AU8339187A - Enhancing gamma-carboxylation of recombinant vitamin k-dependent proteins - Google Patents

Description

ENHANCING GAMMA-CARBOXYLATION OF RECOMBINANT VITAMIN K-DEPENDENT PROTEINS

Background of the Invention This invention relates to vitamin K-dependent proteins.

Vitamin K-dependent proteins require vitamin-K for their complete synthesis. These proteins include prothrombin. Factor IX, Factor VII, protein C,protein S, Factor X, and protein Z, some of which are involved in blood clotting; and osteocalcin and bone matrix 61a protein, which are found in the bone. As a group, the blood clotting proteins share marked homology in amino acid sequence, and are activated by limited proteolysis from the zymogen to active enzyme form (with the exception of Protein S). All of the vitamin K-dependent proteins contain the novel metal binding amino acid γ-carboxyglutamic acid.

The glutantic acid residues in a vitamin K-dependent protein as originally present in the cell are not carboxylated. Carboxylation occurs sometime prior to the secretion of the protein from the cell. Vitamin K-dependent proteins are deficient, on an acquired basis, in liver disease, in vitamin K deficiencies, and in the presence of vitamin K antagonist drugs such as sodium warfarin (Coumadin). Hemophilia B is a disorder characterized as a hereditary deficiency of Factor IX; of the 25,000 persons in the United States with hemopailia, approximately 10-12% are afflicted with Hemophilia B.

The treatment of persons whose disorders comprise acquired or congenital deficiencies of blood clotting proteins continues to be a high risk and costly therapy. For example, Hemophilia B is currently treated in two ways: with fresh frozen plasma, or with a commerical preparation of Factor IX obtained by partial fractionation of normal human plasma. The Factor IX produced by the latter procedure is only of intermediate purity. The human Factor IX gene has been cloned into plasmids and transfected into mammalian cells. The Factor IX expressed, however, is only partially carboxylated as compared to that circulating in the blood. The partially carboxylated Factor IX does not have the full activity of naturally occurring, fully carboxylated Factor IX.

The vitamin K-dependent proteins are, when initially synthesized, composed of the mature protein (e.g., prothrombin), an amino acid sequence upstream of and adjacent to the mature protein, (the "propeptide"); and an amino acid sequence upstream of and adjacent to the propeptide (the "prepeptide"). As the protein is being synthesized in the rough endoplasmic reticulum, the prepeptide is cleaved during the secretion of the proprotein (the propeptide coupled to the mature-protein, e.g., proprothrombin) into the endoplasmic reticulum of the cell. Following this secretion, gamma carboxylation occurs, and finally the propeptide is cleaved and the mature protein is releasedfrom the cell. (Herein, the prepeptide and propeptide together are referred to as the "prepropeptide.")

Summary of the Invention In general, the invention features a DNA sequence including a first DNA sequence encoding a human vitamin K-dependent protein having fused to its 5' end a second DNA sequence not identical to the propeptide encoding sequence naturally associated with the DNA sequence encoding said protein, the non-naturally occurring propeptide encoding sequence being capable of encoding a propeptide sequence which is capable of enhancing the gamma-carboxylation of the protein when the protein is expressed in a recombinant eukaryotic cell. (A recombinant cell is a cell into which the gene for the protein has been introduced by means of an expression vector.)

In preferred embodiments, the second DNA sequence encodes a propeptide closer in amino acid sequence to the propeptide of prothrombin than to the propeptide naturally associated with the protein. Most preferably, a sequence encoding the propeptide of prothrombin itself is used.

The invention is based in part on our discovery that recombinant prothrombin is fully gamma-carboxylated, in contrast to other recombinant vitamin K-dependent proteins, which are poorly gamma-carboxylated and thus exhibit low biological activity. This difference is, we believe, a function of the propeptide of prothrombin, which in some way serves as a superior gamma-carboxylation recognition site in recombinant cells than the propeptides of other vitamin K-dependent proteins.

The prepeptide is not involved in gamma-carboxylation, and therefore, according to the invention, the prepeptide can be that naturally associated with either the protein or the propeptide, or can be any other suitable prepeptide.

In other preferred embodiments, the vitamin K-dependent protein is Factor IX; Factor VII; protein C; protein S; Factor X; protein Z; osteocalcin; or bone matrix Gla protein.

