NO863976L - CYSTEIN-PROCESSED MUTEINS OF BIOLOGICALLY ACTIVE HUMAN TUMOR CANCER FACTOR PROTEINS. - Google Patents

Description

The present invention generally relates to the area of recombinant DNA technology. More particularly, it relates to mutationally altered biologically active tumor necrosis factor proteins that differ from the original analogues by one or more substituents or omissions of cysteine residues. The present application corresponds to EP 109,748, published May 30, 1984.,

Human tumor necrosis factor (hTNF) has been known for many years as an antitumor substance found in sera from bacillus Calmette-Guerin (BCG)-infected mice treated with endotoxin. TNF was first reported by Carswell et al., Proe. Nat'l. Acad. Pollock. (US) (1975) 72:3666; and has been shown to be cytotoxic selective against neoplastic cells. TNF has been purified from cell culture by Matthews et al., Brit. J. Cancer (1981) 44:418 (from mononuclear phagocytes derived from BCG-injected rabbits) and by Mannel et al., Infect. Immune. (1980) 30^523; ibid (1981) 33^156 from cultures of macrophage-enriched peritoneal exudate cells from BCG-infected mice. It was disclosed in EP 131,789 of January 23, 1985 in the name of Old et al. that hTNF could be purified from TNF-sensitive L-M cells derived from cloned mouse cells deposited at ATCC under accession number L929.

In recent times, it has been reported that the hTNF cDNA and the hTNF gene have been cloned and that recombinant hTNF has been produced using them. Reference should be made to D. Pennica et al., (1984) Nature (London) 312, 724-729;

Shirai et al., (1985) Nature (London), 313, 803-806, A.M. Wang et al.,

(1985) Science 228, 149-154 and M. Yamada et al., (1985), J. Biotechnology, 3_, 141-153. Corresponding descriptions can be found in the patent literature, e.g. EP-PS 155,549 of September 29, 1985, EP-PS 158,286 of October 16, 1985 and EP-PS 168,214 of January 15, 1986. From these disclosures, hTNF is a protein containing two cysteine residues. Assuming that hTNF has 157 amino acids (see D. Pennica et al. and A.M. Wang et al., supra), the cysteine residues are located at positions 69 and 101. Shirai et al. and M. Yamada et al., supra, report the cloning of a hTNF of 155 amino acids, i.e., where the first two amino acids Val. and Arg. missing. For the purposes of the invention, reference will be made to hTNF with 157 amino acids. As used herein, the term "hTNF" denotes "cysteine-deprived muteins of hTNF" and "hTNF muteins" a protein produced by a microorganism transformed with a human TNF DNA sequence or a modification of the human TNF DNA sequence which encodes a protein with: (a) an amino acid sequence which is at least in the substantially identical to the amino acid sequence of mature native human tumor necrosis factor and (b) has the biological activity common to native human TNF. Substantial identity of the amino acid sequences means that the sequences are identical or differ from each other by one or more amino acid changes (deletions, additions or substitutions) that do not cause any adverse functional dissimilarity between the synthetic protein and the native human TNF. Specific examples include the N-terminal deletion of hTNF, where any one or more of the first 1-11 amino acids are omitted and where those in the 4-, 7- and 8-position are most preferred. Also included are hTNF proteins in which the amino acid residues in positions 35-66, 110-133, 150-157 or 67-109 have been replaced with other amino acids or omitted (see EP-PS 168,214).

Numerous biologically active proteins such as human tumor necrosis factor, which are produced microbially via recombinant DNA, rDNA technology, contain cysteine residues which are not essential for activity, but which are free and can form unwanted intermolecular bonds. One such protein is microbially produced human tumor necrosis factor. During the development of hTNF or rDNA techniques, dimers and oligomers of microbially produced hTNF have been observed to form in E. coli extracts containing high concentrations of hTNF. This multimeric formation makes the purification and separation of hTNF very laborious and time-consuming and necessitates several additional steps in the purification and isolation procedure such as reducing the protein during purification.

The present invention is directed to the production by directed mutagenesis techniques of mutationally altered biologically active tumor necrosis factor proteins (such proteins are called "mutines", "Glossary of Genetics and Cytogenetics", 4th ed. p. 381, Springer-Verlag (1976)), which retain the activity to the parent analogs, but lack the ability to form intermolecular bonds or undesired intramolecular disulfide bonds. See also EP-PS 109-748 of May 30, 1984.

Directed mutagenesis techniques are well known and are described by R.F. Lather

and J.P. Lecoq in Genetic Engineering, Academic Press (1983) pp. 31.50. Oligonucleotide-directed mutagenesis is specifically described by M. Smith et

S. Gillam in Genetic Engineering: Principles and Methods, Plenum Press

(1981) 3:1-32.

A feature of the invention is a synthetic mutein of a biologically active hTNF protein, the mutein having at least one of its two cysteine residues removed or replaced with another amino acid.

Another feature of the invention relates to synthetic structural genes with DNA sequences that are specifically engineered ("constructor genes") to encode the above-described synthetic hTNF muteins. Subfeatures of this feature are expression vectors that include the structural constructor genes, host cells or organisms transformed with such vectors, and processes for producing the synthetic mutein by culturing such transformants or their progeny and recovering the mutein from the culture. In the case of hTNF muteins having therapeutic use, therapeutic preparations containing therapeutically effective amounts of the muteins and therapeutic methods are other features of the invention.

A further feature of the invention is a method for producing the above-described synthetic structural gene by oligonucleotide-directed mutagenesis comprising the following steps: (a) hybridization of single-stranded DNA comprising a strand of a structural gene encoding the parent protein with a mutant oligonucleotide primer that is complementary to a region of the strand that includes the codon for the cysteine to be omitted or replaced or the antisense triplets paired with the codon, as the case may be, except for a mismatch with the codon or the antisense triplet, as the case may be, that defines an omission of the codon or a triplet that encodes said second amino acid; (b) extending the primer with DNA polymerase to form a mutational

heteroduplex; and

(c) replicating the mutational heteroduplex.

