PROTEIN

EXPRESSION

3, 290-294 (19%)

AND PURIFICATION

High-Level Expression and Production of Recombinant Human Interleukin-6 Analogs Shlomo

Dagan,

Department

Received

Charles

of Molecular

February

11, 1992,

Tackney,l

Biology,

and Susan M. Skelly

ImClone

and in revised

form

Systems Incorporated,

May

8, 1992

We have constructed and analyzed different mutant forms of interleukin-6 (IL-6) expressed in Escherichia coli that can be divided into two groups. The first group contains four full-length IL-6 molecules that differ in the presence of cysteine residues involved in disulfide bridges. The second group contains 22 N-terminal amino acid deletions in addition to the differences in the cysteine residues. The different IL-6 muteins were extracted and their expression levels and solubility were compared. We found that the production levels of IL-6 can be dramatically improved by deleting the first 22 N-terminal amino acids of the molecule. We have also found that the production of IL-6 containing the four cysteine residues is lower than the production of the mutant molecules that lack one or both pairs of cysteines. The yield of soluble and properly refolded IL-6 was the highest when the disulfide bond between the cysteines at positions 74 and 84 was present in the mutein form, which also lacked the 22 N-terminal amino acids. 0 1992 Academic Press, Inc.

Interleukin-6 (IL-6) is a multipotential cytokine that has been shown to mediate a wide variety of biological activities. It is thought to play an important role in modulating immune and acute-phase responses to infection or injury, as well as to play a role in hematopoiesis (1,2). The mature full-length protein contains 185 amino acid residues starting with alanine at position 1 and contains 4 cysteine residues at position 45,51,74, and 84. IL-6 is also produced as an 184-amino acid protein in certain cell types starting with proline instead of alanine as the first amino acid (3). The IL-6 molecule has been expressed in Escherichia coli both as a fusion protein, linked to the carrier 1 To whom correspondence should tems Inc., 180 Varick St., New York, 290

be addressed NY 10014.

New York, New York 10014

at ImClone

Sys-

through a factor Xa cleavage site (4), and as a nonfusion protein (5-7). Structure-function analysis of IL-6 revealed that deletions of up to 28 N-terminal amino acids do not adversely affect its biological activity (8), while deletions of up to 5 carboxy-terminal amino acids reduced the activity by a thousandfold (9). It has been suggested that the secondary structure of IL-6 consists of two disulfide bridges between Cys 45-51 and Cys 7484 (10). In order to analyze the effects of modifications to the IL-6 protein with respect to its biological activity, we have produced a series of cysteine replacement mutants. In addition, both wild-type and mutein forms of the protein in which 22 N-terminal amino acids were deleted were produced. In this study, we show that both specific alterations of disulfide bridges and removal of amino acids from the amino terminus have dramatic effects on production levels as well as recovery of soluble protein from the bacterial cell extract. We found that the replacement of the Cys 45-51 pair with serine residues in a 22-N-terminal amino acid-deleted background yields a biologically active protein that is produced in greater quantities than the full-length wild-type IL-6. The ability to produce a specific recombinant mutant that allows the isolation of large quantities of highly active IL-6 mutein is encouraging. A clearer understanding of the physical organization of the IL-6 molecule, as well as its interaction with its receptor, will ultimately make it possible to design a unique biomolecule of high specific activity. MATERIALS

AND

METHODS

Construction of mutant IL-6 molecules. The genes coding for cysteine-free IL-6 (four cysteines replaced by serines at positions 45,51,74, and 84) and the wild-type IL-6 were obtained from Dana Fowlkes (University of North Carolina at Chapel Hill). These genes were originally assembled from synthetic oligonucleotides and 1046~5928192 $5.00 Copyright 0 1992 by Academic Press, Inc. All rights of reproduction in any form reserved.

