Methods for refolding proteins containing free cysteine residues

This application is a divisional of U.S. patent application Ser. No. 10/276,358, filed Apr. 10, 2003, now U.S. Pat. No. 7,306,931, which is a national stage application under 35 U.S.C. § 371 of PCT Application No. PCT/US01/16088, filed May 16, 2001, which claims the benefit of priority under 35 U.S.C. § 119(e) to U.S. Provisional Application Ser. No. 60/204,617, filed May 16, 2000. The entire disclosure of each of the above-identified applications is incorporated herein by reference.

This invention was made with government support under Grant Nos. 1 R43 CA086577, 2R44 CA086577, 1R43 CA090003, 1R43CA099217, 2R44 CA099217, 1R43 AR051609, 2R44 AR051609, each awarded by the National Institutes of Health, and under Grant No. DAMD17-00-1-01-58, awarded by the Department of the Army. The government has certain rights in the invention.

This application contains a Sequence Listing submitted on a compact disc, in duplicate. Each of the two compact discs, which are identical to each other pursuant to 37 CFR § 1.52(e)(4), contains the following file: “4152-4-PCT_Sequence_Listing.txt”, having a size in bytes of 15 kb, and recorded on 11 Dec. 2007. The information contained in this electronic file is hereby incorporated by reference in its entirety pursuant to 37 CFR §1.52(e)(5) and 37 CFR §1.77(b)(5).

The present invention relates generally to methods of making proteins and more specifically to recombinant proteins containing at least one “free” cysteine residue, i.e., a cysteine residue that does not participate in a disulfide bond.

Protein therapeutics generally must be administered to patients by injection. Most protein therapeutics are cleared rapidly from the body, necessitating frequent, often daily, injections. There is considerable interest in the development of methods to prolong the circulating half-lives of protein therapeutics in the body so that the proteins do not have to be injected frequently. Covalent modification of proteins with polyethylene glycol (PEG) has proven to be a useful method to extend the circulating half-lives of proteins in the body (Abuchowski et al., 1984; Hershfield, 1987; Meyers et al., 1991). Covalent attachment of PEG to a protein increases the protein\'s effective size and reduces its rate of clearance from the body. PEGs are commercially available in several sizes, allowing the circulating half-lives of PEG-modified proteins to be tailored for individual indications through use of different size PEGs. Other documented in vivo benefits of PEG modification are an increase in protein solubility and stability, and a decrease in protein immunogenicity (Katre et al., 1987; Katre, 1990).

One known method for PEGylating proteins covalently attaches PEG to cysteine residues using cysteine-reactive PEGs. A number of highly specific, cysteine-reactive PEGs with different reactive groups (e.g., maleimide, vinylsulfone) and different size PEGs (2-40 kDa, single or branched chain) are commercially available. At neutral pH, these PEG reagents selectively attach to “free” cysteine residues, i.e., cysteine residues not involved in disulfide bonds. Cysteine residues in most proteins participate in disulfide bonds and are not available for PEGylation using cysteine-reactive PEGs. Through in vitro mutagenesis using recombinant DNA techniques, additional cysteine residues can be introduced anywhere into the protein. The newly added “free” or “non-natural” cysteines can serve as sites for the specific attachment of a PEG molecule using cysteine-reactive PEGs. The added “free” or “non-natural” cysteine residue can be a substitution for an existing amino acid in a protein, added preceding the amino-terminus of the mature protein or after the carboxy-terminus of the mature protein, or inserted between two normally adjacent amino acids in the protein. Alternatively, one of two cysteines involved in a native disulfide bond may be deleted or substituted with another amino acid, leaving a native cysteine (the cysteine residue in the protein that normally would form a disulfide bond with the deleted or substituted cysteine residue) free and available for chemical modification. Preferably the amino acid substituted for the cysteine would be a neutral amino acid such as serine or alanine. For example, human growth hormone (hGH) has two disulfide bonds that can be reduced and alkylated with iodoacetamide without impairing biological activity (Bewley et al., (1969). Each of the four cysteines would be reasonable targets for deletion or substitution by another amino acid.

Several naturally occurring proteins are known to contain one or more “free” cysteine residues. Examples of such naturally occurring proteins include human Interleukin (IL)-2 (Wang et al., 1984), beta interferon (Mark et al., 1984; 1985), G-CSF (Lu et al., 1989) and basic fibroblast growth factor (bFGF, Thompson, 1992). IL-2, Granulocyte Colony-Stimulating Factor (G-CSF) and beta interferon (IFN-β) contain an odd number of cysteine residues, whereas basic fibroblast growth factor contains an even number of cysteine residues.