The mutant oligonucleotide primers used in this process are another feature of the invention.

The invention shall be explained in more detail with reference to the drawings, there

Fig. 1 shows the complete nucleotide sequence of pE4 and the deduced one

amino acid sequence; Fig. 2 shows a restriction map of the pE4 insert; Fig. 3a and 3b show the complete nucleotide sequence of the insert encoding the mature TNF protein in pAW731 and the deduced amino acid sequence, respectively; and

Fig. 4 shows the single 17,000 dalton protein band corresponding to

ser 69 TNF in the extracts of clone pAW711 and clone pAW731.

The invention provides: muteins of biologically active hTNF proteins in which cysteine residues found to be non-essential for biological activity are intentionally omitted or replaced with other amino acids to eliminate sites of intermolecular cross-linking; mutant genes encoding such muteins; and means for making such muteins.

Proteins that can be mutationally changed according to the invention can be identified from available information with regard to the cysteine content of biologically active proteins and the roles played by the cysteine residues with regard to activity and tertiary structure. For proteins for which such information is not available in the literature, the information can be determined by systematically changing each of the cysteine residues of the protein by the procedures described herein, and testing the biological activity of the resulting muteins and their tendency to form undesirable intermolecular disulfide bonds. While this invention is specifically described and exemplified below with respect to muteins of human tumor necrosis factor, it will be clear that the following teachings apply to all other biologically active hTNF proteins that contain functionally non-essential amino acid residues, e.g. as

contains other modifications for the protein.

The cysteine residues are either omitted or replaced with amino acids that do not affect the biological activity of the resulting hTNF mutein. Suitable amino acid residues are chosen from serine, threonine, glycine, alanine, valine. leucine, isoleucine, histidine, tyrosine, phenylalanine, tryptophan and methionine. Preferred in this group are the residues serine, threonine, alanine and valine. Particularly preferred is replacement with a serine or alanine residue.

As indicated above, mature human TNF is a 157 amino acid protein containing two cysteine residues, one in position 69 and the other in position 101. The sequence encoding TNF produced by the human promyelocytic leukemia cell line (HL-60, ATCC #CCL240) has been cloned and expressed in E. coli and is shown to have the sequence indicated in FIG. 1. Reference is also made to A.M. Wang et al., supra.

As will be seen below, none of the cysteine residues in the TNF sequence appear to be involved in the disulfide bonds and both can be replaced or omitted according to the method of the invention to obtain a stable and biologically active mutein. This is surprising because hTNF has only the two cysteine residues that would normally be thought to be necessary for the retention of biological activity.

Using the oligonucleotide-directed mutagenesis procedure with a synthetic oligonucleotide primer complementary to the region of the hTNF gene in the codon for cys 69, but containing single or multiple base changes in this codon, a constructor gene can be produced that results in the replacement of cys 69 with another amino acid of choice. When omission is desired, the oligonucleotide primer lacks the codon for cys 69. Conversion of cys 69 to neutral amino acids such as glycine, valine, alanine, leucine, isoleucine, tyrosine, phenylalanine, histidine, tryptophan, serine, threonine, and methionine is the preferred method. Serine, threonine, and alanine are the most preferred substitutions because of their chemical analogy to cysteine. When cysteine is omitted, the mature mutein is one amino acid shorter than the native parent protein or the microbially produced hTNF.

The size of the oligonucleotide primer is determined by the requirement for stable hybridization of the primer to the region of the gene in which the mutation is to be induced, and by the limitations of the currently available methods for synthesizing oligonucleotides. The factors that must be taken into account when designing oligonucleotides for use in oligonucleotide-directed mutagenesis (ie, overall size, size of the portions surrounding the mutation site) are described by M. Smith and S. Gillam, supra. In general, the total amount of the nucleotide will be such as to optimize stable unique hybridization at the mutation site with 5' and 3' extensions from the mutation site of sufficient size to avoid editing of the mutation due to the exonuclease activity of DNA polymerase. Oligonucleotides used for mutagenesis according to the present invention usually contain from approx. 12 to approx. 24 bases, preferably from approx. 14 to approx. 20 bases, and preferably from approx. 15 to approx. 18 bases. They will usually contain at least approx. three bases 3' of the changed or missing codon.

The method for making the modified IFN-3 gene generally involves inducing site-specific mutagenesis in the hTNF gene at codon 69 (TGT) using a synthetic nucleotide primer that omits the codon or changes it to encode a different amino acid. It should be pointed out that when omission is introduced, the correct reading frame for the DNA sequence must be maintained for expression of the desired protein.

The primer is hybridized to single-stranded phage such as M13, fd or 0X174 into which one strand of the hTNF gene is cloned. It should be clear that the phage can be either the sense strand or the anti-sense strand of the gene. When the phage carries the anti-sense strand, the primer is identical to the region of the sense strand containing the codon to be mutated except for a mismatch with that codon that defines an omission of the codon or a triplet encoding a different amino acid. When the phage carries the sense strand, the primer is complementary to the region of the sense strand containing the codon to be mutated except for a suitable mismatch in the triplet paired with the codon to be omitted. Conditions that can be used in the hybridization are described by M. Smith and S. Gillam, supra. The temperature will usually be within the range of approx. 0-70° C, and preferably approx. 10-50 °C. After the hybridization, the primer is extended on the phage DNA by reaction with DNA polymerase I, T4 DNA polymerase, reverse transcriptase or other suitable DNA polymerases. The resulting dsDNA is converted to closed circular dsDNA by treatment with a DNA ligase such as T4 DNA ligase. DNA molecules containing single-stranded regions can be destroyed by Sl endonuclease treatment.

Oligonucleotide-directed mutagenesis is used to obtain a TNF gene that encodes a mutein with TNF activity, but with cys09 changed to an alternative or an omitted amino acid, and/or where the cysgi residue is replaced or omitted. For the exemplified conversion of cys09 to serQ9, the preferred oligonucleotide primer is 5'-CATGGGTGCTCGGGCTGCCTT-3'. This oligonucleotide has a T ...-+ A change in the triplet which is paired with codon 69 in the TNF gene. Likewise, cysjQj can be converted to serial with a primer CAAGAGCCCCTCTCAGAGGGAG containing a corresponding change in the triplet paired with the codon at 101, using ssM13 phage DNA containing the appropriate strand of the human TNF cDNA sequence.