EXPRESSION

AND PRODUCTION

subcloned into an appropriate expression vector (11,12). A 0.43-kb CZaIlZ3gZII IL-6 fragment representing a 142-amino acid fragment, starting with Ile (aa 26) and ending with Leu (aa 168), was taken from these clones and adapted at the carboxy end with an oligonucleotide to restore the natural carboxy sequence of IL-6. Oligonucleotides were made for the amino-terminal end to give either a full-length native IL-6 starting with AlaPro-Val, etc., or to give a truncated version of IL-6 starting at the twenty-third amino acid with Ser-GluArg, etc. Both the short- and the full-length forms of these genes were inserted into the pKK-233-2 expression vector (Pharmacia LKB Biotechnology Inc., Piscataway, NJ), which contains the P-trc IPTG (isopropyl-fiD-thiogalactopyranoside)-inducible promoter. In both cases, translation of the bacterial transcript places a methionine residue at amino acid position 1. The N-terminal methionine is removed by a methionine aminopeptidase in the cell. Starting with the cysteine-free IL6 molecule, the first two serine residues at positions 45 and 51 were replaced with cysteines (referred as Cys 1,2 IL-6) or the two serines at positions 74 and 84 were replaced by cysteines (referred as Cys 3,4 IL-6). The replacements were accomplished by site-directed oligonucleotide mutagenesis using the “Altered Sites” in uitro mutagenesis system (Promega Corp., Madison, WI). The DNA from these new clones were sequenced by the chain-termination method of Sanger (13) to confirm the codon alteration and then transferred back to the expression vector pKK233-2 for protein production in E. coli. Fifty-milliliter cultures of Induction and extraction. each IL-6 clone were grown at 37°C with shaking in M9 media supplemented with casamino acids and thiamine. At an absorbance of 0.4 at 600 nm, the cultures were induced with 1 mM IPTG for 4 h. The cells were harvested at 3000 rpm and the pellet was resuspended in 1 ml of 50 mM Tris-HCl, pH 8.0, containing 1 mM PMSF (phenylmethylsulfonyl fluoride). Lysozyme was added to the cells at a final concentration of 75 pg/ml and then left on ice for 15 min. Pancreatic DNAse and MgCl, were then added at a final concentration of 75 pg/ml and 65 mM, respectively, and incubation was continued at 37°C for another 45 min. The total cell extract was brought up to 6 M urea by adding solid urea and then briefly sonicated. The mixture was then centrifuged at 13,000 rpm and the supernatant containing the solubilized proteins was collected. To facilitate refolding, the 6 M urea-solubilized protein solutions were immediately diluted to 1 M urea with phosphate-buffered saline (PBS) containing 1 mM DTT (dithiothreitol) and then dialyzed against PBS/DTT at room temperature overnight with three changes of dialysis buffer (1 liter each). The protein solution was clari-

OF INTERLEUKIN-6

291

ANALOGS FULLLENGTHMOlEC”LE6

A

6gl II

c!d I

166

I

185 1CYSI +llL-6

~-00-0-0

I

1 cyst-) IL-6

,-..--.4-0

I

1cys1,2lL-6

I

1cyst-1 IL-6

45 51 .-.-.

‘,2q.

74

64

~LOO----0-O ~+-..--OO-G*

1 cys3,4lL-6 I

~p------*G-...--0-O . cys

(

cysl,2lL-6

0 ser

B ~P”PPGEDS’~D”AAPHR(1P2~TSSERIDK(13lo~~lLD~lS*4L0~ KGT~NKsN~~EssKEALn6~NNLNLPKMA~KD~~F~S~F6~ EET~L”KIlqrOGLL,F,“,L’~,LoN~FESS’:OEOARA”aM~:OK 130 140 VLIQFLQKKAKNLDAITTPDPTTNASLLTKLQAONQWLQD 170 190 MTTHLILRSFKEFLCISSLRALRQM

160

160

FIG. 1. Schematic representation of the different IL-6 constructs (A). The first group contains four full-length IL-6 molecules that differ in the presence of specific Cys residues. The second group contains truncated forms of IL-6 which lack the first 22 N-terminal amino acids. The first molecule in each group contains the four natural Cys residues at positions 45, 51, 74, and 84 (Cys(+) IL-6). In the second, the four cysteines were replaced by four serines (Cys(-) IL6). The third molecule has two cysteines at positions 74 and 84 (Cys 3,4 IL-6), while the fourth contains cysteines at positions 45 and 51 (Cys 1,2 IL-6). (B) Amino acid sequence of the native full-length IL-6. The cys residues are underlined.

fied by centrifugation at 10,000 rpm, and protein concentrations in the supernatants were determined by a Bio-Rad protein assay (Bio-Rad, Richmond, CA). The total cell extracts and reIL-6 determination. folded protein were assayed by ELISA using the Quantikine IL-6 assay kit (R&D Systems, Minneapolis, MN). Briefly, a monoclonal antibody specific for IL-6, coated onto a microtiter plate, had been used to capture any IL-6 contained in the applied sample. An equal amount of total protein from each sample of IL-6 construct was applied to each well. After washing, an enzyme-linked polyclonal antibody specific for IL-6 was added to allow detection of any bound IL-6. Optical density values of samples were recorded (using a microtiter plate reader from Hewlett-Packard) and compared to those of an IL-6 standard curve (10-2000 pg/ml). RESULTS