The term vector, as used herein, includes plasmids, viruses, cosmids, or phages into which heterologous DNA (DNA not naturally present in the vector) can be inserted. The vector may be capable of automatically replicating or may allow insertion of the heterogenous DNA into chromosomal DNA. Vectors usually have a replication origin and at least one selectable gene, i.e., a gene which encodes a product which is readily detected or the presence of which is essential for cell growth. Other characteristics of vectors are well known to those skilled in the art of molecular biology. The invention also features a method for producing a human vitamin K-dependent protein with improved gamma-carboxylation by .providing the above-described DNA sequence; inserting the DNA sequence into a eukaryotic, e.g., mammalian, expression vector; transfecting the vector into a eukaryotic, e.g., mammalian, cell; and culturing the cell to produce the vitamin K-dependent protein with improved gamma-carboxylation.

The invention features, in another aspect, a purified DNA sequence (particularly a cDNA sequence) encoding human prothrombin; the DNA sequence includes the prepeptide-encoding region, which is necessary for expression. The term purified, as used herein, means separated from the DNA that is naturally associated with the gene in the human cell. The prothrombin-encoding DNA sequence includes the DNA sequence encoding the mature structural protein as well as the DNA which encodes the prepropeptide of prothrombin. The prothrombin-encoding sequence can be inserted into an expression vector which, when inserted into cultured eukaryotic cells, effects the production by those cells of fully gamma-carboxylated human prothrombin.

Description of the Preferred Embodiments The drawings will first be described. Drawings

Fig. 1 is the amino acid sequence of the prepropeptide of human prothrombin.

Fig. 2 is the DNA sequence encoding part of the amino acid sequence of Fig. 1.

Fig. 3 is the DNA sequence of the untranslated 5'-end of the prepropeptide of human prothrombin.

Fig. 4 is a set of graphs showing the interaction of recombinant prothrombin with four distinct anti-prothrombin antibodies. Structure

Fig. 1 gives the amino acid sequence of the prepropeptide of human prothrombin. The amino acid at position -1 (Arg) is adjacent the amino-terminal (N-terminal) end of mature human prothrombin. The amino acid sequence of the prepropeptide from position -36 to position -1 was previously elucidated by Degen et al. 22 Bipchem. 2087 (1983). All but the 21 5' terminal nucleotides of cDNA encoding human prothrombin are also given in Degen et al. The complete cDNA, including the prepeptide-encoding region, is necessary for expression.

The prepropeptide shown in Fig. 1 is composed of a propeptide and a prepeptide, or "signal" peptide. The prepeptide is the portion from position -1 to position -18; the propeptide is the remaining portion (position -19 to position -43). The prothrombin structural gene encodes the mature human prothrombin protein which begins at position 1. "Proprothrombin" refers to prothrombin having linked to its N-terminal end the propeptide, and "preproprothrombin" refers to prothrombin having the prepropeptide linked to its N-terminal end.

The DNA sequence encoding for residues -35 to -43 of the signal peptide of human prothrombin is shown in Fig. 2. As mentioned above, the DNA sequence of the prothrombin structural gene and of the DNA encoding residues -1 to -36 of the prepropeptide are given in Degen et al., id. The cDNA encoding preproprothrombin was cloned into a mammalian expression vector and transfected into Chinese hamster ovary (CHO) cells. Recombinant prothrombin was expressed in a form that is fully active and completely γ-carboxylated (in comparison to naturally-occurring human prothrombin). The thorough γ-carboxylation observed is in marked contrast to that observed in the expression of human Factor IX (as compared to naturally occurring human Factor IX) when the human Factor IX gene (including the prepro portion) was introduced into rat hepatoma cells (Anson et al., 315 Nature 683 (1985)); human hepatoma cells (de la Salle at al., 316 Nature 268 (1985)); mouse fibroblasts (Anson, supra; de la Salle, supra)); baby hamster kidney cells (Busby et al. (1985) Nature 316, 271); and Chinese hamster ovary cells (Kaufman et al., 261 J. Biol . Chem. 9622 ( 1986) ) .

The propeptides of the vitamin K-dependent proteins are responsible for targeting these proteins for γ-carboxylation of glutamic acid residues. Expressed recombinant prothrombin is fully carboxylated, we believe, because the propeptide of prothrombin is efficient at targeting the protein for carboxylation. In contrast, expressed recombinant Factor IX is not fully carboxylated because the propeptide of pro-Factor IX is not as efficient at targeting.

