ..v
`V . . . . . ..R.
`..SS..FI.. S.A.Y..
`.G..
`I
`Q—- HYA . . . .
`. ..
`V . . . . . ..R.
`..SS..FI .
`.
`. . . . . . . . . .
`.G . . .
`. . . . .
`. . . . . ..--
`. .
`. . . . . . ..
`
`129
`120
`110
`100
`90
`80
`TISRDNGKST LYLQ‘lDSLRS EDTATYYCAR QG———————— TIAG—IRHWG QGTTVTVSS
`.MLV.TS.NQ FS.RLS.VTA A...V..... N--------- L...C.DV..
`..SL.....
`.HLV.TS.NQ FS.RLS.VTA A...V..... E.H------- -T.APFDY..
`..SL.....
`.MLV.'1‘S.NQ FS.RLS.V'.l'A A...V ................. ..- . . .
`. .
`..SL.....
`......S.N.
`.F.......P ...GV.F
`D.GHGFCSSA SCF.P-DY.. ...P.....
`......S.N.
`.F.......P ...GV.F
`---------- ..- . . . . .
`. . .
`. . . . ..
`
`70
`60
`50
`40
`30
`20
`10
`DIQLTQSPVS LSASLGETVN IECLASEDIV1 SDLAWYQQKP GKSPQLLIYN TDTLQNGVPS RFSGSGSGTQ
`...M....S.
`....V.DR.T .T.Q..Q..I KY.N....T.
`..A.K....E ASN..A . . . .
`. .
`. . . . . ..D
`...M....S.
`....V.DR.T .'.l.‘.K..QN.D KY.N......
`..A.K.....
`.NN..T . . . .
`. .
`.
`. . . . ..D
`...M....S.
`....V.DR.T .1 . . . . . . .
`.
`. . . . . . . . . .
`..A.l( . . . . .
`. . . . . . .
`. . .
`. .
`. . . . . ..D
`
`107
`100
`90
`80
`YSLKINSLQS EDVATYFCQQ YNNYPWTPGG GTKLEIK
`.'1‘FT.S...P ..I...Y...
`.QSL.Y...Q ....Q.T
`FTF'.l.'.S...P ..I...Y.L. HISR.R...Q
`FTFT.S...P ..I...Y .
`. .
`. . . . ..
`
`Fro. 1.
`Comparison of the amino acid sequences of the heavy (A) and light (B) chain V regions described in the text. Dots indicate residues
`that are identical to the corresponding residue in Campath-9. Hyphens represent spaces introduced in the sequences by GAP (25) to aid the
`alignment. CDRs of Campath-9 are underlined and residues encoded within the amplification primers and cloning vectors are overlined. resh.,
`Reshaped. Sequences of NEW, KOL, and REI are from the Swiss-Prot protein sequence data base, release 14. It should be noted that there
`are some minor sequence ditferences of NEW and KOL as recorded in the various data bases—for example, Swiss-Prot and Brookhaven (26).
`The actual framework sequences of the NEW- and REI-based reshaped V regions described here are identical to those of the Campath-1H
`antibody (9), which difier only slightly from the reported framework sequences of NEW (27) and REI (28). For consistency, and given the
`demonstrable antigen binding of this reshaped antibody, identical framework sequences were used here.
`
`This was determined by a computer search of several data
`bases. By contrast, the NEW VH region sequence has only
`47% identical residues. We reasoned that since the primary
`function of the framework sequence is to hold the CDRs in
`their correct spatial orientation, we could maximize the
`chances of retaining correct CDR structure (and hence anti-
`gen affinity) by deriving framework sequences from a human
`VH region that is most homologous to that of the rodent. Of
`the several homologous human VH regions available, the
`choice of KOL was made because its three-dimensional
`structure is well characterized. A plasmid was thus con-
`structed, pVHKOL/C51,
`that encodes a reshaped heavy
`chain consisting of a KOL—based VH region with Campath-9
`V3 region CDRs (Fig. 1A) and a human CG1 region.
