IE19950301A1 - Recombinant fibroblast growth factors - Google Patents

Abstract

ABSTRACT DNA sequences encoding human basic fibroblast growth factor (FGF) can be recombinantly expressed to obtain practical amounts of protein useful in effecting wound healing and related tissue repair.

Description

‘INANT FIBROBLAST GROWTH FACTORS Technical Field 'E 9 5 0 3 O 1 The invention relates to recombinant production of growth factors important for constructing vascular systems in healing tissues. In particular. the genes encoding bovine and human basic and acidic fibroblast growth factors (FGF) are cloned and expressed.

Background Art The process of healing when tissue is subjected to trauma, such as wounding or burns. is an extremely complex one. but it is known to be mediated by a number of protein factors. These factors are essential to the growth and differentiation of the cells which serve to replace the tissue destroyed. A number of candidate factors have been identified on the basis of the ability of extracts from various tissues. such as brain. pituitary. and hypothalamus. to stimulate the mitosis of cultured cell lines. Numerous shorthand names have been applied to the active factors in these extracts. including platelet-derived growth factor (PDGF), macrophage—derived growth factor (MDGF). epidermal growth factor (EGF), tumor angiogenesis factor (TAF), endothelial cell growth factor (ECGF). fibroblast growth factor (FGF). hypothalamus-derived growth factor (HDGF). retina-derived growth factor (RDGF). and heparin—binding growth factor (HGF). (See. Hunt. T.K.. Q Trauma (1984) ;5:s39—s49; Lobb. R.R.. et al.

Biochemistry (1984) g;:629S-6299).

Folkman. J.. et al. Science (1983) ;g;:719—725. reported that one of the processes involved in wound for example. healing. the formation of blood vessels. is profoundly affected in tumors by heparin. From this and other or not to bind to heparin is one measure of The capacity to bind differentiation between the activities in the various EGF and PDGF do not bind strongly to heparin: in fact. EGF does not bind to extracts. For example. heparin at all. The other factors above do show strong heparin binding. However, it is believed that acidic brain FGF. ECGF, RDGF.

Similarly. it is also believed_that pituitary cationic brain FGF. TAF.

(Lobb. R.R.. comparison of thirteen endothelial growth factors which and HGF-a are in fact the same factor.

FGF. protein. and HGF-B are the same et al (supra)). A summary and have been purified using heparin affinity is found in Lobb. R., et al. J Biol Chem (1986) 26l:1924—l928.

Using heparin affinity chromatography. basic fibroblast growth factors exhibiting a potent mitogenic activity for capillary endothelium have been isolated from rat chondrosarcoma (Shing. Y.. et al, supra) and from bovine cartilage (Sullivan. R.. et al. J Biol Chem (1985) ggg:2399-2403). Thomas. K.A. et al. Proc Natl Acad Sci (USA) (1984) 81:357-361. U.S. Patent 4.444.760. purified two heterogeneous forms of an acidic bovine brain fibroblast growth factor having molecular weights of l6.600 and 15.800 daltons. Gospodarowicz and collaborators showed the presence in both bovine brains lE950330 IE 950301 These factors also have similar factor isolated from human placenta (cospodarowicz. D.. et al. Biochem Biophys Res Comm (1985) 1281554-S62).

The complete sequence for basic FGF derived from bovine pituitary has been determined (Esch. P.. et al. Proc Natl Acad Sci (USA) (1985) 5;: 6507-6511).

Homogeneous preparations were obtained and showed potent mitogenic activity in in vitro assays with endothelial cells (basic FGF has an EDS0 of 60 pg/ml).

Acidic FGF has an EDSO of about 6.000 pg/ml.

An N-terminal sequence for acidic FGF derived from bovine brain tissue was determined by Bohlen. P.. et al.

EMBO J (1985) g:19S1—1956. determined the N-terminal sequences for both acidic and Gimenez-Gallego. G. et al. basic FGF prepared from human brain. and compared them to the bovine sequences (Biochem Biophys Res Comm (1986) 1352541-S48). disclosed herein.

Their results are consistent with those Also. the complete amino acid sequence of bovine brain-derived acidic FGF was determined from the isolated protein (Gimenez-Gallego.

G. et al. Science (1985) 3;g:l38S—l388: Esch, F, et al.

Biochem Biophys Res Comm (1985) l33:SS4—562). determinations are in agreement with the exception of a These two single amino acid. Subsequent to much of the work herein, the complete amino acid sequence of human acidic FGF was deduced from the gene (Jaye, M.. et al. in press). lE9503(M The FGF proteins described above and other growth factors are clearly effective in promoting the healing of tissue subjected to trauma (see. e.g.. Sporn.

M.B.. et al. Science (1983) ;;g:1329—1331: Davidson.

J.M.. J.C.B. (1985) LQQ:l2l9-1227: Thomas. K.A.. et al. Proc Natl Acad Sci (USA) (1985) g;:s4o9—6413).

Davidson. et al, et al. (supra) specifically discloses the The basic FGF native proteins have been alleged to be useful in treatment of efficacy of FGF in wound healing. myocardial infarction (Svet-Moldavsky, G.J.. et al.

Lancet (April 23. 1977) 913: 4.296.100 and 4.378.347). In addition. has been found to increase neuronal survival and neurite that this factor may also be useful in the treatment of Proc suggesting degenerative neurological disorders. such as Alzheimer's disease and Parkinson's disease.

It would. availability of these FGF proteins in large quantities therefore. be desirable to insure the and in a form free from any toxic or infectious impurities. The human form of the protein is preferred. and perhaps required. for therapeutic use. Clearly practical availability cannot be achieved from natural human sources. as obtaining a pure preparation involves an approximately 3S.000—fold purification. Even if human cadavers were otherwise a practical source. »complete purification would be crucial due to the possibility of transmitting AIDS. hepatitis. or other disease. The recent experience with Creutzfeld-Jacob Syndrome (Powell-Jackson et al. Lancet (1985) ii:244—246) precludes the use of human pituitaries as a recombinant techniques are source. Therefore. particularly suitable to apply to the production of these proteins. The invention herein provides the means whereby acidic and basic FGF can be obtained in practical quantities and in pure. uncontaminated form.

Disclosure of the Invention The invention provides the tools for synthesis and manipulation of fibroblast growth factors useful in effecting accelerated healing of wounds. bone fractures. burn tissue. damaged myocardial tissue. degenerated other trauma. neurological tissue. or Cloning and expression of the genes encoding these factors are provided by the methods and materials of the invention.

In one aspect. the invention relates to recombinant DNA sequences encoding bovine and human acidic and basic FGF (acidic bFGF. acidic hFGF. basic bFGF. and basic hFGF). In particular. these include the human or bovine genomic sequences. In other aspects. the invention relates to recombinant vectors bearing these DNA sequences. to host cells transformed with such vectors and harboring these DNA sequences. and to the recombinant proteins produced by these transformed cells. In other aspects. the invention relates to methods of producing these fibroblast growth factors using recombinant techniques.

Brief Description of the Drawings- ‘. Figures 1-4 show the DNA sequences encoding. and the deduced amino acid sequences of. acidic bFGF. acidic hFGF. basic bFGF. and basic hFGF. shows the partial sequence for the acidic bovine FGF; Figure la Figure lb shows the complete amino acid sequence of this protein. Figures 2a. 2b. and 2c show the nucleotide sequence and deduced amino acid sequence corresponding to the three exons of the human'acidic FGF gene IEQEOSQ1 contained in X phage RHAG-9.1. XHG—3, and XHAG-3. respectively.

Figure 2d shows the complete amino acid sequence and cDNA sequence encoding human acidic FGF as disclosed by Jaye et al.

Figure 5 shows the oligonucleotide probes /890.

N—termina1 sequence. and 853-856 designed from the acidic bFGF Figure 6 gives restriction maps of the inserts for genomic acidic bFGF clones XBAZ and XBA3.

Figure 7 shows the DNA sequence of the bovine acidic FGF genomic probe 250/Alul.

Figure 8 is a restriction map of the insert in acidic hFGF genomic clone XHAG-9.1.

Figure 9 shows the partially synthetic gene for acidic hFGF. "Syn-acidic hFGF".

Figure Figure bFGF cDNA clone Figure X382. 12 shows the shows basic FGF probes 1097/1098. is a restriction map of the basic results of transient expression of basic hFGF in CV-1 cells.

Figure 13 shows the used to construct basic hFGF sequence.

Figure 14 shows the hGH/FGF fusion junctions for recombinant proteins.

Figure 15 shows the hGH/FGF fusion junctions for recombinant proteins. synthetic oligonucleotides for fusions to hGH signal amino acid sequence at the several basic hFGF amino acid sequence at the several acidic hFGF Figure 16 shows DNA sequences used to encode portions of certain of the proteins of Figure 15.

Figure 17 shows a map of the human basic FGF encoding gene.

E95030? Modes of Carrying out the Invention A. The Fibroblast Growth Factors angiogenic ig_x;gg in : Purified basic bFGF has been shown to be a chicken chorioallantoic membrane assay. (Gospodarowicz. D. in Hormonal Proteins and Peptides xII:20S-230 (Academic Press). Purified acidic bFGF has been shown to be angiogenic in vivo in the same assay (Thomas, K.Ai. et al. Proc Natl Acad Sci (supra)). ‘ Bovine pituitary basic FGF has been completely sequenced and is shown in Figure 3: the human sequence determined herein from genomic and cDNA is shown in Figure 4. The primary sequences contain 146 amino acids, beginning with the proline residues numbered "1" in the figures. and are in agreement with the sequence reported for the N—terminus of the native bovine protein by Giminez-Gallego et al. Biochem Biophys Res Comm IEQ50330 . gland.

E95030? Figures 3 and 4 back to an ATG start codon in both human Translation of the upstream sequences of and bovine DNA shows that it is likely that an additional form of each protein containing the amino acids upstream of the proline shown as residue 1 in Figures 3 and 4 is also produced. There are 9 upstream including the ATG. certain that the methionine encoded by the ATG will be codons in the DNAs. It is reasonably processed when the gene is expressed in eucaryotic systems. Such processing may or may not occur when the gene is expressed recombinantly in bacterial systems.

Thus, the long form of the protein contains an additional eight amino acid pro-sequence. or a total of 154 amino acids. It has also been shown that this extended FGF as isolated from SK-HEP—l cells is blocked et al.