Cloning of DNA encoding Factor IX that has attached at its N-terminal end an improved targeting propeptide, e.g., the propeptide of prothrombin (instead of the propeptide of Factor IX) into a mammalian expression vector, followed by transfection of the vector into suitable eukaryotic cells, can result in expression of Factor IX exhibiting improved gamma-carboxylation, because of the substitute or improved propeptide's greater efficiency at targeting the protein for γ-carboxylation.

The expression of completely γ-carboxylated recombinant human prothrombin in recombinant cells is described below, followed by a description of the preparation of a prothrombin propeptide-Factor IX cDNA. Expression of Recombinant Human Prothrombin General Description

A human hepatoma cDNA expression library was screened using affinity-purified anti-prothrombin antibodies. One positive clone was shown by restriction enzyme mapping and nucleotide sequencing to contain an insert of 650 base pairs comprising the 3'-end of prothrombin cDNA. In order to obtain the full coding sequence for prothrombin, this fragment was used to screen a fetal liver cDNA library. Restriction enzyme analysis of the numerous positive clones which were isolated revealed that these inserts also were incomplete at their 5'-ends. A restriction fragment was prepared from the 5'-end of the insert which contained the most complete 5'-sequence, and this fragment was used to rescreen the library. Several clones were recovered which encoded the entire 5'-end including the initiator codon. The full-length prothrombin coding sequence (which includes the propeptide and prepeptide) was reconstructed from two overlapping cDNA inserts. The cDNA obtained is 2.0 kilobases in length and contains approximately 100-150 nucleotides at the 3'-untranslated region and 12 nucleotides at the 5'-untranslated region. The nucleotide sequence of the 5'-end and the corresponding predicted amino acid sequence are shown in Figure 3. The human prothrombin leader sequences begins with residue -43.

The prothrombin expression vector, pMT2-PT, contains the SV40 origin of replication, the adenovirus major late promoter, the prothrombin coding region, the dihydrofolate reductase coding region, the SV40 early polyadenylation site, the adenovirus virus-associated genes, and the pBR322 sequences needed for propagation in Escherichia coli. Details of the components of this vector have been described by Kaufman et al., 261 J. Biol. Chem. 9622 (1986), and by Wong et al., 228 Science 810 (1985). Plasmid pMT2-PT was introduced into dihydrofolate reductase-deficient Chinese hamster ovary cells, and cells were selected for the dihydrofolate reductase-positive phenotype by subculturing in selective medium. Culture media harvested when the cells reached confluency were assayed by competition radioimmunoassay using anti-total prothrombin antibodies. Total prothrombin antigen was found to be expressed by these primary transfectants at varying concentrations up to 0.55 μg/ml. The same samples were assaying using conformation-specific antibodies, anti-prothrombin:Ca(II), which bind to specific determinants expressed on prothrombin in the presence of metal ions. These determinants are present only when prothrombin is sufficiently γ-carboxylated to undergo its metal-induced conformational transition, and their presence correlates closely with coagulant activity. Levels of prothrombin measured using this assay were equivalent to the levels determined for total prothrombin antigen, suggesting that all of the prothrombin expressed was carboxylated and biologically active. Indeed, no des-γ-carboxy (abnormal) prothrombin could be detected, using anti-abnormal prothrombin antibodies, to a limit of 0.03 μg/ml.

Recombinant prothrombin vas isolated from conditioned medium by immunoaffinity chromatography using conformation-specific antibodies, as described by Liebman et al., 82 Nat. Acad. Sci. USA 3879 (1985) and Borowski et al., 260 J. Biol. Chem. 9258 (1985). The tissue culture supernatant was applied to a column of anti-prothrombin:Ca(II)-Sepharose in the presence of Ca(II). All prothrombin antigen was removed from the culture medium by this process; no prothrombin antigen was detected, using anti-total prothrombin antibodies, in the material that failed to bind to the column. The bound prothrombin was eluted with EDTA, and prothrombin was recovered quantitatively in this eluate.

Purified recombinant: prothromoin migrated as a single major band, on dodecyi sulfate gels in the presence of 2-mercaptoethanol. Its electrophoretic mobility was identical to that of prothrombin derived from human plasma.

The coagulant activity of recombinant prothrombin was determined directly using prothrombin-deficient plasma. Purified plasma prothrombin of known concentration was used to prepare a standard curve. Recombinant prothrombin was found to have 99 ± 4% of the coagulant activity of plasma prothrombin, as shown in Table 1.