The resulting mutational heteroduplex is then used to transform a competent host organism or cell. Replications of the hetero-duplex at the host produce progeny from both strands. After replication, the mutant gene can be isolated from the progeny of the mutant strand, inserted into a suitable expression vector and the vector used to transform a suitable host organism or cell. Preferred vectors are plasmids pBR322, pCR1 and variants thereof, synthetic vectors and the like. Suitable host organisms are E. coli, Pseudomonas, Bacillus subtilis, Bacillus thuringiensis, various yeast strains, Bacillus thermophilus, animal cells such as mouse, rat or Chinese hamster ovary (CHO) cells, plant cells, animal and plant hosts and the like. It should be understood that when a selected host is transformed with the vector, appropriate promoter-operator sequences are also introduced for the mutein to be expressed. The hosts can be prokaryotic or eukaryotic (methods for inserting DNA into eukaryotic cells are described in PCT application US81/00239 and US81/00240 of September 3, 1981). E. coli and CHO cells

are the preferred hosts. The muteins obtained according to the invention can be glycosylated or non-glycosylated depending on the glycosylation that is desired in the mutein. If desired, non-glycosylated mutein can be added

obtained when E. coli or a Bacillus is the host strain, optionally glycosylated in vitro by chemical, enzymatic or other modifications of a known type.

The following examples shall illustrate the invention in more detail. They should not be restrictive. Examples 1-9 describe the production of a TNF mutein and its analysis.

Preparation and purification of human TNF:

1. Induction of TNF

High-density (> 2 x 10^ cells/ml) stationary HL-60 cells were centrifuged, washed with RPMI 1640 medium in the absence of serum and then resuspended at a density of 1 x 10? cells/ml. The cells were then treated with 100 ng/ml phorbol ester, 12-0-tetradecanorylphorbol-13-acetate (TPA) for 30 min at 37°C in a suspension culture with constant agitation. The cultures were centrifuged, the supernatant decanted, the cells resuspended with 1x 10 3 cells/ml in RPMI, containing 10 pg/ml bacterial lipopolysaccharide (LPS) and 10 µM Ca ionophore (A23817) for 4 hours at 37°C with constant agitation.The cells were spun down at 1200 rpm. for 10 minutes and the supernatants recentrifuged at 8000 rpm for 20 minutes.The resulting supernatants were used in the purification scheme below to obtain native TNF.

2. Purification of TNF

About 4-8 liters of supernatant prepared from induced HL-60 in the above section was concentrated via Amicon hollow fiber (1 f<2>cartridge; 10,000 molecular weight cut-off) to ca. 300 ml. The concentrated culture fluid was centrifuged to remove cell debris and the supernatant adjusted with 30 mM ammonium bicarbonate buffer (pH 812) to a conductance of 6.2 mS. The solution was further concentrated by ultracentrifugation using a PM10 (Amicon) membrane and the concentrated fluid clarified by centrifugation (20,000 x g for 10 min).

The supernatant was then applied to a DEAE ion exchange column equilibrated in 30 mM ammonium bicarbonate/1 mM NaCl pH 8.2, and the column washed with the same buffer. The fractions were pooled and the protein monitored at 280 nm. These unbound fractions were analyzed using the L-929 cytotoxicity assay and those with TNF activity collected and concentrated again by ultrafiltration.

The concentrate was applied to Sephadex G75 Superfine (Pharmacia) equilibrated in 30 mM ammonium bicarbonate buffer (pH 7.4). Unbound fractions obtained by washing with the same buffer were monitored at 280 nm and analyzed for TNF. Fractions containing peak TNF bioactivity were lyophilized.

The lyophilized protein was resuspended in Laemmli SDS sample buffer and electrophoresed on an SDS-polyacrylamide gel. The gel was cut into 2 mm sections and the protein from each section was eluted by immersion in 1 ml of 30 mM ammonium bicarbonate buffer (pH 7.4) and shaken overnight at room temperature.

Sections containing TNF bioactivity were applied to a Vydac C-4 reverse-phase HPLC column equilibrated in 0.1% trifluoroacetic acid, TFA, and the activity eluted using a linear gradient of 0-60% acetonitrile in 0.1% TFA . The protein was monitored at 280 nm and 214 nm and the fractions bioanalyzed after lyophilization and suspended in 30 mM ammonium bicarbonate buffer, pH 7.4. Fractions containing TNF activity were lyophilized again.

The resulting protein had sufficient purity to be used in sequence analysis. The sequence was determined using a gas phase sequencer (Applied Biosystems Inc.). The sequence obtained from the first 22 amino acids is shown below:

In addition, the purified protein from the G-75 gel was tested with a modification of the L-929 cytotoxicity assay using another human tumor and normal cell lines as substrate. The G-75 fractions that were cytotoxic in this assay against L-929 cells were also cytotoxic against Hs939T (a melanoma line), BT-20 (breast carcinoma), A427 (lung carcinoma), HT-1080 (colon carcinoma), and HT-29 ( colon carcinoma). These fractions were not cytotoxic against Hs939sk (skin fibroblasts), HeLa cells (cervical carcinoma), Hs27F (sheep skin fibroblasts) or COS7 (SV40-transformed monkey cells).

Example 2

Preparation of the coding sequence

An intronless DNA sequence encoding human TNF was prepared by the procedure described here. A human promyelocytic leukemia cell that produces large amounts of TNF upon induction, the HL-60 line, available from ATCC under the number CCL 240, was used as an mRNA source to obtain a cDNA library. Using oligomeric probes constructed on the basis of the protein sequence determined from TNF purified from these cells, this cDNA library was probed to provide the entire coding sequence for the protein.

