Vascular Endothelial Growth Factor-B and Dna Coding Therefor

Angiogenesis, or the proliferation of new capillaries from pre-existing blood vessels, is a fundamental process necessary for normal growth and development of tissues. It is a prerequisite for the development and differentiation of the vascular tree, as well as for a wide variety of fundamental physiological processes including embryogenesis, somatic growth, tissue and organ repair and regeneration, cyclical growth of the corpus luteum and endometrium, and development and differentiation of the nervous system. In the female reproductive system, angiogenesis occurs in the follicle during its development, in the corpus luteum following ovulation and in the placenta to establish and maintain pregnancy. Angiogenesis additionally occurs as part of the body's repair processes, e.g. in the healing of wounds and fractures. Angiogenesis is also a factor in tumor growth, since a tumor must continuously stimulate growth of new capillary blood vessels in order to grow.

Capillary blood vessels consist of endothelial cells and pericytes. These two cell types carry all of the genetic information to form tubes, branches and entire capillary networks. Specific angiogenic molecules can initiate this process. In view of the physiological importance of angiogenesis, much effort has been devoted to the isolation, characterization and purification of factors that can stimulate angiogenesis. A number of polypeptides which stimulate angiogenesis have been purified and characterized as to their molecular, biochemical and biological properties. For reviews of such angiogenesis regulators, see Klagsbrun et al., "Regulators of Angiogenesis", Ann. Rev. Physiol., 53:217-39 (1991); and Folkman et al., "Angiogenesis," J. Biol. Chem., 267:10931-934 (1992). Recent results have implicated several endothelial receptor tyrosine kinases (RTKs) in the establishment and maintenance of the vascular system.

One such growth factor, which is highly specific as a mitogen for vascular endothelial cells, is termed vascular endothelial growth factor (VEGF). See Ferrara et al., "The Vascular Endothelial Growth Factor Family of Polypeptides," J. Cellular Biochem., 47:211-218 (1991); Connolly, "Vascular Permeability Factor: A Unique Regulator of Blood Vessel Function," J. Cellular Biochem., 47:219-223 (1991). VEGF is a potent vasoactive protein that has been detected in media conditioned by a number of cell lines including bovine pituitary follicular cells. VEGF is a glycosylated cationic 46-48 kD dimer made up of two 24 kD subunits. It is inactivated by sulfhydryl reducing agents, resistant to acidic pH and to heating, and binds to immobilized heparin. VEGF is sometimes referred to as vascular permeability factor (VPF) because it increases fluid leakage from blood vessels following intradermal injection. It also has been called by the name vasculotropin.

Four different molecular species of VEGF have been detected. The 165 amino acid species has a molecular weight of approximately 46 kD and is the predominant molecular form found in normal cells and tissues. A less abundant, shorter form with a deletion of 44 amino acids between positions 116 and 159 (VEGF.sub.121) a longer form with an insertion of 24 highly basic residues in position 116 (VEGF.sub.189), and another longer form with an insertion of 41 amino acids (VEGF.sub.206), which includes the 24 amino acid insertion found in VEGF.sub.189, are also known. VEGF.sub.121 and VEGF.sub.165 are soluble proteins. VEGF.sub.189 and VEGF.sub.206 appear to be mostly cell-associated. All of the versions of VEGF are biologically active. For example, each of the species when applied intradermally is able to induce extravasation of Evans blue.

The various species of VEGF are encoded by the same gene and arise from alternative splicing of messenger RNA. This conclusion is supported by Southern blot analysis of human genomic DNA, which shows that the restriction pattern is identical using either a probe for VEGF.sub.165 or one which contains the insertion in VEGF.sub.206. Analysis of genomic clones in the area of putative mRNA splicing also shows an intron/exon structure consistent with alternative splicing.

Analysis of the nucleotide sequence of the VEGF gene indicates that VEGF is a member of the platelet-derived growth factor (PDGF) family. VEGF and PlGF are ligands for two endothelial RTKs, flt-1 (VEGF receptor 1, VEGFR1) and flk-1/KDR (VEGF receptor 2, VEGFR2). The amino acid sequence of VEGF exhibits approximately 20% homology to the sequences of the A and B chains of PDGF, as well as complete conservation of the eight cysteine residues found in both mature PDGF chains. VEGF.sub.165, VEGF.sub.189 and VEGF.sub.206 also contain eight additional cysteine residues within the carboxy-terminal region. The amino-terminal sequence of VEGF is preceded by 26 amino acids corresponding to a typical signal sequence. The mature protein is generated directly following signal sequence cleavage without any intervening prosequence. The existence of a potential glycosylation site at Asn.sup.74 is consistent with other evidence that VEGF is a glycoprotein, but the polypeptide has been reported to exist in both glycosylated and deglycosylated species.

Like other cytokines, VEGF can have diverse effects that depend on the specific biological context in which it is found. VEGF and its high affinity receptors flt-1 and KDR/flk-1 are required for the formation and maintenance of the vascular system as well as for both physiological and pathological angiogenesis. VEGF is a potent endothelial cell mitogen and directly contributes to induction of angiogenesis in vivo by promoting endothelial cell growth during normal development or during wound healing. A most striking property of VEGF is its specificity. It is mitogenic in vitro at 1 ng/ml for capillary and human umbilical vein endothelial cells, but not for adrenal cortex cells, corneal or lens epithelial cells, vascular smooth muscle cells, corneal endothelial cells, granulosa cells, keratinocytes, BHK-21 fibroblasts, 3T3 cells, rat embryo fibroblasts, human placental fibroblasts and human sarcoma cells. The target cell specificity of VEGF is thus restricted to vascular endothelial cells. VEGF can trigger the entire sequence of events leading to angiogenesis and stimulates angiogenesis in vivo in the cornea and in a healing bone graft model. It is able to stimulate the proliferation of endothelial cells isolated from both small and large vessels. Expression of VEGF mRNA is temporally and spatially related to the physiological proliferation of capillary blood vessels in the ovarian corpus luteum or in the developing brain. VEGF expression is triggered by hypoxemia so that endothelial cell proliferation and angiogenesis appear to be especially stimulated in ischemic areas. VEGF is also a potent chemoattractant for monocytes. In addition, VEGF induces plasminogen activator and plasminogen activator inhibitor in endothelial cells.