Proteins having FGF activity in the at the N—terminus (Klagsbrun. M., (supra)). above-mentioned in vitro assays and sharing a similar putative N-terminal sequence with the bovine pituitary basic FGF shown in Figure 3 (the 146 amino acid form) have also been isolated from bovine brain. adrenal Kidney. and from human corpus luteum. retina. placenta. The native protein obtained from certain of these tissues is heterogeneous —- a second form missing the putative fifteen N-terminal amino acids retains Meth Enz (1986) in that bovine and activity. (Gospodarowicz. D., press.) It is considered. therefore. human basic FGFs exist in three forms--those indicated as mature forms in Figures 3 and 4. longer forms containing eight additional amino acids at the N-terminus, and shorter forms lacking fifteen amino Thus. there is believed to be natively produced Wlong" basic acids of the putative mature sequences shown.

FGF containing 154 amino acids. "primary" basic FGF containing 146 amino acids. and "short" basic FGF containing 131 amino acids. These FGFs are designated "basic" FGF. because they contain a high number of basic amino acid residues (lysine. arginine, histidine) and are therefore cations at neutral pH.

A protein is defined herein as basic FGF if it shows FGF activity in the foregoing assays. binds to heparin. is a cation at neutral pH. and reacts immunologically with antibodies prepared using a synthetic analog of the amino terminal sequence [tyrl (if appropriate) or to other antibodies raised against ] FGF (l-10) conjugated to bovine serum albumin bovine (or human) FGF or synthetic or native peptides thereof. see Baird. A.. et al. Regulatory Peptides (1985) ;g:3o9—317.

Acidic FGF has been isolated from bovine brain by others. and the first 34 amino acid residues determined. The cloning herein or the genes for bovine and human acidic FGF has permitted amino acid sequences additional to 1-34 for acidic bFGF. to be deduced as shown in Figure la, and a partial sequence for acidic hFGF has been obtained. as shown in Figure 2a.

Subsequent to much of the work described below. the complete amino acid sequence for acidic bFGF was disclosed by Esch. et al. giochem Biophys Res Comm (supra) and by Gimenez-Gallego. G.. et al. Science Also. most of the present work. the complete coding sequence (supra), as shown in Figure lb. subsequent to lE950330 IEQEOISQ1 for acidic hFGF was determined by the Maciag group. as shown in Figure 2b.

The acidic protein also has two known active forms, one having the l40 amino acid sequence beginning at the phenylalanine residue numbered "1" in the figures. and a second shorter form corresponding to amino acids 7-140. Both the bovine and human proteins may also occur in N—terminal extended forms.

Translation of DNA upstream of the codon for the amino acid numbered "1" in the figures (back to the ATG start codon at -15. shown in parentheses) represents the additional sequence of the extended protein. As is the case for basic FGF. the N-terminal methionine is almost certainly processed off in eucaryotic expression hosts. although it may not be if the gene is expressed in Therefore. like the basic FGF described above. the native acidic protein may exist in three one truncated. i.e., "short." acidic FGf containing 134 amino acids: one N-terminal extended. bacteria. active forms: i.e., "long" form containing 154 amino acids: and the other "primary" acidic FGF containing 140 amino acids beginning at the residue numbered "1" in the figures.

It has been shown by Burgess, w.H.. et al. (in press) that the bovine brain long form is blocked by an acetyl residue. These proteins contain a disproportionate number of acidic amino acid residues. i.e., glutamic and aspartic acids and the proteins are therefore anions at neutral pH.

A protein is defined herein as acidic FGP if it shows FGF activity in in vitro assays, binds to heparin. is an anion at neutral pH. and is immunologically reactive with antibodies prepared against human or bovine acidic FGF or against synthetic or native peptides thereof. I IE9503-{:1 _]_]_..

Acidic FGF and basic FGF are thus used herein to designate the foregoing proteins or proteins having amino acid sequences represented by those shown in Figures 1-4. of course. these definitions are not restricted to the specific sequences shown. but include proteins which contain accidentally or deliberately induced alterations. such as deletions. additions, or exchanges of amino acid residues. so long as the biological activity. as measured by the foregoing in vitro and immunological assays, and respective anionic or cationic character at neutral pH does not change. Of course. modified forms may have slightly altered quantitative activity and specificity.

"Purified" or "pure" refers to material which is free from substances which normally accompany_it as Thus acidic hFGF. example. refers to acidic hFGF which does not contain found in its native state. "pure" for materials normally associated with its in sign environment in human brain or pituitary. of course. "pure" acidic hFGF may include materials in covalent association with it, such as glycoside residues. "operably linked" refers to a juxtaposition wherein the components are configured so as to perform Thus. promoters operably linked to a coding sequence are their usual function. control sequences or capable of effecting the expression of the coding :sequence.

"Control sequence" refers to a DNA sequence or sequences which are capable. when properly ligated to a desired coding sequence, of effecting its expression in hosts compatible with such sequences. Such control sequences include at least promoters in both procaryotic and eucaryotic hosts. and optionally. transcription termination signals. Additional factors necessary or E95030?’ _.]_2_ helpful in effecting expression may also be identified.

As used herein. "control sequences" simply refers to whatever DNA sequence may be required to effect expression in the particular host used.

"Cells" or "cell cultures" or "recombinant host "host cells" are often used interchangeably as will be clear from the context. cells" or These terms include the immediate subject cell. and, of course. the progeny It is understood that not all progeny are exactly identical to the parental cell. mutations or differences in environment. thereof. due to chance However. such altered progeny are included in these terms. so long as the progeny retain the characteristics relevant to those In the present case. for example, such a characteristic might conferred on the originally transformed cell. be the ability to produce recombinant FGF.

B. General Description UtilityflggggAdministration The invention provides DNAS encoding growth factor proteins which are useful in encouraging the healing of wounds and which further may be supplied in sufficiently pure amounts to permit the design of inhibitors specific to them. ‘The purified growth factors are generally applied topically to the traumatized tissue in order to stimulate vascularization and healing. Appropriate substrates are burns, wounds. bone fractures. surgical abrasions such as those of plastic surgery. or others requiring repair. Because application of these factors accelerates healing. they also reduce the risk of infection.

Indications wherein FGF is of value in‘ encouraging neovascularization include musculo-skeletal IEQ50301 conditions such as bone fractures. ligament and tendon repair. tendonitis. and bursitis: skin conditions such as burns. Cuts. lacerations. bed sores. and slow—healing ulcers such as those seen in diabetics: and in tissue repair during ischaemia and myocardial infarction.

Formulations of the recombinantly produced growth factors using available excipients and carriers are prepared according to standard methods known to those in the art. The proteins can be formulated as lotions. gels. as part of a controlled release system. or ointments with additional active ingredients. such as antibiotics. if desired.

For topical administration. which is the most appropriate with regard to superficial lesions. standard topical formulations are employed using. for example. .l—l0% solutions. Such solutions would be applied 3-6 times a day to the affected area. The concentration of the ointment or other formulation depends. of course. on the severity of the wound and nature of the subject. In most protocols. the dose is lowered with time to lessen For bone and tissue repair. administration is preferred locally, but by means of subcutaneous implant or slow release formulation implanted directly proximal the target. Surgery may be required for such conditions as bone injuries, thus making implantation directly practical. Slow—release forms can be formulated in IE950.3a1 et al. Nature and a formulation but not As with topical administration, for sustained- the concentration of FGF in the formulation depends on a number of factors. release delivery. including the severity of the condition and the rate of FGF release from the polymer. In general. the formulations are constructed so as to achieve a constant local concentration of about 100 times the serum level of hormone or 10 times the tissue concentration. as described by Buckley et al (Proc Natl Acad Sci USA (supra)). Based on an FGF concentration in tissue of -50 ng/g wet weight (comparable to EGF weight), at 60 ng/g wet release of 50-5000 ng FGF per hour is acceptable. The initial concentration. depends on the severity of the wound.

It is expected that FGF may act in concert. even synergistically. with other of course. and growth factors such as epidermal growth factor (EGFf, factors (TGF-a or TGF—B). the transforming growth insulin-like growth factors (IGF-l and IGF-2). and/or platelet-derived growth factor (PDGF). In addition. specifically for bone repair. it may act in synergy with antagonists hormone, of parathyroid since parathyroid hormone promotes bone resorption. Therefore. also included within the compositions and administration protocols of the invention are embodiments wherein the FGF of the IE9503o1 invention is administered in the same composition with, or in the same protocol with. one or more of the foregoing factors, thus more effectively to achieve the desired tissue repair.

Since FGF is effective in promoting neurite outgrowth, nerve regeneration. and neuronal survival. it may be useful for treatment of certain neurological disorders such as Alzheimer's and Parkinson's diseases. amyotrophic lateral sclerosis. and general aging of the nervous system. as well as traumatic injury to the spinal cord and peripheral nerves.

Administration of the drug for these indications is preferably by implant in formulations similar to those set forth above in connection with wound healing. The drug may also be delivered by means of implants of cell cultures as in transplant therapy by treating the cultures prior to transplantation with the the FGF may be injected directly into the spinal fluid. or may FGF preparations of the invention. In addition. be applied systemically. systemic formulations are generally as are known in the art and include formulation in buffer or physiological saline. or other appropriate excipient. Dosage levels are approximately those of wound healing: however. for tissue culture or explant maintenance. it may be supplied at 0.1-lo ng/ml of serum or culture medium.

FGF proteins are particularly useful. also. in aiding the reformation and repair of tissues traumatized it may be helpful to embed the FGF proteins in polymers used as surgical during surgery. For this use. staples. The proteins are thus able to supplement biologically the mechanical suturing effected by the. staples. and to,augment and abet the "natural" healing processes in the repaired tissues.

IE950301 In additicn. it has been shown that angiogenic stimuli. such as those provided by the FGF proteins discussed herein. result in the release of tissue plasminogen activator (tPA) and of collagenase in vitro (Gross. J.L., et al, Proc Natl Acad Sci USA (1983) gg:2623- 2627). Therefore, the FGF proteins of the invention are also useful in treatment of conditions which respond to these enzymes. While it may be necessary in acute situations (such as the presence of a blood clot associated with stroke or heart attack) directly to administer large doses of tPA to dissolve the clot. embolisms. for treatment of chronic propensity to form administration of FGF to maintain a suitable level of tPA in the blood stream may be desirable.

Therefore. for this indication, systemic administration of the drug. using conventional means such as intramuscular or intravenous injection, is preferred.