Figure 1A describes schematically eight different IL-6 constructs that can be divided into two groups. The first

292

DAGAN, TABLE The

Effect of 22 N-Terminal on the Relative Levels

IL-6

construct

Cys(+) Cys(-) Cys 3,4 Cys 1,2

IL-6 IL-6 IL-6 IL-6

TACKNEY,

AND

SKELLY

1 Amino of IL-6

TABLE Acid Deletions Production

Ratio of yields truncated:full-length

The

Effect

Relative levels of full-length IL-6 mutants

of IL-6

2.5 18 11 47

Note. Proteins were extracted from cell pellets, solubilized in 6 M urea, and tested for IL-6 production using ELISA. Equal amounts of total cell protein from each sample of the IL-6 constructs were added to the wells. Values are expressed as relative amounts of the truncated form as compared to its full-length counterpart.

group contains four full-length IL-6 molecules that differ in the presence of specific Cys residues. The second group contains the same combination of IL-6 molecules as the first group but also lacks the first 22 N-terminal amino acids. The first molecule in each group contains all four natural Cys residues (Cys(+) IL-6). In the second, the four Cys residues were replaced by four serines (Cys(-) IL-6). The third contains Cys residues at positions 74 and 84 (Cys 3,4 IL-6), while the fourth contains cysteines at positions 45 and 51 (Cys 1,2 IL-6). The amino acid sequence of the native IL-6 is shown in Fig. 1B. These IL-6 constructs were inserted into the pKK-233-2 expression vector and transfected into the HBlOl strain of E. cob. After induction, proteins were extracted from the cell pellets, solubilized, and then tested for IL-6 production by ELISA. We have found, as shown in Table 1, that the deletion of 22 amino acids from the wild-type IL-6 as well as from any mutein form in which cysteine residues were replaced by serines resulted in production levels relatively higher than those in their respective full-length molecules (2- to 50-fold). In comparing the relative amounts of the full-length IL6 mutein forms to the native molecule, it is shown in Table 2 that higher levels of the IL-6 mutants are produced as measured by ELISA (8- to 35-fold). However, when comparing levels of the truncated muteins to levels of the truncated cysteine-containing molecule, we observed a relative level of production much higher (loo- to 160-fold) than that of the truncated Cys(+) IL6. Considering the data presented in Table 1, in which the short form of IL-6 is produced 2.5-fold more than the native full-length protein, the truncated muteins are produced 250- to 400-fold more than the wild-type fulllength IL-6. These values are based on the relative total amounts of IL-6 extracted and solubilized from three different experiments. It should be noted that the levels of expression for both the wild-type and the 22-amino

2

of Cysteine Residues on the Levels of IL-6 Production

Cys(+) Cys(-) Cys 3,4 Cys 1,2

IL-6 IL-6 IL-6 IL-6

Relative

Relative levels of 22-N-terminal-aa IL-6 mutants

1 8 34 14

1 160 147 109

Note. Proteins were extracted from cell pellets, solubilized in 6 M urea, and tested for IL-6 production using ELISA. Levels of Cys-containing IL-6 (full-length and truncated forms) were each assigned a value of 1. Levels of IL-6 mutein forms were expressed in reference to the respective Cys(+) IL-6 in each group.

acid-deleted Cys-containing IL-6 constructs are 9.6 and 24.5 pg/mg total protein, respectively. It is apparent from these results that the yield of IL-6 is affected by the 22-amino acid deletion as well as by the number of cysteine residues in the molecule. It should be noted that after the cell proteins of all eight IL-6 constructs are solubilized in 6 M urea, IL-6 could not be detected in the residual pellet from each. This indicates that successful solubilization of the recombinant protein occurred. When the eight different IL-6 molecules were allowed to refold upon dilution and dialysis, we found that cysteine residues also affect the recovery of the refolded IL-6 from its denatured form. Table 3 shows the recovery of refolded IL-6 as compared to the levels extracted in 6 M urea. It can be seen that both full-length Cys(+) IL-6 and Cys 3,4 IL-6 molecules can be refolded to the same extent (35 and 44%, respectively), whereas only 13 to 15% of both the Cys(-) IL-6 and the Cys 1,2 IL-6 can

TABLE Recovery from

of Soluble and Its Denatured Yields

of soluble

Full-length Cys(+) Cys(-) Cys3,4 Cysl,B

IL-6 IL-6 IL-6 IL-6

Note. Six molar urea with PBS and tested by ELISA. amount measured

3

IL-6 35 15 44 13

Refolded IL-6 Form IL-6

after

refolding

(%)