Tumor cells release angiogenic molecules such as VEGF, and monoclonal antibodies to VEGF have been shown to inhibit the growth of certain types of tumor such as rhabdomyosarcoma. See Kim et al., "Inhibition of Vascular Endothelial Growth Factor-Induced Angiogenesis Suppresses Tumor Growth in vivo," Nature, 362:841-844 (1993). This suggests that blocking VEGF action is of potential therapeutic significance in treating tumors in general, and highly-vascularized, aggressive tumors in particular.

It is an object of the invention to provide a new growth factor having the property of promoting proliferation of endothelial cells.

Another object of the invention is to provide isolated DNA sequences which encode a new growth factor which promotes proliferation of endothelial cells.

It is also an object of the invention to provide new products which may be useful in diagnostic and/or therapeutic applications.

These and other objects are achieved in accordance with the present invention by providing an isolated DNA which codes for a protein exhibiting the following characteristic amino acid sequence

Pro-Xaa-Cys-Val-Xaa-Xaa-Xaa-Arg-Cys-Xaa-Gly-Cys-Cys

(SEQ ID NO:16) and having the property of promoting proliferation of endothelial cells or mesodermal cells, the DNA being selected from the group consisting of the DNA of FIGS. 1 and 2 (SEQ ID NO:1), the DNA of FIG. 3 (SEQ ID NO:4), the DNA of FIG. 5 (SEQ ID NO:6); the DNA of FIG. 7 (SEQ ID NO:8), the DNA of FIG. 10 (SEQ ID NO:10), the DNA of FIG. 12 (SEQ ID NO:12), the DNA of FIG. 14 (SEQ ID NO:14), and DNA's which hybridize under stringent conditions with at least one of the foregoing DNA sequences.

In accordance with further aspects of the invention, the objects are also achieved by providing a protein exhibiting the following characteristic amino acid sequence

Pro-Xaa-Cys-Val-Xaa-Xaa-Xaa-Arg-Cys-Xaa-Gly-Cys-Cys

(SEQ ID NO:16) and having the property of promoting proliferation of endothelial cells or mesodermal cells, which protein comprises a sequence of amino acids substantially corresponding to an amino acid sequence selected from the group consisting of the amino acid sequence of FIG. 1 (SEQ ID NO:2), the amino acid sequence of FIG. 2 (SEQ ID NO:3), the amino acid sequence of FIG. 4 (SEQ ID NO:5), the amino acid sequence of FIG. 6 (SEQ ID NO:7), the amino acid sequence of FIG. 8 (SEQ ID NO:9), the amino acid sequenc of FIG. 11 (SEQ ID NO:11), the amino acid sequence of FIG. 13 (SEQ ID NO:13), and the amino acid sequence of FIG. 15 (SEQ ID NO:15).

In further aspects of the invention, the objects are achieved by providing pharmaceutical preparations which comprise such proteins; and by providing antibodies which react with such proteins.

The novel growth factor of the present invention, referred to hereinafter as vascular endothelial growth factor B or VEGF-B, has close structural similarities to VEGF and to placenta growth factor (PlGF). All of the VEGF-B forms contain the characteristic amino acid sequence

Pro-Xaa-Cys-Val-Xaa-Xaa-Xaa-Arg-Cys-Xaa-Gly-Cys-Cys

(SEQ ID NO:16) (wherein Xaa represents a variable residue) , which is an earmark of the PDGF/VEGF family of growth factors. This characteristic amino acid sequence can be found at amino acids 70 to 82 in FIGS. 4, 6, 8, 11, 13 and 15.

Clinical applications of the invention include diagnostic applications, acceleration of angiogenesis in wound healing, and inhibition of angiogenesis. Quantitation of VEGF-B in cancer biopsy specimens may be useful as an indicator of future metastatic risk. Topical application of VEGF-B preparations to chronic wounds may accelerate angiogenesis and wound healing.

FIG. 1 shows the nucleotide sequence of the (partial) cDNA clone of VEGF-B (SEQ ID NO:1) and the amino acid sequence of the protein segment (SEQ ID NO:2) coded by the first reading frame of the cDNA;

FIG. 2 repeats the nucleotide sequence of the (partial) cDNA clone of VEGF-B (SEQ ID NO:1) and the amino acid sequence of the protein segment (SEQ ID NO:3) coded by the second reading frame of the cDNA;

FIG. 3 shows the nucleotide sequence of the coding region of a full length cDNA clone of VEGF-B.sub.167 (SEQ ID NO:4);

FIG. 4 shows the amino acid sequence of VEGF-B.sub.167 (SEQ ID NO:5);

FIG. 5 shows the nucleotide sequence of the coding region of a cDNA clone of VEGF-B.sub.174 (SEQ ID NO:6);

FIG. 6 shows the amino acid sequence of VEGF-B.sub.174 (SEQ ID NO:7);

