IE62257B1 - A process for the preparation of antibodies by genetic engineering - Google Patents

Abstract

Functional antibody fragments can be produced in bacteria by coupling each of the genes for the individual chains to signal sequences which bring about the transport of the prepeptides through the cytoplasmic membrane, expressing the gene structures in a bacterium, and isolating the functional protein from the periplasmic space or the medium. The functional protein is advantageously isolated by affinity chromatography on an adsorbent which is loaded with the hapten or antigen, and elution with a solution of the hapten or antigen.

Description

A process for the preparation of antibodies by genetic engineering The expression of antibodies in yeast has been described (C.R. Wood, M.A. Boss, J.H. Kenten, J.E. Calvert, N.A. Roberts and J.S. Emtage, Nature 314, 446 (1985)), but only a very small proportion of the expressed protein proved to be functional. In E. coli, to date it has been possible to obtain antibody proteins only in denatured form (M.A. Boss, J.H. Kenten, C.R. Wood and J.S. Emtage, Nucleic Acids Res. 12, 3791 (1984); S. Cabilly, A.D.

Riggs, H. Pande, J.E. Shively, W.E. Holmes, M. Rey, L.J. Perry, R. Wetzel and H.L. Heyneker, Proc. Natl. Acad.

Sci. U.S.A. 81, 3273 (1984)). The purification of active antibodies or antibody fragments from yeast or other microorganisms has not been disclosed. Attempts at protein folding have to date resulted in only a very small percentage of correctly folded recombinant antibody proteins. In addition, it is difficult to separate the desired functional proteins from undesired and non-functional proteins, which impedes accurate measurement of binding constants, yields of folding, spectral properties and the like, and use in therapy and industry. Optimization of the folding conditions is thus extremely difficult, especially since there are no proven processes and measurement methods for the refolding.

The expression of complete functional antibodies or functional binding domains of antibodies in bacterial expression systems has not been disclosed to date, and the prospects of bringing it about have been assessed pessimistically (S.L. Morrison, Science 229, 1202 (1985), M.A. Boss and C.R. Wood, Immunol. Today 6, 12 (1985)).

Such a system would be very desirable because the genetic - 2 engineering processes have been thoroughly worked out, especially for E. coli, and mass production is facilitated by the rapid growth, which is of considerable economic importance.

Hence the invention relates to the preparation of functional antibodies, functional fragments thereof, or fusion proteins composed of antibody domains and other proteins in bacteria, preferably Gram-negative bacteria, especially in E. coli. The process according to the invention comprises coupling the genes for the individual chains of the antibody molecule or fragment each to a signal sequence which brings about the transport of the polypeptide chains through the cytoplasmic membrane and which can be split off, bringing about the expression of the gene structures, and isolating the functional protein from the periplasmic space or the medium. Preferred embodiments of this invention are explained in detail hereinafter and defined in the patent claims.

The coupling of the genes for the individual chains provided with signal sequences preferably takes place in the manner of a regulatable operon system which brings about simultaneous expression by a common control region. In this way, the individual protein chains are expressed together in approximately the stoichiometric ratio, and are transported into the periplasmic space, where joining to form a functional molecule takes place. Proteins are transported out of the cytoplasm by methods known per se, as are described, for example, in European Patent Specification 0,006,694.

Suitable as control region is every suitable regulatable gene regulatory region, for example lac, tac, trp or synthetic sequences. Particularly preferred are regulatory regions which can be reliably turned off.

The proteins are preferably isolated from the periplasmic space by exerting a mild osmotic shock on the harvested cells, and subjecting the liquid phase obtained by this to concentration by ultrafiltration or precipitation, for example with salts such as ammonium sulfate.

A mild osmotic shock brings about the ejection of the periplasm with, however, the cytoplasmic membrane remaining intact.

The protein concentrate is, expediently after a dialysis, applied to an adsorbent, advantageously in the form of an affinity column, which is loaded with the appropriate antigen or hapten. The antibody or the functional antibody fragment is then obtained by suitable elution, advantageously with the antigen or hapten.