The invention provides practical quantities of pure FGF growth factors for use in connection with the foregoing indications. Four specific endothelial growth factors are exemplified, each of which is apparently and active in three forms: bovine acidic and basic FGF. their human counterparts. Both acidic and basic factors are considered to occur in long. primary. and short forms, as described herein. It is considered that the N-terminal methionine of the long forms is processed off when the protein is produced in eucaryotic systems. and that the subsequent amino acid residue is derivatized. probably by acetylation. post—translation.

While FGF in its various forms does not have a recognized signal sequence. it must somehow be secreted. since it acts outside the cells producing it at a while it is membrane-bound receptor. Therefore. probably not secreted by the‘recognized constitutive secretion pathway. its secretion is accomplished by other means, such as by cell lysis or by exocytosis.

For most tissues from which FGF is naturally derived. and Eor many mammalian expression systems, such release may be achieved by securing exocytosis with a calcium ionophore. such as the commonly employed A2318? (CalBiochem). which, is added to the culture medium at 1-10 uM in the presence of 1 mM CaCl2. macrophages or monocytes, in in vitro conditions.

For expression systems derived from other activation methods have been shown to be effective. such as the addition of lipopolysaccharide (LPS) at 10 ug/ml or the addition also be employed in releasing recombinantly produced FGF These techniques can proteins when produced intracellularly without added Additional stimulators for release of intracellularly produced signal sequences. as described below. proteins include the phorbol esters and the triglycerides.

Gene Retrieval The general strategy whereby the illustrated FGF-encoding sequences were obtained herein is as follows. The known N-terminal sequence of bovine acidic FGF was used to design a series of probes for use with a bovine genomic library ligated into phage. Phage recombinants which hybridized to the probes were isolated from the library and digested into smaller fragments suitable fior cloning into M13 cloning vectors "natural" This resulted in in order to obtain a probe. an M13 probe containing a 250 bp sequence corresponding lE950301 to a portion of the bovine acidic protein: this probe is central to recovering the complete coding sequences for the acidic forms of both bovine and human sources. as well as to obtaining the genes for the basic forms in these species.

Briefly. the fragments obtained by AluI digestion of a selected acidic bFGF gene fragment cloned into phage were shotgun cloned into M13 and a 250 bp fragment which hybridized to appropriate probe DNA The above. designated 250/AluI. was transferred into pBR322 and was used to probe a selected and sequenced. bovine brain. hypothalamus or pituitary cDNA library (to obtain the complete acidic bFGF sequence uninterrupted by introns) and a human genomic library (to obtain the first exon of the human acidic FGF genomic sequence).

The middle and third exon of the human gene encoding acid FGF were obtained using oligomer probes. as described in the examples below. These probes were designed on the basis of a synthetic human acidic FGF In addition. this to design probes for the gene. same 250 bp fragment was used basic form. taking advantage of the available amino acid sequence information to alter the DNA to correspond to the basic rather than acidic form. The modified probe. thus designed on the basis of a comparison of the acidic bFGF N—terminal coding sequence and the basic bPGF amino acid sequence. was used to probe the same bovine pituitary cDNA library for the basic bFGF cDNA. then used to probe human genomic and cDNA libraries to The recovered bovine clone was recover the genomic sequence encoding human basic FGF-encoding DNA.

XBBZ was mutagenized to convert the DNA sequence to Alternatively. the bovine cDNA clone one encoding the human form of the basic FGF protein.

For both acidic and basic FGF. the cDNA and genomic clones described hereinbelow are useful in probing DNA libraries prepared from various species to obtain the analogous coding sequences from these mammalian libraries: in addition. the genomic clones are capable of expression in mammalian systems and may give better CDNA libraries prepared from various tissues such as pituitary. brain. results than the corresponding cDNAs. hypothalamus, or Kidney can also be screened in this manner .

Expression of FGF Genes The cloned genomic or CDNA sequences can be expressed in appropriate expression systems. *Of course. for the DNAS disclosed herein. the foregoing protocol for retrieving them need not be repeated. but conventional chemical synthesis methods can suitably be employed. This permits adjustment of the DNA to obtain any desired form of the protein. CDNA sequences can be provided with appropriate controls suitable for any host, including bacteria, yeast. or eucaryotic cells.

Reconstruction of the genomic sequences for human acidic FGF can be obtained from the three deposited X phage harboring the three exons. Genomic sequences containing introns can be expressed using eucaryotic control sequences and eucaryotic hosts which are capable of splicing the transcripts. Vaccinia-based expression systems may also be used. Exemplary control sequence DNAs and hosts are given in paragraph C.l below.

In particular. complete DNA encoding full length FGF can be constructed, for example. using a combination of recombinant and synthetic methods to obtain any of the long, primary or short forms of acidic or basic FGF. Heterologous signal sequences may also be fused to thesef and advantage taken of the Known relationship of the signal sequence to cleavage site to obtain the protein in the desired form. Intracellularly produced forms of the proteins can be obtained by cell lysis. or their release from the cells can be stimulated as described above. Particularly preferred are expression systems for either the cell-associated or putatively secreted (fused to signal sequence) forms which utilize control systems compatible with mammalian such as CHO cells. vaccinia-based systems, which can be used for stable or cells, Also preferred are transient expression in susceptible cells.

The recombinant FGF proteins thus produced are then purified in a manner similar to that utilized for purification of FGF from natural sources. but purification is considerably simpler. as the proteins form a much larger proportion of the starting material.

Polymorphism .

It has also been shown that human genomic DNA exhibits a polymorphism in the region of the second exon of the gene. Existence of the polymorphism is'a predictor of the tendency to solid tumors. as FGF is secreted by them. and probably is necessary for their survival. as it promotes blood vessel growth that keeps nutrients flowing to the tumor.

To detect the polymorphism. human genomic DNA is obtained by conventional methods. from a blood sample, for example. and subjected to size separation on polyacrylamide gels and probed using standard Southern blot techniques. An effective probe is the 1.4 kb EcoRI fragment obtained from the 2.1 Kb insert into XBB2 described hereinbelow. when such a probe or its equivalent is used to hybridize to gels containing Hindlll digests of the isolated human DNA. a 2.7 kb -21.. fragment is normally detected. In some individuals, an additional 2.9 kb fragment is also found. These fragments map to the region of the gene surrounding exon 2. as shown in Figure 17. of three individuals tested. two exhibited only the 2.7 kb fragment: one exhibited both the 2.7 and 2.9 Kb fragments. The hybridization intensity showed that the individual with both fragments contains both alleles, which is supported by results obtained by Southern blot analysis of DNA from mouse/human hybrid cell lines. In such hybrids, wherein only one chromosome is transferred. only one of the two fragments appears in each line.

C. Standard Methods Most of the techniques which are used to transform cells. construct vectors, extract messenger RNA. prepare cDNA libraries. and the like are widely practiced in the art. and most practitioners are familiar with the standard resource materials which describe specific conditions and procedures. However. for convenience. the following paragraphs may serve as a guideline.

C.l.

Both procaryotic and eucaryotic systems may be Hosts ang_9ontrol Sequences used to express the FGF encoding sequences; procaryotic hosts are, of course. the most convenient for cloning procedures. Procaryotes most frequently are represented by various strains of E. coli: however. other microbial strains may also be used. Plasmid vectors which contain replication sites. selectable markers and control sequences derived from a species compatible with the host are used: for example. E. coli is typically $950330 E95030? such as yeast, may also be used as hosts. Laboratory strains of Sacchagomyces cerevisiae. Baker's yeast. are most used although a number of other strains or species are the used. u origin of replication of Broach, J. R.. replication (see. Stinchcomb.

Control sequences for yeast vectors include promoters for the synthesis of glycolytic enzymes (Hess. et al. J Adv Enzyme Reg (1968) 1:149: Holland. et al.

Biochemistry (1978) l1:4900). known in the art include the promoter for Additional promoters -phosphoglycerate kinase (Hitzemah. et al. J Biol Chem (1980) 255:2073). additional advantage of transcription controlled by Other promoters. which have the Nature growth conditions and/or genetic background are the promoter regions for alcohol dehydrogenase 2. isocytochrome C. acid phosphatase. degradative enzymes associated with nitrogen metabolism. the alpha factor system and enzymes responsible for maltose and galactose utilization. It is also believed terminator sequences are desirable at the 3' end of the coding sequences.

Such terminators are found in the 3' untranslated region following the coding sequences in yeast-derived genes.

It is also. of course. possible to express genes encoding polypeptides in eucaryotic host cell cultures derived from multicellular organisms. See. for Axel, 4.399.216. additional advantage of the ability to splice out example. et al. These systems have the introns and thus can be used directly to express genomic Useful host cell lines include VERO and HeLa cells, and Chinese hamster ovary (CHO) cells. fragments.

Expression vectors for such cells ordinarily include promoters and control sequences compatible with mammalian cells such as. for example. the commonly used early and late promoters from Simian Virus 40 (SV40) (Fiers. et al. Nature (1978) ;1;:ll3). or other viral promoters such as those derived from polyoma. Adenovirus 2. bovine papilloma virus, or avian sarcoma viruses.

The controllable promoter. hMTII (Karin. M..

Nature (1982) 2995797-802) may also be used. et al.

General aspects of mammalian cell host system transformations have been described by Axel (supra). It now appears. also that "enhancer" regions are important in optimizing expression: these are. generally. sequences found upstream or downstream of the promoter region in noncoding DNA regions. Origins of replication may be obtained. if needed. from viral sources. However, IE95o3o1 IE950307 integration into the chromosome is a common mechanism for DNA replication in eucaryotes.

C.2.

Depending on the host cell used. transformation Transformations is done using standard techniques appropriate to such cells. The calcium treatment employing calcium chloride. S.N.. Proc Natl Acad sci gUSA) (1972) ggzziio, or the Rbclz method described in Maniatis. et al. Molecular Cloning: A as described by Cohen.

Transformations C.3.

Construction of suitable vectors containing the Vector Construction desired coding and control sequences employs standard ligation and restriction techniques which are well Isolated plasmids. DNA sequences. or synthesized oligonucleotides are cleaved. understood in the art. tailored. and religated in the form desired.