`Reshaped Antibody Light Chain Construct. We have de-
`signed a reshaped VL region that derives its CDRs from the
`Campath-9 VL region and its framework sequences from the
`REI-based framework that has been used previously for the
`reshaped antibody Campath-1H (9). Again, given the demon-
`strable antigen binding of this antibody, it was reasonable to
`try the same framework as well. A plasmid was constructed,
`pVLREI/CK, that encodes a reshaped light chain consisting of
`an REI-based VL region with Campath-9 VL region CDRs
`(Fig. 1B) and a human CK region. A second reshaped VL
`version was not created as with the reshaped V3 region
`because REI is already highly homologous (67% identical
`residues) to the rat VL region of Campath-9. Thus this
`reshaped light chain was coexpressed with the reshaped
`heavy chains in Chinese hamster ovary cells to produce two
`reshaped antibodies (KOL- and NEW-based) differing only
`in their human-derived VH region framework sequences.
`
`Properties of Chimeric and Reshaped Antibodies. The abil-
`ities of the chimeric and reshaped antibodies to bind the
`CD4+ cell line HCD4-NB2 were compared by immunofluo-
`rescence staining (Fig. 2). The chimeric and KOL-based
`reshaped antibodies stained CD4+ cells well. The titration
`curves of these two antibodies were fitted to a sigmoid curve,
`and the concentrations (mean : SEM) of chimeric and
`KOL-based reshaped antibodies needed to achieve 50%
`antigen saturation were determined to be 2.21 : 0.16 and 7.16
`1 0.45 pg/ml, respectively. Thus the avidity of the KOL-
`based reshaped antibody is only slightly reduced as it only
`
`
`
`Meanfluorescence
`
`800
`700
`600
`500
`400
`300
`200
`
`I00
`
`Antibody concentration, pg/ml
`FIG. 2. Fluorescence of CD4+ cells stained with chimeric and
`reshaped antibodies. 0, Campath-9 chimeric antibody; 0, KOL-
`based reshaped antibody; A, NEW-based reshaped antibody; A,
`Campath-1H antibody (isotype-matched negative control). The
`KOL- and NEW-based reshaped antibodies have the same REI-
`based reshaped light chain.
`
`

`
`4184
`
`Immunology: Gorman et al.
`
`Proc. Natl. Acad. Sci. USA 88 (I991)
`
`NO
`
`
`
`Percentlysis
`
`.. (II
`
`o-5
`
`Antibody concentration, pg/ml
`
`FIG. 3. ADCC with chimeric and reshaped antibodies. Symbols
`are the same as in Fig. 2.
`
`takes three times the amount of this antibody to give the same
`50% binding as the chimeric antibody. All KOL V5 region
`framework residues were incorporated into this reshaped VH
`region except for the proline at position 124 (Fig. 1A), which
`was kept as threonine. A second version that incorporates
`this proline residue gave similar results (not shown). By
`contrast, the NEW-based reshaped antibody stained CD4+
`cells only poorly even at the higher concentrations. The
`control Campath-1H antibody did not stain cells at any
`concentration. Also, the chimeric and KOL-based reshaped
`antibodies were effective in cell-mediated lysis, whereas the
`NEW-based reshaped and control Campath-1H antibodies
`were ineffective (Fig. 3).
`
`DISCUSSION
`
`We have described here the successful reshaping of the CD4
`antibody Campath-9. This result, together with the previ-
`ously described reshaping of other therapeutic antibodies (9,
`32), demonstrates the feasibility of applying this concept to
`the many rodent-derived monoclonal antibodies with clinical
`potential. The Campath-1H antibody has already been used
`successfully in clinical studies of lymphoma therapy (33) and
`in the treatment of an autoimmune disorder (10). In this later
`case,
`the Campath-1H antibody was combined with the
`Campath-9 antibody to give a long-term remission in a case
`of systemic vasculitis that appeared intractable prior to
`Campath-9 antibody treatment. Although the relative contri-
`butions and importance of the Campath-1H and Campath-9
`antibodies cannot be ascertained from this single case study,
`the availability of Campath-9 as a reshaped antibody, Cam-
`path-9H (the KOL-based version), should provide for further
`useful clinical studies.
