NO310029B1 - Process for Preparation of huBMP-2 Class I (BMP-2) and II (BMP-4) - Google Patents

Description

The present invention relates to methods for the production of new proteins. These proteins have the possibility of inducing cartilage and bone formation.

Bone is a highly specialized tissue characterized by extensive matrix structure formed by fibrous bundles of the protein collagen, and proteoglycans, non-collagenous proteins, lipids and acidic proteins. The processes of bone formation and renewal/preparation of bone tissue, which occur continuously throughout life, are carried out by specialized cells. Normal formation of the long shaft skeletal parts in fetuses precedes the formation of a cartilage model. Bone growth is presumably carried out by "osteoblasts" (bone-forming cells), while bone remodeling is presumably carried out by bone-resorbing cells, called "osteoclasts" and osteoblasts. Various bone-forming, cartilage-inducing and bone-inducing factors have been described. See e.g. European Patent Application 148,155 and 159,016 regarding discussions thereof.

New proteins have been produced in purified form. Four of the new proteins are designated BMP-1, BMP-2 class I (or BMP-2), BMP-3, and BMP-2 class II (or BMP-4) where BMP is bone morphogenic protein. Two of these, BMP-2 and BMP-4, are proteins produced according to the invention. The proteins are characterized by peptide sequences that are the same as or substantially homologous to the amino acid sequences illustrated in Tables II through VIII below. They have the possibility of inducing bone formation at a predetermined location. These bone-inducing factors are further characterized by biochemical and biological properties that include an activity at a concentration of 10 to 1000 ng/gram of bone in an in vivo rat bone-forming assay described below. The proteins produced in this invention can be encoded by the DNA sequences described in the tables or by sequences that have the possibility of hybridizing thereto and that encode polypeptides with bone growth factor biological properties or other modified sequences that demonstrate such properties.

One of the proteins is called BMP-1. A portion of the human BMP-1 or hBMP-1 is characterized by the same or substantially the same peptide sequence as amino acids #1 through #37 in Table V below, which stand for a genomic hBMP-1 fragment or amino acid # 1 through amino acid #730 in Table VI which stands for hBMP-1 cDNA. hBMP-1 or a related bone-inducing factor can further be characterized by at least part of these sequences. These peptide sequences are encoded by the same or substantially the same DNA sequence as set forth in nucleotide #3440 through nucleotide #3550 in Table V and in nucleotide #36 through nucleotide #2225 in Table VI, respectively. These hBMP-1 polypeptides are further characterized by the fact that they can induce bone formation. hBMP-1 demonstrates activity in an in vivo rat bone formation assay at a concentration of 10 to 1000 ng/gram of bone.

The homologous bovine growth factor is designated bBMP-1, and is characterized by a peptide sequence that contains the same or substantially the same sequence as that of amino acid #1 through amino acid #37 in Table II which stands for a genomic bBMP-1 fragment. This peptide sequence is encoded from the same or substantially the same DNA sequence as provided in nucleotide #294 through nucleotide #404 of Table II. The bovine peptide sequence identified in Table II below is also 37 amino acids in length. bBMP-1 is further characterized by the fact that it can induce bone formation.

Another bone-inducing protein preparation prepared in this invention is designated BMP-2 class I (or BMP-2). It is characterized by at least a portion of a peptide sequence that is the same or substantially the same as that of amino acid #1 through amino acid #396 in Table VII representing cDNA hBMP-2 class I. This peptide sequence is encoded by the same or substantially the same sequence as set forth in nucleotide #356 through nucleotide #1543 of Table VII. The human peptide sequence identified in Table VII is 396 amino acids in length. hBMP-2 or related bone-inducing proteins can also be characterized by at least part of this peptide sequence. hBMP-2 class I is further characterized by the ability to induce bone formation.

The homologous bovine bone-inducing protein designated bBMP-2 class I (or bBMP-2) has a DNA sequence identified in Table III that represents the genomic sequence. This bovine DNA sequence has a 129 amino acid coding sequence followed by approximately 205 nucleotides (presumptive 3' non-coding sequence). bBMP-2, class I is further characterized by the ability to induce bone formation. A bone-inducing protein preparation in the invention is designated BMP-2 class II or BMP-4. The human protein hBMP-2 class II (or hBMP-4) is characterized by at least part of the same or substantially the same peptide sequence between amino acid #1 through amino acid #408 in Table VIII, which represents the cDNA of hBMP-2 class II. This peptide sequence is encoded by at least a portion of the same or substantially the same DNA sequence as set forth in nucleotide #403 through nucleotide #1626 of Table VIII. This factor is further characterized by the ability to induce bone formation.

Yet another bone-inducing factor is represented by the bovine homolog bBMP-3. bBMP-3 is characterized by the DNA sequence and the amino acid sequence in Table IV A and B which stand for the bovine genome sequence. It is characterized by at least a portion of a peptide sequence that is the same or substantially the same as amino acid #1 through amino acid #175 in Table IV A and B. BMP-3 is further characterized by the ability to induce bone formation. The bovine factor can be used as a tool to obtain the analogous human BMP-3 protein or other mammalian bone-inducing proteins. A proper characterization of this bovine bone-inducing factor provides the essential "starting point" for the method using this sequence. This method, using techniques known in genetic engineering, involves using the bovine DNA sequence as a probe to screen a human genomic or cDNA library; and identifying the DNA sequences that hybridize to the probes. A clone with a sequence that hybridizes is purified by plaque formation and the DNA isolated therefrom, subcloned and subjected to DNA sequence analysis. Human hBMP-3 protein is prepared using this method.

The present invention comprises a method for the production of huBMP-2 class I (BMP-2), characterized in that it includes the cultivation of a microorganism transformed with a vector containing a DNA sequence that codes for huBMP-1 in a suitable culture medium, where the DNA sequence is the nucleotide sequence as follows:

and isolating huBMP-2 class I from said culture medium.

The invention also includes a method for the production of huBMP-2 class II (BMP-4), characterized in that it includes the cultivation of a microorganism transformed with a vector containing a DNA sequence that codes for huBMP-2 class II in a suitable culture medium, wherein said DNA sequence is the nucleotide sequence as follows: and wherein said DNA sequence is in relative association with an expression control sequence therefore, and isolating huBMP-2 class II from said culture medium.

Furthermore, the invention comprises cDNA sequence, characterized in that it is selected from the group consisting of: a. cDNA sequence that codes for huBMP-2 class I that includes the nucleotide sequence according to claim 1 or a sequence that hybridizes to it under stringent conditions and that upon expression encodes a protein that has the same properties as huBMP-2 class I; and

b. cDNA sequence that codes for huBMP-2 class II that includes the nucleotide sequence according to claim 2 or a sequence that hybridizes thereto under stringent conditions and that upon expression codes for a protein that has the same properties as huBMP-2 class II.

Also included is vector, characterized in that it contains and can express a DNA sequence that codes for a human bone growth-inducing protein, whereupon the DNA sequence codes for the protein that is selected from the group consisting of: a. a DNA sequence that codes for huBMP-2 class I comprising the nucleotide sequence of claim 1; and b. a DNA sequence that codes for huBMP-2 class II that includes the nucleotide sequence according to claim 2 or a sequence that hybridizes thereto under stringent conditions and that upon expression codes for a protein that has the same properties as huBMP-2 class II, as well as microorganism, characterized by the fact that it contains and can express the vector as previously mentioned, and that it is selected from the group consisting of a bacterial cell, a yeast cell and a mammalian cell.

