Proc. Nat/. Acad. Sci. USA
`Vol. 81, pp. 3273-3277, June 1984
`Biochemistry
`
`Generation of antibody activity from immunoglobulin polypeptide
`chains produced in Escherichia coli
`
`( carcinoembryonic antigen/monoclonal antibody /Fab fragment/ chain recombination/modified immunoglobulin)
`
`SHMUEL CABILLYt:J:, ARTHUR D. RIGGSt, HEMA PANDE§, JOHN E. SHIVELY§, WILLIAM E. HOLMES~,
`MICHAEL REY~, L. JEANNE PERRYII, RONALD WETzELII, AND HERBERT L. HEYNEKER:j:~
`
`Divisions of tBiology and §Immunology, Beckman Research Institute of the City of Hope, Duarte, CA 91010; and Departments of ~Molecular Biology and
`IIChemical Sciences, Genentech, Inc., 460 Point San Bruno Boulevard, South San Francisco, CA 94080
`
`Communicated by Rachmiel Levine, January 3, 1984
`
`Plasmids have been constructed that direct
`ABSTRACT
`the synthesis in Escherichia coli of heavy chains and/or light
`chains of an anti-carcinoembryonic antigen (CEA) antibody.
`Another plasmid was constructed for expression of a truncated
`form of heavy chain (Fd' fragment) in E. coli. Functional
`CEA-binding activity was obtained by in vitro reconstitution in
`E. coli extracts of heavy chain or Fd' fragment mixed with
`extracts containing light chain.
`
`Hybridomas have an intrinsic limitation; they can only pro(cid:173)
`duce the natural sequence antibodies found in the B-cell pop(cid:173)
`ulation. Production of antibodies by recombinant DNA tech(cid:173)
`niques, on the other hand, makes possible the construction
`of immunoglobulin derivatives designed for specific pur(cid:173)
`poses. For this and other reasons, such as stability problems
`of hybridomas during large-scale growth, we sought to
`clone, characterize, and express antibody genes derived
`from eDNA.
`Carcinoembryonic antigen (CEA) is a well-characterized
`human tumor marker for colon cancer (1, 2), and hybridoma
`lines producing anti-CEA monoclonal antibodies of high af(cid:173)
`finity have been obtained (3-8). As a model system to test
`the feasibility of antibody production based on recombinant
`techniques, we used an anti-CEA mouse hybridoma as a
`source of mRNA for eDNA preparation, and plasmids were
`constructed that direct the expression of either heav~ (H; y1)
`or light (L; K) chains, or both H and L chains in Escherichia
`coli. A plasmid was also constructed that allows expression
`of a truncated fragment of H chain (Fd') in E. coli. In vitro
`reconstitution experiments with bacterially produced immu(cid:173)
`noglobulin polypeptide chains have yielded molecules with
`CEA-binding specificity.
`
`EXPERIMENTAL
`Cloning of Anti-CEA H- and L-Chain Genes. Hybridoma
`CEA 66-E3, which originated as a fusion of a SP2/0-Ag14
`lymphocytoma cell with a mouse spleen cell (7), secretes
`only an anti-CEA lgG antibody with a K L chain and y1 H
`chain. Total RNA from CEA 66-E3 was extracted as report(cid:173)
`ed by Lynch eta/. (9) and enriched for mRNA by oligo(dT)
`cellulose chromatography. Five micrograms of unfraction(cid:173)
`ated poly(A) mRNA was used as a template for oligo(dT)(cid:173)
`primed preparation of double-stranded eDNA as described
`(10, 11). The eDNA was fractionated according to size by
`electrophoresis on a 6% polyacrylamide gel, and DNA frag(cid:173)
`ments >600 base pairs long were cloned into the Pst I site of
`pBR322 using the G·C tailing method (12). DNA hybridiza(cid:173)
`tion probes were prepared (13) complemeQtary to coding se(cid:173)
`quences in the constant region of L chain (5' T-C-C-A-T-C-
`
`T-T-C-C-C-A-C-C) (14) or H chain (5' C-T-G-G-G-A-T-G-C(cid:173)
`C-T-G-G-T-C) (15). Approximately 8500 E. coli strain 294
`(ATCC 31446) transformants were screened (16), of which
`200 colonies contained L-chain and 40 colonies contained H(cid:173)
`chain eDNA inserts. Further analysis of these clones re(cid:173)
`vealed that the CEA 66-E3 cell line produces at least three
`different L-chain mRNAs and at least two different H-chain
`mRNAs (unpublished data).
`Determination of Anti-CEA H-Chain Sequences. Two plas(cid:173)
`mids (py298 and pyll) were found in which the combined
`nucleotide sequences did account for the entire coding re(cid:173)
`gion, and these were used in the construction of vectors for
`expression of H chain and Fd' fragment.
`The nucleotide sequence that we determined (17, 18) for
`the variable regions of the mouse anti-CEA H-chain mRNA,
`as well as the corresponding amino acids, is shown in Fig.
