Proc. Nati. Acad. Sci. USA
`Vol. 81, pp. 6851-6855, November 1984
`Immunology
`
`Chimeric human antibody molecules: Mouse antigen-binding
`domains with human constant region domains
`(transfection/protoplast fusion/calcium phosphate transfection/intronic controlling elements/transfectoma)
`
`SHERIE L. MORRISON*, M. JACQUELINE JOHNSONt, LEONARD A. HERZENBERGt, AND VERNON T. 01i
`*Department of Microbiology and the Cancer Center, Institute for Cancer Research, College of Physicians and Surgeons, Columbia University,
`New York, NY 10032; tDepartment of Genetics, Stanford University School of Medicine, Stanford, CA 94305; and tBecton-Dickinson
`Monoclonal Center, 2375 Garcia Avenue, Mountain View, CA 94043
`Contributed by Leonard A. Herzenberg, August 1, 1984
`
`ABSTRACT
`We have created mouse-human antibody
`molecules of defined antigen-binding specificity by taking the
`variable region genes of a mouse antibody-producing myeloma
`cell line with known antigen-binding specificity and joining
`them to human immunoglobulin constant region genes using
`recombinant DNA techniques. Chimeric genes were construct-
`ed that utilized the rearranged and expressed antigen-binding
`variable region exons from the myeloma cell line S107, which
`produces an IgA (K) anti-phosphocholine antibody. The
`heavy chain variable region exon was joined to human IgG1 or
`IgG2 heavy chain constant region genes, and the light chain
`variable region exon from the same myeloma was joined to the
`human ic light chain gene. These genes were transfected into
`mouse myeloma cell lines, generating transformed cells that
`produce chimeric mouse-human IgG (K) or IgG (K) anti-phos-
`phocholine antibodies. The transformed cell lines remained tu-
`morigenic in mice and the chimeric molecules were present in
`the ascitic fluids and sera of tumor-bearing mice.
`
`The capability to transfer immunoglobulin genes into lym-
`phoid cells where they produce protein in quantities suffi-
`cient for structural studies (1-3) provides us with the oppor-
`tunity to generate and characterize novel immunoglobulin
`molecules. Cloned variable (V) region genes from mouse or
`rat hybridoma cell lines can be ligated to human constant (C)
`region genes and we would expect that these chimeric genes
`can be transfected into mouse myeloma cells, which then
`will produce novel human antibody molecules. We would
`thus produce antibodies that are largely human but which
`have antigen-binding specificities generated in mice. The ad-
`ditional potential for in vitro manipulation and alteration of
`both the antigen-binding site and the structures correlated
`with biological effector functions of these antibody mole-
`cules using recombinant DNA techniques would introduce a
`powerful approach for further understanding antibody struc-
`ture, function, and immunogenetics.
`As we show here, both chimeric mouse heavy chain V re-
`gion exon (VH)-human heavy chain C region genes and chi-
`meric mouse light chain V region exon (VK)-human K light
`chain gene constructs are expressed when transfected into
`mouse myeloma cell lines. When both chimeric heavy and
`light chain genes are transfected into the same myeloma cell,
`an intact tetrameric (H2L2) chimeric antibody is produced.
`In this study we used VH and VK exons from the mouse phos-
`phocholine (PCho)-binding antibody-producing S107 myelo-
`ma cell line (4, 5). Chimeric mouse-human anti-PCho anti-
`bodies were produced in culture by appropriate transfected
`cell lines or by "transfectomas" obtained when such cell
`lines are injected into mice.
`
`MATERIALS AND METHODS
`Chimeric Genes. The cloned S107 VH and S107 VK genes
`were gifts from Matthew Scharff (Albert Einstein College of
`Medicine, Bronx, NY). The S107 VH gene was spliced to
`human IgG1 and IgG2 C region genes by using Sal I linkers
`as shown in Fig. 1A. Both constructs were inserted into the
`vector pSV2AH-gpt (1, 6). The S107 V/K gene was spliced to
`the human K gene at a unique HindIII site located in the large
`intron between the K light chain joining and C (J.K and CK)
`region exons as shown in Fig. 1B. This chimeric light chain
`gene construct was inserted into both PSV2AH-gpt and
`pSV2-neo plasmid vectors (7).