`Our reshaping of Campath-9 into Campath-9H raises in-
`teresting questions regarding general strategies for the re-
`shaping of rodent antibodies. At present, there are only four
`additional reports of reshaped antibodies and in no case has
`the effectiveness of two different human antibody frame-
`works regions been compared (9, 30-32). Three of these
`antibodies had reshaped VH regions based on NEW (9, 30,
`31), two of which had reshaped VL regions based on REI (9,
`31), regardless of whether closer homologies existed in the
`sequence data bases. In the fourth example, the antibody was
`reshaped based on a homologous human framework, but 12
`of the 29 residues that differed between the human and mouse
`VH region frameworks were left as in the mouse sequence,
`and no data were presented that compared this reshaped
`antibody with a form in which all of the framework residues
`were derived from the human sequence (32). For the reshap-
`ing of Campath-9, we have shown that the selection of a
`particular human framework can be important in the reten-
`tion of antibody avidity. We have made two reshaped anti-
`bodies that differ only in their usage of human VH region
`
`framework sequences, KOL and NEW, and found one form
`to be far superior to the other. The KOL-based CD4 antibody
`retained biological activity in ADCC assays and had a relative
`binding avidity only slightly reduced from the unaltered V
`region sequences of the chimeric CD4 antibody. In contrast,
`the NEW-based CD4 antibody, though still retaining speci-
`ficity for CD4, had a considerably reduced relative binding
`avidity and had no biological activity. The KOL V3 region
`has a 72% homology to Campath-9 VH region, whereas the
`NEW VH framework has only 47% homology. In this case
`then, it would seem that the selection of a human V region
`framework that was highly homologous to the rodent V
`region was the best strategy for framework selection. We
`have also successfully reshaped a CD3 antibody by the same
`approach (E.G.R., unpublished data), so this strategy may
`prove to be generally applicable to antibody reshaping.
`Arguably, although such a bestfit strategy has been applied
`here to whole antibody V regions, it might also be applied
`separately to individual variable, diversity, and joining re-
`combining segments of an antibody V region.
`The different avidities we observe when reshaping with
`KOL- and NEW-based frameworks are likely to be due to the
`complex intrachain associations between CDR and frame-
`work residues. Alternatively, difl'erent interchain V region
`associations between these two heavy chains and the RBI-
`based reshaped light chain may also play a role. One possible
`structural explanation for the differences between the KOL-
`and NEW-based reshaped antibodies is described by Tra-
`montano et al. (34). In studies of a series of solved immu-
`noglobulin structures they observed that the conformation of
`the heavy chain CDR 2, “H2 loop,” is dependent upon the
`length and sequence of this loop and its interaction with the
`framework residue at position 71 (Kabat numbering system).
`KOL and NEW have distinctly different structures in this
`region and a different residue at position 71. The H2 loop of
`the Campath-9 antibody is very similar to that of KOL and
`both contain an arginine residue at the relevant framework
`position (this is position 74 in Fig. 1A). Perhaps the change of
`this residue upon reshaping to the NEW-based framework
`accounts in part for its low avidity. However, it should be
`noted that Campath-1H was successfully reshaped to the
`NEW-based framework despite having a substantially difi'er-
`ent H2 loop length and sequence and also a different frame-
`work residue at position 71 (34). Clearly the ability to use
`structural features like this to reliably predict a suitable
`strategy for reshaping will benefit from more examples where
`different strategies are experimentally compared.
`In previous studies, genes encoding reshaped antibodies
`were produced either by total synthesis of the desired se-
`quence (32) or by in vitro mutagenesis of a human V region
`sequence to incorporate the rodent CDRs (9, 30, 31). We
`propose the following strategy based on our successful re-
`shaping experiments whereby the isolated rodent V region is
`used as a starting point for constructing reshaped V regions.