The bone growth factors provided herein also include factors encoding from the sequences similar to the sequences of

Tables II - VIII, but on which modifications are naturally included (eg allelic variations in nucleotide sequence which can result in amino acid changes in the polypeptide) or which are intentionally engineered. For example, synthetic polypeptides may fully or partially duplicate contiguous sequences of the amino acid residues in Tables II - VIII. By virtue of sharing primary, secondary or tertiary structural and conformational features with bone growth factor polypeptides in tables II - VIII, these sequences may therefore contain common bone growth factor biological properties. They can therefore be used as biologically active substances for naturally occurring bone growth factor polypeptides in therapeutic methods.

Other specific mutations in the sequences of the bone growth factors described herein include modifications of one or both of the glycosylation sites. No glycosylation or only partial glycosylation results from amino acid substitution at one or both asparagine-linked glycosylation recognition sites present in the sequence of the bone growth factors shown in Tables II-VIII. The asparagine-linked glycosylation recognition sites include the tripeptide sequences that are specifically recognized by appropriate cellular glycosylation enzymes. These tripeptide sequences include either asparagine-X-threonine or asparagine-X-serine, where X is usually any amino acid. Different amino acid substitutions or deletions at one or both of the first or third amino acid positions of a glycosylation recognition site (and/or amino acid deletion at the second position) result in non-glycosylation at the modified tripeptide sequence.

This invention also includes the new DNA sequences, which have no connection with DNA sequences which code for other proteinaceous materials, and which code, when expressed, for bone growth factors. These DNA sequences include those prepared in Tables VII and VIII in a 5' to 3' direction and those sequences that hybridize under stringent hybridization conditions [see T. Maniatis et al, Molecular

Cloning (A Laboratory Manual), Cold Spring Harbor Laboratory

(1982), pages 387 to 389] to the DNA sequences in Tables VII and VIII.

DNA sequences that hybridize to the sequences in Tables VII and VIII under other hybridization conditions and that code upon expression for bone growth factors having bone growth factor biological properties also code for bone growth factors of this invention. For example, a DNA sequence that shares regions of significant homology, e.g. glycosylation sites or disulfide bond sites, with the sequences in Tables VII and VIII and encoding a bone growth factor having one or more bone growth factor biological properties clearly encodes a member of this new growth factor family, even though such a DNA sequence would not hybridize strictly to the sequence in Tables VII and VIII.

The DNA sequences encoding bone growth factor polypeptides such as. encoded by the sequences in Tables VII and VIII, but which have different codon sequences due to degeneration of the genetic code or allelic variations (naturally occurring base changes in the species population that may or may not result in amino acid change) also encode the new the growth factors described herein. Variations in the DNA sequences in Tables VII and VIII which are due to point mutations or which are due to induced modifications to increase the activity, half-life or production of the polypeptides encoded from this DNA sequence are also included in this invention.

Production of new bone growth-inducing factors includes cultivation of suitable cells or cell lines, which have been transformed with a DNA sequence which, upon expression, codes for a new bone growth factor polypeptide of the invention, under the control of known regulatory sequences. Suitable cells or cell lines may include mammalian cells, such as Chinese hamster ovary (CHO) cells. Selection of suitable mammalian host cells and methods for transformation, cultivation, amplification, screening and product manufacture and purification are known within the field. See, e.g., Gething and Sambrook, Nature, 293:620-625 (1981), or alternatively, Kaufman et al., Mol. Cell. Biol., 5(7 ):1750-1759 (1985) or Howley et al, U.S. Patent 4,419,446. Another suitable mammalian cell line, which is described in the attached examples, is the monkey COS-1 cell line. Another useful mammalian cell line is the CV-1 cell line.

Bacterial cells are suitable hosts. For example, the various E. coli strains (eg, HB101, MC1061) are well-known host cells in the biotechnology field. Various B. subtilis, Pseudomonas strains, other bacteria and the like can also be used in this method.

Many yeast cell strains known in the art are also available as host cells for expression of the polypeptides of this invention. In addition, when desired, insect cells can be used as host cells in the methods of this invention. See e.g. Miller et al, Genetic Engineering, 8:277-298 (Plenum Press 1986) and references therein.

The vectors according to the invention preferably contain the entire new DNA sequence described above which codes for new factors in this invention. These vectors additionally contain appropriate expression control sequences that allow expression of bone-inducing protein sequences. Vectors containing modified sequences as described above are also included in this invention and useful in the production of bone-inducing proteins. The vectors can be used in the method which transforms cell lines and which contain selected regulatory sequences in operative association with the DNA coding sequences of the invention which can direct the replication and expression thereof in selected host cells. Suitable regulatory sequences for such vectors are known to those skilled in the art and can be selected depending on the selected host cells. Such selection is routine and does not form part of this invention.

A protein produced according to the invention, which induces bone growth in cases where bones are not normally formed, can be used for healing bone fractures. A bone-forming preparation that includes one or more of the proteins in this invention can have prophylactic use for closed or open fracture reduction and also for improved fixation of artificial joints. New bone formation induced by a bone-forming agent participates in the repair of congenital, injury-induced or oncological excision-induced craniofacial defects, and can also be used in cosmetic plastic surgery. A bone-forming factor of the invention may be valuable in the treatment of periodontal diseases, and in other tooth repair procedures. Such agents can provide an environment to attract bone-forming cells, stimulate growth of bone-forming cells or induce differentiation of stem cells into bone-forming cells. The proteins produced according to the invention can also be used in other therapies.

BMP-2 class I may be used individually in a pharmaceutical preparation. BMP-2 class I may also be used in conjunction with one or more of the other proteins of this invention. BMP-2 class I can be combined with BMP-2 class II. It can also be combined with BMP-3. Furthermore, BMP-2 class I can be combined with BMP-2 class II and BMP-3.

BMP-2 class II can be used individually in pharmaceutical preparations. In addition, it can be used together with other proteins as defined above. It can also be used together with the BMP-3.

A therapeutic method may include local administration of the preparation as an implant or device. When administered, the therapeutic composition used in this invention is, of course, in a pyrogen-free, physiologically acceptable form. The preparation can, if desired, be encapsulated or be injected in a viscous form to deliver it to the bone injury site. The bone growth-inducing factor preparation preferably includes a matrix that can deliver the bone-inducing factor to the site of bone damage, and provides a structure for bone and cartilage that develops and which is eventually resorbed in the body. Such matrices can be formed from other materials now in use in other implanted medical conditions.

The choice of material is based on, for example, biocompatibility, biodegradability, mechanical properties, cosmetic production and interphase properties. Use of the bone growth-inducing factors will define the appropriate formulation. Potential matrices for bone growth-inducing factors may be biodegradable and chemically defined, such as, but not limited to, calcium sulfate, tricalcium phosphate, hydroxyapatite, polyacetic acid, polyanhydrides; biodegradable and biologically well-defined, such as bone or skin collagen, other pure proteins or extracellular matrix components; non-biodegradable and chemically defined, such as sintered hydroxyapatite, bioglass, aluminate, or other ceramics; or combinations of any of the aforementioned material types, such as polyacetic acid and hydroxyapatite or collagen and tricalcium phosphate. The bioceramics can also be changed in composition, such as in calcium-aluminate-phosphate and manufacturing to change, for example, pore size, particle size, particle shape and biodegradability.