`1A. Sequence data for the constant region are not shown but
`are in agreement with the sequence determined by Honjo et
`a/. (15). The deduced amino acid sequence of the NH2-termi(cid:173)
`nal region of mature y1 chain corresponds perfectly with the
`sequence of the first 23 NH2-terminal amino acids of hybrid(cid:173)
`oma-derived anti-CEA y1 chain, which we determined by
`protein microsequence analysis (19). The CEA y-chain
`eDNA codes for a putative 19 amino acid signal sequence
`and 447 amino acids of mature H chain. The mature unglyco(cid:173)
`sylated H chain (Mn 49,200) has a variable region of 123 ami(cid:173)
`no acids, including an assumed diversity region of 12 amino
`acids, a J4 joi11ing region of 13 amino acids, and a constant
`region of324 amino acids. H-chain Fab fragment or Fd' frag(cid:173)
`ment to be used in the construction of the Fab portion of lgG
`consists of the first 226 amino acids of mature H chain with a
`Mr of 24,156.
`Determination of Anti-CEA L-Chain Sequences. Several
`plasmids reacting with the K-chain probe were found to con(cid:173)
`tain eDNA inserts large enough to encode full-length K
`chain. The complete nucleotide sequence of the L-chain
`gene was determined by the dideoxynucleotide chain-termi(cid:173)
`nation method (17), after subcloning restriction endonucle(cid:173)
`ase cleavage fragments into M13 mp8 or mp9 vectors (18).
`Fig. 1B shows the nucleotide sequence for the variable re(cid:173)
`gion of the mouse anti-CEA L-chain mRNA. The eDNA in(cid:173)
`sert codes for a tentative 24 amino acid signal sequence and
`214 amino acids of mature L chain. The deduced amino acid
`sequence of the NH2-terminal region of mature K chain cor(cid:173)
`responds perfectly with the first 23 NH2-terminal amino ac(cid:173)
`ids of the hybridoma-derived mouse anti-CEA K chain as de(cid:173)
`termined by our amino acid sequence analysis of the purified
`protein (19). The mature L chain (Mr 23,598) has a variable
`region of 107 amino acids, including a J1 joining region of 12
`amino acids, and a constant region of 107 amino acids.
`
`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: CEA, carcinoembryonic antigen; H chain, heavy
`chain of immunoglobulin; L chain, light chain of immunoglobulin.
`*To whom reprint requests should be addressed.
`
`3273
`
`BEQ 1034
`Page 1
`
`

`
`3274
`
`Biochemistry: Cabilly et aL
`
`A anti-CEA -y-1 chain
`
`Proc. Nat/. Acad. Sci. USA 81 (1984)
`
`10
`1
`-10
`met aen phe g~y ~eu ser ~eu ik tyr ~eu va~ ~eu va~ ~eu ~ys va~ va~ g~n eys glu val met leu val glu ser gly gly val leu met glu
`GAGUCAGCACUGAACACGGACCCCUCACG AUG AAC UUC GGG CUC AGC UUG AUU UAC CUU GUC CUG GUU UUA AAA GUU GUC CAG UGU GAA GUG AUG CUG GUG GAG UCU GGG GGA GUC UUA AUG GAG
`so
`roo
`I
`
`w
`m
`~
`~
`pro gly gly ser leu lys leu ser cys ala ala ser gly phe thr phe ser arg tyr ala met ser trp val arg gln thr pro glu lys arg leu glu trp val ala thr ile ser ser
`.
`a
`~-~~~AAA~~~D~~-~~~~~~~~~~~~~~~~m~~~~GUCD~~~~
`~
`
`w
`ro
`~
`~
`gly gly ser ser his leu pro ser arg gln cys glu gly arg phe thr ile ser arg asp asn ala lys asn thr leu tyr leu gln met ser ser leu arg ser glu asp thr ala met
`~~~~m~ru~~~~w~ru~~-~~~~~m•~~~~~~-~~~~~~a~~
`2!0
`300
`3!0
`
`no
`120
`100
`tyr tyr cys ala arg pro pro leu i le ser leu val ala asp tyr ala met asp tyr trp gly gln gly thr ser val thr val ser ser
`UAU UAC uGU GCA AGA CCC CCU CUU AUU UCG UUA GUA GCG GAC UAU GCU AUG GAC UAC UGG GGU CAA GGA ACC UCA GUC ACC GUC UCC UCA • • • • • • • • • • • • • • • • •
`4 - - - o region- 400
`45o - - - Col)stant Region ----
`J~ region
`
`B anti-CEA K chain
`
`10
`1
`-10
`-20
`met g~y Ue ~Y• met g~u thr hisser g~n va~ phe va~ tyr met ~eu ~ trp ~eu ser g~y va~ g~u g~y asp ile val met thr gln ser his lys phe
`CAAAUAGCGAGUCAGACCAGC AUG GGC AUC AAG AUG GAG ACA CAU UCU CAG GUC UUU GUA UAC AUG UUG CUG UGG UUG UCU GGU GUU GAA GGA GAC AUU GUG AUG ACC CAG UCU CAC AAA UUC
`so
`I
`10"0
`
`w
`w
`~
`~
`met ser thr ser val gly asp arg val ser ile thr cys lys ala ser gln asp val gly ala ala ile ala trp tyr gln gln lys pro gly gln ser pro lys leu leu ile tyr trp
`AUG UCC ACA UCA GUA GGA GAC AGG GUC AGC AUC ACC UGC AAG GCC AGU CAG GAU GUG GGU GCU GCU AUA GCC UGG UAU CAA CAG AAA CCA GGA CAA UCU CCU AAA CUA CUG AUU UAC UGG
`1!0
`~0
`
`w
`ro
`~
`~
`ala ser thr arg his thr gly val pro asp arg phe thr gly ser gly ser gly thr asp phe thr leu thr ile ~er asn val gln ser asp asp leu ala asp tyr phe cys gln gln
`GCA UCC ACC CGG CAC ACU GGA GUC CCU GAU CGC UUC ACA GGC AGU GGA UCU GGG ACA GAU UUC ACU CUC ACC AUU AGC AAU GUG CAG UCU ~AU GAC UUG GCA GAU UAU UUC UGU CAA CAA
`250
`.