22.N-terminal-aadeleted IL-6 19 5 26 6

urea cell extracts were immediately diluted then dialyzed. Levels of IL-6 in the solutions The yield of refolded IL-6 was compared in the original 6 M urea extract.

to 1 M were to the

EXPRESSION

AND

PRODUCTION

be refolded properly. The truncated versions of these molecules follow the same pattern. The Cys(+) IL-6 and the Cys 3,4 IL-6 molecules were refolded to a level of 19-26% while the Cys 1,2 IL-6 or the Cys(-) IL-6 showed only 5% recovery. These values, together with the fact that truncated Cys 3,4 IL-6 produces up to 400fold more IL-6 than the wild type, make this molecule preferable for large-scale IL-6 production. These results demonstrate the importance of the disulfide bridge between the two cysteines at positions 74 and 84 in both protein folding and solubility.

DISCUSSION The results presented in this study show that production levels of recombinant IL-6 can be improved by deleting 22 N-terminal amino acids. Moreover, our study demonstrates that the production of IL-6 containing four natural cysteine residues is lower than the production of mutant molecules that lack one or both pairs of cysteines. We also demonstrated that in order to recover higher yields of solubilized, properly folded IL-6 from the cell extract, one disulfide bond between Cys 74 and Cys 84 must be present. Earlier studies by Brakenhoff et al. have demonstrated that the 28 N-terminal amino acids of IL-6 are dispensable in terms of biological activity (8). It was suggested that expression of such mutant structures of IL-6 is considerably higher than that of wild type because deletion mutants change the hydrophobicity of the molecule and consequently render it more favorable for isolation (7). It is also possible that low levels of expression of certain IL-6 constructs can be explained by mRNA secondary structure constraints which prevent efficient translation (14-16). A more favorable mRNA secondary structure in the deleted forms of IL-6 may contribute to the higher levels of truncated IL-6 protein production. However, to obtain the most soluble and correctly folded IL-6 molecule, one disulfide bond (Cys 74-84) must be present. Whether this is due to secondary structure requirements or the necessity of a specific redox group is at present unclear. IL-6 and human granulocyte colony stimulating factor (G-CSF) are 30% homologous in the N-terminal region and they both contain cysteine residues in similar topological positions within each molecule except that G-CSF contains five cysteine residues instead of four in IL-6 (10). The disulfide bonds of G-CSF are required to maintain biological activity (17). The relationship of the integrity of the disulfide bonds in IL-6 to its multiple biological actions remains unclear. It was shown by Snouwaert et al. that Cys-free IL-6 had a greatly reduced biological activity based on in vitro human and