`Here, human framework sequences are transferred to the
`rodent V regions by means of in vitro mutagenesis. When
`coupled with the bestfit method from above, mutagenic
`oligonucleotides can be highly homologous to the rodent
`frameworks, and hence the efficiency of mutagenesis is high.
`In the case of Campath-9H, this was accomplished in a single
`mutagenesis reaction with five oligonucleotides 33-58 bases
`in length. This strategy should be readily applicable to any
`monoclonal antibody for which the cDNA has been cloned
`and for which a homologous human framework can be
`identified in the sequence data bases. It is also interesting to
`note that expression in the nonlymphoid Chinese hamster
`ovary cell line results in an antibody with demonstrable
`activity in ADCC. The capacity of this cell line for high—level
`expression of reshaped antibodies (16) should facilitate the
`large-scale production of this antibody for clinical studies.
`
`

`
`Immunology: Gorman et al.
`
`Proc. Natl. Acad. Sci. USA 88 (1991)
`
`4185
`
`We thank the following for their helpful discussions and assistance:
`G. Winter and P. Jones for the vectors M13-VKPCR1 and M13-
`VHPCR1, J. S. Crowe for Campath-1H CDNA, M. Page for the
`plasmids pLD9 and pNH316, J . Ivanyi and J. Howard for the cell line
`NB2-6TG, A. Lesk, and C. Chothia. We thank H. Spence for
`synthetic oligonucleotide synthesis, H. Kruger-Gray for flow cytom-
`etry assistance, and M. Frewin for technical assistance. This work
`was supported by the Medical Research Council, United Kingdom,
`Wellcome Biotech PLC, and the Gilman Foundation. S.D.G. is a
`recipient of a Special Fellowship from the Leukemia Society of
`America.
`
`1.
`
`2.
`
`P
`
`I-|.°
`
`t—4>—Ibd>—dyugsmu:
`
`Qin, S. X., Cobbold, S., Benjamin, R. & Waldmann, H. (1989)
`J. Exp. Med. 169, 779-794.
`Qin, S. X., Wise, M., Cobbold, S. P., Leong, L., Kong, Y. M.,
`Parnes, J. & Waldmann, H.
`(1990) Eur. J. Immunol. 20,
`2737-2745.
`Cobbold, S. P., Martin, G. & Waldmann, H. (1990) Eur. J.
`Immunol. 20, 2747-2755.
`Madsen, J. C., Superina, R. A., Wood, K. J. & Morris, P. J.
`(1988) Nature (London) 332, 161-164.
`Benjamin, R. J. & Waldmann, H. (1986) Nature (London) 320,
`449-451.
`_
`Gutstein, N. L. & Wofsy, D. (1986) J. Immunol. 137, 3414-
`3419.
`Herzog, C., Walker, C., Muller, W., Rieber, P., Reiter, C.,
`Riethmuller, G., Wassmer, P., Stockinger, H., Madic, 0. &
`Pichler, W. J. (1989) J. Autoimmun. 2, 627-642.
`Hafler, D. A., Ritz, J., Schlossman, S. F. & Weiner, H. L.
`(1988) J. Immunol. 141, 131-138.
`Riechmann, L., Clark, M., Waldmann, H. & Winter, G. (1988)
`Nature (London) 332, 323-327.
`Mathieson, P. W., Cobbold, S. P., Hale, G., Clark, M. R.,
`Oliveira, D. B. G., Lockwood, C. M. & Waldmann, H. (1990)
`N. Engl. J. Med. 323, 250-254.
`Orlandi, R., Gussow, D. H., Jones, P. T. & Winter, G. (1989)
`Proc. Natl. Acad. Sci. USA 86, 3833-3837.
`Crowe, J. S., Smith, M. A. & Cooper, H. J. (1989) Nucleic
`Acids Res. 17, 7992.
`Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl.
`Acad. Sci. USA 74, 5463-5467.