The dosage regimen will be determined by the attending physician which depends on various factors that modify the effect of such growth factors, e.g. amount of bone mass desired to be formed, site of bone injury, condition of the injured bone, patient's age, gender and diet, severity of infections, time of administration and other clinical factors. The dose may vary with the types of matrix used for reconstruction and the composition of the BMPs. Addition of other known growth factors, such as IGF 1 (insulin-like growth factor 1), to the final preparation may also affect the dose. In general, the dosage regimen should be in the range of approximately 10 to 10<6> nanograms of protein per gram of bone weight which is desirable. Progress can be read by periodic determination of bone growth and/or repair, e.g. by X-ray. Such therapeutic preparations are also now valuable for veterinary medical use due to the lack of species specificity in bone-inducing factors. In particular, livestock and thoroughbred horses in addition to humans are desirable patients for such treatment with bone-inducing factors in this invention.

The following examples illustrate the use of this invention in the extraction and characterization of bovine proteins and their use in extracting human proteins, obtaining these human proteins and expressing these proteins via recombinant techniques.

The examples that follow below also include other bone-inducing factors than those covered by the invention.

Example I

Isolation of bovine bone-inducing factor.

Ground bovine bone powder (20-120 mesh, Helitrex) is prepared according to the method of M.E. Urist et al., Proe. Nati. Acad. Sei USA, 70:3511 (1973) with the elimination of some of the extraction steps as identified below. Ten kg of ground powder is demineralized in successive changes of 0.6N HC1 at 4°C over a 48-hour period with vigorous stirring. The resulting suspension is extracted for 16 hours at 4°C with 50 liters of 2M CaCl 2 and 10M ethylenediaminetetraacetic acid (EDTA), followed by extraction for 4 hours in 50 liters of 0.5M EDTA. The residue is washed three times with distilled water before being resuspended in 20 liters of 4M guanidine hydrochloride

[GuCl] , 20 mM Tris (pH 7.4), ImM N-ethylmaleimide, ImM iodo-acetamide, ImM phenylmethylsulfonyl-fluorine as described in Clin. Orthop. Rel. Res., 171: 213 (1982). After 16 to 20 hours, the supernatant is removed and replaced with a new 10 liters of GuCl buffer. The residue was extracted for another 24 hours.

The crude GuCl extracts are pooled, concentrated approximately 20-fold on a Pellicon apparatus with a 10,000 molecular weight cut-off membrane, and then dialyzed into 50mM Tris, 0.IM NaCl, 6M urea (pH7.2), which is the starting buffer for the first column. After long dialysis, the protein is applied to a 4 liter DEAE cellulose column and the unbound fractions are collected.

The unbound fractions are concentrated and realized against 50mM NaAc, 50mM NaCl (pE 4.6) in 6M urea. The unbound fractions were applied to a carboxymethyl cellulose column. Protein that does not bind to the column is removed by extensive washing with the starting buffer, and the material containing bone-inducing factor is desorbed from the column with 50mM NaAc, 0.25mM NaCl, 6M urea (pH 4.6). The protein from this step is concentrated 20- to 40-fold, then diluted 5-fold with 80mM KPO4, 6M urea (pH6.0). The pH of this solution is adjusted to 6.0 with 500mM K2HPO4. The sample is applied to a hydroxyapatite column (LKB) equilibrated in 80mM KPO4, 6M urea (pH6.0) and all unbound proteins are removed by washing the column with the same buffer. Bone-inducing factor activity is eluted with 100 mM KPO 4 (pH 7.4) and 6 M urea.

The protein is concentrated approximately 10 times, and solid NaCl is added to a final concentration of 0.15M. This material is applied to a heaprin-Sepharose column equilibrated in 50m KP04, 150mM NaCl, 6M urea (pH7.4). After extensive washing of the column with the starting buffer, a protein with bone-inducing factor activity is eluted with 50mM KPO4, 700mM NaCl, 6M urea (pH7.4). This fraction is concentrated to a minimal volume, and 0.4ml aliquots are applied to Superose 6 and Superose 12 columns connected in series, equilibrated with 4M GuCl, 20mM Tris (pH7.2) and the columns formed at an elution rate of 0.25ml/mln. The protein demonstrating bone-inducing factor activity has a relative migration corresponding to approximately 30,000 daltons of protein.

The above fractions are pooled, dialyzed against 50mM NaAc, 6M urea (pH4.6), and applied to a Pharmacia MonoS HR column. The column is developed with a gradient of 1.OM NaCl, 50mM NaAc, 6M urea (pH4.6). The active fractions are combined and brought to pH 3.0 with 10% trifluoroacetic acid (TFA). The material is applied to a 0.46 x 25 cm Vydac C4 column in 0.1$ TFA and the column developed with a gradient of 90* acetonitrile, 0.1* TFA (31.5* acetonitrile, 0.1* TFA to 49.5* acetonitrile, 0.1* TFA for 60 minutes at lml per minute). Active material is eluted at about 40-44* acetonitrile. Aliquots of the appropriate fractions are iodinated using one of the following methods: P.J. McConahey et al., Int. Arch. Allergy, 29:185-189 (1966); A. E. Bolton et al, Biochem J., 133-529 (1973); and D.F. Bowen-Pope,

<J. Biol. Chem. , 237:5161 (1982). The iodinated proteins present in these fractions are analyzed by SDS gel electrophoresis and urea Triton X 100 isoelectric focusing. At this stage, the bone-inducing factor is calculated to be approximately 10-50* pure.

Example II

Characterization of bovine bone-inducing factor.

A. Molecular weight.

Approximately 20 µg of protein from Example I is freeze-dried and redissolved in 1X SDS sample buffer. After 15 minutes of heating at 37°C, the sample is applied to a 15* SDS polyacrylamide gel and then electrophoresed with cooling. The molecular weight is determined relative to colored molecular weight standards (Bethesda Research Labs). Immediately after it is finished, the gel file containing the bone-inducing factor is cut into 0.3cm pieces. Each part is mashed and 1.4 ml of 0.1* SDS is added. The samples are gently shaken overnight at room temperature to elute the protein. Each gel piece is desalted to prevent interference in the biological analysis. The supernatant from each sample is acidified to pH 3.0 with 10* TFA, filtered through 0.45 micron membrane and applied to a 0.46cm x 5cm C4 Vydac column developed with a gradient of 0.1* TFA to 0.1* TFA, 90* CH3CN. The appropriate bone-inducing-factor-containing fractions are pooled and reconstituted with 20 mg of rat matrix. In this gel system, the majority of the bone-inducing factor fractions have a mobility as a protein having a molecular weight of approximately 28,000 - 30,000 daltons.

B. Isoelectric focusing.

The isoelectric point of bone-inducing factor activity is determined in a denaturing isoelectric focusing system. The Triton X100 urea gel system (Hoeffer Scientific) is modified as follows: 1) 40* of the ampholytes used are Servalyte 3/10; 60* is Servalyte 7-9. 2) The catholyte used is 40mM NaOH. Approximately 20 µg of protein from Example I is freeze-dried, dissolved in sample buffer and applied to the isoelectrofocusing gel. The gel is run at 20 watts, 10°C for approximately 3 hours. Upon completion, the file containing bone-inducing factor is cut into 0.5 cm pieces. Each piece is crushed in 1.0ml 6M urea, 5mM Tris (pH 7.8) and the samples are shaken at room temperature. The samples are acidified, filtered, desalted and analyzed as described above. The major part of the activity as determined in the analysis described in Example III migrates in a manner indicative of a pl of 8.8 - 9.2.

C. Subunit characterization.

The subunit composition of bone-inducing factor is also determined. Pure bone-inducing factor is isolated from a preparative 15* SDS gel as described above. A portion of the sample is then reduced with 5mM DTT in sample buffer and electrophoresed again on a 15* SDS gel. The approximately 30kd protein gives two major bands at approximately 20kd and 18kd, as well as a minor band at 30kd. The width of the two bands indicates heterogeneity likely due to glycosylation, other post-translational modifications, proteolytic degradation or carbamylation.