`300
`. 3!0
`
`100
`tyr ser gly tyr pro leu thr phe gly ala gly thr lys leu glu leu lys
`UAU AGC GGG UAU CCU CUC ACG UUC GGU GCU GGG ACC AAG CUG GAG CUG AAA • • • • • • • • • • • • • • • • • • •
`-Constant Region - - - -
`--'---JL1 REGION-400
`
`fiG. 1. Nucleotide sequence and deduced amino acid sequence of the variable region of anti-CEA y1-chain mRNA (A) and anti-CEA K(cid:173)
`chain mRNA (B). Putative presequences of 19 amino acids and 24 amino acids have been assigned to Hand L chain, respectively. Diversity
`region (D) and joining regions (JH4 and JLl) are underlined. The numbers above each line refer to amino acid position of mature H or L chain.
`The numbers below each line refer to nucleotide position.
`
`Construction of Plasmids for Expression of Antibody Chains
`in E. coli. Details of the constructions are outlined in the
`legend to Fig. 2. It was our strategy to replace the human
`growth hormone (HGH) gene in plasmid pHGH207-1* with
`either the gene for ll or L chain of Fd' fragment. The plas(cid:173)
`mid pHGH2Q7-1 * resembles in all aspects the expression
`vector pHGH207,1 (21), except that the EcoRI restriction
`site 5' from the trp promoter was eliminated. The nucleotide
`sequence between the trp leader Shine-Dalgarno sequence
`and the initiation codon for the immunoglobulin chains is as
`follows:
`
`ter I operator sequences were placed upstream from the
`structural gene. The most significant feature of this con(cid:173)
`struction was the use of a synthetic deoxyoligonucleotide
`primer to position an A TG initiation codon directly in front
`of the DNA sequence coding for mature H chain (20). In the
`final construction, the tetracycline resistance (TetR) gene
`froQI pBR322 was introduced to yield pyCEAtrp207-1 *.
`Construction of Vector pyCEAFd'trp207-l* for the Direct
`Expression ofFd' Fragment. The gene for Fd' was construct(cid:173)
`ed by introducing a stop codon in the hinge region of the H(cid:173)
`chain gene. The TGT triplet coding for cysteine (nucleotide
`
`initiation
`S-D
`5' ... trp promoter ... A-A-G-G-G-T-A-T-C-T-A-G-A-A-T-T-A-T-G ... antibody chain ...
`Xba I
`
`Construction of this efficient ribosome binding site was
`based on observations that (i) an AT -rich region between the
`Shine-Dalgarno sequence (S-D) and ATG results in effi(cid:173)
`cient expression of HGH (22) and bovine growth hormone
`(BGH) (23); (ii) efficiency of J3-interferon synthesis is opti(cid:173)
`mal when a distance of7-ll bases is maintained between the
`Shine-Dalgarno sequence and ATG (24).
`Construction of Vector pyCEAtrp207-l* for the Direct
`Expression of H Chain. Since neither of the eDNA plasmids
`included a full-length H-chain gene, it was necessary to con(cid:173)
`struct the entire gene from two plasmids: py298, which con(cid:173)
`tains the 5' end of the H-chain gene; and pyll, which com(cid:173)
`pletes its 3' end. Two intermediate plasmids were required
`for the construction of the final expression vector. The first
`intermediate plasmid (pyCEAintl) encodes the COOH-ter(cid:173)
`minal region of H chain. In the construction of the second
`intermediate plasmiq (pyCEAint2), the gene for mature full(cid:173)
`length H chain was reconstructed and the E. coli trp promo-
`
`position 764-766; see figure 5 of ref. 15) was converted to a
`TGA l)top codon by primer repair (20). The cysteine at nucle(cid:173)
`otide position 758-760 (see figure 5 of ref. 15) was retained to
`allow disulfide-bond formation with K chain on in vitro reas(cid:173)
`sociation, as is found in Fab antibody fragments generated
`by papain treatment (25).