The DNA sequences which form the vectors are available from a number of sources. Backbone vectors and control systems are generally found on available "host" vectors which are used for the bulk of the sequences in construction. Typical sequences have been lE950301 set forth in 1C.l above. For the pertinent coding sequence, initial construction may be. and usually is, a matter of retrieving the appropriate sequences from CDNA or genomic DNA libraries. However. once the sequence is disclosed it is possible to synthesize the entire gene sequence in vitro starting from the individual nucleoside derivatives. The entire gene sequence for 500-1000 bp may be prepared by synthesizing individual overlapping genes of sizeable length. e.g.. complementary oligonucleotides and filling in single stranded nonoverlapping portions using DNA polymerase in the presence of the deoxyribonucleotide triphosphates.

This approach has been used successfully in the Synthetic oligonucleotides are prepared by either the phosphotriester method as described by Edge, et al. Nature (supra) and Duckworth. et al, Nucleic Acids Res (1981) g:1691 or the phosphoramidite method as described by Beaucage. S.L., and Caruthers. M.H.. 2;; ggggg (1981) ;;:18S9 and Matteucci. M.D.. and Caruthers.

M.H., J Am Chem Soc (1981) 1Q;:318S using commercially available automated oligonucleotide and can be prepared synthesizers. Kinasing of single strands prior to annealing or for labeling is achieved using an excess, e.g.. approximately 10 units of polynucleotide kinase to 1 nmole substrate in the presence of 50 mM Tris. pH 7.6, mM Mgclz. 5 mm dithiothreitol, 1—2 mM ATP. 1.7 pmoles 132?-ATP (2.9 mci/mmole). 0.1 mM spermidine. .1 mM EDTA.

IEQEOIW1 Once the components of the desired vectors are thus available. they can be excised and ligated using standard restriction and ligation procedures.

Restriction cleaved fragments may be blunt ended by treating with the large fragment of E. coli DNA ' polymerase I (Klenow) in the presence of the four deoxynucleotide triphosphates (dNTPs) using incubation times of about 15 to 25 min at 20 to 25°C in 50 mM Tris pH 7.6.50 mM NaC1. 5 mM MgCl2. 6 mM DTT and 0.1-1.0 mM dNTPs. single—stranded overhangs but chews back protruding 3‘ The Klenow fragment fills in at 5' single strands, even though the four dNTPs are present.

If desired, selective repair can be performed by supplying only one of the. or selected. dNTPs within the limitations dictated by the nature of the overhang.

After treatment with Klenow, the mixture is extracted with phenol/chloroform and ethanol precipitated.

Treatment under appropriate conditions with S1 nuclease or BAL—3l results in hydrolysis of any single-stranded portion. _ Ligations are performed in l5-S0 ul volumes under the following standard conditions and for example. 20 mM Tris-Cl pH 7.5, 10 mM Mgclz. 10 mM DTT. 33 ug/ml BSA. 10 mM-S0 mM NaCl. and either 40 uM ATP. 0.01-0.02 (Weiss) units T4 DNA ligase at 0°C (for "sticky end" ligation) or 1 mM ATP. 0.3-0.6 (Weiss) units T4 DNA ligase at 14°C (for "blunt end" ligation). temperatures: Intermolecular "sticky end" ligations are usually performed at 33-100 uq/ml total DNA concentrations (S-100 nM total end concentration).

Intermolecular blunt end.ligations are performed at 1 uM total ends concentration.

In vector construction employing "vector fragments". the vector fragment is commonly treated with bacterial alkaline phosphatase (BAP) or calf intestinal alkaline phosphatase (CI?) in order to remove the 5' phosphate and prevent self-ligation of the vector.

Digestions are conducted at pH 8 in approximately 10 mM Tris-HC1. 1 mM EDTA using about 1 unit of BAP or CIP per ug of vector at 60° for about one hour. In order to recover the nucleic acid fragments. the preparation is extracted with phenol/chloroform and ethanol precipitated. Alternatively, religation can be prevented in vectors which have been double digested by additional restriction enzyme digestion and separation of the unwanted fragments.

IE950330 E95030? For portions of vectors derived from CDNA or genomic DNA which require sequence modifications, site specific primer directed mutagenesis may be used (Zoller, M.J., and smith. M. Nucleic Acids Res (1982) ;g:6487-6500 and Adelman. J.P.. et a1. DNA (1983) ;:l83—l93). This is conducted using a primer synthetic oligonucleotide complementary to a single stranded phage DNA to be mutagenized except for limited mismatching, the synthetic oligonucleotide is used as a primer to direct representing the desired mutation. Briefly, synthesis of a strand complementary to the phage. and the resulting partially or fully double-stranded DNA is transformed into a phage-supporting host bacterium.

Cultures of the transformed bacteria are plated in top agar. permitting plaque formation from single cells which harbor the phage.

Theoretically, 50% of the new plaques will contain the phage having. as a single strand. the resulting plaques are washed after hybridization with mutated form: 50% will have the original sequence. by ampicillin, tetracycline or other antibiotic Plaques which hybridize with the probe IEE Q 5 0 3 0 1 commonly used. Anal Biochem et al.

C.5. Hosts Exemplified Host strains used in cloning and procaryotic expression herein are as follows: For cloning and sequencing. and for expression of construction under control of most bacterial promoters. E. coli strains such as Mcloél. Dfil, RRl, csoonrl. K803. H3101. JA22l, and JMl0l were used.

D. Illustrative Procedure The following examples are intended to The DNA encoding the illustrated FGF sequences is obtained illustrate but not to limit the invention. initially by screening a bovine genomic library and obtaining a pivotal probe. followed by retrieval of additional DNA. repeat this procedure. as the sequence of the pivots; However. it would not be necessary to probe is now known and could thus be constructed IE950301 chemically ig_g;g£g. In addition. bacteriophage harboring the four illustrated sequences are deposited at the American Type Culture Collection.

Example 1 Construction of the 250/Alul Probe; Preparation of Acidic bFGF Genomic DNA A 250 bp A1uI bovine genomic fragment was used to probe both human and bovine libraries in order to obtain complete coding sequences for the illustrated This probe. designated 250/Alul. was obtained as follows. acidic FGF proteins.

The N-terminal amino acid sequence for residues library obtained from Dr. Friti Rottman. Case Western The probes were used to screen a bovine genomic Reserve. which had been prepared as a partial Mbol digest and was cloned into BamHI treated phage vector Charon 28 (Woychik. R.F.. et al. Nucleic Acids Res (1982) $917197-7210).

Hybridization was conducted on denatured DNA replicated onto filters using a modification of the method described by Ullricht A.. et al. EMBO J (1984) :361-364: and the washing conditions were those of x Denhardt's solution (lO0x Denhardt's equals 2% bovine serum albumin. 2% polyvinyl % Ficoll); 6x SSC (20x SSC equals 3 M .3 M Na citrate); PH 5,3; formamide. pyrollidone: NaCl. 50 mM sodium phosphate, 100 ug/ml herring sperm DNA; hybridization buffer further included 10% dextran sulfate and about lOS—lO6 Prehybridization and hybridization were at 42°C for 1 hr cpm/ml kinased probes 891 or 889/890.

The filters were then washed 2x min with 1x SSC. 0.1% SDS at 22°C. followed by 1 ten minute wash in 1x SSC. 0.1% SDS at 55°C. and 16 hr respectively.

After washing, the filters were exposed for 1 day using intensifying screens.

The screened bovine genomic library contained 50 phage out of 106 recombinants which hybridized to both probes. These 50 phage were further screened with mixtures of probes 853-856. In this screen. prehybridization/hybridization buffer contained 6x SSC. lx Denhardt's. 0.1% SDS, 0.05% Na pyrophosphate. and 100 ug/ml salmon sperm DNA: hybridization buffer further contained 105-106 Probes 853-856 are pools of 16 sequences each of the 64 (total) 17—mers cpm/ml probe. corresponding to amino acids 7-12. synthesized using the phosphotriester method. However. 46 of the 50 clones This hybridization was performed at between 65°C to 35°C for further hybridized to the shorter probes.

(XBA2 and XBA3) and the phage inserts were -32— restriction mapped, as shown in Figure 6, and partially sequenced as shown in Figure la. Comparison of the deduced amino acid sequence with that published for the N—terminal 34 residues of the bovine acidic FGF native protein confirmed that these clones are correct. From the nature of the coding sequence it is apparent that amino acid residues l-41 (as shown in Figure la) are immediately subsequent The length but it is encoded in these clones: nucleotides appear to represent an intron. of this intron is, at present, uncertain. possible that the complete acidic bFGF encoding sequence resides on these XBAZ and XBA3 DNAs. additional DNA required to obtain the complete coding However any sequences for this protein can be obtained from the same gene library using the XBAZ or XBA3 in "walking" techniques. The codons preceding the N-terminal residue are believed to encode the indicated fifteen amino acid prosequence. or, as discussed above. the Elong" form of the native protein extended by fifteen amino acids at the N-terminus (or by fourteen if the N—terminal methionine is cleaved) as compared to isolated "primary" form.

To prepare the 250/Alul probe. XBAZ was partially digested with AluI and shotgun cloned into M13 et al. mg (1982) l9:269—276). The M13 plaques were hybridized in duplicate with 853-856 and /890. Phage hybridizing to both probes were (Messing. J.. sequenced. The resulting 280 bp Alul probe is shown in Figure 7 along with the corresponding deduced amino acid sequence: its location on the XBA2 and XBA3 inserts of Figure 6 corresponds to the site of probes 889/890 and 891.

N—terminal portion of the acidic bFGF protein.

The 250/Alul probe corresponds to the IE; 9 5 O 3 O 1 Example 2 Recovery of Acidic bFGF cDNA The 250/Alul probe is used to retrieve the cDNA A cDNA library is obtained from bovine pituitary. brain. or hypothalamus sequence encoding acidic bFGF. mRNA using the Xgtlo vector of Huynh. V.T., et al. DNA Cloning Techniques: A Practical Approach (IRL Press.

Oxford. 1984). recovery of the entire sequence encoding acidic bFGF.

The resulting hybridizing clones permit Comparable CDNA libraries constructed using the analogous mRNA from other mammalian species is probed with the 250/Alul probe to obtain. for example. rat. ovine. bovine. feline. canine. equine. or porcine basic FGF.

Example 3 Preparation of Acidic hFGF Genomic DNA A human fetal liver genomic library in Charon 4A (Lawn. R.M.. et al. gel; (1978) l5:ll57-1174) was used as a source of the human sequences. The library The conditions of prehybridization/hybridization were the was probed with nick—trans1ated 250/Alul probe. same as those for the 889/890 and 891 probes of Example 1. except that 40% formamide was used. Hybridization was at 42°C for 16 hr. room temperature with 1x SSC. 0.1% SDS.