Example III

Biological activity of the bone-inducing factor.

A rat ossification assay according to the general methods of Sampath and Reddi, Proe. Nati. Acad. Pollock. U.S.A., 80:6591-6595 (1983) is used to determine the bone-forming activity of the bovine bone-inducing factor of this invention as obtained in Example I. This assay can also be used to determine bone-inducing factors in other species. The ethanol precipitation step is replaced by dialysis of the fraction to be analyzed against water. The solution or suspension is then redissolved in a volatile solvent, e.g. 0.1 - 0.2* TFA, and the resulting solution is then added to 20mg rat matrix. This substance is frozen and lyophilized and the resulting powder is placed into #5 gelatin capsules. The capsules are implanted subcutaneously in the thoracoabdominal area in 21-49 day old male Long Evans rats. The implanted substances are removed after 7-10 days. Half of each substance implanted is used in alkaline phosphatase assays [see A.H. Reddi et al., Proe. Nati Acad Sei., 69:1601 (1972)] and half are fixed and prepared for histological analysis. Lum glycol methacrylate sections are usually stained with Von Kossa and acid fuschin to detect new bone mineral. Alkaline phosphatase, an enzyme formed by chondroblasts and osteoblasts during matrix formation, is also measured. New bone and bone formation are often associated with alkaline phosphatase levels. Table I below illustrates the dose response in rat matrix samples including a control not treated with inducing factor.

Bones or bones that are formed are physically limited to the area occupied by the matrix. Samples are also analyzed by SDS gel electrophoresis and isoelectric focusing as described above, followed by autoradiography. The analyzes show a correlation between activity and protein binding at 28 - 30kd and at pl 9.0. An extinction coefficient of 1 OD/mg-cm is used as an estimate for protein and for an approximate determination of the purity of the bone-inducing factor in a particular fraction. In the in vivo rat bone formation assays done at dilutions as described above, the protein is active in vivo at 10 to 200 ng protein/gram bone to probably higher than ear protein/gram bone.

Example IV

Bovine bone-inducing factor protein composition.

The protein composition in example IIA with molecular weight 28 - 30 kd is reduced as described in example IIC and cut with trypsin. Eight tryptic fragments are isolated using the standard method which have the following amino acid sequences:

Fragment 1: AAFLGDIALDEEDLG

Fragment 2: AFQVQQAADL

Fragment 3: NYQDMVVEG

Fragment 4: STPAQDVSR

Fragment 5: N Q E A L R

Fragment 6: LSEPDPSHTLEE

Fragment 7: F D A Y Y

Fragment 8:LKPSN?ATIQSIVE

A less pure protein preparation from bovine bone is prepared according to a purification scheme similar to that described in Example I. The purification is somewhat different from that previously described because it omits the DE-52 column, the CM-cellulose column and the mono S column, likewise a reversal in the order of hydroxylapatite and heparing sepharose columns. The concentrated crude 4 M extract is brought to 85* final ethanol concentration at 4 degrees. The mixture is then centrifuged, and the precipitate is then redissolved in 50 mM Tris, 0.15 M NaCl, 6.0 M urea. This material is then fractionated on Heparin Sepharose as described. The Heparin-bound material is fractionated on hydroxyapatite as described. The active fractions are pooled, concentrated and fractionated by high-resolution gel filtration (TSK 30000 in 6 M guanidine chloride, 50 mM Tris, pH 7.2). The active fractions are pooled, dialyzed against 0.1* TFA, and then fractionated on a C4 Vydac reverse-phase column as described. The preparation is reduced and electrophoresed on an acrylamide gel. The protein corresponding to the 18K band is eluted and cut with trypsin. Tryptic fragments are isolated and have the following amino acid sequences:

Fragment 9: SLKPSNHATIQS7V

Fragment 10: SFDAYYCS7A

Fragment 11: VYPNMTVESCA

Fragment 12:VDFADI?W

The tryptic fragments 7 and 8 are substantially similar to fragments 10 and 9, respectively.

A. bBMP-1

Probes consisting of oligonucleotide pools (or unique oligonucleotides) are designated according to the method of R. Lathe, J. Mol. Biol., 183 (1):1-12 (1985) and prepared on an automated DNA synthesis machine. A probe consists of a relatively long (32 nucleotide) "guessmer" [see J.J. Toole et al., Nature, 312:342-347 (1984)] with the following nucleotide sequence.

Because the genetic code is degenerate (more than one codon can code for the same amino acid), the number of nucleotides in a probe pool is reduced based on the frequency with which the codon is used in eukaryotes, the relative stability of the G:T base pairs, and the relative infrequency of the dinucleotide CpG in eukaryotic coding sequences [see Toole et al., supra]. The second probe set consists of shorter oligonucleotides (lengths of 17 nucleotides) that contain all possible sequences from which the amino acids can be coded. The second probe set has the following sequence:

(a) A [A/G] [A/G] TC [T/C] TC [T/C] TC [A/G] TC [T/C] AA

(b) A [A/G] [A/G] TC [T/C] TC [T/C] TC [A/G] TCNAG

The nucleotides in parentheses stand for alternatives. "N" means either A, T, C or G.

In both cases, the regions of amino acid sequences used for probe generation are selected by avoiding codons that are highly degenerate where possible. Oligonucleotides are prepared on an automated DNA synthesis machine; the probes are then radioactively labeled with polynucleotide kinase and <32p_,>

ATP.

These two probe sets are used to screen a bovine genomic recombinant library. The library is prepared as follows: Bovine liver DNA is partially cut with the restriction endonuclease enzyme Sau 3A and sedimented through a sucrose gradient. Size fractionated DNA in the range of 15-30kb is then ligated into the bacteriophage Barn HI vector EMBL3

[Frischauf et al, J. Mol. Biol., 170:827-842 (1983)]. The library is then plated with 8000 recombinants per dish. Duplicate nitrocellulose replicates of the plaques are performed and amplified according to a modification of the method of Woo et al, Proe. Nati. Acad. Pollock. USA, 75:3688-91 (1978).

The 32-mer probe is kinase treated with <32>P-gamma-ATP and hybridized to a filter set 5X SSC, 0.1* SDS, 5X Denhardt's, 100 µg/ml salmon sperm DNA at 45 degrees C and washed with 5X SSC, 0.1* SDS at 45 degrees C. The 17-mer probes are kinased and hybridized to the second filter set in 3M tetramethylammonium chloride (TMAC), 0 . IM sodium phosphate pH6.5, ImM EDTA, 5X Denhardt's, 0.6* SDS, 100ug/ml salmon sperm DNA at 48 degrees C, and washed in 3M TMAC, 50mM Tris pH8.0 at 50 degrees C. Under these conditions, detection of mismatched nucleotides to the 17-mer probe pool [see Wood et al, Proe. Nati. Acad. Sei, U.S.A., 82:1585-1588 (1985)]. 400,000 recombinants are screened using this method and a positive duplicate is plaque purified. DNA is isolated from a dish lysate from this recombinant bacteriophage designated lambda bP-50. bP-50 was deposited December 16, 1986 at the American Type Culture Collection, 12301 Parklawn Dr., Rockville, MD, USA (hereafter "ATCC") under accession number 40295. This and the others deposited herein uphold the Budapest Treaty as regards "International Recognition of the Deposit of Microorganisms for the Purpose of Patent Procedure and Regulations thereunder". This bp-50 clone encodes at least part of the bovine bone growth factor designated bBMP-1.