`Construction of Vector p~eCEAtrp207 -1* for Direct Expres(cid:173)
`sion of L Chain. Details of the assembly are not presented in
`Fig. 2, but techniques used were similar to those described
`for the construction of the H-chain expression vector, in(cid:173)
`cluding primer repair to position an initiation codon in front
`of the mature L-chain gene.
`
`RESULTS
`Production of Immunoglobulin Polypeptide Chains. E. coli
`strain W3110 (ATCC 27325) was transformed with either
`pyCEAtrp207-l*, pKCEAtrp207-l*, or pyCEAFd'trp207-
`
`BEQ 1034
`Page 2
`
`

`
`Biochemistry: Cabilly et aL
`
`Proc. Nat/. Acad. Sci. USA 81 (1984)
`
`3275
`
`~I
`
`EcoRI
`
`Pstl GEcoRI
`
`Xbal
`
`Pstl
`
`Apll
`
`Xbal
`
`11
`
`Tc"
`
`Pill
`
`T4DNA
`litOII
`
`Pol I
`1101010 -850bp ,,.. .... ,.,
`.... u
`Klenow Poll + 4 dNTP's
`... I
`l106ate 375bp Awl II,
`Toql frogMHt
`
`Ava I
`KlenowPoll
`+ 4 dNTP's
`BaoiHI
`ISOlate Iaroe
`wector fratiMnt
`
`T40NA1ita•
`
`T4NII, Ba•HI
`,~ .... 500 bp frqiMftt
`
`Psll
`ltolato 628 bp trae....m
`BamHI, Hpall
`ltolote 380bp frotiMRI
`
`EcaRI
`Klonow Poll
`+4 dNTP'a
`Balli HI
`laolate lorve
`vector frotiMIIt
`
`Pst I
`
`Ap"
`
`ao :
`
`Mal partial
`K-Poll
`+4dNTp'l
`Ia• HI
`J:,l69bp
`
`Pstt,lciMHI
`loolate Ul43bp
`
`I I
`
`EcoRI
`K - Poll
`+ 4dNTP's
`Pit I
`IIOiatolorve
`vector ftOQMiftf
`
`,,.. ....
`....... ----r:=::-::---' '-----.-=::-::--oiJ ....... ===== r~
`
`Nco• • ..,.,
`:-..::, :~...:~ fratment
`5' C.UCCAC.UTCCC
`Klenow Poll+ 4dNTP's
`lomHI
`1101ote 179bp frotment
`SowtHI
`
`,.
`Poll, BomHI
`Isolate f543bp
`frot.....,
`
`T4 DNA litott
`
`Xbal
`
`I
`
`_!.,bal Hpall
`
`EcoRI
`
`r-------------------,
`~
`I
`:
`I
`I
`I
`I
`:
`I
`I
`I
`I
`I
`I
`I
`
`Pstl
`
`Ap"
`
`I
`I
`I
`I
`:
`I
`I
`I
`I
`I
`I
`l _________________ J
`I
`
`FIG. 2. Construction of vectors for expression in E. coli of immunoglobulin chains. An account of the enzymatic steps involved in the
`generation of new plasmids is presented in chronological order. Purification of DNA and ligation reactions were done as described (20). DNA
`restriction nuclease fragments used in these constructions are indicated by arrows inside the plasmid circles. Primer repair (20) was used to
`create initiation and termination codons. The primer 5' A-T-G-G-A-A-G-T-G-A-T-G-C-T-G, which hybridizes to positions 87-99 of the anti(cid:173)
`coding strand of eDNA (Fig. lA), was used to insert an ATG 5' to position 87. The primer 5' C-A-A-C-C-A-C-A-A-T-C-C-C, which hybridizes
`to the coding strand of n-chain eDNA from position 752-765 (see figure 5 of ref. 15), was used to facilitate conversion of the TGT cysteine
`codon at position 227 to the stop codon TGA.
`
`1*. Transformants were grown in L broth containing 5 p.g of
`tetracycline per ml to an A550 value of 5, and they were then
`induced with 50 p.g of indoleacrylic acid per mi. Total cellu(cid:173)
`lar protein of induced cells was analyzed by NaDodS04/
`polyacrylamide gel electrophoresis. Cells transformed with
`either pyCEAtrp207-1* or pyCEAFd'trp207-1* had promi(cid:173)
`nent new protein bands at the sizes expected for H chain or
`Fd' fragment (data not shown). Cells transformed with
`pKCEAtrp107-1* had a band of increased intensity in the ex(cid:173)
`pected region, but other E. coli proteins of similar size pre(cid:173)
`cluded positive identification by protein staining. It was con(cid:173)
`firmed by immunoblotting (26) that immunoglobulin L(cid:173)
`chain, H-chain, and Fd' fragment were made in good yield
`(Fig. 3). Rough quantitation was done by comparing the
`amounts made in E. coli with different amounts of authentic
`CEA Hand L chains. Estimated yield (percent of total cellu(cid:173)
`lar protein) of immunoglobulin polypeptides was 3% H chain
`and 0.5% L chain. Yield ofFd' fragment could only be deter(cid:173)
`mined from Coomassie-blue stained gels and was =4%.