1. Preparation of enriched mRNA

Total messenger RNA was extracted and purified from HL-60 cells as follows: HL-60 cells were induced for TNF production as indicated in section D.l.a. and the 4-hour cell suspension harvested by centrifugation. Total cytoplasmic ribonucleic acid (RNA) was isolated as follows: All steps occurred at 4°C. The cells were washed twice in PBS (phosphate buffered saline) and resuspended in IHB (140 mM NaCl, 10 mM Tris, 1.5 mM MgCl 2 , pH 8) containing 10 mM vanadyladenosine complex (S.L. Berger et al., Biochem. (1979) 18 :5143).

A nonionic detergent of the ethylene oxide polymer type, NP-40, was added at 0.3% to lyse the cellular but not nuclear membranes. The nuclei were removed by centrifugation at 1000 x g for 10 min. The post-nuclear supernatant was added to an equal volume of TE (10 mM Tris, 1 mM ethylenediaminetetraacetic acid (EDTA), pH 7.5) saturated phenol/chloroform (1:1) containing 0.5% sodium dodecyl sulfate (SDS) and 10 mM EDTA. The supernatant was extracted again four times and phase separated by centrifugation at 2000 x g for 10 min. RNA was precipitated by adjusting the sample to 0.25 M NaCl, adding 2 volumes of 100% ethanol and storing at -20 °C. RNA was pelleted at 5000 x g for 30 min, washed with 70% and 100% ethanol and dried. Polyadenylated (poly A<+>) messenger RNA, mRNA, was obtained from the total cytoplasmic RNA by chromatography on oligo-dT-cellulose (J. Aviv et al., Proe. Nati. Acad. Sei. (1972) 69^ 1408-1412). This RNA was dissolved in ETS (10 mM Tris, 1 mM EDTA, 0.5% SDS, pH 7.5) at a concentration of 2 mg/ml. This solution was heated to 65 °C for 5 minutes, then rapidly cooled to 4 °C. After the RNA solution was brought to room temperature, it was adjusted to 0.4 M NaCl and slowly passed through an oligo-dT-cellulose column as on pre-equilibrated with a binding buffer (500 mM NaCl, 10 mM Tris, 1 mM EDTA, pH 7.5). The flow-through was passed over the column twice more and the column washed with 10 volumes of binding buffer. Poly A<+>mRNA was eluted with aliquots of ETS, extracted once with TE-saturated phenol chloroform and precipitated by addition of NaCl to 0.2 M and 2 volumes of 100% ethanol RNA was precipitated again twice, washed once in 70% and then in 100% ethanol before drying.

This poly A<+>mRNA was fractionated on a sucrose gradient in 10 mM Tris-HCl, pH 7.4, 1 mM EDTA, 10 mM NaCl and 0.1% SDS. After centrifugation in a Beckman SW40 rotor at 38,000 rpm. for 17 h the mRNA fractions were recovered from the gradient by ethanol precipitation. The fractions containing TNF mRNA were identified by injecting the mRNA into oocytes and analyzing the oocyte extracts for cytotoxic activity. Fractions containing peak activity were pooled for use in cDNA library constructions.

2. Construction of a cDNA library

cDNA was prepared from an enriched 16 S mRNA fraction using oligo-dT priming of the poly A tails and AMV reverse transcriptase using the method of H. Okayama et al., Mol. Cell. Biol. (1983) 3^280. This method results in a higher proportion of full-length clones and effectively uses as host vector portions of two vectors which are described there and which can be easily obtained from the authors, pcDV1 and pL1. The resulting vectors contain the insert between vector fragments containing proximal BamHI and XhoI restriction sites; the vector contains the pBR322 origin of replication and the Amp resistance gene.

Other methods for preparing cDNA libraries are of course well known. One, now classic method, uses oligo-dT primer, reverse transcriptase, tailing the double-stranded cDNA with poly dG and joining it to a suitable vector such as pBR322 or a derivative thereof that is cleaved at the desired restriction site and tailed with poly dC. A detailed description of this alternative method can be found e.g. in US-PS 4,578,584.

In the method used here, 5 pg of enriched mRNA was denatured by treatment with 10 mM methylmercury at 22°C for 5 minutes and detoxified by the addition of 100 mM 2-mercaptoethanol (F. Payvar et al., J. Biol. Chem (1979) ) 254:7636-7642). Plasmid pcDV1 was cleaved with KpnI, tailed with dTTP and added to the denatured mRNA. This oligo-dT-primed mRNA was treated with reverse transcriptase and the fully synthesized DNA strand tailed with dCTP. Finally, the unwanted part of the pcDV1 vector was removed by digestion with HindIII. Separately, pL1 was cleaved with Pstl, tailed with dGTP, cleaved with HindIII and then mixed with poly-T-tailed mRNA/cDNA complex extended with the pcDVl vector fragment, ligated with E. coli ligase and the mixture treated with DNA polymerase I ( Klenow) E. coli ligase and RNase H. The resulting vectors are transformed in E. coli K12 MM294 into Amp^.

3. Selection of probe

Oligomers complementary to the coding sequence for amino acids 8-12 of the purified TNF sequence were prepared. Due to codon abundance, there are a total of sixty-four 14-mers as candidates for complementation to the promoter that encodes this part. All sixty-four 14-mers were prepared and divided into four groups of 16. Each group was mixed with the sucrose gradient size-fractionated, enriched mRNA preparation, prepared as above, and the mixture was injected into the oocyte translation system. Controls were run using untranslated reporter RNA. The proteins produced in the oocyte system were subjected to the L-929 cytotoxicity assay (^S release), and the proteins derived from the oocytes injected with control and with a mixture of mRNA with three of the oligomer groups showed activity. In this "hybrid arrest" assay, only the oocyte injected with the reporter was treated with the group with the sequence inactive. The specificity of this oligomeric pool was further determined by dot blot hybridization with enriched mRNA prepared as above from both induced and uninduced HL-60 cells, as well as the corresponding mRNA fraction obtained from cells known to be producers of lymphotoxin . This group hybridized well to the induced mRNA, but did not hybridize to the corresponding fractions from non-induced or lymphotoxin-producing cells. However, Northern blot samples using the kinased group as a probe showed that it contained sequences that cross-hybridized to the 18S (ribosomal) RNA fraction and to pBR322

DNA.