The oligonucleotide hybridization region of this bBMP-1 clone is localized to an approximately 800bp Eco RI fragment that is subcloned into M13 and sequenced using standard techniques. The partial DNA sequence and the deduced amino acid sequence of lambda bP-50 are shown below in Table II. Amino acid sequences corresponding to the tryptic fragments isolated from bovine bone 28 to 30kd material are underlined in Table II. The first underlined part of the sequence corresponds to tryptic fragment 1 above from which the oligonucleotide probes are constructed. The second underlined part corresponds to tryptic fragment 2 above. The putative amino acid sequences indicate that tryptic fragment 2 is preceded by a basic residue (R) which is expected according to the specificity of trypsin. The nucleic acid sequence predicted by the linked CT at nucleotide positions #292-293 in Table II is believed to be an intron (non-coding sequence) based on the presence of a consensus acceptor sequence (ie, a pyrimidine-rich region, TCTCTCTCC, followed by AG) and lack of the basic residue in the appropriate position of the derived amino acid sequence. This bBMP-1 genome sequence appears in Table II. The presumptive bBMP-1 peptide sequence from this genomic clone is 37 amino acids in length and is encoded from the DNA sequence from nucleotide #294 through #404 in Table II.

b. bBMP-2

Two probes consisting of oligonucleotide pools are constructed on the basis of the amino acid sequence of fragment 3 and prepared on an automated DNA synthesis machine as described above.

Probe #1: ACNACCAT [A/G] T C [T/C] T G [A/G] A T Probe #2: C A [A/G] G A [T/C] ATGGTNGTNGA

These probes are radiolabeled and are used to screen the bovine genomic library constructed as described in Part A with the exception of the vector being lambda Jl Bam Hl arms [Mullins et al Nature 308: 856-858 (1984).] The radiolabeled 17- mer probe #1 is hybridized to the filter sets according to the procedure for the 17-mer probe described in Sec

A.

400,000 recombinants are screened by the method described above in Part A. A positive duplicate is plaque-purified and the DNA is isolated from a plate lysate of the recombinant bacteriophage designated lambda bP-21. Bacteriophage bP-21 was deposited with ATCC under accession number ATCC 40310 on March 6, 1987. The bP-21 clone encodes bovine growth factor designated bBMP-2.

The oligonucleotide hybridizing region of this bBMP-2 clone is localized to an approximately 1.2 kb Sac I restriction fragment that is subcloned into M13 and sequenced using standard techniques. The partial DNA sequence and DNA-derived amino acid sequence of this Sac I fragment and the adjacent Hind III-Sac I restriction fragment of bP-21 are shown below in Table III. The bBMP-2 peptide sequence from this clone is 129 amino acids in length and is encoded from the DNA sequence from nucleotide #1 through nucleotide #387. The amino acid sequence corresponding to the tryptic fragment isolated from bovine bone 28 to 30kd material is underlined in Table III. The underlined part of the sequence corresponds to tryptic fragment 3 above from which the oligonucleotide probes for bBMP-2 are constructed. The putative amino acid sequence indicates that the tryptic fragment 3 is preceded by a basic residue (K) which is expected on the basis of the specificity of trypsin. The arginine residue encoded by the CGT triplet is thought to be the carboxy-termini of the protein based on the presence of an adjacent stop codon (TAG).

C. bBMP-3

Probes consisting of oligonucleotide pools are constructed based on the amino acid sequence of tryptic fragment 9 (probe #3), 10 (probe #2), and 11 (probe #1), and prepared on an automated DNA synthesis machine.

Probe #1: ACNGTCAT [A/G] T T N G G [A/G] T A

Probe #2: C A [A/G] T A [A/G] T A N G C [A/G] T C [A/G] A A

Probe #3: T G [A/G/T] ATNGTNGC [A/G] T G [A/G] T T

A recombinant bovine genomic library constructed in EMBL3 is screened using TMAC hybridization procedures described above in Part A. 400,000 recombinants are screened in duplicate with probe #1 which has been labeled with <32>P. All recombinants that hybridize to this probe are replated to open secondary. Triplicate nitrocellulose replicates are made from the secondary dishes, and amplified as described. The three filter sets are hybridized to probe #1, #2 and #3, again under TMAC conditions. One clone, lambda bP-819, hybridizes to all three probes and is plaque purified and the DNA is isolated from a dish lysate. Bacteriophage lambda bP-819 was deposited with ATCC on June 16, 1987 under accession number 40344. This bP-819 clone encoded the bovine growth factor designated bBMP-3.

The region of bP-819 that hybridizes to probe #2 was located and sequenced. The partial DNA and the deduced amino acid sequence of this region are shown in Table IVA. The amino acid sequences corresponding to tryptic fragments 10 and 12 are underlined. The first underlined sequence corresponds to fragment 12 while the second corresponds to fragment 10. This region a^ bP-819, which hybridizes to probe #2 encodes at least 111 amino acids. This amino acid sequence is encoded from the DNA sequence from nucleotide #414 through #746.

The bP-819 region that hybridizes to probe #1 and #3 is located and sequenced. The partial DNA and the deduced amino acid sequence of this region are shown in Table IVB. The amino acid sequences corresponding to the tryptic fragments 9 and 11 are underlined. The first underlined sequence corresponds to fragment 9, while the second underlined sequence corresponds to fragment 11. The peptide sequence of this region of bP-819 that hybridizes to probe #1 and #3 has a length of 64 amino acids encoded by nucleotide #305 even #493 in Table IVB. The arginine residue encoded by the AGA triplet is thought to be the carboxy-termini of the protein based on the presence of a stop codon (TAA) adjacent to it. The nucleic acid sequence preceding the linked TC

(positions 305-306) is believed to be an intron (non-coding sequence) based on the presence of a consensus acceptor sequence (ie, a pyrimidine-rich region, TTCTCCCTTTCGTTCCT, followed by AG) and the presence of a stop in instead of a basic residue at the appropriate position of the derived amino acid sequence.

bBMP-3 is therefore characterized by the DNA and amino acid sequence in Table IV A and Table IV B. The peptide sequence of this clone is 175 amino acids in length and is encoded by the DNA sequence from nucleotide #414 through nucleotide #746 in Table IV A and nucleotide #305 through nucleotide #493 in Table IV B.

Example V

Human bone-inducing factors.

A. hBMP-1

Because bovine and human bone growth factor genes are believed to be significantly homologous, the bovine bBMP-1 DNA sequence in Table II (or portions thereof) is used as a probe to screen a human genomic library. The 800bp EcoEI fragment of the bovine genome clone is labeled with <32>P by nick translation. A humanized genomic library (Toole et al., supra) is plated on 20 dishes with 40,000 recombinants per dish. Duplicate nitrocellulose filter copies are made from each dish and hybridized to nick-translated probe in 5 X SSC, 5 X Denhardfs, 100 pg/ml denatured salmon sperm DNA, 0.1* SDS (standard hybridization solutions) at 50 degrees C for approximately 14 hours. The filters are then washed in 1 X SSC, 0.1* SDS at 50 degrees and exposed to autoradiography. Five positive duplicates are isolated and plaque purified. DNA is provided from a dish lysate to one of these recombinant bacteriophages, designated LP-H1. LP-H1 was deposited at ATCC on March 6, 1987 under accession number 40311. This clone encodes at least a portion of the human genome bone growth promoter designated hBMP-1. The hybridization region of LP-H1 is localized to a 2.5 kb XbaI/HindIII restriction fragment.