`Expression of H and L Chains in the Same CeO. To obtain
`transformants producing both chains in the same cell, E. coli
`strain W3110 was first transformed with pKCEAtrp207-1*a,
`which is an ampicillin sensitive (Aps) derivative of
`pKCEAtrp207-1* made by deleting the Pst I/Pvu I fragment
`from the J3-lactamase gene. Cells containing pKCEAtrp207-
`1* a were used as recipients for a second transformation us(cid:173)
`ing pyCEAint2 (Fig. 2). Successful double transformants
`were identified by their ApR TetR phenotype. Double-trans(cid:173)
`formed cells showed protein bands for both H and L chains
`when induced with indoleacrylic acid and analyzed by im(cid:173)
`munoblotting as described above (Fig. 3).
`
`Reconstitution of Antibody Activity. E. coli extracts were
`prepared by suspension of 1 g of frozen cells in 9 ml of lysis
`buffer (7.6 M guanidine hydrochloride/50 mM Tris·HCl, pH
`8/1 mM EDTA). 2-Mercaptoethanol was added to 0.1 M,
`and the suspension was incubated at 37°C for 1 hr, then cen(cid:173)
`trifuged at 15,000 rpm (Sorvall SS-34 rotor) for 30 min. The
`clarified supernatant was stored at 4°C. Simple mixing of cell
`lysates containing 'Yl and K chains does not generate detect(cid:173)
`able antibody activity, nor is activity observed in a lysate of
`cells coproducing the two chains (Table 1). Our reconstitu(cid:173)
`tion procedure was designed to completely reduce and solu(cid:173)
`bilize the 'Yl and K chains in mixtures of crude extracts, fol(cid:173)
`lowed by a gradual return to native conditions in the contin(cid:173)
`ued presence of a mild redox buffer (29) that promotes thiol(cid:173)
`disulfide interchange. Hybridoma anti-CEA was converted
`to denatured disulfide-free chains in the form of the protein
`S-sulfonates (30). This was accomplished by reacting hybrid(cid:173)
`oma anti-CEA (0.5 mg/ml) in 7 M guanidine hydrochloride/1
`mM EDT A/50 mM Tris·HCl, pH 8.5, for 6-16 hr at room
`temperature with sodium sulfite (20 mg/ml) and sodium tet(cid:173)
`rathionate (10 mg/ml), followed by dialysis WC) against 25
`mM Tris·HCl (pH 8.5) in 9 M urea. A nonreducing NaDod(cid:173)
`S04/polyacrylamide gel (not shown) indicated that chain
`separation was complete within the limit of detection, and no
`anti-CEA activity was detectable (Table 1). For reconstitu(cid:173)
`tion, the above extracts were diluted to give a potential IgG
`content of 25 p.g/ml in 8 M guanidine hydrochloride/50 mM
`Tris·HCl, pH 8/1 mM EDTA, and dialyzed 1-2 hr at 4°C
`against 20 vol of 8 M urea (deionized)/50 mM sodium glycin(cid:173)
`ate, pH 10.8/10 mM glycine ethyl ester/1 mM EDTA/1 mM
`reduced glutathione/0.1 mM oxidized glutathione. Dialysis
`
`BEQ 1034
`Page 3
`
`

`
`3276
`
`Biochemistry: Cabilly et al.
`
`Proc. Natl. Acad Sci. USA 81 (1984)
`
`Table 1. Anti-CEA activities of chain combination reactions
`Goat anti-mouse IgG (H and L chains)
`Goat anti-mouse IgG F(ab'h fragment
`Anti-CEA activity, ng/ml
`Anti-CEA activity, ng/ml
`With CEA Without CEA Yield,% With CEA Without CEA Yield,%
`
`Preparation
`
`4
`4
`
`199
`
`163
`454
`880
`
`0
`0
`
`14
`
`8
`28
`11
`
`0.02
`0.02
`
`0.7
`
`0.6
`1.7
`3.5
`
`0
`0
`
`119
`
`143
`717
`1412
`
`0
`0
`
`2
`
`3
`20
`10
`
`0
`0
`
`0.5
`
`0.6
`2.8
`5.6
`
`No treatment
`Hybridoma 'Y" and K-chain S-sulfonates
`E. coli, -y1- and K-chain extract
`Denaturation/renaturation
`Hybridoma -y1- and K-chain S-sulfonates
`Hybridoma -y1- and K-chain S-sulfonates and E.