In the eukaryotic cell, antibodies are formed in the lumen of the endoplasmic reticulum - probably with the cooperation of disulfide isomerases, proline cis-transisomerases and possibly other enzymes or proteins. It was surprising that the bacterial cell is also able to prepare the two chains in approximately the same stoichio metric amount, to transport the two precursor proteins into the periplasmic space or the medium surrounding it, to eliminate the signal sequences correctly, to fold the globular and soluble domains correctly, to form the intra molecular disulfide linkages, and to associate the two chains to give a heterodimer, because it is regarded as improbable that the bacterial cell is equipped with enzymes which are of this nature or have the same effect Thus, it has surprisingly emerged that, in the case of Gram-negative bacteria, the bacterial periplasm is in this respect functionally equivalent to the lumen of the eukaryotic endoplasmic reticulum.

The process according to the invention has a number of advantages: apart from the easy and low-cost large-scale bacterial - 4 fermentation already mentioned, there is direct -formation of -functional proteins, which thus means that the cleavage of fusion proteins, with subsequent isolation of the desired protein or protein fragment, its oxidation or in vitro refolding, are avoided.

Nor have problems with cellular proteases been found in the process according to the invention. It is, after all, known that, because of these cellular proteases, proteins are normally expressed in bacteria in the form of fusion proteins, especially as insoluble inclusion bodies, which, however, is associated with the stated elaborate further processing steps. In contrast, the separation and purification to homogeneity is rapid and straightforward in the process according to the invention.

Thus the invention permits easy access to antibodies, functional fragments thereof and modified antibodies which differ from the natural antibodies by the insertion, elimination and/or exchange of amino acids. Thus, for example, cysteines can be eliminated or replaced by other amino acids in order to suppress undesired folding. It is possible in just the same way to convert a murine into a human antibody, or to introduce other mutations. Thus, besides the possible pharmacological and industrial applications, access is facilitated for research into antibody structure and function and into the fundamentals of enzymatic catalysis (V. Raso and B.D. Stollar, Biochemistry 14, 584 (1975); V. Raso and B.D. Stollar, Biochemistry 14, 591 (1975); A. Tramontano, K.D. Janda and R.A. Lerner, Science 234, 1566 (1986); S.J. Pollack, J.W. Jacobs and P.G. Schultz, Science 234, 1570 (1986); J. Jacobs, P.G. Schultz, R. Sugasawara and M. Powell, J. Am. Chem. Soc. 109, 2174 (1987)).

The invention furthermore permits the application of test systems (assays) directly to the bacterial cell in which the functional antibody is formed, and thus rapid investigation for possibly mutated antibodies.

Besides the stated variations in the antibody molecule by altering individual, or a few, amino acids, it is also possible to insert into the antibody gene other gene regions, or to exchange parts for non-critical gene regions. It is possible in this way to couple marker enzymes (R.S. Neuberger, G.T. Williams and R.O. Fox, Nature 312, 604 (1984)), toxins (G. Roller (ed.) Antibody carriers of drugs and toxins in tumor therapy, Immunol. Rev. 62, Runksgaard, Copenhagen (1982)) or immunoglobulin regions of another class (R.S. Neuberger, G.T. Williams, E.B. Ritchell, S.S. Jouhal, J.G. Flanagan, and T.H. Rabbitts, Nature 314, 268 (1985)) or of another species (P.T. Jones, P.H. Dear, J. Foote, R.S. Neuberger, and G. Winter, Nature 321, 522 (1986)) to the antibody molecule.

The process according to the invention is explained hereinafter taking the example of the variable domains of the phosphorylcholine-binding antibody myeloma protein RcPC603. The three-dimensional structure of this mouse immunoglobulin A is known (D.R. Segal, E.A. Padlan, G.H. Cohen, S. Rudikoff, R. Potter and D.R. Davies, Proc. Natl. Acad.

Sci. 71, 4298 (1974)). Synthetic genes for the variable light chain V|_ and heavy chain were used. Such synthetic genes are proposed in German OffenIegungsschrift 3,715,033 and published European Patent Application 0,290,005 and published Australian Patent Application 15631/88 (Tables 1 and 2 therein). A particularly expedient embodiment of such synthetic genes is shown in the Table, in which the DNA sequence of the complete expression plasmid is depicted, and the genes for the two chains are emphasized by indicating the amino acids.

Codons which are rarely used by E. coli were avoided in the construction of these DNA sequences. Furthermore, unique restriction enzyme cleavage sites were incorporated, and account was taken of the secondary structure of the RNA. ό In the functional antibody fragment serving as model, each domain has an intramolecular disulfide bridge (from Cys 23 to Cys 94 in VL and from Cys 22 to Cys 98 in V^).

There is no disulfide bridge between the chains, nor is there any free cysteine.