The partial DNA sequence and deduced amino acid sequence of lambda LP-E1 is shown below in Table V. The peptide sequence from this clone is 37 amino acids in length and is encoded by the DNA sequence from nucleotide #3440 through nucleotide #3550. The coding sequence in Table V is flanked by about 28 nucleotides (a putative 5' non-coding sequence) and about 19 nucleotides (a putative 3' non-coding sequence). A comparison of the bBMP-1 sequence in Table II with the hBMP-1 genomic sequence in Table V indicates the significant homology between the two.

On the basis that the size of the coding region ends and the positions of the non-coding regions are generally conserved in homologous genes from different species, the locations of the coding and non-coding regions of the bone-inducing factor genes can be identified. Regions of homology between the genes of the two species, flanked by RNA processing signals at homologous sites, indicate a coding region.

A probe specific for the human coding sequence given in Table V is used to identify a human cell line or tissue that produces bone-inducing factor. The sample is made according to the following procedure. Two oligonucleotides having the following sequences:

(a) GGGAATTCTGCCTTTCTTGGGGACATTGCCCTGGACGAAGAGGACCTGAG

(b) CGGGATCCGTCTGAGATCCACAGCCTGCTGTACCTGGAAGGCCCTCAGG

prepared on an automated synthesizer, fused, extended using the Klenow fragment of E. coil DNA polymerase I, cut with restriction enzymes Eco RI and Bam HI, and inserted into an M13 vector. A single-stranded <32>P-labeled probe is then templated from this subcloned using standard techniques. Polyadenylated RNA<*>s from various cell and tissue sources are then electrophoresed on formaldehyde-agarose gels and transferred to nitrocellulose using the method of Toole et al., supra. The probe is then hybridized to nitrocellulose blotted in 50* formamide, 5 X SSC, 0.1* SDS, 40 mM sodium phosphate pH 6.5, 100 µg/ml denatured salmon sperm DNA, and 5 mM vanadyl ribonucleosides at 42°C overnight and washed at 65°C in 0.2X SSC, 0.1* SDS. By autoradiography, the file containing RNA from the numan osteosarcoma cell line U-2 OS contains hybridizing bands corresponding to RNA species of approximately 4.3 and 3.0 kb.

cDNA is prepared from U-2 OS polyadenylated RNA and cloned into lambda gtlO using established techniques (Toole et

al., above). 20,000 recombinants from this library are plated on each of 50 dishes. Duplicate nitrocellulose copies are made of the dishes. The above-described oligonucleotides are kinase-treated with ^ ?- sexDma.- k1? and hybridized to the two replicate sets at 55°C in standard hybridization solution overnight. The filters were then washed in 1 X SSC, 0.1* SDS at 55°C and subjected to autoradiography. A positive duplicate, designated lambda U20S-1, is plaque purified. Lambda U20S-1 was deposited with ATCC on June 16, 1987 under accession number 40343.

The entire nucleotide sequence and deduced amino acid sequence of the insert of lambda U20S-1 is given in Table VI. This cDNA clone encodes a Met followed by a hydrophobic leader sequence characteristic of a secreted protein, and contains a stop codon at nucleotide positions 2226-2228. This clone contains an open reading frame of 2190bp, which encodes a protein of 730 amino acids with a molecular weight of 83kd based on this amino acid sequence. The clone contains a sequence identical to the coding region given in Table V. This protein is believed to represent a primary translation product that is cleaved upon secretion to form hBMP-1 protein. This clone is therefore a cDNA for hBMP-1 which corresponds to the human gene fragment in the genomic hBMP-1 sequence lambda LP-H1. Amino acids #550 to #590 of BMP-1 are homologous to epidermal growth factor and the "growth factor" domains of protein C, factor X, and factor

IX.

B. hBMP-2; Class I and II

The HindiII-SacI bovine genome bBMP-2 fragment described in Example IV B. is subcloned into the M13 vector. A <32>P-labeled single-stranded DNA probe is made from a template preparation of this subclone. This probe is used to screen polyadenylated RNAs from various cell and tissue sources as described above in part A. A hybridized band corresponding to an mRNA form of approximately 3.8 kb is detected in the file containing RNA from the human cell line U- 2 OS. The HindIII-SacI fragment is labeled with <32>P by nick translation and used to screen the nitrocellulose filter copies of the above-described U-2 OS cDNA library by hybridization in standard hybridization buffer at 65° overnight followed by washing 1 in X SSC , 0.1* SDS at 65°C. Twelve positive duplicate clones are picked and replated to obtain secondaries. Duplicate nitrocellulose probes are made from the secondary dishes and both sets hybridized to the bovine genome probe as the primary screening was performed. A filter set was then washed in 1 X SSC, 0.1* SDS; and the second 0.1 X SSC, 0.1* SDS at 65°C.

Two classes of hBMP-2 cDNA clones are evident based on strong (4 recombinants) or weak (7 recombinants) hybridization signals under the more stringent washing conditions (0.1 X SSC, 0.1* SDS). All 11 recombinant bacteriophages are plaque-purified, small-scale DNA preparation from the dish lysates of each, and the inserts subcloned into pSP65 and into M13 for sequence analysis. Sequence analyzes of the strongly hybridizing clones designated hBMP-2 class I (also known as BMP-2) indicate that they have extensive sequence homology with the sequence given in Table III. These clones are therefore cDNAs that encode the human equivalent of the protein encoded by the bBMP-2 gene, the partial sequence of which is given in Table III. Sequence analyzes of the weakly hybridizing recombinants designated hBMP-2 class II (also known as BMP-4) indicate that they are also quite homologous to the sequence given in Table III at the 3' end of their coding regions, but less so in the more 5 ' areas. They thus code for a human protein with a similar, but not identical, structure to the above.

Full-length hBMP-2 class I cDNA clones are obtained as follows. The 1.5 kb insert of one of the class II subclones (II-10-1) is isolated and radioactively labeled by nick translation. A set of the nitrocellulose copies of the U-2 OS cDNA library screened above (50 filters, corresponding to 1,000,000 recombinant bacteriophage) is rehybridized with this probe under stringent conditions (hybridization at 65<4> in standard hybridization buffer ; wash at 65° in 0.2 X SSC, 0.1* SDS). All recombinants that hybridize to the bovine genome probe that do not hybridize to class

The II probe is picked and plaque-purified (10 recombinants). Sticks are made from the bowls and bacteriophage DNA preparations are made on a small scale. After subcloning into M13, sequence analysis indicates that 4 of these represent clones overlapping the original class I clone. One of these, lambda U20S-39, contains a deposit of approximately 1.5 kb and was deposited at ATCC on June 16, 1987 under accession number 40345. The partial DNA sequence (prepared from lambda U20S-39 and several other hBMP-2 class I cDNA recombinants) and derived amino acid sequence are shown below in Table VII. Lambda U20S-39 is expected to contain the entire nucleotide sequence necessary to encode the entire human counterpart of protein BMP-2 class II which is encoded from the bovine gene segment whose partial sequence is presented in Table III. This human cDNA hBMP-2 class II contains an open reading frame of 1188 bp, which codes for a protein of 396 amino acids. This 396 amino acid protein has a molecular weight of 45kd based on this amino acid sequence. It is believed that this sequence represents the primary translation product. The protein is preceded by a 5' untranslated region of 342 bp with stop codons in all frames. The 13 bp region that precedes this 5' untranslated region represents a linker that is used in the cDNA cloning procedures. Full-length hBMP-2 class II human cDNA clones are provided as follows. The 200 bp EcoRI-SacI fragment from the 5' end of class II recombinant II-10-1 is isolated from its plasmid subclone, tagged by nick translation, and hybridized to a set of duplicate nitrocellulose copies of the U-2 OS cDNA library ( 251 filters/set; representing 500,000 recombinants). Hybridization and washing are performed under stringent conditions as described above. 16 positive duplicates are picked and replated to obtain secondaries. The nitrocellulose filter copies of the secondary dishes are prepared and hybridized to an oligonucleotide that was prepared to correspond to the sequence of II-10-1 and has the following sequence:

Hybridization is done in standard hybridization buffer at 50°C with washing at 50<*> in 1 X SSC, 0.1* SDS. 14 recombinant bacteriophages that hybridize to this oligonucleotide are plaque-purified. Sticks from the dishes are made and bacteriophage DNA preparations made on a small scale. After subcloning 3 of these into M13, sequence analysis indicates that they represent clones overlapping the original class II clone. One of these, lambda U20S-3, was deposited with ATCC under accession number 40342, June 16, 1987. U20S-3 contains a deposit of approximately 1.8 kb. The partial DNA sequence and deduced amino acid sequence of U20S-3 is shown below in Table VIII. This clone is expected to contain the entire nucleotide sequence necessary for it to encode the entire human BMP-2 class II protein. This cDNA contains an open reading frame of 1224 bp, encoding a protein of 408 amino acids, which is preceded by a 5' untranslated region of 394 bp with stop codons in all frames, and contains a 3' untranslated region of 308 bp after the stop codon in the correct reading frame. The 8 bp region preceding the 5' untranslated region represents a linker used in the cDNA cloning procedure. This protein of 408 amino acids has a molecular weight of 47 kd and is believed to represent the primary translation product.

The sequences of BMP-2 classes I and II, and BMP-3 as shown in

tables III, IV, VII and VIII have significant homolog! to the beta (B) and beta (A) subunits of the inhibins. The inhibins are a family of hormones that are now being investigated for use in contraception. See A.J. Mason et al., Nature, 318:659-663

(1985). They are also to a lesser degree homologous to Muller i an inhibitory substance (MIS), which is a testicular glycoprotein belonging to regression of the Mullerian duct during development of the male embryo and transforming growth factor-beta (TGF-b ) which can inhibit or stimulate the growth of cells or cause them to differentiate. The sequence in Table VII which codes for hBMP-2 class II has significant homolog! to the Drosophila decapentaplegic (DPP-C) locus transcript. See J. Massague, Cell, 49:437-438 (1987); R. W. Padgett et al., Nature, 325:81-84 (1987); R. L. Cate et al, Cell 45: 685-698

(1986). It is therefore possible that BMP-2 class II is the human protein homolog made from this transcript from this developmental mutant locus.

C. BMP-3

Because bovine and human bone growth factor genes are believed to be significantly homologous, oligonucleotide probes shown to hybridize to bovine DNA sequences from Tables IV.A and IV.B are used to screen a human genomic library. A human genomic library (Toole et al., supra) is screened using these probes, and presumptive positives are isolated and the DNA sequence provided as described above. Evidence that this recombinant encodes part of the human bone-inducing factor molecule lies in the bovine/human protein and gene structure homologies.

When a recombinant bacteriophage containing DNA encoding a part of the human BMP-3 molecule is obtained in the human coding sequence is used as a probe as described in Example V (A) to identify a human cell line or tissue that produces BMP -3. mRNA is selected by oligo (dT) cellulose chromatography and cDNA is prepared and cloned into lambda gtlO using established techniques (Toole et al., supra).

Alternatively, the entire gene encoding this human bone-inducing factor can be identified and provided in multiple recombinant clones if necessary. Multiple recombinants containing multiple 3' or 5' regions of this human bone-inducing factor gene can be provided by identifying unique DNA sequences at the end(s) of the original clone and then using these probes to re-screen the human genome library. The gene can then be reassembled into a single plasmid using standard molecular biology techniques and amplified in bacteria. The entire human BMP-3 factor gene can then be transferred into an appropriate expression vector. The expression vector containing the gene is then transfected into a mammalian cell, e.g. monkey COS cells, where the human gene is transcribed and the RNA spliced correctly. Media from the transfected cells are assayed for bone-inducing factor activity as described herein as an indication that the genes are complete. mRNA is obtained from these cells and cDNA is prepared from this mRNA source and cloned. The method described above can similarly be used to isolate other species of bone-inducing factor of interest using the bovine bone-inducing factor and/or the human bone-inducing factor as a probe source. This other kind of bone-inducing factor turns out to be able to be used for, among other things, the repair of fractures.

Example VI

Expression of bone-inducing factor

To produce bovine, human or other mammalian bone-inducing factors, the DNA encoding it is transferred into an appropriate expression vector and introduced into mammalian cells using conventional genetic engineering techniques.

One skilled in the art can construct mammalian expression vectors using the sequence in Tables II-VIII or other modified sequences and known vectors, such as pCD (Okayama et al., Mol. Cell Biol., 2:161-170 (1982) ]. and pJL3, pJL4 [Gough et al., EMBO J., 4:645-653 (1985 )]. Transformation of these vectors into appropriate host cells can result in the expression of bone growth-inducing factors. One skilled in the art can manipulate the sequences in Tables II - VIII by eliminating or replacing mammalian regulatory sequences flanking the coding sequence with bacterial sequences to form bacterial vectors for intracellular or extracellular expression by bacterial cells. For example, the coding sequences can be further manipulated (eg ligated to other known linkers or modified by deletion of non-coding sequences therefrom or alteration of nucleotides therein using other known techniques). the coding sequence of the modified bone-inducing factor could then be inserted into a known bacterial vector using methods such as those described in T. Taniguchi et al., Proe. Nat Acad. Pollock. USA, 77:7230-5233 (1980). This bacterial vector could then be transformed into bacterial host cells and thus bone-inducing factor could be expressed. A strategy for the production of extracellular bone-inducing factor in bacterial cells, see e.g. European Patent Application EPA 177,343.

Similar manipulations can be performed in the construction of an insect vector [see e.g. methods described in published European patent application 155,476] for expression in insect cells. A yeast vector can also be constructed using yeast regulatory sequences for intracellular or extracellular expression of the factors of this invention by yeast cells. [See e.g. methods described in published PCG application W086/00639 and European patent application EPA 123,289].

A method for producing large amounts of a bone growth stimulating factor in the invention from mammalian cells involves the construction of cells containing multiple copies of the heterologous bone inducing factor gene. The heterologous gene can be linked to an amplifiable marker,> e.g. dihydrofolate reductase (DHFR) gene upon which cells containing increasing gene copies can be selected for propagation in increasing concentrations of methotrexate (MTX) according to the method of Kaufman and Sharp, J. Mol. Biol., 159:601-629 (1982). This method can be used for a large number of different cell types.

For example, a plasmid containing a DNA sequence for a bone-inducing factor of the invention that is in operative association with other plasmid sequences enabling expression thereof and DHFR expression plasmid pAdA26SV(A)3 [Kaufman and Sharp, Mol. Cell. Biol., 2:1304 (1982)] can be co-introduced into DHFR-deficient CHO cells, DUKX-BII, by calcium phosphate co-precipitation and transfection. Transformants expressing DHFR are selected for growth in alpha medium with dialyzed fetal calf serum, and then selected for amplification by growth in increasing concentrations of MTX (subsequent steps in 0.02, 0.2, 1.0 and 5uM MTX) as described in Kaufman et al ., Mol Cell Biol., 5:1750 (1983). Transformants are cloned, and biologically active bone-inducing factor expression is determined by rat bone formation assay. Bone-inducing factor expression should increase with increasing levels of MTX resistance. Similar procedures can be followed to produce other bone-inducing factors.