`coli/IFN-aA extract
`E. coli, -y1-chain extract and E. coli, K-chain extract
`E. coli, -y1- and K-chain extract double-transformed
`E. coli, Fd' fragment extract and E. coli, K-chain
`extract
`E. coli, -y1-chain extract
`E. coli, K-chain extract
`E. coli, Fd' fragment extract
`E. coli, IFN-aA extract
`
`164
`17
`7
`6
`10
`
`0
`10
`0
`0
`0
`
`0.6
`0.03
`O.o3
`0.02
`0.04
`
`360
`1
`10
`0
`0
`
`5
`1.4
`0.004
`0
`0.04
`0
`0
`0
`0
`0
`Reactions were analyzed in an ELISA using polyvinyl chloride rnicrotiter plates coated with CEA antigen by incubation for 16 hr at room
`temperature with a solution of 5 1-11/ml. Nonspecific sites were sealed with 5% gelatin, and then samples of hybridoma anti-CEA (for the
`standard curve) and folding reactions were incubated in the wells for 90 min at 3"rC. After washing with phosphate-buffered saline, wells were
`incubated with goat anti-mouse lgG (H and L chains) or goat anti-mouse lgG F(ab'h conjugated with alkaline phosphatase (both from Zymed
`Laboratories, South San Francisco), then with phosphatase substrate (Sigma). Samples were read according to a standard curve (with CEA).
`Nonspecific background was assessed on wells treated only with 5% gelatin (without CEA). Association yields were calculated as follows.
`Based on estimates of levels of immunoglobulin chains in E. coli, the theoretical yield of lgG in experiments involving K and y chains was 25
`1-11/ml; yield of Fab was 17 1-11/ml. In experiments containing only K chain, y chain, or interferon (IFN) extracts, a hypothetical yield of 25
`1-11/ml was assigned; for Fd' alone it was 17 1-11/ml. For IgG percent yield, the amount of specific binding (anti-CEA activity with CEA minus
`anti-CEA activity without CEA) was multiplied by 100 and divided by 25,000 ng/ml. For Fab, the apparent specific binding (ng/ml) from the
`standard curve was multiplied by 0.67 to reflect that two-thirds of the molecular weight of the CEA-associated ligand in the standard curve was
`Fab fragment sensitive to goat "anti-mouse IgG F(ab'h. This number was multiplied by 100 and divided by 17,000 ng/ml. The results shown are
`the averages of two dilutions of each of two independently conducted folding reactions.
`
`bags were placed in 20 vol of fresh buffer in a graduated
`cylinder at 4°C, and N2-saturated native buffer (the above
`buffer minus urea) was delivered over a 15-hr period to give
`a final urea concentration of 1 M. At this time, serum albu(cid:173)
`min was added to each dialysis bag to a concentration of 0.5
`mg/ml, and the bags were transferred to phosphate-buffered
`saline/0.1 mM phenylmethylsulfonyl fluoride, not degassed,
`and dialyzed at 4°C for another 2 hr.
`The final dialysate was assayed for CEA-specific binding
`activity using an ELISA procedure. Table 1 shows that no
`significant anti-CEA activity is obtained for E. coli extracts
`producing 1'1 chain, K chain, Fd' fragment, or subtype A of
`human a-interferon (31) alone. On the other hand, significant
`binding is obtained when K-chain extract is mixed with either
`1'1 chain or Fd' fragment extract. Extract from E. coli ex(cid:173)
`pressing both 1'1 and K chains generates comparable activity.
`Since we had observed that H- and L-chain specific anti(cid:173)
`mouse IgG antibodies were directed primarily toward Fe de(cid:173)
`terminants (see legend to Fig. 3), we independently assayed
`the extracts with a goat anti-mouse IgG conjugate specific
`for F(ab'h (Table 1).
`
`DISCUSSION
`The expression of cloned genomic K-chain immunoglobulin
`genes in mouse lymphoid cell lines has been reported (32-
`34), but the expression of complete H- and L-chain genes
`derived from eDNA has not previously been obtained. Our
`results demonstrate that complete antibody chains can be
`produced efficiently in E. coli. There is, however, no detect(cid:173)
`able antibody activity in extracts of E. coli coproducing sub(cid:173)
`stantial levels of lgG H and L chains (Table 1). This may be
`due to the highly reducing intracellular environment (35, 36),
`which inhibits disulfide-bond formation, and to the accumu(cid:173)
`lation of gene products in insoluble "refractile bodies" in the
`cell, a phenomenon noted in many cases of exogenous gene
`
`expression in E. coli (37). It is also possible that E. coli is
`lacking a protein that might be required for in vivo formation
`of IgG (28).
`Most of the work described in the literature on in vitro
`antibody reconstitution from reduced denatured H and L
`chains was done with poly clonal antibodies. While formation
`of insoluble aggregates is a significant side reaction in such
`studies, reconstitution of at least several percent was ob(cid:173)
`tained for some immunoglobulin derivatives (38). Better
`yields (17%) have been reported for reconstitution of re(cid:173)
`duced denatured Fab fragment rabbit anti-ribonuclease (39).
`Since it was necessary to solubilize E. coli-produced immu(cid:173)
`noglobulin polypeptides using denaturant, and since anti(cid:173)
`body activity has been regenerated in modified immunoglob(cid:173)
`ulins from completely reduced and denatured molecules, we
`devised a method of lgG reconstitution starting from dena(cid:173)
`turing conditions. Under these conditions, the yield of anti(cid:173)
`CEA activity from E. coli extracts was 3-5% when H and L
`chain were coexpressed in E. coli. For reasons not yet un(cid:173)
`derstood, reconstitution has been consistently better with E.