FIG. 7 shows the nucleotide sequence of a cDNA clone of VEGF-B.sub.112 (SEQ ID NO:8);

FIG. 8 shows the amino acid sequence of VEGF-B.sub.112 (SEQ ID NO:9);

FIG. 9 shows a comparison of the amino acid sequences of mVEGF-B.sub.167 (SEQ ID NO:5), mVEGF.sub.164 (SEQ ID NO:20), hPlGF (SEQ ID NO:21), mPDGF A (SEQ ID NO:22), and mPDGF B (SEQ ID NO:23);

FIG. 10 shows the nucleotide sequence of a clone of human VEGF-B.sub.167 (SEQ ID NO:10);

FIG. 11 shows the amino acid sequence of human VEGF-B.sub.167 (SEQ ID NO:11); and

FIG. 12 shows the nucleotide sequence of murine VEGF-B.sub.186 (SEQ ID NO:12);

FIG. 13 shows the amino acid sequence of murine VEGF-B.sub.186 (SEQ ID NO:13);

FIG. 14 shows the nucleotide sequence of human VEGF-B.sub.186 (SEQ ID NO:14);

FIG. 15 shows the amino acid sequence of human VEGF-B.sub.186 (SEQ ID NO:15);

FIG. 16 shows an amino acid sequence comparison of murine and human VEGF-B.sub.167 and VEGF-B.sub.186 isoforms (SEQ ID NOS:5, 11 13 and 15).

FIG. 17 shows the schematic structure of mouse and human genes for VEGF-B;

FIG. 18 shows a hydrophilicity analysis of murine VEGF-B.sub.167 and VEGF-B.sub.186 isoforms;

FIG. 19 shows a phylogenetic analysis of the VEGF/PDGF family of growth factors; and

FIG. 20 is a graph showing the induction of ›.sup.3 H!thymidine incorporation by VEGF-B, VEGF and bFGF for human umbilical vein endothelial cells (HUVEC) and bovine capillary endothelial (BCE) cells.

The present invention thus is directed to new vascular endothelial growth factors, hereinafter referred to as VEGF-B growth factors, which share the angiogenic and other properties of VEGF, but which are distributed and expressed in tissues differently from VEGF.

VEGF-B growth factors are members of the family of platelet derived growth factors and are a growth factors which promote mitosis and proliferation of vascular endothelial cells and/or mesodermal cells. They are produced by expression of DNA sequences which correspond to, or which are hybridizable under stringent conditions with, any one of the DNA sequences depicted in FIGS. 1 and 2 (SEQ ID NO:1), FIG. 3 (SEQ ID NO:4), FIG. 5 (SEQ ID NO:6), FIG. 7 (SEQ ID NO:8), FIG. 10 (SEQ ID NO:10), FIG. 12 (SEQ ID NO:12) or FIG. 14 (SEQ ID NO:14). It is intended to include within the scope of the invention all angiogenic proteins encoded by DNA sequences which hybridize under stringent conditions to any one of the foregoing DNA sequences. Suitable hybridization conditions include, for example, 50% formamide, 5.times. SSPE (=0.9M sodium chloride, 0.05M sodium phosphate, pH 7.7 and 0.005M ethylene diamine tetraacetic acid (EDTA)) buffer, 5.times. Denhardts solution, 0.5% SDS (=sodium dodecyl sulfate) and 100 .mu.g/ml of salmon sperm DNA at 42.degree. C. overnight, followed by washing 2.times.30 minutes in 2.times. SSC (=0.75M sodium chloride and 0.075M sodium citrate) at 55.degree. C.

The invention is also directed to an isolated and/or purified DNA which corresponds to, or which hybridizes under stringent conditions with, any one of the foregoing DNA sequences.

In a further aspect, the invention is directed to antibodies of VEGF-B growth factors, and particularly to monoclonal antibodies.

VEGF-B proteins are believed to interact with protein tyrosine kinase growth factor receptors. Details of such receptors are known in the art ›See e.g. Wilks, A. F., "Protein Tyrosine Kinase Growth Factor Receptors and Their Ligands in Development, Differentiation, and Cancer," Adv. Cancer Res., 60:43-73 (1993)!.

Various adult mouse tissues were tested for expression of transcripts corresponding to VEGF-B by Northern blotting. The size of the mRNA was 1.3-1.4 kb. A mouse multiple tissue Northern blot (MTN, Clontech) was probed with the 0.89 kb Sall-Not1 fragment derived from the pPC67 yeast expression vectors described above. The probe was labelled with .sup.32 P-dCTP using random priming (specific activity 10.sup.8 -10.sup.9 cpm/.mu.g of DNA). The blot was hybridized overnight at 42.degree. C. using 50% formamide, 5.times. SSPE buffer, 2% SDS, 10.times. Denhardts solution, 100 .mu.g/ml salmon sperm DNA and 1.times.10.sup.6 cpm of the labelled probe/ml. The blot was washed at room temperature for 2.times.30 min in 2.times. SSC containing 0.05% SDS and then for 2.times.20 min at 52.degree. C. in 0.1.times. SSC containing 0.1% SDS. The blot was then exposed at -70.degree. C. for three days using intensifying screens. Kodak XAR film was used. The relative expression levels as determined by visual examinations of the film are listed in the following table:

A human multiple tissue Northern blot (MNT) from Clontech was probed using the murine partial cDNA to determine relative VEGF-B expression levels in various human tissues. The size of the transcript was 1.3-1.4 kb. The conditions were identical to those used for the mouse Northern blot described above. The relative VEGF-B transcript levels for the human Northern blot are listed in the following Table 2. For comparison purposes, Table 2 also lists relative expression level data from the literature for VEGF in various mammalian systems.