The successful group was therefore further fractionated by synthesis of the participants as eight pairs of 14-mers, each of which was used in the "hybrid-arrest" assay indicated above. Only the pair with the sequence

inhibited the synthesis of TNF in the oocytes. Dot-blot experiments using the fractionated induced HL-60 mRNA fraction induced total HL-60 poly A<+>RNA, uninduced HL-60 poly A<+>RNA and pBR322 DNA confirmed the specificity for the preceding 14- more pairs and the inability of the remaining pairs to hybridize to the desired communicator.

4. Recovery of the coding sequence

The cDNA library was probed with the 14-mer pair as identified above.

28 colonies that hybridized with the probe were picked, cultured and plasmid DNA isolated. The plasmid containing the insert of sufficient length to encode the entire sequence was selected and several were examined for correct sequence using hybrid translation in combination with the ^S release version of the cytotoxicity assay as described below. Hybrid translation assays use the sample sequence to obtain the correct mRNA from unfractionated preparations as verified by analyzing the protein produced by the oocyte translation system injected with the recovered mediator.

Plasmid cDNA to be processed is bound to the filter and the filter treated with poly A<+>RNA isolated from induced HL-60 cells. The filters are then eluted and the eluates are injected into the oocyte translation system. The oocytes are extracted for protein which is then tested in the <35>s_version of the L-929 cytotoxicity assay. The results for several hybridizing clones, named E2-E4, E6 and E8, are shown below:

Restriction analysis and partial sequencing of the inserts suggested that two candidate plasmids, pE4 and pBll, likely possessed the total TNF coding sequence. The results of this analysis for pE4 are shown in fig. 19. pE4 has been deposited with ATCC on 15 October 1984 under the number 39,894.

pE4 was sequenced and the correct reading frame for TNF identified by alignment of the amino acid sequence deduced from translation of all three possible reading frames with the known N-terminal sequence of the native mature TNF as determined by N-terminal sequencing of the purified protein, see fig. 18. The amino acids in the mature protein are numbered from valine in position 1. As indicated above, the homology was not total. However, the high degree of homology suggested that the correct cDNA had been selected. Verification of the experimentally determined restriction cleavage

seats shown in fig. 19 is also displayed. The HindIII site upstream of the 3' Pstl site in the 1.1 kb Pstl fragment is downstream of the stop codon and thus allows excision of the coding sequence as a HindIII cassette after modification of the upstream region as described below. 5. Characteristics of human TNF, determined from the DNA sequence As deduced from the cDNA sequence as indicated in fig. 18, the mature TNF protein contains 157 amino acid residues and has a molecular weight without glycosylation of approx. 17,354. The leader sequence apparently contains roughly 76 amino acids from the first available Met start codon. There are two cysteine residues, at positions 69 and 101.

Example 3

Modification of the N-terminal codons in pE4

It was appropriate, when carrying out the expression of the mature protein, to introduce an ATG start codon immediately before the GTC sequence which encoded the N-terminal valine in the mature protein (indicated by 1

in fig. 18), as well as providing a HindIII site immediately upstream of the ATG for ligation into appropriate host expression vectors. This was carried out by site-directed mutagenesis in a manner analogous to that described in examples 1, 2 and 5 of US-PS 4,578,584.

The DNA fragment containing the upstream part of the coding sequence was excised from pE4 by digestion with Pstl, isolated by agarose gel electrophoresis, removed by electroelution and ligated into the Pstl site of bacteriophage M13mpl8.

The ligated phage was transduced into frozen competent E. coli K12 strain DG 98 (ATCC #39768) and grown by plating on media containing 5 x 10~<4>M isopropylthiogalactoside (IPTG) obtained from Sigma Chem. (St. Louis, Missouri) and 40 pg/ml X-gal. Non-a-complementing white plaques were picked on fresh media. Minicultures were examined for recombinant single-stranded phage DNA containing inserts of the 1.1 kb expected size. The structure of the desired recombinant phage,

named clone 4.1, was confirmed by restriction analysis.

A chemically synthesized, purified, 33-mer oligodeoxyribonucleotide with

the sequence:

5'-GAAGATGATCTGACCATAAGCTTTGCCTGGGCC-3'

to introduce the HindIII restriction enzyme site and an ATG initiation codon before the GTC codon encoding the first amino acid, valine, in the mature TNF protein.

10 picomole oligonucleotide was hybridized to 2.6 Mg ss-clone 4.1 DNA for 15

ul of a mixture consisting of 100 mM NaCl, 20 mM Tris-HCl, pH 7.9,

20 mM MgCP? and 20 mM ε-mercaptoethanol by heating to 67°C for 5 minutes and 42°C for 25 minutes. The combined mixtures were cooled on

ice and then adjusted to a final volume of 25 µl with a reaction mixture consisting of 0.5 mM of each dNTP, 17 mM Tris-HCl, pH 7.9, 17 mM MgCP), 83 mM NaCl, 17 mM β-mercaptoethanol, 5 units of DNA polymerase

In Klenow fragment, incubated at 37°C for one hour. The reactions were terminated by heating to 80 °C and the reaction mixture used to transform competent DG98 cells, plated on agar plates and incubated overnight to obtain phage plaques.