`coli-derived antibody chains than with hybridoma-derived
`material. It is possible that S-sulfonation inhibits reconstitu(cid:173)
`tion. The yield of activity with E. coli-produced Fab frag(cid:173)
`ment was 1.4% when assayed in the ELISA using as anti(cid:173)
`body conjugate an anti-mouse lgG Fab antibody (Table 1).
`The approach we have taken to express immunoglobulin
`chains from eDNA opens up new avenues toward produc(cid:173)
`tion of recombinant antibodies. For example, this approach
`makes possible the construction at the DNA level of modi(cid:173)
`fied immunoglobulins with unique properties such as the Fab
`fragment described here. The construction of chimeric anti(cid:173)
`CEA antibody chains by recombining the variable antigen(cid:173)
`binding region from mouse hybridoma eDNA with the con(cid:173)
`stant region from human IgG eDNA could produce antibod(cid:173)
`ies with the advantage over mouse hybridomas of being less
`immunogenic when used for human in vivo diagnostic or
`
`BEQ 1034
`Page 4
`
`

`
`Biochemistry: Cabilly et a/.
`
`Proc. Nat/. Acad. Sci. USA 81 (1984)
`
`3277
`
`H
`
`L
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7 8
`
`Immunoblots of bacterially produced immunoglobulin
`FIG. 3.
`peptide chains. Frozen E. coli cells were dispersed in 10 vol of 10
`mM Tris·HCI, pH 7.5/1 mm EDTA/0.1 M NaCI/0.1 mM phenyl(cid:173)
`methylsulfonyl fluoride, disrupted by sonication, and partially clari(cid:173)
`fied by centrifugation at 3000 rpm for 20 min in an SS-34 rotor. Ali(cid:173)
`quots of these extracts and of hybridoma-derived anti-CEA anti(cid:173)
`body were boiled for 5 min in 10 mM Tris · HCI/1 mM EDT A/0.1 M
`2-mercaptoethanol/2.0% NaDodS04 • The proteins were precipitat(cid:173)
`ed with 10 vol of acetone for 10 min, centrifuged for 5 min at 12,000
`rpm, boiled for 5 min in O'Farrell's buffer (27), and size-fractionated
`on a 10% polyacrylamide gel in the presence of NaDodS04 • Using
`the immunoblot technique (26), the proteins were transferred toni(cid:173)
`trocellulose and the immunoglobulin chains were identified radioim(cid:173)
`munochemically. Lanes 1-6, rabbit anti-mouse IgG and then ml(cid:173)
`labeled protein A were used. Lanes 7 and 8, 12~1-labeled sheep anti(cid:173)
`mouse IgG was used, because we found that pooled rabbit anti(cid:173)
`mouse IgG did not react with Fd'. Even sheep anti-mouse IgG reacts
`less well with Fd' than with 'Y chain, probably because the major
`antigenic determinants are in the Fe region, which has been re(cid:173)
`moved. Lanes 1-3 contain K (H) and 'Y (L) chains from hybridoma(cid:173)
`derived anti-CEA; the amounts are 1.0, 0.1, and 0.01 ,...g, respective(cid:173)
`ly. Lane 4, E. coli K chain; lane 5, E. coli 'Y chain; lane 6, E. coli K
`and 'Y chains produced in the same cell; lane 7, E. coli K and 'Y
`chains; lane 8, E. coli Fd' fragment.
`
`therapeutic purposes. In addition, the expression of pre(cid:173)
`heavy chain (or pre-Fd' fragment) and pre-light chain in E.
`coli, yeast, and mammalian cells could be pursued to estab(cid:173)
`lish conditions for in vivo assembly of secreted antibodies.
`These approaches could lead to alternative ways of prepar(cid:173)
`ing stable human monoclonal antibodies, as well as antibod(cid:173)
`ies having properties custom-designed for particular re(cid:173)
`search and therapeutic applications.
`
`We thank Charles Todd (City of Hope), Charles Benton, Mike
`Geier, and Vince Anicetti (Genentech) for helpful discussions and
`encouragement; the Organic Chemistry Department for synthesis of
`DNA fragments; and the Process Development Department for fer(cid:173)
`mentation of the E. coli strains. We also thank Jeanne Arch (Genen(cid:173)
`tech) for preparation of the manuscript and Alane Gray (Genentech)
`for artwork.
`
`1. Gold, P. & Freedman, S. 0. (1965)1. Exp. Med. 121,439-448.
`2. Shively, J. E. & Todd, C. W. (1981) in Handbook of Cancer
`Immunology. Tumor Antigens: Structure and Function, ed.
`Waters, H. (Garland, New York), Vol. 8, pp. 321-354.
`3. Acolla, R. S., Carrel, S. & Much, J.-P. (1980) Proc. Nat/.
`Acad. Sci. USA 71, 563-567.
`4. Kupchik, H. Z., Zurawski, V. R., Jr., Hurreii,.J. G. R., Zam(cid:173)
`chek, M. & Black, P. H. (1981) Cancer Res. 41, 3306-3310.