From a comparison of Table 1 and Table 2 it can be seen that mouse and human tissue expression levels of VEGF-B transcripts are relatively similar with the highest expression levels being found in heart and skeletal muscle. Significant differences may be seen in brain and kidney tissue. It should also be noted that tissues containing a large proportion of both muscular and epithelial cells, such as prostate, pancreas and colon from which some of the most common human tumors originate, express relatively high levels of VEGF-B.

A comparison of the relative expression levels of VEGF and VEGF-B in human tissues shows some striking differences. VEGF is expressed rather weakly by human heart tissue, but VEGF-B is very strongly expressed by the same tissue. On the other hand, VEGF is strongly expressed by human lung tissue, but VEGF-B is only weakly expressed by human lung tissue. In a similar vein, human liver tissue expresses VEGF at a moderate level, but VEGF-B is expressed only very weakly. These data evidence that despite their general similarities, the actions of VEGF and VEGF-B are not completely identical.

The expression of VEGF-B transcripts was further analyzed in mouse and human tissues by Northern blotting and compared with the expression of VEGF transcripts. Mouse and human multiple tissue Northern (MTN) blots (Clontech) were hybridized with a 32P-labelled mouse VEGF-B probe (0.9 kb Sal I/Not I insert of the clone pcif 2). VEGF expression was analysed with .sup.32 P-labelled VEGF.sub.165 cDNA as the probe. The hybridizations were carried out at 42.degree. C. in 50% deionized formamide, 5.times. SSC pH 7.0, 1% SDS, 5.times. Denhardt's solution and 100 .mu.g/ml of denatured salmon sperm DNA. The filters were washed 2.times.30 min at 52.degree. C. in 2.times. SSC containing 0.5% SDS and exposed to Kodak XAR film for 2-5 days at -70.degree. C. using intensifying screens. In situ hybridization analysis of adult mouse tissues from CBA mice and of embryos derived from matings of CBA and NMRI mice were carried out essentially as previously described by Korhonen et al., Blood, 80, 2548-55 (1992). The RNA probes (a 383 bp antisense probe and a 169 bp sense probe) were generated from a linearized plasmid containing a 440 bp Sal I/Sac I fragment derived from the pcif 2 cDNA clone. Radiolabelled RNA was synthesized using T7 and SP6 RNA polymerases and ›.sup.35 S!UTP (Amersham Inc.). Alkaline hydrolysis of the probes was omitted. Hematoxylin was used for counterstaining. Control hybridizations with sense strand and RNAse A-treated sections did not give signals above background.

In mouse tissues the most abundant expression of the 1.4 kb VEGF-B transcript was detected in heart, brain, skeletal muscle, and kidney. The major 3.7 kb VEGF transcript was is expressed in heart, brain, lung, skeletal muscle and kidney. In human tissues, the most abundant expression of the 1.4 kb VEGF-B transcript and the major 3.7 and 4.5 kb VEGF transcripts were detected in heart, skeletal muscle, pancreas and prostate. Thus, although clear quantitative differences exist, it appears that VEGF-B and VEGF are coexpressed in many human and mouse tissues.

The expression of VEGF-B transcripts was further examined by in situ hybridization in sections from adult mouse heart and skeletal muscle and from the early (E 10) mouse embryo. In the adult heart, VEGF-B transcripts are prominently expressed in the myocardium, while no specific signal is detected in arterial smooth muscle. In adult striated muscle, VEGF-B transcripts are expressed by some of the myofibers whereas others seem to lack the transcript. In the E 10 mouse embryo, VEGF-B transcripts are detected mainly in the developing heart. The myocardium of the adult mouse heart has a prominent signal. In striated muscle, VEGF-B expression is seen in subpopulations of myofibers. Strong signals were also obtained in the developing heart of the E10 mouse embryo. Other embryonic structures expressed lower or undetectable levels of transcripts for VEGF-B. Taken together, these tests indicate that VEGF-B transcrips are expressed primarily in muscular tissues. VEGF-B is particularly abundant in heart and skeletal muscle and is co-expressed with VEGF in these and other tissues. In transfected cells, VEGF-B forms cell surface associated, disulfide-linked homodimers and heterodimers with VEGF when coexpressed.

A partial cDNA clone encoding murine VEGF-B was identified as follows. A cDNA library (E 14.5) derived from poly A+ mRNA isolated from 14.5 day old mouse embryos ›Chevray P. and Nathans D., "Protein interaction cloning in yeast: Identification of mammalian proteins that react with the leucine zipper of Jun," Proc. Natl. Acad. Sci. USA, 89:5789-93 (1992)! was screened for cellular proteins which potentially might interact with cellular retinoic acid-binding protein type 1 (CRABP-I) using a yeast two-hybrid interaction trap screening technique as described by Gyuris J., Golemis E., Chertkov H. and Brent R., "Cdil, a Human G1 and S Phase Protein Phosphatase That Associates with Cdk2," Cell, 75:791-803 (1993). This screening technique involves a fusion protein that contains a binding domain and that is known to be transcriptionally inert (the "bait"); reporter genes that have no basal transcription and that are bound by the bait; and an expression library which encodes proteins expressed as chimeras and whose amino termini contain an activation domain and other useful moieties (the "prey"). The screened library was a plasmid library in the yeast expression vector pPC67 obtained from Dr. Pierre Chevray of the Johns Hopkins University, School of Medicine, 725 North Wolfe St., Baltimore, Md. 21205. A positive cDNA clone (pcif-2) was recovered from the screening. The positive clone was sequenced using well known, conventional techniques and found to encode a protein highly homologous to VEGF and the other members of the PDGF family of growth factors. The 890 base pair SalI-NotI insert in the plasmid pPC67 was cloned into pBluescript and fully sequenced using T7 and T3 vector primers together with internal primers. The plasmid pBluescript is commercially available from Stratagene Inc., LaJolla, Calif. The cDNA insert was found to be 886 base pairs long and to encode two polypeptides in different reading frames which were homologous to the N-terminal end and the C-terminal end, respectively, of VEGF. This novel growth factor is referred to hereinafter as VEGF-B. The clone is partial and lacks several amino acids in the amino terminal region and the entire signal sequence.