Plates containing mutagenized clone 4.1 plaques as well as two plates containing non-mutagenized clone 4.1 phage plaques were cooled to 4°C and the phage plaques from each plate were transferred to two nitrocellulose filter circles by coating a dry filter on the agar plate for 5 minutes for the first filter and 15 minutes for the second filter. The filters were then placed on thick filter papers soaked in 0.2 N NaOH, 1.5 M NaCl and 0.2% Triton

X-100 for 5 minutes and neutralized by coating onto filter papers soaked with 0.5 M Tris-HCl, pH 7.5 and 1.5 M NaCl for a further 5 minutes. The filters were similarly washed twice on filters soaked in 2 x SSC, dried and then heat treated in a vacuum oven at 80°C for 2 hours. The duplicate filters were pre-hybridized at 42 °C for 4 hours with 10 ml per filter DNA hybridization buffer (5 x SSC, pH 7.0, 4 x Denhardt's solution (polyvinylpyrrolidine, ficoll and bovine serum albumin, lx = 0.02% each), 0.1% SDS, 50 mM sodium phosphate buffer, pH 7.0 and 100 ug/ml denatured salmon sperm DNA. ^p<->meri^e probes were prepared by kinased the primer with labeled ATP. The filters were hybridized to 5

x 10^ cpm/ml with <32p_>labeled primer in 1-5 ml per filter and DNA hybridization buffer at 64 °C for 8 h.

The filters were washed once at room temperature for 10 min in 0.1% SDS, 20 mM sodium phosphate (buffer), and 6 x SSC; once at 37°C for 20 minutes in buffer and 2 x SSC; once at 50°C for 20 minutes in buffer and 2 x SSC and finally at 60°C for 20 minutes in buffer and 1 x SSC. The filters were air dried and autoradiographed at -70°C for 4 hours.

Because the oligonucleotide primer was designed to create a new HindIII restriction site in the mutagenized clones, the RF DNA from a number of the clones that hybridized with the primer was degraded with this restriction enzyme. One of the mutagenized clone 4.1 plaques having a new HindIII restriction site (M13-AW701) was picked and inoculated into a culture of DG98, ssDNA was prepared from the culture supernatant and dsRF-DNA was prepared from the cell pellet. The correct sequence is confirmed by dideoxy sequencing.

The correctly synthesized strands were isolated and cleaved with PstI and HindIII (partially) or with HindIII alone for ligation into expression vectors.

Example 4

Expression of TNF

For prokaryotic expression, the coding sequence (along with some 3' untranslated nucleotides) was excised from dsM13-AW701 in two ways: In the first method, dsM13-AW701 was digested with Pstl and then partially digested with HindIII to obtain HindIIIH- Pstl the TNF coding sequence.

(Partial HindIII digestion is necessary because there are several HindIII sites in M13-AW701.) The partial digestion of the DNA fragment can be achieved by using one-tenth of the amount of restriction enzyme required for total digestion of DNA. The mixture was incubated at the appropriate temperature for the enzyme and aliquots of the degradation mixture were removed at 10 minute intervals up to one hour. The aliquots were then loaded onto a gel and the DNA fragments analyzed. The time point which gave the highest yield for the DNA fragment was chosen as preparative digestion with the restriction enzyme and the suitable fragment was purified from the gel by electroelution.

The Pstl/BamHI fragment containing the 3' non-coding sequence of the TNF gene (see Fig. 2) was purified from pE4 after digestion of DNA with the enzymes Pstl and BamHI.

Together, the HindIII/Pstl and Pstl/BamHI fragments comprised the coding sequence plus a 600 bp 3' untranslated portion of DNA. The two fragments were ligated into HindIII/BamHI digested host vector pTRP3 as follows: pTRP3 (ATCC 39,946) contains the E. coli trp promoter and ribosome binding site. pTRP3 was digested with HindIII and BamHI and the vector fragment purified on agarose gel. The isolated fragment was then ligated with the above HindIII/Pstl and Pstl/BamHI segments in a 3-way ligation and the mixture was used to transform E. coli MM294

to Amp<R>, giving pAW701.

In a second method, dsM13-AW701 was digested with HindIII and the fragment containing the gene isolated on agarose gel. The isolated fragment was ligated with HindIII cleaved BAPped pTRP3 and transformed into E.

coli MM294 to obtain pAW702.

pFC54.t (ATCC 39789) or pPLOP (ATCC 39947) containing the PL promoter and bacillus-positive retroregulatory sequence can also be used as host vectors. These vectors are digested with HindIII and BamHI and the large plasmid fragments containing the control sequences purified on agarose gel. HindIII/Pstl and Pstl/BamHI portions of the TNF gene, prepared as indicated above, are ligated in a three-way ligation into HindIII and BamHI sites in these vectors, resulting in plasmids pAW711 and pAW712, respectively. Alternatively, purified HindIII fragment from mutagenized pE4 is ligated into HindIII cleaved BAPped pFC54.t or pPLOP

to give pAW713 and pAW714, respectively. Plasmid pAW711 has been deposited with ATCC on 8 November 1984 under the number 39,918.

pAW701 and pAW702 were transferred into E. coli MM294 and the cultures grown under conditions repressing the trp promoter. Upon induction with tryptophan depletion, the production of TNF was initiated. In an analogous manner, pAW711 was constructed and transferred into E. coli MC1000-39531 and the cells were

induced at high temperature. After several hours of cultivation under induction conditions, the cells were sonicated and the sonicates verified to contain TNF by the L-929 cytotoxicity assay. The results are:

Units of the TNF activities are as indicated below in Example 9.

Vector pBll isolated from the above cDNA library contains the SV40 promoter in operable linkage to the TNF coding sequence. All of the 28 positively hybridizing colonies would be expected to contain this binding including especially pE4 and pBll, and are thus able to express suitable mammalian hosts. Accordingly, pBll was used to transform COS-7 monkey kidney cells and the cells were cultured under conditions causing induction of the SV40 promoter. As the signal sequence is still present in pBll and works in mammalian cell systems, TNF was secreted into the medium. TNF was assayed in the supernatant of the monocoated COS-7 cells by ^S release from L-929 cells, with the following results:

Example 5

Preparation of coding sequence and expression vectors for TNF muteins

Clone 4.1, prepared by Pstl treatment of pE4 as described in Example 23, was subjected to site-specific mutagenesis essentially as described in Example 23, but, by using as a primer

5'-CATGGGTGCTCGGGCTGCCTT-3' which is complementary to the sequence adjacent to the cysteine at position 69, but contains nucleotides complementary to that codon to effect the change from TGC to AGC. Mutagenicity-

serte plaques were identified and confirmed by sequencing as described above. A plaque containing the desired mutation, MP-AW731, was digested with Aval and Pstl and the fragment ligated into Pstl/Aval digested pAW711. The ligation mixture was transferred into E. coli MC1000-39531 to Amp<R> and the transformants scored with the primer probe for correct sequence. A colony called pAW731 was used to express the modified sequence. pAW731 has been deposited at ATCC on January 25, 1985

and has the access number 53.007.