The expression vector used, pASK 22, is depicted diagrammatically in the following formula In it the synthetic genes for the V|_ and domains are coupled to gene fragments for the bacterial signal sequence of the outer membrane protein A (ompA) on the one hand, and of alkaline phosphatase (phoA) on the other hand. The genes for the two precursor proteins are located in a synthetic operon-like structure downstream of the lac promoter, which ensures that both genes undergo simultaneous induction, coexpression and cosecretion.

After induction of gene expression, the cells are harvested and exposed to a mild osmotic shock. The liquid - 7 phase obtained by this, which contains the periplasmic proteins, is concentrated by ultrafiltration, dialyzed and applied directly to an affinity column which contains a phosphorylcholine derivative (B. Chesebro and H. Metzger, Biochemistry 11, 766 (1972)) as affinity ligand. Elution with phosphorylcholine results in a homogeneous Fy fragment which is gel electrophoretica11 y homogeneous. It can be deduced from the SDS polyacrylamide gel that the two chains of the purified Fy fragment are present in a molar ratio of 1:1, and the expected molecular weights of the mature proteins (V^: 13600 D, VL: 12400 D) are present. To demonstrate correct elimination of the two signal sequences, the six N-terminal amino acids of both chains were sequenced. It emerged that both chains have the correct N-termini for the mature proteins. Thus, both heterologous preproteins have been correctly cleaved by the bacterial signal peptidase, and there is no detectable evidence that imprecise processing or an N-terminal degradation reaction has taken place.

The affinity constant of the recombinant Fy fragment of McPC603 was measured by equilibrium dialysis. The conditions used for this were the same as were applied in the determination of the affinity constant of the natural antibody McPC603 isolated from mouse ascites. The value of 1.21 ♦ 0.06 x 10^ m-1 found for the Fy fragment is identical, within the accuracy of the experiment, to the value of 1.6 + 0.4 x 10^ M"1 reported for the natural antibody. The Scatchard binding plot (Ann. N.Y. Acad. Sci. 51, 660 (1949)) is linear, and extrapolation indicates that approximately 1 mol of hapten is bound per mol of Fy fragment. This confirms that there is only one type of binding site per Fy fragment, and that there are no inactive components present in the isolated protein.

Thus, it has emerged, surprisingly, that it is possible to prepare the Fy fragment of the antibody McPC603 as - 8 completely functional and stable protein in E coli. This demonstrates the functional equivalence of the transport into the periplasm of the bacterial cell with the transport into the lumen of the endoplasmic reticulum of the eukaryotic cell. This equivalence has not been disclosed or even suspected, because the bacterial protein which has hitherto been characterized best and which could be defined as soluble heterodimeric protein in the periplasm - E. coli penicillin acylase - is produced by proteolytic processing from a single-chain precursor in the periplasm (G. Schumacher, D. Sizmann, H. Hang, P. Buckel und A.

Bock, Nucleic Acids Res. 14, 5713 (1986)). In addition, it has already been pointed out that, according to present knowledge, bacteria have no enzymes or proteins which might in eukaryotes be involved in the folding.

It was also surprising that the F y fragment of H c P C 6 0 3 has essentially the same affinity constant for phosphorylcholine as has the intact antibody McPC603 itself. This finding is unexpected because the functionality of Fy fragments is controversial in the literature (J.. Sen and S. Beychok, Proteins 1, 256 (1986)). It emerges from this that the functionality can be maintained on modification or even complete deletion of the constant domains.

In contrast to the only method used hitherto for the preparation of Fy fragments, namely by proteolysis of an antibody, there are no problems in the process according to the invention with non-functional, incorrectly folded, incorrectly reassociated or chemically modified proteins. Moreover, the preferred isolation process using an antigenor hapten-loaded adsorbent is also suitable for purifying those Fy fragments obtained by known processes, because such impurities would either not be bound or not be eluted.

Example 1: Preparation of the plasmid pASK22: The expression plasmid pASK22 is constructed from the large EcoRI-Hind111 fragment of pUCl2 (C. Yanisch-Perron , J. Vieira, and J. Messing, Gene 33, 103 (1985)), from fragments of the vectors pIN III-0mpA1 (Y. Masui, J. Coleman and M. Inouye in Experimental manipulation of gene expression, M. Inouye, ed., Academic Press 1983) and pHI61 (H. Inouye, W. Barnes and J. Beckwith, J. Bacteriol. 149, 434 (1982)) as well as various synthetic DNA fragments, in several stages using methods known per se (T. Maniatis, E.F. Fritsch and J. Sambrook, Molecular Cloning, Cold Spring Harbor 1982). The complete DNA sequence of the resulting vector pASK22 is depicted in the Table.