FIG. 1 shows the nucleotide sequence (SEQ ID NO:1) of this partial cDNA clone of VEGF-B and the amino acid sequence (SEQ ID NO:2) encoded in the first reading frame thereof. The DNA sequence of FIG. 1 was obtained by conventional sequencing of a clone (pcif-2) in the yeast expression vector pPC67. The clone comprised 886 base pairs and encoded a part of murine VEGF-B.

The isolated cDNA sequence will hybridize with the mammalian genomic DNA, e.g. either murine or human, which contains the VEGF-B gene. In addition to the coding sequence, the genomic DNA will contain one or more promoter sequence(s) which give and direct expression of VEGF-B in one or more specific tissues. Thus the coding sequence of VEGF-B may be linked to a muscular specific promoter which is specific to a certain type or types of tissue.

The nucleotide sequence is translated in two different reading frames into two different amino acid sequences. There is a stop codon (TGA) within the coding sequence at base pairs 309-311. Thus, VEGF-B comes in several splicing variants. The 5' end of the cloned cDNA sequence encodes an 102 amino acid long protein with significant homology to the N-terminal domains of VEGF, PlGF and PDGF A and B. In particular, a number of cysteine residues are perfectly conserved within this group of proteins. In addition to the nucleotide sequence (SEQ ID NO:1), FIG. 1 further depicts the deduced amino acid sequence (SEQ ID NO:2) of this first protein.

Translation of the C-terminal end of the cDNA (base pairs 308-475) in a different reading frame results in a protein which is highly homologous to the C-terminal part of VEGF.sub.165, VEGF.sub.189 and VEGF.sub.206. FIG. 2 again shows the nucleotide sequence (SEQ ID NO:1) of FIG. 1, but this time includes the deduced amino acid sequece (SEQ ID NO:3) of the second protein, which is encoded in the second reading frame and is 54 amino acids long. It thus appears that the VEGF-B gene encodes different proteins using alternative splicing of the primary transcript. The last part of the clone, encoding the second peptide might be expressed as a functional protein in other spliced variants of VEGF-B.

The aforedescribed proteins may exist in combined association with an additional N-terminal sequence of approximately five (5) to ten (10) amino acids, as well as a further leader sequence of approximately twenty-one (21) to twenty-eight (28) amino acids. Inasmuch such combined amino acid sequences exhibit the property of promoting the proliferation of endothelial cells and the DNA sequences which code for such combined peptide sequences will hybridize under stringent conditions with the DNA sequence of FIGS. 1 and 2, such amino acid sequences and the DNA which codes for them are expressly contemplated to be within the scope of the present invention.

Using the approximately 0.9 kb Sal-I/Not-I cDNA insert of the previously identified cDNA clone of Example 1 as a probe, an adult mouse heart lambda ZAP-II cDNA library obtained from Stratagene Inc., of La Jolla, Calif. was screened using standard techniques. The library was titrated and plated as recommended and filters were prepared. Following prehybridization at 42.degree. C. in 50% formamide, 5.times. SSPE, 5.times. Denharts solution, 1% SDS and 100 ug of salmon sperm DNA/ml, the filters were hybridized at the same temperature and in the same solution containing the denatured radiolabelled probe using 10.sup.6 cpm/ml of hybridization solution. The probe was labelled using a random priming kit (Amersham). After 16 hours the filters were washed in 2.times. SSC containing 0.5% SDS for 2.times.30 mins at 52.degree. C. The filters were exposed overnight using intensifying screens at -70.degree. C. Positive clones were rescreened two times until all plaques on a plate were positive. The inserts from several positive clones were subcloned into the plasmid pbluescript SK+ by in vivo excision as recommended by the supplier.

Several clones were mapped by restriction enzyme analysis and were found to fall into two distinct groups characterized by the length of a Spe1/BamH1 restriction fragment. The first of these groups comprised three of the restriction mapped clones which each had a 240 bp Spe1/BamH1 restriction fragment. The other group comprised a clone which had a 340 bp Spe1/BamH1 fragment. Analysis of this fragment is described in Example 5.

The three clones which exhibited the 240 bp Spe1/BamH1 restriction fragment were fully or partially sequenced (Sequenase 2.0, U.S. Biochemicals), and the characteristics of the clones are summarized as follows:

Nucleotide sequence analyses revealed that two of the cDNA clones were substantially identical, although they differed in length, and one has a mutation. One of the clones was full length and contained an open reading frame encoding 188 amino acid residues in which the first 21 amino acids are a clevable signal sequence. The other of the two substantially identical clones terminated at the G of the start initiation codon. It could be inferred by sequence analysis of additional clones that the sequence preceeding the G reads ACCAT. Both of the clones were found to have the same coding region nucleotide sequence, which is depicted in FIG. 3 (SEQ ID NO:4). The deduced amino acid sequence of the open reading frame of the coding region of both of these two cDNA clones is shown in FIG. 4 (SEQ ID NO:5). The resulting protein encoded by this sequence is referred to hereinafter as VEGF-B.sub.167. In each of the protein names used herein, the subscript number refers to the number of amino acids in the mature protein without the signal sequence.