The interactions of recombinant prothrombin with four distinct populations of anti-prothrombin antibodies were assessed by competition radioimmunoassay. Anti-total prothrombin antibodies bind to prothrombin. regardless of its extent of carboxylation or conformation. As expected, recombinant prothrombin, plasma-derived prothrombin, and abnormal (des-γ-carboxy) prothrombin equally displaced

¹²⁵ I-labeled prothrombin from these antibodies, as shown in Fig. 4A. Anti-abnormal prothrombin antibodies bind to antigenic determinants present only on des-γ-carboxy forms of prothrombin. As shown in Fig. 4B, recombinant prothrombin and plasma prothrombin did not displace ¹²⁵I-labeled abnormal prothrombin from these antibodies, demonstrating the absence of even trace quantities of des-γ-carboxy prothrombin in the recombinant preparation. Two antibody populations specific for different metal-dependent conformers of prothrombin have been described by Borowski et al., 261

J. Biol. Chem. (1986) (in press). Anti-prothrombin:

Mg(II) antibodies bind to antigenic determinants expressed on prothrombin only when it is sufficiently carboxylated to undergo the first conformational change which is induced by most divalent and trivalent metal ions . Anti-prothrombin:Ca ( II)-specific antibodies are directed against antigenic determinants exposed when prothrombin undergoes a second conformational change associated with expression of the lipid binding site.

This change is supported only by Ca(II), not by most other metal ions. As shown in Figs. 4c and 4D, recombinant prothrombin and plasma prothrombin equally displaced ¹²⁵I-labeled prothrombin from both of these antibodies, demonstrating essential equivalence of recombinant prothrombin and plasma prothrombin by these criteria.

The amino-terminal sequence of recombinant prothrombin was established by automated Edman degradation. As shown in Table II, the amino acid sequences of the first 16 amino acid residues of recombinant prothrombin and plasma-derived prothrombin are identical. Of particular interest is the fact that no secondary sequences, representing incompletely processed pre- or propeptide, were detected. Additionally, as the γ-carboxyglutamic acid derivative is not released from the filter during the standard sequencing cycle, the undetectable levels of glutamic acid found at residues 6, 7, 14, and 16 are consistent with complete γ-carboxylation of these residues. The γ-carboxyglutamic acid content of recombinant prothrombin was determined by amino acid analysis of the alkaline hydrolysate. Purified recombinant prothrombin contained 9.9 ± 0.4 moles of γ-carboxyglutamic acid per mole of protein, using as the standard the value of 10 moles per mole for plasma-derived prothrombin, as shown in Table I. Specific Procedures Screening of cDNa Libraries:—A human hepatoma cDNA expression library, prepared in the λgtll vector of Young and Davis, 80, Proc. Nat. Acad. Sci. USA 1194 (1983), was screened for clones expressing prothrombin antigen using the chromogenic immunodetection system described by de Wet et al., 3 DNA 437 (1984). Approximately 100,000 plaques were transferred to nitrocellulose filters and incubated overnight in Tris-buffered saline containing 3% bovine serum albumin, 0.02% sodium azide, and 0.5 μg/ml of immunoaffinity-purified rabbit anti-total prothrombin antibodies. After washing several times in

Tris-buffered saline, the filters were incubated for 4 hours in Tris-buffered saline containing 3% bovine serum albumin and goat anti-rabbit immunoglobulin conjugated with horseradish peroxidase (Bio-Rad, 1:1000 dilution). After further washing, positive plaques were detected upon addition of the substrate 1-chloro-2-naphthol (Bio-Rad). Seven positive clones were isolated by three rounds of rescreening, and phage DNa was purified from plate lysates by the procedure described by Helms et al., 4 DNA 39 (1985). The cDNA inserts were excised with EcoRl and subcloned for analysis by restriction enzyme mapping. One 650-base pair insert was tentatively identified as the 3'-end of prothrombin cDNA based upon its restriction pattern. This identity was confirmed by nucleotide sequencing using the chemical cleavage method of Maxam and Gilbert, 65 Methods Enzymol, 499 (1980). To obtain the full-length coding sequence of prothrombin, a human fetal liver cDNA library in Charon 21A (described by Toole et al., 313 Nature 342 (1984)) was screened using the method of Benton and Davis, 196 Science 180 (1977). Restriction fragments of prothrombin cDNA, derived from the cDNA insert obtained from the human hepatoma expression library, were radiolabeled to a specific activity of approximately