`Transfection. Protoplast fusion and calcium phosphate
`precipitation techniques (1, 8, 9) were used to transfect these
`chimeric immunoglobulin genes into the J558L myeloma cell
`line (a A light chain-producing mouse myeloma cell line) and
`an immunoglobulin nonproducing derivative of the P3 my-
`eloma cell line. Mycophenolic acid (GIBCO) was used for
`selection of cells transfected with pSV2A-gpt vectors as de-
`scribed (1, 3). G418 (GIBCO) at 1.0 mg/ml was used for se-
`lection of cells transfected.with pSV2-neo vectors (7).
`Light and heavy chain chimeric immunoglobulin genes
`were transfected sequentially by protoplast fusion using
`G418 selection for the chimeric light chain gene vector and
`mycophenolic acid for the chimeric heavy chain gene vector.
`The protoplast fusion transfection procedure used was as de-
`scribed (1).
`Transfection using the calcium phosphate precipitation
`procedure was done by transfecting a mixture of 40 pg each
`of the chimeric light and chimeric heavy chain pSV2AH-gpt
`vectors into 5 x 106 cells. Mycophenolic acid was used to
`select for transformed cell lines as described (1).
`Antigen Binding. PCho binding of antibody secreted into
`the culture supernates of transfected cell lines was analyzed
`by using a solid-phase radioimmunoassay described previ-
`ously (10). PCho-keyhole limpet hemocyanin antigen was
`bound to 96-well polyvinyl plates; binding of chimeric anti-
`PCho antibodies was detected by using 115I-labeled protein
`A or 125I-labeled-anti-human IgG. PCho-binding antibodies
`in biosynthetically labeled culture supernates and cell ly-
`sates of transfected cell lines also were analyzed by binding
`the biosynthetically labeled antibody to PCho-coupled Seph-
`arose 4B (Pharmacia) and then eluting the bound antibody
`with PCho-hapten. The bound and eluted antibody was ex-
`amined by NaDodSO4/polyacrylamide gel electrophoresis
`(NaDodSO4/PAGE). Biosynthetic-labeling procedures were
`as described (11).
`Idiotope Analysis. Three hybridoma anti-idiotope antibod-
`ies, also kindly provided by Matthew Scharff and Angela
`
`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.
`
`Abbreviations: V, variable; C, constant; J, joining; H' heavy chain;
`L, light chain; PCho, phosphocholine; kb, kilobase(s); NEPHGE,
`nonequilibrium pH gradient electrophoresis.
`
`6851
`
`PETITIONER'S EXHIBITS
`
`Exhibit 1031 Page 1 of 5
`
`

`
`6852
`
`Immunology: Morrison etaLP
`Proc. Natl. Acad Sci. USA 81 (1984)
`
`A
`
`Dam HI
`
`(Sad) Sal I (Hind II)
`
`Bam Hi
`
`pSV2AH-S107 HuGI
`
`0
`
`V111E"1JA'EI
`
`Mouse VDJ
`
`Human constant region
`
`pSV2AH-S107 HuG2
`
`pBR322 ori
`
`Pvull
`SV40 ori
`
`Eco-gpt
`
`B
`
`w
`am Hi
`
`Mlouse VJ
`
`Hind III
`
`Human kappa
`
`ha Hi
`
`pSV2AH-S107 Huk
`
`pBR322 ori
`
`Pvull
`SV40 or!
`
`Eco-gpt or
`Eco-neo
`
`Schematic diagrams (not drawn to scale) of the chimeric mouse-human heavy chain gene vector (A) and the chimeric light chain
`FIG. 1.
`gene vectors (B). DNA fragment sizes are as follows: HindIII-BamHI human IgG1 (HuGi) or IgG2 (HuG2) heavy chain gene, 7 kilobases (kb);
`mouse S107 VH gene, 4.5 kb; HindIII-BamHI human CK gene (HuK), 10 kb; and mouse S107 VK gene, 3.5 kb. The HindI1 sites in the human
`IgG1 and IgG2 heavy chain genes were ligated to the Sac I site of the mouse S107 VH gene with Sal I linkers.