As would be expected, a comparison of the amino acid sequence encoded by these two clones with the partial amino acid sequence deduced from cDNA clone of Example 1 showed a striking similarity. However, the two open reading frames in the clone of Example 1, each of which encoded an amino acid sequence homologous to a different portion of VEGF, are both present in the same reading frame in each of these two clones according to Example 2. The frame shift in the clone of Example 1 is caused by an insertion of two extra adenine units which displace the C-terminal part of the clone of Example 1 out of frame. The reason for this is not presently understood, but may be due to a cloning artifact.

The coding part of the third clone had a nucleotide sequence identical to those of the preceding two clones except for a 21 bp insertion. FIG. 5 shows the nucleotide sequence of this third clone (SEQ ID NO:6). To facilitate identification, the 21 extra bases are underlined in the Figure. This insertion gives rise to 7 additional amino acid residues in the mature protein. Thus the resulting protein encoded by this longer cDNA is termed VEGF-B.sub.174. The amino acid sequence of the protein encoded by the cDNA of FIG. 5 is depicted in FIG. 6 (SEQ ID NO:7). The seven additional amino acids also are underlined in the figure for ease of identification. The additional amino acids are inserted into the sequence in a splice site, and sequencing of mouse genomic DNA clones indicates that these additional amino acids are the result of true alternative splicing. Furthermore, based on what is known about the receptor binding site locations of PDGF, the insertion occurs in a position in the protein which is probably part of a receptor binding site. The insertion is thus likely to affect receptor binding and could be of functional importance in influencing antagonist and/or different receptor specificity.

As previously pointed out this original cDNA clone of Example 1 was not full length and may contain an artifact. However, if the extreme 5' nucleotide sequence of the clones which encode VFGF-B.sub.167 and/or VEGF-B.sub.174 is added, the open reading frame encodes a protein of 133 amino acids, yielding a mature protein which is 112 amino acids long and hence is named VEGF-B.sub.112. The hybrid cDNA sequence encoding VEGF-B.sub.112 is shown in FIG. 7, and the amino acid sequence of the corresponding protein is illustrated in FIG. 8.

FIG. 9 shows a multiple amino acid sequence alignment for comparison purposes of the 167 amino acid variant of mouse Vascular Endothelial Growth Factor B (mVEGF-B.sub.167) , mouse Vascular Endothelial Growth Factor (mVEGF.sub.164) , human Placenta Growth Factor (hPlGF), mouse Platelet Derived Growth Factor A (mPDGF A), and mouse Platelet Derived Growth Factor B (mPDGF B). Amino acid residues identical to mouse VEGF-B.sub.167 are boxed. The homologous relationship of the sequences is apparent, and the figure clearly demonstrates the conserved structure of the growth factors belonging to the PDGF/VEGF family of growth factors, and that VEGF-B is a structural homolog of the other growth factors of this group. Pairwise comparisons of the amino acid sequences show that mouse VEGF-B is approximately 43% identical to mouse VEGF.sub.164, approximately 30% identical to human PlGF, and approximately 20% identical to mouse PDGF A and B. The conserved cysteine residues are particularly noteworthy. It can be seen that the first eight cysteine residues in the N-terminal domains (i.e. the PDGF-like domains) of the five growth factors are shared by all members of this family, and it is thus evident that the eight cysteine residues, which are involved in intramolecular and intermolecular disulfide bonding, are invariant among these growth factors. Furthermore, the C-terminal domains of mouse VEGF-B.sub.167 and VEGF.sub.164 also display a significant similarity with eight additional conserved cysteine residues and several stretches of basic amino acids.

Cloning of Human VEGF-B cDNA

10.sup.6 .lambda.-clones of human fibrosarcoma cDNA library HT1080 in .lambda.gt11 (Clontech) were screened with the 0.9 kb insert of the mouse VEGF-B clone pcif 2 according to standard procedures. Among several positive clones, one, termed H.1 was analyzed more carefully and its nucleotide sequence was determined. The nucleotide sequence indicated that a 207 amino acid isoform of human VEGF-B was encoded. Analysis of this isoform is described subsequently in Example 6. Based on the H.1 sequence two oligonucleotides were designed that would amplify the whole coding region of putative human cDNA corresponding to mouse VEGF-B.sub.167 form.

5'-CACCATGAGCCCTCTGCTCC-3' (forward) (SEQ ID NO:17)

5'-GCCATGTGTCACCTTCGCAG-3' (reverse) (SEQ ID NO:18)

These oligonucleotides were used to amplify by polymerase chain reaction (PCR) the whole coding region of human VEGF-B from oligo-dT primed human erythroleukemia cell (HEL) RNA. The amplified product was cloned into the pCR II-vector of TA cloning kit (Invitrogen) and sequenced using standard techniques. The nucleotide sequence of the human VEGF-B cDNA clone is shown in FIG. 10 (SEQ ID NO:10) , and the deduced amino acid sequence of human VEGF-B.sub.167 is shown in FIG. 11 (SEQ ID NO:11).

The full length mouse cDNA clone of Example 2 and the full length human cDNA clone of Example 4 each encode a polypeptide of 188 amino acids containing an N-terminal hydrophobic putative signal sequence. In analogy with VEGF, the signal peptidase cleavage site is believed to be located between Ala 21 and Pro 22. The putative cleavage site of the signal peptidase is indicated in FIG. 16 by an arrow. Accordingly, the processed VEGF-B polypeptides of these two clones each contain 167 amino acids.