Expression of recombinant proteins containing free cysteine residues has been problematic due to reactivity of the free sulfhydryl at physiological conditions. Several recombinant proteins containing free cysteines have been expressed cytoplasmically, i.e., as intracellular proteins, in bacteria such as E. coli. Examples include natural proteins such as IL-2, beta interferon, G-CSF, and engineered cysteine muteins of IL-2 (Goodson and Katre, 1990), IL-3 (Shaw et al., 1992), Tumor Necrosis Factor Binding Protein (Tuma et al., 1995), Insulin-like Growth Factor-I (IGF-I, Cox and McDermott, 1994), Insulin-like Growth Factor binding protein-1 (IGFBP-1, Van Den Berg et al., 1997) and protease nexin and related proteins (Braxton, 1998). All of these proteins were predominantly insoluble when expressed intracellularly in E. coli. The insoluble proteins were largely inactive and needed to be refolded in order to regain significant biological activity. In some cases the reducing agent dithiothreitol (DTT) was used to aid solubilization and/or refolding of the insoluble proteins. Purified, refolded IL-2, G-CSF and beta interferon proteins are unstable and lose activity at physiological pH, apparently due to disulfide rearrangements involving the free cysteine residue (Wang et al., 1984; Mark et al., 1984; 1985; Oh-eda et al., 1990; Arakawa et al., 1992). Replacement of the free cysteine residue in these proteins with serine, resulted in a protein that was more stable at physiological pH (Wang et al., 1984; Mark et al., 1984; 1985; Arakawa et al., 1993).

A second known method for expressing recombinant proteins in bacteria is to secrete them into the periplasmic space or into the media. It is known that certain recombinant proteins such as GH are expressed in a soluble active form when they are secreted into the E. coli periplasm, whereas they are insoluble when expressed intracellularly in E. coli. Secretion is achieved by fusing DNA sequences encoding GH or other proteins of interest to DNA sequences encoding bacterial signal sequences such as those derived from the stII (Fujimoto et al., 1988) and ompA proteins (Ghrayeb et al., 1984). Secretion of recombinant proteins in bacteria is desirable because the natural N-terminus of the recombinant protein can be maintained. Intracellular expression of recombinant proteins requires that an N-terminal methionine be present at the amino-terminus of the recombinant protein. Methionine is not normally present at the amino-terminus of the mature forms of many human proteins. For example, the amino-terminal amino acid of the mature form of human GH is phenylalanine. An amino-terminal methionine must be added to the amino-terminus of a recombinant protein, if a methionine is not present at this position, in order for the protein to be expressed efficiently in bacteria. Typically addition of the amino-terminal methionine is accomplished by adding an ATG methionine codon preceding the DNA sequence encoding the recombinant protein. The added N-terminal methionine often is not removed from the recombinant protein, particularly if the recombinant protein is insoluble. Such is the case with hGH, where the N-terminal methionine is not removed when the protein is expressed intracellularly in E. coli. The added N-terminal methionine creates a “non-natural” protein that potentially can stimulate an immune response in a human. In contrast, there is no added methionine on hGH that is secreted into the periplasmic space using stII (Chang et al., 1987) or ompA (Cheah et al., 1994) signal sequences; the recombinant protein begins with the native amino-terminal amino acid phenylalanine. The native hGH protein sequence is maintained because bacterial enzymes cleave the stII-hGH protein (or ompA-hGH protein) between the stII (or ompA) signal sequence and the start of the mature hGH protein.

hGH has four cysteines that form two disulfides. hGH can be secreted into the E. coli periplasm using stII or ompA signal sequences. The secreted protein is soluble and biologically active (Hsiung et al., 1986). The predominant secreted form of hGH is a monomer with an apparent molecular weight by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) of 22 kDa. Recombinant hGH can be isolated from the periplasmic space by using an osmotic shock procedure (Koshland and Botstein, 1980), which preferentially releases periplasmic, but not intracellular, proteins into the osmotic shock buffer. The released hGH protein is then purified by column chromatography (Hsiung et al., 1986). A large number of GH mutants have been secreted into the E. coli periplasm. The secreted mutant proteins were soluble and could be purified using procedures similar to those used to purify wild type GH (Cunningham and Wells, 1989; Fuh et al., 1992). Unexpectedly, when similar procedures were used to secrete GH variants containing a free cysteine residue (five cysteines; 2N+1), it was discovered that certain recombinant GH variants were insoluble or formed multimers or aggregates when isolated using standard osmotic shock and purification procedures developed for GH. Very little of the monomeric GH variant proteins could be detected by non-reduced SDS-PAGE in the osmotic shock lysates. Insoluble or aggregated GH variants have reduced biological activities compared to soluble, properly folded hGH. Methods for refolding insoluble, secreted Growth Hormone variants containing a free cysteine residue into a biologically active form have not been described.