`
`Guisti (Albert Einstein College of Medicine), were usea to
`analyze the VH-VL (L is light chain) domain structure of the
`chimeric human anti-PCho antibodies. These antibodies
`(TC102.1.2, T139.2, and T156.1.1), recognizing three inde-
`pendent idiotopes (12), were used to immunoprecipitate bio-
`synthetically labeled material eluted with PCho from the
`PCho-Sepharose 4B matrix. Immunoprecipitates were ana-
`lyzed by NaDodSO4/PAGE.
`Immunoglobulin Chain Composition. Monoclonal anti-hu-
`man IgG and anti-human K antibodies (Becton-Dickinson)
`were used to immunoprecipitate biosynthetically labeled chi-
`meric human anti-PCho antibodies for analyses using two-
`dimensional nonequilibrium pH gradient PAGE (NEPHGE)
`(11, 13). PCho-coupled Sepharose 4B also was used for af-
`finity purification.
`Imniunoglobulin Heavy Chain Glycosylation. Tunicamycin
`(Calbiochem-Behring) was used to inhibit asparagine-linked
`glycosylation of biosynthetically labeled antibody from
`mouse cell lines producing mouse-human chimeric immuno-
`globulins (11). PCho-binding antibody from tunicamycin-
`treated cells was analyzed by NaDodSO4/PAGE. Proce-
`
`dures for tunicamycin treatment were as described (11).
`Chimeric Mouse-Human Antibody Production in Mice.
`Transfected J558L cells producing chimeric mouse-human
`antibody were injected intraperitoneally into BALB/c mice
`(106 cells per mouse). Sera and ascitic fluids from tumor-
`bearing mice were analyzed for human anti-PCho antibody
`by a solid-phase radioimmunoassay (10) and by immunoelec-
`trophoresis using a polyclonal anti-human IgG antiserum.
`
`RESULTS
`We obtained expression of chimeric mouse V region-human
`C region genes in transfected J558L and the immunoglobulin
`nonproducing P3 myeloma cell lines. When both light chain
`and heavy chain chimeric genes (see Fig. 1) were expressed
`in the same cell, tetrameric (H2L2) antigen-binding antibod-
`ies were obtained. Biosynthetically labeled antibody mole-
`cules secreted by J558L cells expressing chimeric genes
`were bound and hapten-eluted from PCho-Sepharose 4B.
`Autoradiograms of two-dimensional NEPHGE analyses
`(Fig. 2) show immunoglobulin polypeptide chains of the ex-
`
`PETITIONER'S EXHIBITS
`
`Exhibit 1031 Page 2 of 5
`
`

`
`Immunology: Morrison et aL
`
`Proc. Nadt Acad Sci USA 81 (1984)
`
`6853
`
`a
`
`b
`
`C
`
`d
`
`H2L2-- __ _u __
`
`Autoradiograph of nonreduced NaDodSO4/PAGE anal-
`FIG. 3.
`yses of PCho-binding material from a chimeric IgG2 (K) antibody-
`producing J558L cell line immunoprecipitated with Staphylococcus
`aureus protein A (lane a), monoclonal anti-human K antibody (lane
`b), and anti-idiotope antibodies TC102.1.2 and T156.1.1 (lanes c and
`d).
`
`on the presence of the asparagine-linked carbohydrate
`moeity in the CH2 domain of the molecule (4). The loss of
`this carbohydrate chain affects profoundly the overall do-
`main structure of this part of the immunoglobulin molecule.
`Concomitant with this structural change, the catabolism rate
`of the molecule is increased and biological effector functions
`such as complement-fixation are lost (14). Glycosylation of
`the mouse-human chimeric antibodies in mouse myeloma
`cells was deduced by determining the molecular masses of
`antibodies synthesized in the presence and absence of tuni-
`camycin, an antibiotic inhibitor of asparagine-linked glyco-
`sylation. Fig. 4 is an autoradiogram of NaDodSO4/PAGE
`analysis of glycosylated and nonglycosylated chimeric heavy
`and light chains produced in transfected mouse myeloma
`cells. Though we cannot be certain that in the absence of
`tunicamycin the appropriate asparagine residue in the CH2
`domain is glycosylated, clearly, the lower molecular mass of
`the heavy chain synthesized in the presence of tunicamycin
`is as expected if a single N-linked carbohydrate were absent
`
`a
`
`b
`
`-67
`
`-45
`
`-30
`
`-20
`
`Autoradiograph of reduced NaDodSO4/PAGE analyses
`FIG. 4.