The filters were then washed at and then for 2x min at 50°C with the same buffer. Positively hybridizing clones were cultured. and one. designated XHAG-9.1. contained the desired acidic hFGF sequences. A partial restriction map of this clone is shown in Figure 8: nucleotide and amino acid sequence information is shown in Figure 2a. The nucleotide sequence encoding amino acids 1-4l can be identified: this sequence. and. as in the genomic acidic bFGF.

IE95o3o1 The acidic hFGF and bFGF amino acid sequences differ at positions 5. 21. and 35. be followed by an intron.

Human acidic FGF-encoding DNA also contains 15 codons preceding the N—terminus of the corresponding bovine isolated protein, which encode an amino acid sequence highly homologous to the N-terminal extension of the bovine protein. The translated sequence is shown in parentheses in Figure 2a. bovine DNA. there are nucleotide substitutions in codons -3, -6, -9. and -12. protein. A nucleotide change in codon —l0 results in In comparison to the which are silent in the translated the Thr residue of the bovine protein being an Ile residue in the human protein. Analogous to the bovine acidic FGF. prosequence or this N-terminal extension may represent a "long" form of the isolated. "primary" protein. either containing a fourteen or fifteen amino acid N4terminal extension depending on the fate of the methionine.

X phage clones containing the nucleotide sequences corresponding to the middle and C-terminal encoding exons of the human acidic FGF-encoding genomic Together with XHAG-9.1. described above. these phage provide the complete DNA were also obtained. protein encoding sequence.

The phage containing the middle exon was obtained from a human genomic library prepared as [described by Wyman. A.R. er al. Proc Acad Natl Sci gusng This is a library prepared by insertion of the resulting fragments into the polylinker This places the insert between two EcoRI restriction sites region of kcharon 30 phage at the BamHI site. for easy removal. lE950301 This genomic library was probed with two oligonucleotides which had been used to construct the synthetic human acidic FGF gene. as described in Example The oligonucleotides designated "3" and "4" in that figure immediately below, and illustrated in Figure 9. were those used as probes. The coding region of the recovered phage. designated kHG—3. is shown in Figure 2b. This coding sequence encodes amino acids 42-85 of the Jaye sequence. and corresponds to the exon/intron boundaries of the gene encoding the basic FGF protein. similarly. the third exon was obtained from the Maniatis human genomic library of Lawn. et al, (supra) in Charon-4A. prepared as described above. and probed with oligonucleotides labeled "6" and "7" of the synthetic gene shown in Figure 9. The retrieved X-phage clone. designated XHAG-3, has been Pértially The sequence information also confirms the presence of the sequenced, and the results are shown in Figure 2c.

C-terminal exon sequence in the XHAG—3 insert.

The foregoing three inserts can be recombined to assemble the complete human acidic FGF genomic sequence by digestion of each phage with EcoRI to remove the insert and ligation of the resulting fragments to reconstruct the gene. The genomic sequence can then be used to construct expression vectors in a manner analogous to that described for cfiNA sequences in Example 7 below. Specifically. the reassembled gene can be inserted as an EcoRI(blunt) fragment in a manner similar to that described for the Syn-acidic hFGF NcoI(blunt)/HindIII(blunt).

Example 4 Preparation of Acidic hFGF Coding Sequence A cDNA library prepared from human pituitary, breast carcinoma, brain. brainstem. SK—HEP—l. or !E950301 hypothalamus mRNA by the method of Huynh. for the bovine mRNA in Example 2, as described is probed with the 250/A1uI probe under the conditions described in Example 3 to obtain the cDNA encoding acidic hFGF. An unspliced cDNA containing the first exon was obtained from the breast carcinoma library.

In the alternative, the CDNA sequence information obtained by Jaye. M.. et al. Science (1986), in press (see Figure 2d). was used as a guide for the synthesis of a gene encoding the acidic hFGF. The cDNA clone reported by Jaye et al was obtained using messenger RNA from human brain stem and encodes an acidic hFGF whose deduced amino acid sequence is shown in Figure 2b.

The genomic XHAG-9.1 clone described in Example 3 was used to provide the 5' portion of the gene. To prepare this portion. a 1.9 Kb BamHI fragment was isolated from kHAG—9.l and subcloned into pUCl3 to obtain pCBI—lol. This intermediate plasmid was then digested with Ncol/BamHI and the containing the codons for the 1S bp fragment amino acids of the pro sequence along with the first 25 amino acids of the hFGF was isolated using The location of the Ncol site mature. "primary" form of acidic a 5% polyacrylamide gel. which contains the ATG that is believed to constitute the start codon at amino acid ~15 from the beginning of the primary sequence. is shown in Figure 9. which diagrams the synthetic gene.

The remainder of the coding sequence was synthesized using the synthetic oligonucleotides numbered 1-20 in Figure 9. The synthesis of the individual oligonucleotides uses conventional automated techniques. (The oligos were designed so as to yield the same nucleotide sequence as that reported by Jaye et al (supra) with two exceptions: IE95o3o1 oligonucleotides 4 and 14 were constructed so as to destroy the Ncol site spanning codon 67 by altering the GCC encoding alanine at codon 66 to GCT, as shown by the asterisk; in addition. oligonucleotides 19 and 20 were modified so as to add Hindlll and EcoRI cleavage sites following the TGA termination codon. Neither of the foregoing changes affects the amino acid sequence encoded.

The synthetic oligonucleotides are ligated to obtain the sequence shown in Figure 9 by kinasing 5 ug of each oligonucleotide (except #1 and uzo) using standard reaction conditions. annealing the 10 different complementary oligonucleotide pairs (1 + 11. 2 + 12. etc.). and then ligating the ten oligonucleotide pairs into three segments. These segments are formed sequentially using T4 ligase under standard conditions. the pair 1/11 is ligated with 2/12. followed by ligation with 3/13. followed by ligation with 4/14. 6/16. followed by 7/17. ligating 8/18 with 9/19. followed by ligation of the product with 10/20. 144 bp. 8 a 108 bp. and C = 106 bp) are purified using gel electrophoresis and then sequentially ligated by To obtain segment A.

Segment B is formed by ligation of 5/15 with Segment C is obtained by The three ligated subfragments (A - mixing B and C under standard conditions with T4 ligase. followed by addition of A. extracted with phenol. precipitated with ethanol.

The final reaction is the ethanol precipitate electrophoresed on a 5% acrylamide The fragment spans the BamHI/EcoRI sites. as shown in Figure gel. and the 358 bp fragment A+B+C is eluted. . and its sequence is verified using dideoxy sequencing by subcloning the segment into M13mp19.

To complete the coding sequence, ;xe synthetic 358 bp BamHI/EcoRI synthetic fragment is isolated from E95030! the phage or the polyacrylamide gel. its ends kinased, if neCe583EY. and ligated to the 118 bp NcoI/BamHI fragment from pCBI-101. The resultant partially synthetic nucleotide sequence encoding acidic hFGF is shown in Figure 9. and is designated Syn-acidic hFGF. of course. additional constructs wherein the "primary" and "short" forms of acidic FGF are immediately preceded by an ATG start codon. and contain a suitable restriction site might also be constructed.

Example 5 Retrieval of Basic bFGF Genomic/gndpgpNA Clones The 250/Alul probe was then used to design appropriate probes to obtain the corresponding basic bFGF sequences. Advantage was taken of the finding of Esch. F.. et a1 (supra) that amino acids 4-29 of acidic bFGF align with amino acids 13-38 of the basic hFGF sequence. Probes were designed based on the basic bFGF residues 18-36 and acidic bFGF residues 9-27. which regions are homologous at 14 of the 19 amino acids.

Probes 1097 and 1098. encode this region. were prepared using the -mers designed to The probes are shown in Figure 10: they overlap in the amino phosphoramidite method on an automatic synthesizer. acid 23-31 region of the basic sequence. In designing the probes. the 250/Alul sequence was used where the ,amino acid sequence was the same. and where different. minimum nucleotide differences in order to effect the required change in encoded sequence were incorporated.

The bovine pituitary CDNA library obtained by the method of Huynh. V.T.. as set forth in Example 2. was screened with 1098. Correct conditions for hybridi2ation’were determined using genomic DNA (Example ) for Southern blot as follows: "E'9 5 0 3'0 1 It was. of course. expected that the 1097 and lO98 probes would cross—hybridize with acidic FGF encoding DNA under low stringency conditions. Southern blot analysis showed that genomic sequences known to encode acidic bFGF which hybridized to 1097 and 1098 at 55°C wash temperatures failed to hybridize at 65°C.

(Prehybridization/hybridization buffer and conditions were as for 889/890 and 891 probes in Example 1.) Therefore. a wash temperature of 65°C was chosen. At this temperature. a 10 kb fragment in an EcoRI digest and a 3.4 Kb fragment in a Pstl digest hybridized to probes 1097 and 1098. when the CDNA library was probed as above using a 65°C wash temperature, a single clone designated XBB2. representing a 2.1 kb cDNA with EcoRI linkers. was recovered. A restriction map of this phage is shown in Figure 11. Subfragments of the insert in XBB2 were transferred to M13 for sequencing and a 1.4 Kb EcoRI—digested subfraqment was shown to encode amino acids l-146 (the complete "primary" sequence) of bovine basic FGF. The sequence upstream from the N—terminal codon is believed to encode either a nine amino acid prosequence or an N-terminal extended "long" form of the native protein which retains activity. The N-terminal extension may contain only eight residues. of course. depending on whether the methionine is cleaved during post—translational processing. The portion of this subfragment encoding basic bFGF is shown in Figure 3; amino acid numbering starting at position 1 corresponds to the N—terminus of the isolated "primary" protein.

The upstream nine codons are translated in parentheses.

The possibility that this extension represents an integral part of the native active protein is suggested by the higher MW’form of the human basic FGF prepared IE950301 from hepatoma cells by Klagsbrun. et al, Proc Natl Acad ggi (supra).

The 1.4 kb subfragment is then nick translated and used to screen a bovine genomic library constructed in a manner similar to that of Example 1 for the basic bFGF genomic sequences.

The 1.4 Kb basic bFGF—encoding cDNA fragment is also used to probe alternate mammalian CDNA libraries. pig, feline. equine or murine sources to obtain the basic FGF such as those from rat, or bovine. canine. encoding sequences from these species.