The clone which exhibited the 340 bp Spe1/BamH1 fragment isolated in Example 2 was analyzed, and the major portion was found to be identical to the first two clones of Example 2 which exhibited the 240 bp Spe1/BamH1 fragment. The difference is due to the presence of an insertion in the C-terminal part of the sequence.

This 340 bp Spe1/BamH1 DNA fragement encodes a further isoform of mouse VEGF-B containing 207 amino acids. The coding portion of the DNA encoding this protein is illustrated in FIG. 12, and the translated amino acid sequence is illustrated in FIG. 13. After cleavage of the 21 amino acid leader sequence, the mature protein contains 186 amino acids and is referred to as mVEGF-B.sub.186. This isoform is clearly a result of alternative DNA splicing as described below with reference to FIG. 17.

The H.1 clone isolated as described in Example 4 was found to encode a 207 amino acid isoform of human VEGF-B. The coding portion of the DNA encoding this protein is illustrated in FIG. 14 and the translated amino acid sequence is illustrated in FIG. 15. Again, this isoform, which is designated hVEGF-B.sub.186, appears to be a product of alternative splicing.

FIG. 16 shows the aligned amino acid sequences of mouse and human VEGF-B.sub.167 and VEGF-B.sub.186 in one-letter code. Identical residues are enclosed in boxes, while amino acid residues which differ between mouse and human VEGF-B.sub.167 and VEGF-BE.sub.186 isoforms are outside the boxes. Mouse and human VEGF-B display approximately 88% amino acid sequence identity and are highly basic, especially in their C-terminal regions. Both polypeptides lack the consensus sequence for N-linked glycosylation (N-X-T/S). The arrow indicates the putative cleavage site for the signal peptidase. Excluding the signal sequences, the mouse and human VEGF-B.sub.167 amino acid sequences are highly homologous with only 20 replacements out the the 167 residues. The replacements are clustered in the N-terminus, in two regions around amino acids 60 and 145. All cysteine residues in both VEGF-B.sub.167 proteins are invariant, but the eight cysteine residues in the C-terminal end of VEGF-B.sub.167 are not conserved in the VEGF-B.sub.186 isoforms. It is notable that the mouse and human sequences in the region between residues 66 and 129 are identical apart from one evolutionarily conserved replacement (Q105R). This is of importance since the receptor binding domains are found within this portion of the protein (compared to PDGF structure). From this it can be concluded that it is likely that mouse and human VEGF-B will exhibit cross-reactive binding on the receptor level and thus display identical or similar biological activities.

FIG. 17 is a schematic representation of the structure of mouse and human genes for VEGF-B. Each of the genes contains seven exons. Exon 6 contains an alternative splice acceptor site which enables the gene to produce two different transcripts for VEGF-B. VEGF-B.sub.167 uses exons 1-5, the last part of exon 6, and exon 7 (TGA). VEGF-B.sub.186 uses exons 1 through 5, the first part of exon 6, and terminates in the last part of exon 6 (TAG). Exon 7 is not translated in VEGF-B.sub.186 since the insertion of the first part of exon 6 introduces a frame shift and gives rise to a stop codon in the last part of exon 6.

FIG. 18 shows a comparative hydrophilicity analysis of murine VEGF-B.sub.167 and VEGF-B.sub.186. As would be expected, the pattern of hydrophilicity/hydrophobicity is essentially identical from amino acid 1 through amino acid 115. After amino acid 115, the hydrophilicity/hydrophobicity patterns diverge because of the frame shift introduced by the first part of exon 6. Thus, VEGF-B.sub.167 and VEGF-B.sub.186 can be expected to exhibit both similar and dissimilar activities.

FIG. 19 is a dendrogram showing the phylogenetic relationship of the amino acid sequences of five members of the VEGF/PDGF family of growth factors. The number of replacements or substitutions decreases from the left to the right of the chart. It can be seen that VEGF-B lies between VEGF and the platelet derived growth factor (PDGF) group.

The multiple amino acid sequence alignments of FIGS. 9 and 16 and the phylogenetic analysis of FIG. 19 were carried out accoring to Hein, Method in Enzymology, Vol. 183, pp. 626-45, Academic Press Inc., San Diego (1990) using the PAM 250 distance table.

Antipeptide antiserum to human VEGF-B was generated by immunizing rabbits with a branched 23-mer oligopeptide comprising the following N-terminal region amino acid residue sequence (SEQ ID NO:19):

S-Q-P-D-A-P-G-H-Q-R-K-V-V-S-W-I-D-V-Y-T-R-A-T.

The branched 23-mer oligo peptide was synthesized according to Tam, "Synthetic peptide vaccine design: synthesis and properties of a high-density multiple antigenic peptide system", Proc. Natl. Acad. Sci. USA, Vol. 85, pages 5409-413 (1988). In the first immunization, rabbits were subcutaneously injected with 500 .mu.g of the branched peptide emulsified in Complete Freunds Adjuvant. In the subsequent boosters, 200 .mu.g of the antigen emulsified in Incomplete Freunds Adjuvant was injected. Antisera were collected after the second and third boosters by conventional techniques.