In an analogous manner, pAW741, an expression vector for serjQj TNF was prepared using the primer CAAGAGCCCCTCTCAGAGGGAG.

Example 6

Expression of the coding sequence and activity of the TNF mutein

E. coli MC1000-39531 harboring pAW731 was grown and induced at high temperature as indicated in Example 24. The sonicates from the induced cells were analyzed and found to have approximately the same TNF activity per ml as the pAW711 transformants. However, SDS analysis showed that the amount of 17 kD TNF protein in these extracts is approx. five times less, which shows

that the specific activity for Ser0o. TNF is higher than that of the native or wild-type recombinant TNF protein.

Example 7 Expression of the Coding Sequence for Recombinant Ser^0, Human TNF Mutein

Plasmid pAW731 (ser^oj) is digested with HindIII and HincII, and the small HindIII-HincII fragment containing the ser09 mutation is purified on agarose gel. In the same way, plasmid pAW732 (serjgi) is digested with the enzymes HincII and BamHI, and HincII-BamHI fragment containing the ser101 mutation is purified. The previously purified HindIII-BanHI vector fragment from pFC54.t is then ligated with the HindIII-HincII (ser^) fragment and the HincII-BamHI (serjoi) fragment to form ser^serjqi TNF clone, pAW735 This dimutein, se^serjoi TNF, is less likely to dimerize upon purification due to the absence of free cysteine.

Example 8

Cysteine residues in native or wild-type human TNF are not required for biological activity

95% pure unmodified protein was reduced and alkylated by treatment with DTT and iodoacetate according to the protocol outlined below. While the untreated protein had an activity in units/ml of 2.6 x 10<4>, reduced or reduced alkylated protein had activities of 4.4-4.8 x 10<4>units/ml.

The data shown above show that cysteines in wild-type recombinant human TNF are not necessary for biological activity and thus one or both cysteines can be omitted or replaced with another amino acid as indicated within the framework of the invention.

Example 9

Cytotoxic assay procedure for TNF

The L-929 assay system is an improved convenient in vitro assay that allows rapid measurement of TNF activity. The degree of correlation with the in vivo tumor necrosis assay according to Carswell above is currently unknown; however, as it uses murine tumor cells specifically, the correlation is believed to be high. The protein called lymphotoxin in EPO publication 0100641 also gives activity in this assay. The assay is similar in concept to that described in US-PS 4,457,916 which used murine L-M cells and methylene blue staining. However, the L-929 assay is shown herein to correlate (for HL-60-derived TNF) with human tumor cell line cytotoxicity.

In the L-929 assay system, L-929 cells are prepared overnight as monolayers in microtiter plates. The test samples are diluted twice across the plate, UV irradiated and then added to the prepared cell monolayers. The culture media in the wells is then brought to 1 µg/ml actinomycin D. The plates are allowed to incubate for 18 hours at 37"C and the plates are scored visually under the microscope. Each well is given a 25, 50, 75 or 100% rating indicating the degree of cell death in the well .One unit of TNF activity is defined as the reciprocal of the dilution at which 50% killing occurs.

In addition, a more sensitive version of this assay was developed that monitors the release of 35S-labeled peptides from pre-labeled cells, when treated with the test sample and actinomycin D. This assay version can be used to quantify the potency, i.e. to judge the relative potency of oocyte-translated material. Briefly, actively adult L-929 cultures are labeled with<35>s-methionine (200 MCi/ml) for 3 hours in methionine-free media supplemented with 2% dialyzed fetal calf serum. The cells are then washed and plated in 96 well plates, incubated overnight and treated the next day with twofold dilutions of the test samples and 1 µg/ml actinomycin D. The cultures were then incubated at 37°C for 18 hours. 100 µl supernatant aliquots from each well were then transformed into another 96 well plate, acid (TCA) precipitated and harvested on glass fiber filters. The filters were washed with 95% ethanol, dried and counted. An NP40 detergent control is included in each assay to measure the maximum release of radioactivity from the cells. Percentage 35S release is then calculated by the ratio between the difference in count between the treated cells and untreated controls divided by the difference between NP4Q-treated cells and untreated cells, i.e. by the ratio:

Higher TNF potency results in higher values for this ratio.

The TNF muteins according to the invention are suitably formulated into suitable therapeutic formulations which typically include a therapeutically effective amount of the mutein and a suitable physiologically acceptable carrier as described above with regard to IL-2 muteins. Alternatively, other cytotoxic, antiviral or anti-cancer agents can be used in combination

with the TNF muteins according to the invention such as -γ-interferon.

On 15 October 1984, the applicant deposited the plasmid pE4 described here with ATCC under the number 39,894. On November 8, 1984, pAW711 was deposited with ATCC under the number 39,918. On January 25, 1985, pAW731 was deposited with ATCC under the number 53.007. These depositions took place under the regulations of the Budapest Agreement. This ensures 30 years of maintenance of viable cultures. The organisms are made available by ATCC under the regulations of the Budapest Agreement and which ensure unlimited availability upon issuance of the relevant US patent. This availability shall not be construed as a license to practice the invention against the rights guaranteed under the regulations of any government pursuant to the patent laws of that country.

It should be clear that variations and modifications can be carried out without going outside the scope of the invention as defined in the accompanying claims.