`of chimeric IgG2 (K) antibody synthesized in the presence (lane a)
`and absence (lane b) of tunicamycin. Size markers are shown in
`kDa.
`
`pected charge and relative molecular mass. Identical two-
`dimensional gel analysis results were obtained with immuno-
`precipitates by using monoclonal anti-human K and IgG anti-
`bodies. Similar results were obtained when the chimeric
`IgG2 (K) antibodies produced in the transfected nonproduc-
`ing P3 cell line were analyzed. Since the recipient P3 cell line
`does not produce endogenous immunoglobulin polypeptide
`chains, only the chimeric mouse-human heavy and light
`chains were seen (data not shown). PCho binding by the chi-
`meric antibody produced in the J558L cell line required the
`specific association of both the VH and VL domains of the
`S107 myeloma protein connected to human C region poly-
`peptides. Antibody secreted by transfected J558L cells ex-
`pressing only the chimeric heavy chain and the endogenous
`J558L light chain did not bind PCho (data not shown).
`Further verification of the appropriate polypeptide folding
`of the mouse VH and VL domains attached to human C re-
`gion polypeptide chains was done by analyses detecting the
`presence of idiotopes known to occur on the parental S107
`PCho-binding antibody molecule. Three monoclonal anti-
`idiotope antibodies-two recognizing idiotopes requiring the
`association of light and heavy V region domains (TC102.1.2
`and T139.2) and the third (T156.1.1) recognizing an epitope
`present on the heavy chain V region domain (12)-were
`found to react with the mouse-human chimeric anti-PCho
`antibodies (see Fig. 3). This is good evidence that the mouse
`S107 antigen-binding domains have folded into their intend-
`ed structures.
`The conformation of IgG heavy chain C regions depends
`
`Acidic
`
`-67
`
`-45
`
`-30
`
`-67
`
`-45
`
`-30
`
`0 t
`
`Basic
`A
`
`HuG2--
`
`B
`
`HuG 1 _
`
`0
`
`t
`
`.
`
`HuK
`
`J558X
`
`Autoradiographs of two-dimensional NEPHGE gels of
`FIG. 2.
`mouse-human chimeric IgG2 (K) anti-PC antibody (A) and mouse-
`human chimeric IgG1 (K) anti-PC antibody (B). Antibodies synthe-
`sized by transfected J558L cells were affinity purified with PCho-
`Sepharose 4B. The electrophoretic mobility of the J558L x chain had
`been determined previously (1). Size markers are shown in kDa.
`
`PETITIONER'S EXHIBITS
`
`Exhibit 1031 Page 3 of 5
`
`

`
`6854
`
`Immunology: Morrison et al
`
`anti-human IgG
`
`Immunoelectrophoresis of ascitic fluid from a BALB/c
`FIG. 5.
`mouse bearing a J558L transfectoma producing chimeric IgG2 (K)
`anti-PCho antibodies developed with a polyclonal anti-human IgG
`antiserum.
`from the polypeptide chain. Note that the light chain band is
`not affected by tunicamycin treatment. From these data, we
`conclude that the mouse myeloma cell appropriately glyco-
`sylates the human heavy chain.
`When transfected J558L cells producing the human IgG2
`(K) chimeric anti-PCho antibody were grown as an ascitic
`tumor in BALB/c mice, analyses of sera and ascitic fluids
`showed the presence of significant quantities of anti-PCho-
`binding antibodies by radioimmunoassay. Fig. 5 shows the
`results of immunoelectrophoretic analysis of ascitic fluids
`from these transfectoma-bearing mice. Polyclonal anti-hu-
`man antiserum (a gift of E. A. Kabat, Columbia University,
`New York) was used to demonstrate the presence of human
`immunoglobulin. Based on our previous experience with
`mouse hybridoma-antibody production in mice, we conclude
`that the amounts of immunoglobulin visualized by this im-
`munoelectrophoretic analysis are similar to the levels seen
`with lower-producing mouse hybridomas in vivo.