`Gunning, P., Leavitt, J ., Muscat, G., Ng, S. Y. & Kedes, L.
`(1987) Proc. Natl. Acad. Sci. USA 84, 4831-4835.
`Takahashi, N., Ueda, S., Obata, M., Nikaido, T., Nakai, S. &
`Honjo, T. (1982) Cell 29, 671-679.
`
`3555
`
`N.°
`
`IN?|-‘
`
`N5"’
`
`.‘°s=°_\I'6~’&’$’B
`
`I‘)I\)I\’
`
`w9
`
`la) F1
`
`Page, M. J. & Sydenham, M. A. (1991) Biotechnology 9, 64-
`68.
`Hieter, P. A., Max, E. E., Seidmann, J. G., Maize], J. V., Jr.,
`& Leder, P. (1980) Cell 22, 197-207.
`Schmidt, W. E., Jung, H.—D., Palm, W. & Hilschmann, N.
`(1983) Hoppe-Seyler’s Z. Physiol. Chem. 364, 713-747.
`Ballhausen, W. G., Reske—Kunz, A. B., Tourvieille, B.,
`Ohashi, P. S., Pames, J. R. & Mak, T. W. (1988) J. Exp. Med.
`167, 1493-1498.
`Maddon, P. J., Littman, D. R., Godfrey, M., Maddon, D. E.,
`Chess, L. & Axel, R. (1985) Cell 42, 93-104.
`Wigler, M., Pellicer, A., Silverstein, S., Axel, R., Urlaub, G.
`& Chasin, L. (1979) Proc. Natl. Acad. Sci. USA 76, 1373-1376.
`Oi, V. T. & Herzenberg, L. A. (1980) in Selected Methods in
`Cellular Immunology, eds. Mishell, B. B. & Shiigi, S. M.
`(Freeman, San Francisco), pp. 351-372.
`Hale, G., Hoang, T., Prospero, T., Watt, S. M. & Waldmann,
`H. (1983) Mol. Biol. Med. 1, 305-319.
`Clark, M. R. & Waldmann, H. (1987) J. Natl. Cancer Inst. 79,
`1393-1401.
`Devereux, J ., Haeberli, P. & Smithies, O. (1984) Nucleic Acids
`Res. 12, 387-395.
`Lesk, A. M., Boswell, D. R., Lesk, V. 1., Lesk, V. E. &
`Bairoch, A. (1989) Protein Sequences Data Anal. 2, 295-308.
`Poljak, R. J., Nakashima, Y., Chen, B. L. & Konigsberg, W.
`(1977) Biochemistry 16, 3412-3420.
`Epp, 0., Lattman, E. E., Schifier, M., Huber, R. & Palm, W.
`(1975) Biochemistry 14, 4943-4952.
`Kabat, E. A., Wu, T. T., Reid-Miller, M., Perry, H. M. &
`Gottesmann, K. S. (1987) Sequences of Proteins of Immuno-
`logical Interest (U.S. Dept. of Health and Human Services)
`(GPO, Washington).
`Jones, P. T., Dear, P. H., Foote, J., Neuberger, M. S. &
`Winter, G. (1986) Nature (London) 321, 522-525.
`Verhoeyen, M., Milstein, C. & Winter, G. (1988) Science 239,
`1534-1536.
`Queen, C., Schneider, W. P., Selick, H. E., Payne, P. W.,
`Landolfi, N. F., Duncan, J. F., Avdalovic, N. M., Levitt, M.,
`Junghans, R. P. & Waldmasm, T. A. (1989) Proc. Natl. Acad.
`Sci. USA 86, 10029-10033.
`Ha.le,G.,Dyer,M. J. S.,Clark,M. R.,Phillips,J. M.,Marcus,
`R., Riechmann, L., Winter, G. & Waldmann, H. (1988) Lancet
`ii, 1394-1399.
`Tramontano, A., Chothia, C. & Lesk, A. M. (1990) J. Mol.
`Biol. 215, 175-182.