10⁸ cpm/μg by random hexanucleotide priming using

³² P-dCTP and the Klenow fragment of DNA polymerase described by Feinberg et al., 132 Anal. Biochem. 6 (1983). Hybridization for 16 hours at 68°C in 90 mM sodium citrate, pH 7.0, 900 mM sodium chloride, 5x Denhardt's solution (described by Denhardt, 23 Biochem. Biophys. Res. Comm. 541 (1966)), 10 mM EDTA, and 0 .53 sodim dodecyl sulfate was followed by extensive washing at 68°C in 30 mM sodium citrate, pH 7.0 , 300 mM sodium chloride, and 0.5% sodium dodecyl sulfate. Duplicate positives were plaque-purified and phage DNA was isolated from plate lysates by the procedure described in Helms, supra. Desired restriction fragments of cDNA inserts were subcloned into appropriate M13 vectors (by the general procedure described in Norrandes, 26 Gene 101 (1983)) for restriction enzyme mapping and sequencing by the dideoxynucleotide chain termination method described by Sanger et al., 74 Proc. Nat. Acad. Sci. USA 5463 (1977). Construction of Prothrombin Expression Plasmid PMT2-PT:—The full-length prothrombin coding sequence was reconstructed from two overlapping cDNA inserts, each digested at the single Hindlll site, by cloning the appropriate fragments into mp¹⁸. A 2.0-kilobase EcoRi fragment encoding the complete prothrombin sequence (including the DNA encoding the prepropeptide) was then isolated and inserted into the EcoRl site of the mammalian expression vector, pMT2, described in Toole et al. (1986) P.N.A.S. U.S.A. 83, 5939. The resultant prothrombin expression plasmid, pMT2-PT, was siiown by restriction mapping to contain the prothrombin coding region in the proper orientation with respect to the adenovirus major late promoter. The prothrombin cDNA included the prepeptide encoding sequences given in Fig. 2 fused to the remainder of the prothrombin cDNA given in Fig. 2 of Degen et al., id.

Cell Culture, DNA Transfection, and Cell Line Selection:—The dihydrofolate reductase-deficient Chinese hamster ovary cell line, CHO DUKX-Bll, was grown and maintained as described by Chasin et al., 77 Proc. Nat. Acad. Sci. USA 4216 (1980) and Kaufman et al., 159 Mol. Bio. 601 (1982). The cells were transfected with 20 μg of the prothrombin expression plasmid, pMT2-PT, by the calcium phosphate coprecipitation described in Kaufman et al., 195 Med. Bio. 601 (1982). After transfection, cells were grown in α-modified Eagle's medium lacking nucleosides (Gibco) containing 10% heat-inactivated fetal bovine serum, 5μg/ml vitamin K₁ (Aquamephyton, Merck Sharp and Dohme), and thymidine, adenosine, deoxyadenosine, penicillin, and streptomycin (10 μg/ml each). The cells were subcultured two days later into the same medium except that the nucleosides were omitted and dialyzed serum was used ('selective medium'). Transfected cells were fed every 3-4 days with selective medium until colonies were visible, about 10-12 days after subculturing. These initial transformants were pooled and grown in selective medium until confluent. At confluence, a 10-cm dish contained approximately 1.7 x 10⁷ cells in 10 ml of medium. Large volumes of conditioned media were collected by refeeding flasks of confluent cells every 3-4 days for several weeks. Conditioned media were stored at -20°C until needed.

Preparation of Protein Standards and Antibodies:—Human prothrombin was purified from plasma by barium citrate adsorption, DEAE-cellulose chromatography, and affinity chromatography using dextran-Sepharose by the procedure described in Rosenberg et al., 250 J. Biol. Chem. 1607 (1975), and by Miletich et al., 253 J. Biol. Chem. 6908 (1978). Abnormal (des-γ-carboxy) human prothrombin was purified from plasma by DEAE-Sephacel chromatography and affinity chromatography using anti-prothrombin¬

Sepharose, as described in Blanchard et al., 101 J. Lab.

Clin. Med. 242 (1983). The protein concentration of purified prothrombins was measured using A₁% at 280 nm of 14.4. Prothrombin and abnormal prothrombin were lodmated with Na ¹²⁵I using the lactoperoxidase method of Morrison, 70 Meth. Enzymol. 214 (1980).