The ligation mixture is used to transform competent E. coli cells, and the latter are selected for ampicillin resistance. The plasmids with the desired gene structure are characterized by restriction analysis and sequencing of the critical junctions.

Example 2: Preparation of the Fy fragment A culture of the E. coli strain W3110 transformed with pASK22 is cultivated in LB medium containing 100 mg/l ampicillin to an OD55Q of 0.5. Expression is induced by addition of IPTG to a final concentration of 1mM.

After 45 minutes, the cells are harvested by centrifugation at 4000 x g (10 minutes at 4°C). Cell fractionation is carried out by resuspension of the cell pellet in TES buffer (0.2 M tris.HCl, pH 8.0; 0.5 mM EDTA; 0.5 M sucrose) in 10 ml/l of the original culture. The cells are exposed to a mild osmotic shock by addition of 15 ml/l of the original culture of TES buffer which is diluted 1:4 with water and contains 2 mM phosphorylcholine. After the suspension has been incubated on ice for 30 minutes it is centrifuged at 5000 x g for 10 minutes, and the supernatant is subjected to renewed centrifugation at 48000 x g for 15 minutes. The resulting supernatant, which contains all the soluble periplasmic proteins, is t p ) concentrated by ultrafiltration ( AMICON YM5 membrane) to a volume of about 2.5 ml/l of the original culture, and is dialyzed against BBS buffer (0.2 M borate/NaOH, pH 8.0; 0.16 M NaCl). This concentrated solution is applied to an affinity column loaded with a phosphorylcholine derivative (B. Chesebro and H. Metzger, loc.cit.) (1-4 ml of solution per ml of bed votume), which is washed with BBS buffer, and the pure Fy fragment is eluted with a solution of 1mM phosphoryIcholine in BBS buffer.

Example 3: Equilibrium dialysis In a dialysis chamber with a volume of 100 yl on each side of the membrane, 50 yl of purified Fy fragment in BBS buffer were placed on one side, and a solution of 14 yl of phosphoryI(methyl- Ocholine (50 mCi/mmol) in BBS buffer was placed on the other side. The concentration of the Fy fragment determined from the 0&205 6.85 was 0.22 mg/ml (R.K. Scopes, Protein Purification Principles and Practice, Springer-Verlag, New York, 1982, p. 241). After 22 hours at room temperature, equilibrium had been reached, and 20 yl samples of each solution were measured in a scintillation counter (Beckman LS 1801), and the data were subjected to Scatchard plotting (loc. cit.). The affinity constant derived from the gradient of the resulting line is Ka = 1.21 ♦ 0.06 x 10^ M \

Claims (8)

1. A process for the preparation of functional antibodies, functional fragments of antibodies or fusion proteins composed of functional antibody domains and other proteins, which comprises the genes for the heavy and light chain each being coupled to one signal sequence which bring about the transport of the polypeptide chains through the cytoplasmic membrane of a bacterial cell and are then eliminated, bringing about the expression of the gene structures in a bacterium, and isolating the functional protein from the periplasmic space or the medium.

2. The process as claimed in claim 1, wherein the bacterium is a Gram-negative bacterium.

3. The process as claimed in claim 2, wherein the bacterium is E. coli.

4. The process as claimed in claim 1, 2 or 3, wherein the genes for the two prepeptides are coupled in the form of a regulatable operon system. }

5. The process as claimed in one or more of the preceding claims, wherein, in order to isolate the functional protein from the periplasmic space, the bacteria which have been separated off are exposed to an osmotic shock such that the periplasm is ejected but the cytoplasm remains intact, the liquid phase obtained by this is enriched by centrifugation, and the desired proteins are obtained from this concentrate.

6. The process as claimed in one or more of the preceding claims, wherein the solution containing the functional protein is applied to an adsorbent loaded with the appropriate antigen or hapten, and the desired proteins are eluted by elution with a solution containing this antigen or hapten.

7. A process according to claim 1 for the preparation of a functional antibody, a functional fragment of an antibody or a fusion protein, substantially as hereinbefore described and exemplified.

8. A functional antibody, a functional fragment of an antibody or a fusion protein, whenever prepared by a process claimed in a preceding claim.