Example 6 Preparation of Human Basic FGF Genomic and cDNA Clones A xgtlo CDNA library prepared from human kidney mRNA was also probed using the 1.4 kb bovine basic subfragment. Prehybridization/hybridization .mM Na phosphate. pH .5. Sx Denhardt's, 5x SSC. and 50 ug/ml herring sperm buffer contained 40% formamide, DNA: hybridization buffer also included 10% dextran sulfate and 104-105 and one selected for characterization.

Three clones were This clone, designated XKB7, contained an approximately 1.4 cpm/ml probe. isolated.

Kb EcoRI fragment which was partially sequenced to yield the data shown in Figure 4. along with the deduced amino acid sequence. The sequenced coding region permits deduction of amino acids 1-S0 shown in the Figure; the continuing sequence immediately downstream appears to represent the cDNA copy of an unspliced mRNA. indicating that an intron occurs in the basic FGF gene in a homologous position to the intron after amino acid 41 in The XKB7 clone also provides upstream DNA encoding the nine amino acid the bovine and human acidic FGF genes.

N—terminal extension of the long form shown.i IE95o391 Additional genomic and cDNA libraries were screened using the same 1.4 kb basic bFGF-encoding fragment under precisely the same hybridization conditions as those employed for the human kidney Xgtlo library above. Four additional clones were obtained, which between them encode the entire 146 amino acid protein corresponding to the isolated basic bFGF, as shown in Figure 4. Nine upstream codons included in XKB7 above translate into a sequence having complete homology with the translated upstream codons in the bovine basic FGF clone. although there is a silent This translated nucleotide substitution in codon -8.

N-terminal extension is shown in parentheses in Figure 4: and, as above. may represent a prosequence or the additional amino acids of an N-terminal extended active protein.

In more detail. two positively hybridizing clones from a human genomic library in X Charon 4A. prepared as described by Lawn, R.M.. et al (supra) were designated XMG4 and XMGIO. acids (-9)-Sl: XMGIO encodes amino acids 86-146, representing the third of three exons contained in the (The XMG4 encodes amino mature protein-encoding region of the gene. ‘location of exon/intron boundaries was determined by homology to the bovine sequence.) A slightly different genomic library in X Charon 28. obtained from E.

Fritsch, yielded XHTl which contains the second mature protein exon. encoding amino acids Sl-85. Finally.

XHFLI, a cDNA clone obtained from a human fetal liver library prepared in kgtlo as described above. encodes amino acids 56-146, confirming the location of the relevant intron/exon junction.

There are only two amino acid ._.;erences between basic bFGF and hFGF, at position ll2. where the IE9503o1 bovine protein has Ser and the human protein has Thr. and at position 128. where the bovine protein has Pro and the human has Ser. These differences are the result of a single nucleotide difference in each case; therefore bovine cDNA may conveniently be modified by site directed mutagenesis as described below to encode the human protein. and. indeed. standard site-specific mutagenesis.techniques were used to alter these codons.

The XBBZ clone of Example 5 was digested with EcoRI and the 1.4 Kb region spanning the bFGF protein-encoding The ig_y;;£g mutagenesis was carried out in the presence of portion was ligated into the EcoRI site of Ml3mp8. three oligonucleotides: the "universal" primer. a the mutagenic l6—mer S‘-GAAATACACCAGTTGG—3'; which alters the coding sequence at codon ll2. and the mutagenic l7-mer 5'-ACTTGGATCCAAAACAG—3', which alters The mutagenized phage was l7-mer: the sequence at codon 128. also subjected to a second round of in vitro primer- directed mutagenesis to create a HindIII site 34 bp downstream from the translation termination codon using the mutagenic 25-mer. 5'-TTTTACATGAAGCTTTATATTTCAG—3‘.

The resultant mutated DNA was sequenced by dideoxy sequencing to confirm that the desired mutagenesis had occurred, and the approximately 630 bp fragment spanning the FGF coding region was excised with Hindlll and ligated into pUCl3 to obtain the intermediate plasmid pJJl5-l.

Of course. modified forms of the coding sequence to encode any of the three Known N-terminal modifications of basic FGF may also be prepared by using standard synthesis techniques.

Example 7 Construction of Expression Vectors and Stable Expression of FGF in Mammalian Cells The cDNA clones encoding FGF are most conveniently used to produce the recombinant proteins in a variety of hosts. as set forth in 1C.l above.

However. expression in mammalian systems is favored as the host is capable of post translational processing analogous to that experienced by the natively produced protein. and either cDNA or genomic sequences may be used, as the host is also capable of processing introns.

Thus, a full-length CDNA or genomic FGF encoding clone is prepared for insertion into a host vector. illustrated by. but not limited to. those described below.

To construct the vectors. the cloned FGF-encoding insert is excised with EcoRI (by partial digestion if the insert itself contains EcoRI sites), or other appropriate enzyme, provided with EcoRI or other appropriate linkers if necessary. and then inserted into an appropriate host vector such as pHS1 or its derivatives as described below.

Construction of Host Vectors REEL The plasmid pHSl is suitable for expression of inserted DNA in mammalian hosts. It contains 840 bp of the hMT-II sequence from p84H (Karin. M.. et al. Nature (1982) gggz 297-802) which spans from the HindIII site at position -765 of the hMT-II gene to the BamHI cleavage site at base + 70. To construct pHSl. plasmid p84H was digested to completion with BamHI. treated with exonuclease BAL:3l to remove terminal 1.‘Leotides. and then digested with HindIII. The desired 840 bp fragment IE95o3o1 was ligated into pUC8 (Vieira, J., et al. gene (1982) lg: 259-268) which had been opened with HindIII and Hincll digestion. The ligation mixture was used to transform E. coli HBl0l to AmpR. and one candidate plasmid. designated pHSl. was isolated and sequenced by pHSl contains the hMT-II control sequences upstream of a polylinker containing convenient restriction sites. dideoxy sequencing.

The workable host plasmid pHSl can be further modified to contain additional control elements besides In particular. the such as SV40. included. as well as termination signals associated with the metallothionein promoter. enhancer elements of viral systems. can be the 3' untranslated regions of other proteins such as hGH.

Viral Enhancer T A pair of host expression vectors containing the svao enhancer in operable linkage to the MT-II promoter was constructed by inserting an ll2o bp SV40 DNA fragment into the HindIII site preceding the MT-II The sv4o DNA fragment spans the sv4o origin of replication and includes nucleotide 5171 through nucleotide 5243 (at the origin). the duplicated 72 bp repeat from nucleotide 107-250, and promoter sequences in pHSl. continues through nucleotide l046‘on the side of the origin containing the 5' end of late viral mRNAs. This HindIII 1120 bp fragment is obtained from a HindIII digest of SV40 DNA (Buchman. A.R.. et al, DNA Tumor Viruses. 2d ed (J. Tooze."ed.). Cold Spring Harbor New York (1981), pp. 799-841). into pBR322 for amplification.

Laboratory. and cloned The cloning vector was cut with Hindlll. and the 1120 bp SV40 DNA fragment isolated by gel electrophoresis and ligated into Hindlll-digested. CIP-treated. pHSl. The resulting _45_ lE950301 vectors. designated pHSl—Sv(9) and pHSl-SV(l0). contain the fragment in opposite orientations preceding the MT—II promoter. In pHSl—SV(9). the enhancer is about 1600 bp from the 5' mRNA start site; in the opposite orientation it is approximately 980 bp from the 5' mRNA start site. Both orientations are operable. but the orientation wherein the enhancer sequences are proximal to the start site provides higher levels of expression.

It is believed that deletions which place the enhancer 250-400 bp upstream of the transcription start are optimal.

Additional vectors were constructed which place the SV40 enhancer 3' terminus 190 bp. 250 bp. and 360 bp respectively upstream from the 5' end of the MT promoter TATA box. The constructions were based on the mapping of the upstream regulatory regions of the human MT promoter described by Karin. M.. et al. Nature (1984) gggzsiz-519. All constructions retain the sequences containing the duplicated sites for regulation by heavy metals. but the constructions with the 190 bp and 250 bp separations do not retain the sequence for glucocorticoid regulation which is further upstream from these sites.

These vectors. designated pHS'-SV190. pHS'—SV2S0. and pHS'-SV360 are prepared as follows: all constructions are identical except for the length of sequence containing the metallothionein promoter and upstream region which is supplied as a fragment excised from pHSl.

For pas‘-SVl90. pHSl is digested with Sacll. blunted. and ligated to Kpnl linkers. The DNA is then digested with EcoRI and Kpnl to liberate the appropriate portion of the MT—}I control sequences. Qimilarly. for pHS‘-SVZSO. pHSl is digested with Hgal, blunted. ligated "E 9‘5 0 3'0 1 to Kpnl linkers and digested with EcoRI and KpnI: for pHS'-SV360. Ddel is used in the initial digestion.

‘ An intermediate vector containing the sv4o is prepared by inserting the HindIII/Kpnl of SV40 (which extends from position 5171 to enhancer fragment position 294 and which contains the enhancer element 50 bp from the Kpnl site) into Kpnl/Hindlll digested pUCl9 to obtain pUC-SV. (pUCl9 contains three convenient restriction sites in the polylinker region. in order.

HindIII. KpnI. and EcoRI.) obtained by inserting the Kpnl/EcoRI fragments prepared The finished vectors are as described above into Kpnl/EcoRI digested pUC-SV.

All of the foregoing modified vectors. thus, other be used in an take advantage of the SV40 enhancer element. viral enhancers could. of course. analogous manner.

Transcription Termination Sequences To provide transcription termination control sequences. DNA representing the coding sequence and 3‘ untranslated sequence of human growth hormone was ligated into pfisl. The intermediate vector can provide the hGH 3' untranslated sequence to coding sequences ~subsequently ligated into the vector in place of the hGH coding sequence.

The genomic sequences encoding hGH were isolated from p2.6-3 (DeNoto, et al. Nucleic Acids Res (1981) lg:37l9) by digestion with BamHI. which cuts at the 5' and EcoRI. of the functional gene. followed by polyacrylamide gel end of the first exon. which cuts 3' purification. The isolated fragment was ligated into BamHI/EcoRI digested pHSl and the ligation mixture transformed into E. coli MCl06l to AmpR. Successful transformants were screened by restriction analysis. and IE950301 a strain containing the desired plasmid, pMT—hGHg, was further propagated to prepare quantities of plasmid DNA.