Rabbit anti-prothrombin:Ca(II) antibodies were purified by immunoaffinity chromatography on prothrombin-Sepharose in the presence of Ca(II) followed by elution with EDTA, as described by Blanchard, supra. Anti-prothrombin antibodies which bound to prothrombin-Sepharose in the presence of EDTA were eluted with 4 M guanidine hydrochloride and were termed anti-total prothrombin. Anti-abnormal prothrombin antibodies were prepared by sequential immunoaffinity chromatography using des-γ-carboxy prothrombinSepharose as described by Blanchard et al., 101 J. Lab. Clin. Med. 242 (1983). Anti-prothrombin: Mg(II) and anti-prothrombin:Ca(II)-specific antibodies were prepared by sequential immonoaffinity chromatography using prothrombin-Sepharose in the presence of either Mg(II) or Ca(II) followed by elution with EDTA according to the method described by Bcrαwski, supra.

Radioimmunoassays:—The displacement of

¹²⁵ I-labeled prothrombin from anti-prothrombin antibodies was studied using a competition readioimmunoassay. Anti-total prothrombin antibodies (1.1 x 10^-9 M) were added to a reaction mixture which included ¹²⁵I-labeled prothrombin (1.7 x 10^-10 M) and varying concentrations of competitors. All components were diluted in Tris-buffered saline containing 1 mM benzamidine, 0.1% bovine serum albumin, 3 mM EDTA, and carrier rabbit gammaglobulin. Anti-prothrombin:Ca(II) antibodies (3.4 x 10^-10 M), anti-prothrombin:Mg(II) antibodies (1.0 x 10^-9 M), or anti-prothrombin:Ca( lI)- specific antibodies ( 1.3 x 10^-9 M) were added to the same reaction mixture except that 3 mM calcium chloride replaced EDTA. Anti-abnormal prothrombin antibodies (4.0 x 10^-9 M) were added to a similar mixture containing ¹²⁵I-labeled abnormal prothrombin (2.8 x 10^-10 M) and a 3 mM EDTA. After overnight incubation at 4°C, the bound ¹²⁵I-labeled prothrombin or abnormal prothrombin was precipitated by the addition of goat anti-rabbit immunoglobulin. The precipitate which formed was removed by centrifugation and was assayed for ¹²⁵I in a Beckman Gamma 8000 spectrometer. For determination of unknown antigen concentrations a standard curve was prepared using human plasma-derived prothrombin or abnormbal prothrombin of known concentration.

Purification of Recombinant Prothrombin:—Recombinant prothrombin was purified from conditioned medium by immunoaffinity chromatography using conformation- specific antibodies as described in Borowski, supra, and Liebman et al., 82 Proc. Nat. Acad. Sci. USA 3879 (1985). The antibodies employed, anti-prothrombin: Ca(II), bind to the antigenic determinants expressed on γ-carboxylated prothrombin in the presence of metal ions. Conditioned medium containing 10 mM calcium chloride and 0.02% sodium azide was applied to a column of anti-prothrombin:Ca(II)- Sepharose at 4°C. After exhaustive washing with Tris-buffered saline containing 10 mM calcium chloride and 0.5 M sodium chloride, the column was eluted with 10 mM EDTA in Tris-buffered saline. The protein concentration of purified recombinant prothrombin was measured using A_1% at 280 nm of 14.4. Electrophoresis was carried out on 10% polyacrylamide gels in the presence of sodium deodecyl sulfate. Samples were incubated for 5 minutes at 100°C in sample buffer containing 2-mercaptoethanol prior to loading on the gel. Gels were stained after electrophoresis with Coomassie Blue R-250.

Prothrombin Coagulation Assay:—Prothrombin activity was determined in a two-stage assay using prothrombin-deficient plasma, as described by Blanchard, supra. Human prothrombin of known concentration was used to prepare the standard curve. Amino-Terminal Sequence Analysis:—Automated Edman degradation was performed on an Applied Biosystems Model 470A gas-phase protein sequencer equipped with a Model 120 PTH Analyzer. Recombinant prothrombin was desalted by high pressure liquid chromatography using an RP-300 Brownlee guard cartridge prior to analysis. γ-Carboxyglutamic Acid Analysis:—Amino acid analyses were performed on a Beckman Model 119CL amino acid analyzer equipped with a Beckman Model 126 data system. The proteins were hydrolyzed in 2 M potassium hydroxide for 22 hours at 110°C, as described by Hauschka, 80 Anal. Biochem., 212 (1977). The γ-carboxyglutamic acid composition was quantitated after alkaline hydrolysis by automated amino acid analysis using a ninhydrin detection system. Prepropeptide of prothrombin-Factor IX cDNA

The coding sequence for a hybrid molecule comprising the prepropeptide of prothrombin joined to mature human Factor IX was constructed according to the following steps, in which mutagenesis was carried out in a phage vector.