Alpha interferon (IFN-α2) also contains four cysteine residues that form two disulfide bonds. IFN-α2 can be secreted into the E. coli periplasm using the stII signal sequence (Voss et al., 1994). A portion of the secreted protein is soluble and biologically active (Voss et al., 1994). Secreted, soluble recombinant IFN-α2 can be purified by column chromatography (Voss et al., 1994). When similar procedures were attempted to secrete IFN-α2 variants containing a free cysteine residue (five cysteines; 2N+1), it was discovered that certain of the recombinant IFN-α2 variants were predominantly insoluble or formed multimers or aggregates when isolated using standard purification procedures developed for IFN-α2. Insoluble or aggregated IFN-α2 variants have reduced biological activities compared to soluble, properly folded IFN-α2. Methods for refolding insoluble, secreted IFN-α2 variants containing a free cysteine residue into a biologically active form have not been described.

Human Granulocyte Colony-Stimulating Factor (G-CSF) contains five cysteine residues that form two disulfide bonds. The cysteine residue at position 17 in the mature protein sequence is free. Perez-Perez et al. (1995) reported that G-CSF could be secreted into the E. coli periplasm using a variant form of the ompA signal sequence. However, very little of the ompA-G-CSF fusion protein was correctly processed to yield mature G-CSF. The percentage of correctly processed G-CSF could be improved by co-expressing the E. coli dnaK and dnaJ proteins in the host cells expressing the ompA-G-CSF fusion protein (Perez-Perez et al., 1995). Correctly processed, secreted G-CSF was largely insoluble in all E. coli strains examined (Perez-Perez et al., 1995). Insoluble G-CSF possesses reduced biological activity compared to soluble, properly folded G-CSF. When similar procedures were attempted to secrete wild type G-CSF, G-CSF variants in which the free cysteine residue was replaced with serine [G-CSF (C17S)], and G-CSF (C17S) variants containing a free cysteine residue (five cysteines; 2N+1) using the stII signal sequence, it was discovered that the recombinant G-CSF proteins also were predominantly insoluble. Methods for refolding insoluble, secreted G-CSF proteins into a biologically active form have not been described.

Human Granulocyte Macrophage Colony-Stimulating Factor (GM-CSF) contains four cysteine residues that form two disulfide bonds. Libbey et al. (1987) and Greenberg et al. (1988) reported that GM-CSF could be secreted into the E. coli periplasm using the ompA signal sequence. Correctly processed, secreted GM-CSF was insoluble (Libbey et al., 1987; Greenberg et al., 1988). Insoluble GM-CSF possesses reduced biological activity compared to soluble, properly folded GM-CSF. When similar procedures were attempted to secrete GM-CSF variants containing a free cysteine residue (five cysteines; 2N+1) using the stII signal sequence, it was discovered that the recombinant GM-CSF proteins also were predominantly insoluble. Methods for refolding insoluble, secreted GM-CSF proteins into a biologically active form have not been described.

U.S. Pat. No. 5,206,344 and Goodson and Katre (1990) describe expression and purification of a cysteine substitution mutein of IL-2. The IL-2 cysteine mutein was insoluble when expressed intracellularly in E. coli. The protein was solubilized by treatment with a denaturing agent [either 10% sodium dodecyl sulfate (SDS) or 8M urea] and a reducing agent [100 mM dithiothreitol (DTT)], refolded and purified by size-exclusion chromatography and reversed phase HPLC. Expression and purification of cysteine muteins of IL-3 are described in U.S. Pat. No. 5,166,322. The IL-3 cysteine muteins also were insoluble when expressed intracellularly in E. coli. The proteins were solubilized with a denaturing agent (guanidine) and a reducing agent (DTT), refolded and purified by reversed phase HPLC. The purified IL-3 cysteine muteins were kept in a partially reduced state by inclusion of DTT in the storage buffers. When the inventors used only a denaturing agent agent and a reducing agent (DTT) to denature and refold insoluble cysteine muteins of GH and G-CSF, it was discovered that the refolded proteins were heterogeneous, comprising multiple molecular weight species. Similarly, when the inventors denatured and refolded insoluble, secreted IFN-α2 cysteine muteins with only a denaturing agent and a reducing agent (DTT), undetectable levels of properly folded IFN-α2 cysteine muteins were obtained.