In a manner similar to that described above for constructing pHS1—SV(9) or pHSl-SV(l0), for pHSl, pMT-hGHg. gene under the control of the MT promoter. and operably but substituting a pair of vectors containing the hGH linked to SV40 enhancer. and designated. respectively. phGHg—SV(9) and phGHg—SV(l0). were obtained. The A ligation mixtures were used to transform E. coli 1061 to AmpR, and the correct constructions verified.

Construction of Expression vectors phGHg-SV(l0) was then used as a host vector to accommodate Syn—acidic hFGF. phGHg—SV(10) was digested with BamHI and Smal. blunted with Klenow, with CIP to excise the hGH coding sequence. and treated This opened vector was ligated to the NcoI(b1unt)/EcoRI(blunt) syn—acidic hFGF fragment to obtain the desired expression vector pahFGF—SV(l0), in which the NcoI site of the syn-acidic hFGF fragment is recreated.

Similarly, the remaining FGF-encoding fragments described above are ligated into phGHg—SV(l0) to prepare analogous vectors containing these coding sequences under control of the viral enhancer. MT-II promoter and the hGH 3‘ the ~500 bp NcoI (blunt)/HindIII (blunt) fragment from pJJl5—l of Example 6 is conveniently inserted into BamHI (blunt)/SmaI—digested phGH—SV(10) to obtain pJJl6-2.

In addition. untranslated regions. For example. other host vectors may be used to obtain expression of these sequences. including pHSl and pHS1 modified to contain the various configurations of SV enhancer as above described. Insertion is by analogous means; using BamHI/EcoRI ;;,:stion of the host vector. Also. DNA modified to encode any of the "long", IE95o3o1 "primary" or "short" forms of the acidic or basic FGF may be employed.

These vectors are generically designated pMT-FGF for the purposes of the discussion below.

Production of FGF by Mammalian Recombinants Chinese hamster ovary (CHO)—Kl cells were grown on medium composed of a l:l mixture of F12 medium and DME medium with 12% fetal calf serum. The competent cells were co—transformed with pMT—FGF and pSv2:NEO (Southern. P., et al, J Mol Appl Genet (1982) l: 327-341). conferring resistance to the neomycin analog G418. In 500 ng of pSV2-NEO and 5 ug of pMT-FGF were applied to a 16 mm dish of cells in a psV2:NEo contains a functional gene the transformation. calcium phosphate-DNA co—precipitate according to the protocol of Wigler. M.. et al. QQLL (1979) lg: 777-785. with the inclusion of a two minute "shock" with 15% glycerol after four hours of exposure to the DNA. A day later. the cells were subjected to l mg/ml G418 to provide a pool of G418-resistant colonies. which were assayed for PG? production and then can be cloned out.

Successful transformants. also having a stable inheritance of pMT—FGF, are plated at low density for purification of clonal isolates. Small amounts of these isolates are grown in multi-well plates after exposure to 10-4 M zinc chloride for convenient assay of FGF production. FGF determinations are made by standard ELISA or radio-immunoassays against the antisera prepared against the appropriate FGF protein using standard methods. Clonal isolates which produce large amounts of the desired FGF are selected.

The cells. which have been shown to produce FGF under suitable conditions. are seeded at 1/10 confluency /E9503o1 in basal medium supplemented with 10% fetal calf serum. incubated overnight, and then induced for FGF production by addition of zinc chloride in the concentration range of 1 x 10'" M to 3 x 10'" M. 7-10 days, under optimal inducing conditions, 2 x lo'4 M Znclz.

FGF levels rise for In a particular experiment, CHO cells were transformed using pMT-FGF containing the approximately 500-bp Ncol(blunt)/HindIII(blunt) fragment encoding human basic FGF derived from pJJlS-1 of Example 6. This fragment was inserted into BamHI(blunt)/SmaI—digested phGH—SV(l0). particular form of pMT-FGF (designated pJJl6-2. as described above. to obtain this hereinabove). The cells were cotransformed with this and pHSl-MT. After G418 selection. pooled resistant colonies produced approximately 500 pg vector, pSV~neo, the of human basic FGF per 106 cells.

The amount of FGF produced was determined by affinity-purifying the basic FGF from lysed cells using heparin-Sepharose, followed by assay for growth The heparin affinity purification is performed by standard promotion of endothelial cells in tissue culture. methods such as those described. for example. by Sullivan. R.. et al. J Biol Chem (1985) ggg:2399—24o3, or by Shing, et al. Science (1984? ;;;:l296—l299, and the activity assay was performed using procedures as described by Gospodarowicz, D.. et al. J Cell Physiol (1985) ;;;:323—332. or Gospodarowicz. D.. et al. J Cell gig; (1983) g1:l677—l68S.

The foregoing pools. producing at a level of 500 pg/106 cells. were then selected for cadmium resistance by growing them ‘* *he presence of 10 uM CdCl2 with 100 flM Znclz as inducer. resistant clones were then assayed.

Pools of as described above, Production levels of 5.6 ng/l06 cells were found in one assay.

If desired, the FGF can be obtained from the lysed cells and purified according to the procedures set forth above for the native protein. or by other standard methods known in the art.

In addition. as discussed above. secretion of the FGF proteins produced by the foregoing constructs can be achieved by exocytosis initiated by a calcium ionophore or other suitable stimulant. While it is not expected that proteins produced by CHO cells, specifically, would be released by LPS or phorbol ester stimulation. for example. by substituting for CHO cells. cell lines derived from macrophage as recombinant hosts, Also. construction so as to provide a signal sequence. such secretion can be effected. by altering the such as that exemplified below. derived from hGH, secretion using the normal~constitutive pathways could also be effected using CR0 or other mammalian cell hosts.

Effecting secretion has some advantages. of course, since the protein purification task becomes much simpler.

Transfection with a pMT-FGF vector containing the Syn-acidic hFGF partially synthetic sequence will result in the production of "long" FGF containing the 14 amino acid pro region upstream of the 140 amino acids of the mature primary form: the processed Met residue may also be replaced with a blocking group such as acetyl.

Processing may also occur in mammalian cells to result it is established that the long form containing the leader sequence minus the in the mature form: however. initiating Met. and with the now N—terminal alanine residue acetylated. is active as a mitogen.

In any event. FGF is partially purified by passage over heparin/sepharose} find elution with 1.2 M $950330 [E 9 5 0 3'0 ? Nacl for acidic PG? and 2 M Nacl for basic FGF. The eluate is analyzed for the presence of acidic or basic FDF by SDS—PAGE and by mitogenic activity on endothelial or 3T3 cells.

Example 8 Construction of Vacciniaéyectors for Human FGF and Transient Expression in CV-1 Cells The basic hFGF—encoding sequences were provided with the 3' untranslated region from hGH by digesting phGH-SV(10) (supra) with BamHI and Smal. blunting with Klenow. and inserting the approximately 500 bp Ncol (blunt)/Hindlll (blunt) fragment spanning the FGF from pJJlS-l. used directly as an expression vector. as described The resulting plasmid. pJJl6-2. can be above.

However. the Ncol/EcoRI fragment (approximately l.l Kb) containing the basic bFGF coding region and the hGH poIyA addition signal was purified on a 5% acrylamide gel. eluted. blunted with Klenow. and ligated into smaI—digested phosphatased pGS20 (Mackett et al. Q Eiggi (1984) 131857-864). designated pJVl—l. was amplified in E. coli MC106l. and The resulting plasmid. the plasmid DNA was isolated using a cesium chloride gradient.

The Syn-acidic hFGF DNA fragment synthesized in Example 4 is also ligated into the vaccinia transient expression vector pGS20. The Syn-acidic hFGF gene shown in Figure 9 is cut with Neal and EcoRI. blunted with Klenow and ligated into smal-cut phosphatased pGS20.

The resulting plasmid preparation is purified by centrifugation to equilibrium in cesium chloride to recover the recombinant pl=smid designated pJVl—2.

/E95030’ The pJVl-1 and pJVl-2 vectors are transformed into CV—l cells infected with vaccinia, as described by Cochran. M.A.. et al. Proc Natl Acad Sci (USA) (1985) g2_:19-23. ‘ CV-1 cells transfected with pJVl—l or pGS20 were assayed for the production of basic hFGF using SDS-PAGE autoradiography. The results are shown in Figure 12. Lanes 1 and 2 are the media of cells transfected with pJV1-l and pGS20 respectively, lanes 3 and 4 are samples of the corresponding cell lysates. and lanes 5 and 6 are the same as lanes 3 and 4 except that the samples of lysate were bound to heparin sepharose in the presence of 0.6 M Nacl. washed with 10 mM phosphate. pH 7.4/l.l M Nacl. and eluted from the column with 2 M NaCl in the same buffer. (The eluates were precipitated with TCA before loading on the gel.) Lane 7 is l2sI—labe1ed basic FGF in the 146 amino acid form.

The band at approximately 18 Rd in lane 5, which has a slightly higher molecular weight than the FGF standard. shows that the "long" form of the bovine sequence is formed in preference to the "primary" protein obtained from tissues.

Samples prepared as described for lanes 5 and 6 (except for the TCA precipitation) were also tested for mitogenic activity on bovine brain capillary endothelial No activity was present in the cells. (See Example 9.) pGS20 sample. but the pJVl-l sample contained activity equivalent to 20 pg FGF/ul.

Example 9 In Vjgro Aggy for FGF The assay was performed substantially as described by Esch et al. Proc Natl Acad sci (USA) (1985) cells were maintained in the presence of DMEM supplemented with 10% calf serum. Monolayers were dissociated by exposure to a solution containing 0.9% Nacl. pH 7.4. 0.05% trypsin. and 0.02% EDTA for 2-3 minutes at room temperature. .01 M sodium phosphate.

After the cells had rounded up. they were resuspended in DMEM and 102 calf serum and an aliquot of the cell The cells were seeded at an initial density of 2 x 104 cells per mm dish. each dish containing a total of 2 ml DMEM suspension was counted in a Coulter counter. plus 10% calf serum. Six to twelve hours later. a set of duplicated plates was trypsinized and cells were counted to determine the plating efficiency.