OF

INTERLEUKIN-6

ANALOGS

293

mouse cell assays (12). Snouwaert et al. (18) demonstrated that a full-length Cys 3,4 IL-6 mutein was as active as the wild-type IL-6 but the activity in a fulllength Cys 1,2 IL-6 form was significantly reduced. As seen from our results, one can potentially improve yields of recombinant proteins by altering cysteine residues as well as deleting amino acids from the molecule. The high yield obtained after refolding the IL-6 Cys 3,4 IL-6 mutein suggests that this disulfide bond is essential for the molecule to properly fold in a soluble state. We have also determined that the deleted Cys 3,4 IL-6 molecule is at least as active as the wild-type IL-6 (manuscript in preparation). As reported recently by Mitraki et al. (19), amino acid substitutions, other than cysteine residues, within the bacteriophage protein P22, altered the protein-folding pathway of this protein. This result demonstrates the importance of alternative amino acids within a protein that might affect proper protein folding. Clearly, empirical manipulation and re-engineering of recombinant proteins make it feasible to improve yields of active structures and perhaps alter specificity and biological range. REFERENCES 1. Revel, M., Zilberstein, A., Chen, L., Gothelf, Y., Barash, I., Novick, D., Rubinstein, M., and Michalevicz, R. (1989) Biological activities of recombinant human IFN-B2/IL-6 (E. coli) in “Regulation of the Acute Phase and Immune Responses: Interleukin-6” (Sehgal, P. B., Grieninger, G., and Tosato, G., Eds.), Vol. 557, pp. 144-155, Ann. NY Acad. Sci., New York. 2. Revel, M. (1989) Host defense against infections and inflammations: Role of the multifunctional IL-G/IFN-B2 cytokine. Experientia 45,549-557. 3. Van Damme, J. (1989) Biochemical and biological properties of human HPGF/IL-6, in “Regulation of the Acute Phase and Immune Responses: Interleukin-6” (Sehgal, P. B., Grieninger, G., and Tosato, G., Eds.), Vol. 557, pp. 104-112, Ann. NY Acad. Sci., New York. 4. Asagoe, Y., Yasukawa, K., Saito, T., Maruo, N., Miyata, K., Kono, T., Miyake, T., Kato, T., Kadidani, H., and Mitani, M. (1988) Human B-cell stimulatory factor-2 expressed in Escherichia coli. BiolTechnology 6, 806-809. 5. Weissenbach, J., Chernajovsky, Y., Zeevi, M., Shulman, L., Soreq, H., Nir, U., Wallach, D., Perricaudet, M., Tiollais, P., and Revel, M. (1980) Two interferon mRNAs in human fibroblasts: In vitro translation and Escherichiu coli cloning studies. PFOC. N&Z. Acad. Sci. USA 77, 7152-7156. 6. Yaseuda, H., Nagase, K., Hosoda, A., Akiyama, Y., and Yamada, K. (1990) High-level direct expression of semi-synthetic human interleukin-6 in Escherichia coli and production of N-terminus met-free product. BiolTechnology 8, 1036-1040. 7. Brakenhoff, J. P., De Groot, E. R., Evers, R. F., Pannekoek, H., and Aarden, L. A. (1987) Molecular cloning and expression of hybridoma growth factor in Escherichia coli. J. Immunol. 139, 4116-4121. 8. Brakenhoff, J. P., Hart, M., and Aarden, L. A. (1989) Analysis of human IL-6 mutants expressed in Escherichia coli. J. Immunol. 143,1175-1182.

294

DAGAN,

9. Brakenhoff, J. P., Hart, M., De Groot, Aarden, L. A. (1990) Structure-function J Immunol. 145,561~568. 10.

TACKNEY,

SKELLY

E. R., DiPadova, F., and analysis of human IL-6.

The influence of messenger RNA secondary structure sion of an immunoglobulin heavy chain in Escherichia cleic Acids Res. 12, 3937-3950.

on exprescob. Nu-

Clogston, C. L., Boone, T. C., Crandall, C., Mendiaz, E. A., and Lu, H. S. (1989) Disulfide structures of human interleukin-6 are similar to those of human granulocyte colony stimulating factor. Arch. Biochem. Biophys. 272, 144-151.

15. Hall, M. N., Gabay, J., Debarbouille, M., and Schwartz, A role for mRNA secondary structure in the control tion initiation. Nature 295, 616-618.

M. (1982) of transla-

11. Jambou, R. C., Snouwaert, J. N., Bishop, G. A., Stebbins, J. R., Frelinger, J. A., and Fowlkes, D. M. (1988) High-level expression of a bioengineered, cysteine-free hepatocyte-stimulating factor (interleukin-6)-like protein. Proc. Natl. Acad. Sci. USA 86,94269430. 12.

AND

Snouwaert, J. N., Kariya, of site-specific mutations human IL-6. J. Immunol.

13. Sanger, ing with

K., and Fowlkes, D. M. (1991) Effects on biologic activities of recombinant

146,585-591.

F., Nicklen, S., and Coulsen, chain-terminating inhibitors.

A. R. (1977) Proc. Natl.

DNA Acad.

sequencSci. USA

74,5463-5467. 14. Wood,

C. R., Boss,

M. A., Patel,

T. P., and

Emtage,

J. S. (1984)

16. Tessier, L., Sondermeyer, P., Faure, T., Dreyer, D., Benavente, A., Villeval, D., Courtney, M., and Lecocq, J. (1984) The influence of mRNA primary and secondary structure on human IFN-7 gene expression in E. cob. Nucleic Acids Res. 12, 7663-7675. 17. Lu, H. S., Boone, T. C., Souza, L. M., and Lai, P. H. (1989) Disulfide and secondary structures of recombinant human granulocyte colony stimulating factor. Arch. Biochem. Biophys. 268,81-92. 18.

Snouwaert, of disulfide Biol. Chem.

J. N., Leebeek, F. W., and Fowlkes, bonds in biologic activity of human 266,23097-23102.

D. M. (1991) interleukin-6.

Role J.

19. Mitraki, A., Fane, B., Haase-Pettingell, C., Sturtevant, J., and King, J. (1991) Global suppression of protein folding defects and inclusion body formation. Science 253, 54-58.