Claims (44)

1. Synthetic mutein of a biologically active human tumor necrosis factor protein, characterized in that the protein has at least one cysteine residue that is free to form a disulfide bond and is not essential for the biological activity, the mutein having at least one of the aforementioned cysteine residues removed or replaced by a other amino acid. 2. Mutein according to claim 1, characterized in that there is only one of the mentioned cysteine residues. 3. Mutein according to claim 1, characterized in that the cysteine residues are replaced by serine, threonine, glycine, alanine, valine, leucine, isoleucine, histidine, tyrosine, phenylalanine, tryptophan or methionine. 4. Mutein according to claim 1, characterized in that the cysteine residues are replaced by serine, threonine or alanine. 5. Mutein according to claim 1, characterized in that the mutein is non-glycosylated. 6. Mutein according to claim 1, characterized in that the cysteine residue in position 69 has been replaced. 7. Mutein according to claim 1, characterized in that the cysteine residue in position 101 is replaced. 8. Mutein according to claim 1, characterized in that the cysteine residue in positions 69 and 101 has been replaced. 9. Mutein according to claim 8, characterized in that the cysteine residues are replaced by residues selected from serine, valine, alanine and threonine. 10. Mutein according to claim 9, characterized in that the cysteine residue is replaced by a serine residue. 11. Mutein according to claim 3, characterized in that the protein is TNF and that the cysteine residue in position 101 is replaced by a serine residue. 12. Mutein according to claim 11, characterized in that the cysteine residue is replaced by residues selected from serine, valine, alanine and threonine. 13. Mutein according to claim 12, characterized in that the cysteine residue is replaced by a serine residue. 14. Mutein according to claim 3, characterized in that the protein is TNF, and that each of positions 69 and 101 in TNF is independently replaced. 15. Mutein according to claim 14, characterized in that each cysteine residue is independently replaced by a residue selected from serine, valine, alanine and threonine. 16. Mutein according to claim 15, characterized in that each cysteine residue is replaced by a serine residue. 17. Structural gene, characterized by having a DNA sequence encoding the synthetic mutein according to claim 1, 1, 3, 4, 5, 6, 7, 8, 9 or 10. 18. Expression vector, characterized in that it includes the structural gene according to claim 17 in a position that allows expression thereof. 19. Host cell or organism transformed with the expression vector according to claim 18 or its offspring. 20. E. coli transformed with the expression vector according to claim 18 and its progeny. 21. Method for producing a synthetic mutein, characterized in that it comprises growing the host or the offspring according to claim 19 and harvesting the synthetic mutein from the culture. 22. Method of preventing a protein having at least one cysteine residue which is free to form a disulfide bond from forming such a bond, characterized in that it comprises mutationally changing the protein by omitting the cysteine residue or by replacing it with another amino acid. 23. Method according to claim 22, characterized in that the protein is biologically active and the cysteine is not essential for the biological activity. 24. Method according to claim 22, characterized in that the cysteine residue is replaced by serine or threonine. 25. Method for producing the gene according to claim 17, characterized in that it comprises: (a) hybridizing single-stranded DNA comprising a strand of a structural gene encoding the parent protein with a mutant oligonucleotide primer complementary to a region of the strand that includes the codon for the cysteine to be omitted or substituted or the anti-sense triplets paired with the codon, as appropriate; except for a mismatch with the codon or anti-sense triplet, as the case may be, which defines an omission of the codon or triplet encoding said second amino acid; (b) extending the primer with DNA polymerase to form a mutational heteroduplex; and (c) replicating the mutational heteroduplex. 26. Method according to claim 25, characterized in that the mismatch defines a triplet that codes for serine or threonine. 27. Method according to claim 25, characterized in that the single-stranded DNA is a single-stranded phage which includes the strand and the mutational heteroduplex according to step (b), is converted into a closed circular heteroduplex. 28. Method according to claim 25, characterized in that the replication is carried out with transformation of a competent bacterial host with the closed circular heteroduplex and cultivation of the resulting transformants. 29. Method according to claim 25, characterized in that it further comprises isolating the offspring of the mutant strand of the hetero-duplex, isolating DNA from the offspring and isolating the gene from DNA from the offspring. pp 30. Method according to claim 25, 26, 27, 28 or 29, characterized in that the protein is human TNF, the cysteine residue in position 69 and the mismatch defines a codon for serine. 31. Method according to claim 30, characterized in that the strand is the perception strand of TNF and that the mutant oligonucleotide primer is 5'-CATGGGTGCTCGGGCTGCCTT-3\ 32. Method according to claim 23, characterized in that the protein is human TNF, that the cysteine residue is in position 101 and that the mismatch defines a codon that codes for serine. 33. Oligonucleotide for use in the production of the structural gene according to claim 17 by oligonucleotide-directed mutagenesis, characterized in that it has a nucleotide sequence that is complementary to a region of the strand of the structural gene that includes the codon for the cysteine residue or the anti-sense triplet paired with the codon alt as, apart from a mismatch with said codon defining an omission of the codon or a triplet coding for the second amino acid. 34. Therapeutic preparation with TNF activity, characterized in that it comprises a therapeutically effective amount of the synthetic mutein according to claim 7, 8 or 14, mixed with a pharmaceutically acceptable carrier. 35. Therapeutic preparation with TNF activity, characterized in that it comprises a therapeutically effective amount of the synthetic mutein according to claim 14, mixed with a pharmaceutically acceptable carrier. 36. Plasmid selected from the group pAW731, pAW741 and pAW735. 37. Bacteria transformed with plasmids according to claim 36 and progeny thereof. 38. Bacteria according to claim 37, characterized in that it is E. coli. 39. Human recombinant ser69 tumor necrosis factor (TNF) mutein. 40. Human recombinant serlOl tumor suppressor factor (TNF) mutein. 41. Human recombinant ser69 ser101 tumor necrosis factor (TNF) mutein. 42. Method for treating a patient for a cancer disease comprising administering to the patient a cancer-inhibiting amount of the mutein according to claims 1, 39, 40 or 41. 43. Therapeutic formulation, characterized in that it comprises an effective amount of the mutein according to claim 1, 39, 40 or 41 and at least one other anti-cancer or anti-viral compound. 44. Formulation according to claim 43, characterized in that the anti-cancer or anti-viral compound is "γ-interferon".