`Transfectomas producing chimeric heavy and light chains
`were produced either by cotransfection of both genes using
`calcium phosphate precipitation or by sequential protoplast
`fusion using vectors with different drug markers. For proto-
`plast fusion, the heavy chains were introduced first on a
`plasmid expressing the Eco-gpt gene. One heavy chain-pro-
`ducing cell line then was transfected by protoplast fusion
`with a plasmid containing the light chain gene and the neo
`marker. Transformants resistant to both mycophenolic acid
`and G418 were assayed for their production of anti-PCho
`antibodies. Of 77 transformants analyzed, only 7 were posi-
`tive for antibody by radioimmunoassay and only 2 of these
`were found to be producing significant quantities of K light
`chain. In cotransfection experiments using the calcium phos-
`phate precipitation protocol, the same phenomenon was ob-
`served. Analyses of a large number of independent transfor-
`mants revealed that a minority (<10%) of the transfected cell
`lines produced both chimeric heavy and light chain polypep-
`tides. Since our own unpublished data using the calcium
`phosphate precipitation procedure demonstrate that two in-
`tact mouse immunoglobulin genes cotransfected into the
`same cell, using the same protocol as described here, coex-
`press both gene products in the majority of transformed cell
`lines, these results suggest that appropriate transcriptional
`or translational controlling elements are absent in the chi-
`meric light chain gene construction or that required tran-
`scriptional or translational factors are absent in the mouse
`cell lines transfected here.
`DISCUSSION
`The opportunity to use recombinant DNA techniques to con-
`struct novel antibody genes and then to produce antibody
`molecules in mammalian lymphoid cells transfected with
`these genes in appropriate vectors provides a new approach.
`to understanding the structure, function, and immune prop-
`erties of antibody molecules. The use of chimeric mouse-
`human antibody gene constructs permits us to study "near-
`human" antibodies with desired antigen-binding specific-
`ities. For clinical applications, human antibody molecules
`constructed by using recombinant DNA techniques should
`complement human-human hybridoma antibodies as useful
`immunotherapeutic and diagnostic reagents. The use of hu-
`man or near-human antibodies might decrease or eliminate
`
`Proc. Natl. Acad Sci. USA 81 (1984)
`
`the immunogenicity of antibodies used in vivo, relative to
`monoclonal mouse hybridoma antibodies currently being
`used (15). Further, since the same V region can be joined to
`any C region, it is possible to use the C region of an appropri-
`ate human heavy chain isotype that will exhibit desired bio-
`logical effector functions.
`The use of mammalian lymphoid cells to produce these
`antibodies rather than a prokaryotic expression system as-
`sures that any post-translational modifications of the anti-
`body molecules as they are synthesized, processed, folded,
`and assembled in the eukaryotic endoplasmic reticulum are
`carried out correctly. For example, the immunoglobulins en-
`coded by the transfected mouse-human chimeric immuno-
`globulin genes in both mouse recipient cells were glycosylat-
`ed, bound antigen, and presumably would be treated as nor-
`mal immunoglobulins if injected into humans. This obviates
`the need for any post-synthetic in vitro modification or as-
`sembly that is required with prokaryotic synthesis of immu-
`noglobulin polypeptides to produce functional antibody mol-
`ecules (16). Furthermore, we have shown that transfectomas
`produce useable amounts of human antibodies either in tis-
`sue culture or in mice.
`It is now feasible to shuffle exons or carry out other kinds
`of directed mutagenesis to explore the human antibody mol-
`ecules C region structures required for carrying out diverse
`antibody functions. It is conceivable that an altered antibody
`molecule more effective in specific functions than naturally
`occurring antibody molecule can be created. In a similar
`manner, it should be possible to modify V regions to alter
`their interactions with antigen. Even more exciting would be
`to construct antigen-binding sites de novo.
`A limitation yet to be overcome is the low frequency of
`transformants that produce chimeric light chains. The chi-
`meric light chain gene construct (see Fig. 1B) has the mouse
`V1K gene coupled to the presumed human intronic controlling
`element (ICE) or "enhancer"-that is, this construct con-
`tains the human intron with a sequence homologous to the
`established mouse ICE sequence (17). In contrast, the heavy
`chain chimeric mouse-human polypeptide is expressed effi-
`ciently. In the latter construct (see Fig. lA), the ICE se-
`quence is mouse-derived. Thus, the lower expression fre-
`quency of the chimeric light chain gene may reflect a species
`preference for the effective interaction of the ICE sequence
`with factors or elements controlling immunoglobulin expres-
`sion in mouse lymphoid cells. This suggestion is consistent
`with our earlier finding that the mouse light chain gene, pre-
`sumably due to its ICE sequence, is not expressed efficiently
`in a rat myeloma cell line (1). Construction of chimeric genes
`using entire mouse intronic sequences and elimination of
`most human intronic sequences, to determine if these will be
`expressed efficiently, await further investigation.