Aliquots of the sample to be tested for FGF activity were diluted 1:2. 1:4. and l:8 with DMEM plus 0.5% BSA. and 10 ul of the dilutions were added to triplicate assay plates on days 0 and 2. on day 4. the triplicate plates for each sample dilution were trypsinized and the cell densities determined by Coulter COUIICEZ .

Example 10 Expression of Signal Sequence Fusions Since the recombinant forms of FGF were not found in the medium of CHO or CV-l cells. the PG?-encoding DNA sequences are also ligated to a heterologous signal sequence in order to effect the secretion of the recombinant FGF protein. The fused sequences are then ligated into vaccinia-based vectors to effect transient expres""# and secretion in ’E95O3o1 vaccinia—infected CV—l cells, or into pHSl based vectors for expression and secretion in CHO cells.- _ DNAs encoding four forms of basic hFGF are supplied as partially synthetic fragments each containing a constant C—terminal fragment from the altered XBB2 clone described above and a variable synthetic N-terminal portion. For the C-terminal position. the altered XBBZ clone is digested with Hhal. which cuts 122 bp downstream from the initiating and with Hindlll. to obtain a 377 bp subfragment extending to the HindIII site in the 3' methionine codon. untranslated region. The missing portions upstream from the Hhal site are supplied using synthetic oligonucleotides.

’E95O3gg The synthetic oligonucleotides are shown in Figure 13. They are synthesized and ligated in a manner analogous to that described above in connection with the production of Syn-acidic hFGF. The individual oligonucleotides are synthesized using a standard automated nucleic acid synthesizer. annealed, and the double-stranded portions ligated to form the pertinent entire upstream portion containing a Hhal site at its 3‘ end. These synthetic upstream fragments are then ligated using T4 ligase to the downstream Hhal/HindIII fragment to obtain the entire FGF gene. and then ligated to the hGH BamHI-Ndel fragment to add the hGH signal sequence coding region.

The hGH/basic FGF protein junctions encoded by the synthetic upsteam portions are shown in Figure 14.

Protein A contains the reconstructed upstream portions and ligated C—terminal codons thus encoding amino acids -9 to 146. the total of 155 amino acids thus encoding the long form of basic human FGF. shown in Figure 4.

Protein E contains the same sequence. except that the N—terminal methionine at -9 has been replaced by an alanine residue. Protein C encodes amino acids 12-146 of Figure 4. and protein D encodes amino acids 16-146 of the "short" which is already known to show mitogenic activity. this protein, i.e., form of human basic FGF.

Figure 14 also shows the predicted signal sequence cleavage sites (in heavy arrows) for the immediate expression product. according to the rules set forth by von Heijne. G., Eur J Biochem (1983) ;;;:l7-21.

To complete the constructions. the fragments encoding the four proteins fused to hGH signal sequence set forth in Figure 14 are inserted into carrier plasmid pUC9 for amplification as a 9amHI(partial)/HindIII sequence and corréct construction is confirmed by dideoxy sequencing. The BamHI(partia1)/Hindllf confirmed sequence fragment is then excised from the carrier plasmid. blunted with Klenow. and ligated into the Smal site of pGS20 (supra). and the ligated recombinant plasmid expressed in vaccinia—infected CV1 cells. is described above.

Analogously. constructs are made of acidic hFGF for secretion in similar expression systems. The FGF-encoding sequences are derived from the Syn-acidic hFGF DNA fragment. modified to produce proteins E. F. and G in Figure 15, which represent. respectively. hGH/acidic FGF protein junction regions of the long form of acidic FGF spanning the residues -15 to 140 of Figure 2b. the primary form represented by residues 1-140 of Figure 2b. and the short form spanning residues 7-140 of that figure. all with minor changes in the FGF amino-terminus. as shown in the figure. the Ncol/EcoRI Syn—acidic FGF is blunted with Klenow at the Ncol site and ligated into Smal/EcoRI-digested M13mp19.

The resultant phage is mutagenized with the oligonucleotide: 5'-GTAATTCCCGGGAGGCAGAT-3'. thus creating a Smal site at nucleotide position 62 of Figure To construct these FGF—encoding sequences. immediately before the codon for the glycine. which is residue 6 of the primary acidic hFGF. Digestion of the mutagenized fragment with Smal and EcoRI provides a 414 bp fragment. which is either ligated directly to the BamHI/Nael hGH signal-encoding fragment to obtain a DNA encoding a recombinant form shown as protein G in Figure . or is first ligated to the synthetic oligonucleotides (shown in Figure 16) and then to the BamHI/Nae! fragment to obtain the sequences encoding the peptides designated E and F in Figure 15.

,E950330 ’E'95O3oy y In a manner exactly analogous to that set forth above for human basic FGF. the signal sequence-preceded acidic FGF DNA fragments are disposed in pGS2o and transfected into vaccinia—infected CV-1 cells to assay for transient expression of secreted forms of acidic FGF. proteins are indicated by heavy arrows in Figure 15.

(The expected signal cleavage sites of the treated also deduced according to von Heijne.) It should be noted that the FGF sequences do not contain traditional signal sequences. and accordingly do not have the capacity to effect their own secretion by the signal hypothesis mechanism under constitutive conditions. It is unclear from the art whether fusing a foreign signal sequence to normal cytoplasmic proteins is capable of effecting their secretion. (It has been shown that B-galactosidase _ fused to the malE signal sequence does not reach the periplasm in bacteria, although Lingappa. V.R., et al.

Proc Natlégcad Sci (supra) show that B—globin fused to the B—lactamase signal is processed by dog pancreatic microsomes in vitro. and the processed protein is protected from trypsin digestion.) The ligated signal/FGF-encoding sequences may also be inserted into the MT-II promoter—containing host vectors described above and expressed in CHO cells.

Example 11 gacterial Expression of FGF The CDNA sequences encoding FGF. which are uninterrupted by introns, are also expressible in bacterial systems. A convenient host vector for expression is pKTS2, which contains the "trc" promoter, The contains the upstream portions of the trp promoter and followed by an ATG start c¢‘on. "trc" promoter !E9503o1 the dOWnStIeam. operator—containing. regions of the 13¢ promoter. pKT52. containing this promoter was constructed by a simple manipulation of pKK233—2. which is described by Amman. E.. et al. Qggg (1935) gQ:l83-190: pKK233—2 was digested with EcoRI and Pvull. filled in with dATP and dTTP. and religated to obtain the desired vector. pKTS2 contains in addition to the desired trc promoter and downstream ATG start codon, downstream Ncol. PstI and HindIII sites.

For construction of expression vectors. the FGF-encoding CDNA is obtained by excising with EcoRI or other appropriate enzyme digestion and isolating and, if The 3' end is prepared for insertion into pKTS2 by cutting necessary. modifying the appropriate fragment. downstream of the termination codon at any convenient restriction site and supplying PstI or HindIII linkers.

The 5‘ end is prepared by cutting at a site inside the coding sequence and supplying the missing codons and an Ncol site using a synthetic DNA. or by providing an The resulting Ncol/Hindlll or Ncol/Pstl fragment is then ligated into Ncol/HindIII-digested pKT52 or Ncol/Pstl digested pKTS2 to provide the FGF-encoding cDNA in reading frame with the ATG start codon. The Ncol/HindIII—bordered Syn-acidic hFGF DNA is inserted directly in this_way. appropriately located Ncol site by mutagenesis.

Similar vectors are constructed using the human basic FGF encoding DNA.

For bacterial expression. the resulting expression vectors are used to transform E. coli MClO6l and the transformed cells are then grown on M9 medium containing 1 mM IPTG for 375 hr to an O.D: of 0.2-0.5. (IPTG is a standard inducer for control seguences regulated by the or other appropriate host cells to AmpR, lac operator.) The cells are then harvested. lysed by sonication or treatment with 52 trichloroacetic acid.

FGF be purified from the extracts by methods used for and the cell extracts assayed for the desired FGF. can the native protein or by other procedures known in the art.

Example 12 Activity of FGF in Promotinq Wound Healing al. In the alternative. al. J Surg Res (1982) _3_3_:394—4o1).

In the standard procedure. a total of four rats The sponges were either not treated. treated with heparin received two identically treated sponges each. sepharose beads. treated with FGF bound to heparin sepharose beads using 5 ug FGF per sponge: or treated with 5 ug FGF in after 6 days and solution. The sponges were removed examined histologically for granulation indicative of wound healing. tissue. which is sponges which contained FGF showed a higher amount of granulation. which was centered around the heparin sepharose beads in the case of the sponges where the FGF was supplied bound to these beads.

Similar results are observed whether the FGF is from native or 2eCOmbifi‘T sources and whether the FGF is basic or acidic.

/E95030; lE9503o1 on or before 9 September 1985. Applicant deposited with the American Type Culture Collection (ATCC). Rockville. MD. USA, the X phage XBA2.

XBA3. XHAG-9.1. K882. and XKB-7 which were assigned ATCC accession numbers 40195. 40194. 40197, 40196. and 40198. respectively. made under conditions as provided under ATCC's agreement These deposits were for Culture Deposit for Patent Purposes. on or before 12 September 1986. conditions of deposit or XBB2 (ATCC 40196) and XHAG-9.1 (ATCC 40197) were converted to conform to those specified under the Budapest Treaty on the International Recognition of the Deposit of Microorganisms (Budapest Treaty). on or before September 1986. the X phages designated 1HG—3 and XHAG-3 were deposited at ATCC under the terms of the Budapest Treaty and were assigned ATCC accession numbers 40257 find ‘@258 . of the deposited strains is not to be construed as a respectively. Availability license to practice the invention in contravention of the rights granted under the authority of any government in accordance with its patent laws.

Claims (3)

1. An isolated, cloned recombinant or synthetic DNA sequence encoding basic human FGF.

2. A DNA Sequence according to claim 1 which is obtainable by screening one or more DNA libraries for a sequence hybridising, in a prehybridisation hybridisation buffer containing 40% formamide, 5mM sodium phosphate, 5 x Denhart's (0.1% bovine serum albumin, 0.1% polyvinyl pyrollidone and 0.1% Ficoll), 5 x SSC (0.75M Nacl, 0.075M Na Citrate), Song/ml herring sperm DNA and 10% dextran sulphate, at pH 6.5, to the 1.4kb basic bovine FGF encoding fragment obtained by EcoR1 digestion of ABB2 (deposited as ATCC 40196), and isolating the sequence hybridised.

3. A DNA sequence according to claim 1 or 2 which encodes an FGF protein comprising the amino acid sequence numbered 16-146 in