The cDNA comprising the coding sequence of Factor IX was cloned into the vector M13mp8. The Factor IX coding sequence is given by Kaufman et al., id. The plasmid pMT2-IX was cut with the restriction enzyme Pst I, and the resulting 2.5-kilobase fragment gel purified using standard methods. This fragment was inserted into the Pst I site of the publically available phage vector mp8. Clones containing the Factor IX coding sequence in the desired orientation in the vector were identified by the presence of a 2.0-kilobase fragment upon digestion with the enzymes EcoRV and BamHI. This vector is termed mp8-IX. A cDNA fragment containing the coding sequence of the prepro region of prothrombin was obtained by cutting mp18-PT with the restriction enzyme HindiII and isolating the 350 base pair fragment by standard techniques. Mp18-PT is described above and in Jorgensen et al, J. Biol. Chem., 1987, id. The 350 base pair fragment was inserted into the Hindlll site of mp8-IX. Clones containing the insert in the correct orientation relative to the Factor IX sequence were identified by the presence of a 1.0-kilobase fragment upon digestion with the enzymes Xhol and Bglll. The construct containing the HindiII fragment of the coding sequence for Factor IX is termed mp8-PT/IX.

In order to splice the sequence encoding the prothrombin leader sequence to that encoding the mature Factor IX sequence, the intervening nucleotide sequences were removed by "loopout" mutagenesis using a synthetic oligonucleotide, as described by Oostra et al. (1983) Nature 304, 456. Accordingly, a heteroduplex is formed between single stranded mp8-PT/IX and single stranded mp18 cut with EcoRI. This heteroduplex contains a single stranded segment to which the mutagenic primer was annealed. To prepare single stranded mp8-PT/IX, E. coli (TGI strain) were transformed with the vector, a single plaque was picked, and a single stranded template prepared using standard methods. For the second strand of the heteroduplex, mp18 was linearized with the restriction enzyme EcoRI. To form the heteroduplex, 2 μg of linearized mp18 and 1.5 μg of mp8-PT/IX were mixed in a final volume of 15 μl and denatured by addition of 4 μl of 1 N NaOH. After 10 minutes at room temperature, the mixture was neutralized by the addition of 180 μl of 0.1 M Tris, pH 7.5 and 0.02 M HCl. The denatured DNA (40 μl) was allowed to reanneal by incubating the mixture overnight at 68°C and cooling gradually to room temperature over several hours. The resulting mixture of annealed DNA contains, in part, the desired heteroduplex molecule.

Annealing of the appropriate mutagenic oligonucleotide to the heteroduplex molecule results in formation of a loop which contains the single-stranded sequence to be deleted. The remaining single-stranded gaps were filled by primer extension and end ligation. The mutagenic primer used was a 31-mer, S'-GCGGGTCCGGCGATATAATTCAGGTAAATTG, composed of blocks of 13 and 18 bases complementary to the regions to be brought together to give the desired hybrid coding sequence. The oligonucleotide was synthesized on an Applied Biosystems 380B Synthesizer and gel-purified prior to use. After 5'-phosphorylation using T4 polynucleotide kinase in a standard reaction mixture (Maniatis, supra), 2 μl (10 pmol) of the primer was mixed with 40 μl (0.1 pmol) of the heteroduplex mixture. The annealing reaction was carried out by heating to 68ºC for two and one half hours and cooling slowly to 15°C. The primer was extended and joined to the existing partial second strand during a four hour incubation at 15°C, after addition of 8 μl containing 20 mM MgCL₂; 40 mM dithiothreitol; 2 mM ATP; 1 mM each dATP, dTTP, dCTP, and dGTP; 5 U Klenαw fragment of DNA polymerase I (Pharmacia) and 0.5 U T4 ligase (Bethesda Research Labs). E. coli (strain JM105, Sup F minus) were transformed with the resulting heteroduplex (2 μl). Plaques were obtained only from E. coli transformed with DNA containing the mutagenized mp18 strand, as mpp contains the amber mutation. The DNA from a positive plaque was used to retransform E. coli. A single positive plaque was picked and single stranded template prepared for DNA sequencing. In addition, extensive restriction enzyme mapping was performed to assure that the isolated clone contained the desired mutation and no others.