Alternatively, the human gene is directly expressed, as described above. Active bone-inducing factor can be produced in bacteria and yeast cells. The currently preferred expression system for biologically active recombinant human bone-inducing factor is stably transformed CHO cells.

As a specific example, to prepare the human bone-inducing factor (hBMP-1) of Example V, the insert of U20S-1 is released from the vector arms by cutting with Sal I and subcloned into mammalian expression vector pMT2CX cut with Xho I. Plasmid dNA from this subclone is transfected into COS cells by the DEAE-dextran method [Sompayrac and Danna PNAS 78:7575-7578 (1981); Luthman and Magnusson, Nucl. Acids Res. 11: 1295-1308 (1983)]. Serum-free 24 h conditioned medium is collected from the cells 40-70 h after transfection.

Mammalian expression vector pMT2 Cla-Xho (pMT2 CX) is a derivative of p91023 (b) (Wong et al., Science 228:810-815, 1985) which differs from the latter in that it contains the ampicillin resistance gene instead of tetracycline -the resistance gene and that it further contains an Xhol site for the insertion of cDNA clones. The functional elements of pMT2 Cla-Xho have been described (Kaufman, R.J., 1985, Proe. Nati. Acad. Sei. USA 82:689-693) and include adenovirus VA genes, SV40 origin of replication including the 72 bp enhancer, adenovirus major late promoter included a 5' splice site and most of the adenovirus tripartite leader sequence present on adenovirus late mRNAs, a 3' splice acceptor site, a DHFR insert, SV40 early polyadenylation site (SV40), and pBR322 sequences that is required for propagation in E. coli.

Plasmid pMT2 Cla-Xho is obtained by EcoRI digestion of pMT2-VWF, which has been deposited at the American Type Culture Collection (ATCC), Rockville, MD (USA) under accession number ATCC 67122. EcoRI digestion cuts the cDNA insert present in pMT2 -VWF out, which gives pMT2 in linear form that can be ligated and used to transform E. coli HB 101 or DH-5 to ampicillin resistance. Plasmid pMT2 DNA can be prepared using conventional methods. pMT2CX is then constructed by cutting pMT2 with Eco RV and XbaI, treating the cut DNA with Klenow fragment of DNA polymerase I, and ligation with Cla linkers (NEBiolabs, CATCGATG). This removes bases 2266 to 2421 starting from the Hind III site near the SV40 origin of replication and the enhancer sequence of pMT2. Plasmid DNA is then cut with EcoRI, blunted as above, and ligated to an EcoRI adapter,

5' PO4-AATTCCTCGAGAGCT 3<*>

3' GGAGCTCTCGA 5<*>

cut with XhoI, and ligated, yielding pMT2 Cla-Xho, which can then be used to transform E. coli into ampicillin resistance. Plasmid pMT2 Cla-Xho DNA can be prepared by conventional methods.

Example VII

Biological activity of expressed bone-inducing factor A. BMP-1

For measuring the biological activity of expressed bone-inducing factor (hBMP-1) provided in Example VI above, the factor is partially purified on a Heparin Sepharose column. 4 ml of the transfection supernatant from a 100 mm dish is concentrated approximately 10-fold by ultrafiltration on a YM 10 membrane and then dialyzed against 20 mM Tris, 0.15 M NaCl, pH 7.4 (stock buffer). This material is then applied to a 1.1 ml Heparin Sepharose column in output buffer. Unbound proteins are removed using an 8 mL wash with starting buffer, and bound proteins, including BMP-1, are desorbed using a 3-4 mL wash with 20 mM Tris, 2.0 M NaCl, pH 7.4.

The proteins bound by the Heparin column are concentrated approximately tenfold on a Centricon 10 and the salt is reduced by diafiltration with 0.1* trifluoroacetic acid. The appropriate amount of this solution is mixed with 20 mg of rat matrix and then assayed for in vivo bone or cartilage formation as recently described in Example III. A mock transfection supernatant fractionation is used as a control.

The tissues implanted containing rat matrix to which specific amounts of human BMP-1 have been added are removed from the rats after seven days and processed for histological evaluation. Representative sections from each implant are stained for the presence of new bone mineral with von Kossa and acid-fuschin, and for the presence of cartilage-specific matrix formation using toluidine blue. Types of cells present within the section are evaluated as well as the extent to which these cells exhibit phenotype.

Addition of human BMP-1 to the matrix material resulted in the formation of cartilage-like lumps 7 days after implantation. Cartilaginous-type cells can be recognized by their shape and expression of metachromatic matrix. The amount of activity observed for human BMP-1 was dependent on the amount of human BMP-1 protein added to the matrix. Table IX illustrates the dose-response relationship between human BMP-1 protein and amount of bone induction observed.

Cartilage (c) activity was scored on a scale of O(-) to 5.

Similar activity levels are seen in Heparin Sepharose fractionated COS cell extracts. Partial purification is achieved in a similar manner as described above with the exception that 6 M urea is included in all buffers. In a rat bone formation assay as described above, BMP-2 has similarly demonstrated cartilage-forming activity.

The methods described above can be used to isolate other bone-inducing factors of interest using the bovine bone-inducing factors and/or human bone-inducing factors as a probe source. Such other bone-inducing factors may be useful in, among other things, fracture repair.

The preceding descriptions describe the preferred embodiments of this invention. Multiple modifications and variations in execution thereof are expected to occur to one skilled in the art upon consideration of these descriptions. These modifications and variations are intended to be included within the attached requirements.

Claims (5)

1. Process for the production of huBMP-2 class I (BMP-2), characterized in that it includes the cultivation of a microorganism transformed with a vector containing a DNA sequence that codes for huBMP-2 class I, in a suitable culture medium, where said DNA sequence is the nucleotide sequence as follows: and isolating huBMP-2 class I from said culture medium. 2. Process for the production of huBMP-2 class II (BMP-4), characterized in that it includes the cultivation of a microorganism transformed with a vector containing a DNA sequence that codes for huBMP-2 class II in a suitable culture medium, where said DNA sequence is the nucleotide sequence as follows: and where the DNA sequence is in conjunction with an expression control sequence therefore, and isolating huBMP-2 class II from said culture medium. 3. cDNA sequence, characterized in that it is selected from the group consisting of: a. cDNA sequence that codes for huBMP-2 class I that includes the nucleotide sequence according to claim 1 or a sequence that hybridizes thereto under stringent conditions and that, upon expression, codes for a protein having the same properties as huBMP-2 class I; and b. cDNA sequence which codes for huBMP-2 class II which includes the nucleotide sequence according to claim 2 or a sequence which hybridizes thereto under stringent conditions and which, upon expression, codes for a protein which has the same properties as huBMP-2 class II. 4. Vector, characterized in that it contains and can express a DNA sequence that codes for a human bone growth-inducing protein, whereupon the DNA sequence codes for the protein selected from the group consisting of: a. a DNA sequence that codes for huBMP-2 class I which includes the nucleotide sequence according to claim 1; and b. a DNA sequence that codes for huBMP-2 class II that includes the nucleotide sequence according to claim , 2 or a sequence that hybridizes thereto under stringent conditions and that, upon expression, codes for a protein that has the same properties as huBMP-2 class II . 5. Microorganism, characterized in that it contains and can express the vector according to claim 4, and that it is selected from the group consisting of a bacterial cell, a yeast cell and a mammalian cell.