`In this work we have taken the initial steps toward con-
`structing and efficiently expressing unique immunoglobulin
`molecules. We expect that this technique will permit an in-
`creased understanding of the structural and dynamic require-
`ments of antibody functions. It also will facilitate develop-
`ment of antibody molecules for immunodiagnosis and immu-
`notherapy.
`
`We thank Letitia Wims and Linda Nakamura Roark, who provid-
`ed invaluable assistance with these experiments. We also thank
`Olivia Gagliani for her patience in preparing this manuscript. This
`research was supported in part by Grants K04 AI00408, AI 19042,
`Al 08917, CA 16858, CA 22736, CA 13969, and CA 04681 from the
`National Institutes of Health, Grant IMS-360 from the American
`Cancer Society, and Grant SPO 13712-02-00 from the Becton-Dick-
`inson Monoclonal Center.
`
`1.
`
`Oi, V. T., Morrison, S. L., Herzenberg, L. A. & Berg, P.
`(1983) Proc. Natl. Acad. Sci. USA 80, 825-829.
`
`PETITIONER'S EXHIBITS
`
`Exhibit 1031 Page 4 of 5
`
`

`
`Immunology: Morrison et al.
`
`Proc. NatL Acad. Sci. USA 81 (1984)
`
`6855
`
`10.
`
`11.
`
`12.
`
`13.
`
`14.
`
`15.
`16.
`
`17.
`
`Oi, V. T. & Herzenberg, L. A. (1979) Mol. Immunol. 16,
`1005-1017.
`Oi, V. T., Bryan, V. M., Herzenberg, L. A. & Herzenberg,
`L. A. (1980) J. Exp. Med. 151, 1260-1274.
`Desaymaid, C., Guist, A. M. & Sharff, M. D. (1984) Mol. Im-
`munol., in press.
`O'Farrell, P. Z., Goodman, H. M. & O'Farrell, P. H. (1977)
`Cell 12, 1133-1142.
`Nose, M. & Wigzell, H. (1983) Proc. Natl. Acad. Sci. USA 80,
`6632-6636.
`Miller, R. A. & Levy, R. (1981) Lancet i, 226-229.
`Boss, M. A., Kenton, J. H., Wood, C. R. & Emtage, J. S.
`(1984) Nucleic Acids Res. 12, 3791-3806.
`Morrison, S. L. & Oi, V. T. (1984) Annu. Rev. Immunol. 2,
`239-256.
`
`6.
`7.
`
`8.
`
`9.
`
`2.
`
`3.
`
`5.
`
`Gillies, S. D., Morrison, S. L., Oi, V. T. & Tonegawa, S.
`(1983) Cell 33, 717-728.
`Ochi, A., Hawley, R. G., Hawley, T., Shulman, M. J., Trau-
`necker, A., Kohler, G. & Hozumi, N. (1983) Proc. Natl. Acad.
`Sci. USA 80, 6351-6355.
`4. Kwan, S. P., Rudikoff, S., Seidman, J. G., Leder, P. &
`Scharff, M. D. (1981) J. Exp. Med. 153, 1366-1370.
`Clarke, C., Berenson, J., Governman, J., Boyer, P. D.,
`Crews, S., Siu, G. & Calame, K. (1982) Nucleic Acids Res. 10,
`7731-7749.
`Mulligan, R. C. & Berg, P. (1980) Science 209, 1422-1427.
`Mulligan, R. C. & Berg, P. (1981) Proc. Natl. Acad. Sci. USA
`78, 2072-2076.
`Sandri-Goldin, R. M., Goldin, A. L., Levine, M. & Glorioso,
`J. C. (1981) Mol. Cell. Biol. 1, 743-752.
`Chu, G. & Sharp, P. A. (1981) Gene 13, 197-202.
`
`PETITIONER'S EXHIBITS
`
`Exhibit 1031 Page 5 of 5