`5. Hedin, A., Hammarstrom, S. & Larsson, A. (1982) Mol. lm(cid:173)
`munol. 19, 1641-1648.
`
`6. Primus, F. J., Newell, K. D., Blue, A. & Goldenberg, D. M.
`(1983) Cancer Res. 43, 686-692.
`7. Wagener, C., Yang, Y. H. J., Crawford, F. G. & Shively,
`J. E. (1983) 1. lmmunol. 130, 2308-2315.
`8. Wagener, C., Clark, B. R., Rickard, K. J. & Shively, J. E.
`(1983) 1. lmmuno/. 130, 2302-2307.
`9. Lynch, K. R. , Pennica, D., Ennis, H. L. & Cohen, P. S.
`(1979) Virology 89, 251-254.
`10. Goedde!, D. V., Heyneker, H. L., Hozumi, T., Arentzen, R.,
`ltakura, K. , Yansura, D. G., Ross, M. J., Miozzari, G., Crea,
`R. & Seeburg, P. H. (1979) Nature (London) 281, 544-548.
`11. Wickens, M.P., Buell, G. N. & Schimke, R. T. (1978)1. Bioi.
`Chern. 253, 2483-2495.
`12. Chang, A. C. Y., Nunberg, J. H., Kaufman, R. J., Erlich,
`H. A., Schimke, R. T. & Cohen, S. N. (1978) Nature (Lon(cid:173)
`don) 275, 617-624.
`13. Crea, R. & Horn, T. (1980) Nucleic Acids Res. 8, 7331-7348.
`14. Hamlyn, P. H., Gait, M. J . & Milstein, C. (1981) Nucleic Ac(cid:173)
`ids Res. 9, 4485-4494.
`15. Honjo, T., Obata, M., Yamawaki-Kataoka, Y., Kataoka, T.,
`Kawakami, T., Takahashi, N. & Mano, Y. (1979) Ce//18, 559-
`568.
`16. Wallace, R. B., Johnson, M. J., Hirose, T., Miyake, T.,
`Kawashima, E. H. & ltakura, K. (1981) Nucleic Acids Res. 9,
`879-894.
`17. Smith, A. J. H. (1980) Methods Enzymol. 65, 560-580.
`18. Messing, J., Crea, R. & Seeburg, P. H. (1981) Nucleic Acids
`Res. 9, 309-321.
`19. Shively, J. E. (1981) Methods Enzymol. 79, 31-48.
`20. Goedde!, D. V., Shepard, H. M., Yelverton, E., Leung, D.,
`Crea, R., Sloma, A. & Pestka, S. (1980) Nucleic Acids Res. 8,
`4057-4074.
`21. de Boer, H. A., Comstock, L. J ., Yansura, D. & Heyneker,
`H. L. (1982) in Promoters: Structure and Function, eds. Rodri(cid:173)
`guez, R. L. & Chamberlin, M. (Praeger, New York), pp. 462-
`481.
`22. de Boer, H. A., Comstock, L. J., Hui, A., Wong, E. &
`Vasser, M. (1983) Biochem. Soc. Symp. 48, 233-244.
`23. Matteucci, M. & Heyneker, H. L. (1983) Nucleic Acids Res.
`11, 3113-3121.
`24. Shepard, H. M., Yelverton, E. & Goedde!, D. V. (1982) DNA
`1, 125-131.
`25. Anzel, L. M. & Poljak, R. J. (1979) Annu. Rev. Biochem. 48,
`961-997.
`26. Burnette, W. N. (1981) Anal. Biochem. 112, 195-203.
`27. O'Farrell, P. H. (1975) 1. Bioi. Chern. 250, 4007-4021.
`28. Wabl, M. & Steinberg, C. (1982) Proc. Nat/. Acad. Sci. USA
`79, 6976-6978.
`29. Saxena, V. P. & Wetlaufer, D. B. (1970) Biochemistry 9, 5015-
`5023.
`30. Means, G. E. & Feeney, R. E. (1971) Chemical Modification
`of Proteins (Holden-Day, San Francisco), pp. 152-154.
`31. Goedde!, D. V., Yelverton, E., Ullrich, A., Heyneker, H. L.,
`Miozzari, G., Holmes, W., Seeburg, P. H., Dull, T., May, L.,
`Stebbing, N., Crea, R., Maeda, S., McCandliss, R., Sloma, A.,
`Tabor, J. M., Gross, M., Familletti, P. C. & Pestka, S. (1980)
`Nature (London) 287, 411-416.
`32. Rice, D. & Baltimore, D. (1982) Proc. Nat/. Acad. Sci. USA
`79, 7862-7865.
`33. Ochi, A., Hawley, R. G., Shulman, M. J. & Hozumi, T. (1983)
`Nature (London) 302, 340-342.
`34. Oi, V. T., Morrison, S. L., Herzenberg, L.A. & Berg, P.
`(1983) Proc. Nat/. Acad. Sci. USA 80, 825-829.
`35. Freedman, R. B. & Hillson, D. A. (1981) in Enzymology of
`Post-Trans