The desired construct, the coding sequence of the prepro region of prothrombin immediately adjacent to the amino terminus of the nature Factor IX coding sequence (preproPT/IX), was prepared by the above technique. In order to insert this sequence into the mammalian expression vector pMT2, the preproPT/IX will be excised from the mp18 vector using the restriction enzyme EcoRI and inserted into the EcoRI site of pMT2. Clones containing the proper orientation of the preproPT/IX in the pMT2 are identified by the presence of a 600 base pair fragment in. restriction digest performed with the enzymes Bglll an EcoRV. This plasmid, pMT2-PT/IX can be used to transfect mammalian cells, e.g., CHO cells, as described, to yield the mature, processed Factor IX. The expressed Factor IX is gamma carboxylated and biologically active.

Other Embodiments Other embodiments are within the following claims. For example, DNA encoding other vitamin K-dependent proteins, including Factor VII, protein C, protein S, Factor X, protein Z, osteocalcin and bone Gla matrix protein, can be linked at their N-terminal ends to the propeptide sequence of proprothrombin. Vectors containing the linked DNA can be transfected into mammalian cells, resulting in the protein being expressed in its fully carboxylated form. The vectors of the invention may be transformed into any other suitable mammalian cells other than CHO cells, e.g., mouse C127 cells.

Although Factor IX is linked to the entire leader sequence of prothrombin it is understood that only the region encoding the propeptide portion of proprothrombin peptide need be linked to the Factor IX DNA to obtain the advantage of the invention, i.e., complete carboxylation of the expressed proteins. Accordingly, instead of replacing the DNA encoding the entire leader sequence of Factor IX with the DNA encoding the leader sequence of prothrombin, it is necessary only to replace the DNA encoding pro-Factor IX peptide with the DNA encoding the propeptide of proprothrombin.

Claims (16)

Claims

1. A DNA sequence comprising a first DNA sequence encoding a human vitamin K-dependent protein having fused to its 5' end a second DNA sequence not identical to the propeptide encoding sequence naturally associated with said DNA sequence encoding said protein, said non-naturally occurring propeptide encoding sequence being capable of encoding a propeptide which is capable of enhancing the gamma-carboxylation of said protein when said protein is expressed in a recombinant eukaryotic cell.

2. The DNA sequence of claim 1 wherein said second DNA sequence encodes a propeptide closer in amino acid sequence to the propeptide of prothrombin than to the propeptide naturally associated with said protein.

3. The DNA sequence of claim 2 wherein said second DNA sequence encodes the propeptide of human prothrombin.

4. The DNA sequence of claim 1 wherein said vitamin K-dependent protein is Factor IX.

5. The DNA sequence of claim 1 wherein said vitamin K-dependent protein is Factor VII.

6. The DNA sequence of claim 1 wherein said vitamin K-dependent protein is protein C.

7. The DNA sequence of claim 1 wherein said vitamin K-dependent protein is protein S.

8. The DNA sequence of claim 1 wherein said vitamin K-dependent protein is Factor X.

9 . The DNA sequence of claim 1 wherein said vitamin K-dependent protein is protein Z.

10. The DNA sequence of claim 1 wherein said vitamin K-dependent protein is osteocalcin.

11. The DNA sequence of claim 1 wherein said vitamin K-dependent protein is bone Gla matrix protein.

12. The DNA sequence of claim 1 wherein said first DNA sequence encodes for the prepropeptide of prothrombin.

13. A vector comprising the DNA of claim 1.

14. A purified DNA sequence encoding human prothrombin, said sequence including a sequence encoding the prepeptide of prothrombin.

15. A vector comprising the DNA sequence of claim 14.

16. A method for producing a human vitamin K-dependent protein with improved gamma-carboxylation comprising providing the DNA sequence of claim 1, inserting said DNA sequence into a mammalian expression vector; transfecting said vector into a mammalian cell; and culturing said cell to produce said vitamin

K-dependent protein with improved gamma-carboxylation.