The biochemical properties of human VEGF-B.sub.167 were examined in transfected human embryonic kidney 293EBNA cells (Invitrogen Corp.). cDNA inserts encoding human VEGF-B.sub.167 and human VEGF.sub.165 ›see Keck et al., Science, Vol. 246, pages 1309-312 (1989)! were individually cloned into the pREP7 expression vector (Invitrogen Inc.). Human embryo kidney 293EBNA cells (expressing Epstein-Barr virus nuclear antigen-1) were transfected by transient transfection with the respective expression plasmids using calcium phosphate precipitation, and the cells were incubated for 48 hrs. As a control, cells also were transfected with an expression vector containing the VEGF-B.sub.167 cDNA in reverse orientation. Monolayers of cells were incubated in methionine-free and cysteine-free medium for 30 minutes followed by labeling with 100 .mu.Ci/ml ›.sup.35 S!methionine and ›.sup.35 S!cysteine (Promix, Amersham Inc.) in the same medium for 2 hours. The labeling medium was replaced with normal medium without serum, and labelled proteins were chased for 6 hours. Heparin was included during the chase when indicated (100 .mu.g/ml). Media were collected after the chase period, and cells were solubilized in 10 mM Tris pH 7.5, 50 mM NaCl, 0.5% sodium deoxycholate, 0.5% Nonidet P-40, 0.16 SDS and 0.1 U/ml aprotinin.

VEGF-B.sub.167 was expressed in the cells transfected with the plasmids containing the VEGF-B.sub.167 DNA. Aliquots of the culture supernatants from cells treated or untreated with heparin and detergent solubilized cell lysates were subjected to immunoprecipitation with the specific antipeptide antiserum to VEGF-B described in Example 7 and analyzed by SDS-PAGE under reducing conditions unless otherwise indicated. The data show that VEGF-B.sub.167 homodimers and VEGF-B.sub.167 -VEGF.sub.165 heterodimers are released from cells by heparin. By heparin treatment (1-100 .mu.g/ml) or 1.2M NaCl, VEGF-B.sub.167 was released from cells and found in the supernatant. If cells were not treated with heparin, VEGF-B.sub.167 remained cell-associated and was not released into the culture media. Under the same conditions, VEGF.sub.165 homodimers are secreted from the cells and found in the culture supernatants without heparin treatment.

Under reducing conditions, human VEGF-B.sub.167 migrated with a Mr of 21 kDa. Analysis of culture supernatants under non-reducing conditions showed that VEGF-B.sub.167 migrated as an Mr 42 kDa species indicating a dimeric structure. These results suggest that VEGF-B.sub.167 forms disulfide-linked dimers associated with the cell surface, probably through ionic interactions with extracellular heparan sulfate proteoglycans. The association is likely to be mediated by the C-terminal basic domain, as observed for the longer splice variants of VEGF.

Since VEGF has been shown to form heterodimers with PlGF, it was decided to test whether VEGF.sub.165 could also form heterodimers with VEGF-B.sub.167. For this purpose 293EBNA cells were co-transfected with expression vectors encoding both human VEGF.sub.165 and human VEGF-B.sub.167, and VEGF-B.sub.167 was expressed in combination with VEGF.sub.165. Metabolically labelled proteins were chased in the presence of heparin and immunoprecipitations were carried out with antisera to either VEGF-B.sub.167 or VEGF.sub.165. The antiserum to human VEGF was from R&D Systems. Under non-reducing conditions the VEGF-B.sub.167 -VEGF.sub.165 heterodimers migrated as Mr 42-46 kDa species. The results show that VEGF-B can form disulfide linked heterodimers with VEGF, which, in the absence of heparin, remain cell-associated. Since homodimers of VEGF.sub.165 are efficiently secreted into the media, VEGF-B appears to determine the secretion of the heterodimer.

VEGF-B is synthesized normally in the endoplasmic reticulum of the source cell for subsequent export. Recombinant VEGF-B may be produced by inserting a DNA sequence encoding the VEGF-B protein together with a suitable operatively linked promoter and control sequences into a suitable vector, such as the well known plasmid pBR322 or a derivative thereof, transforming or transfecting a suitable host cell, such as E. coli or a Cos cell, with the resulting vector or other systems well known in the art, screening the resulting transformants or transfectants for VEGF-B expression, and then culturing cell lines or bacterial cell strains which are positive for the expression of VEGF-B. Either a eukaryotic vector or a prokaryotic vector may be used, depending on the type of cell which is to be transfected or transformed therewith. A particularly preferred system for production of recombinant VEGF-B is the baculovirus--insect cell system, which has proved capable of producing excellent yields of recombinant protein.

9.1 VEGF-B with its own signal peptide.

a) Cloning and Transfection.

The complete human VEGF-B.sub.167 gene was inserted into a commercially available plasmid pCRII (Invitrogen Corp.). The HindIII-XbaI fragment from the resulting plasmid pCRII-VEGF-B.sub.167, which encodes the whole open reading frame of VEGF-B.sub.167 then was cloned into pFASTBAC1, and both the 3'- and 5'-junctions were sequenced. Bacmid-DNA was prepared according to the manufacturers instructions for the "Bac-To-Bac.TM. Baculovirus Expression System" (Life Technologies Inc.) and lipofected to Sf900II-adapted Sf9 cells (obtained from Dr. Christian Oker-Blom). Sf9 cells are from the American Type Culture Collection Cell Repository Line Bank, Rockville Md. (ATCC CRL-1711). The transfected cells were then cultured on standard TMN-FH medium in 25 cm.sup.2 culture dishes.

b) Assay for protein expression.

About 72 hours after transfection, the cells were lysed and 1 ml of culture supernatant and the cell lysate were assayed for expressed VEGF-B by immunoprecipitation as described in Example 8 and Western blotting. Lysates from three out of four independently transfected cell cultures were found positive for VEGF-B, although the intensity of the signal in the Western blot varied. The expressed VEGF-B polypeptide in each case was found to correspond in size to the protein expressed in mammalian cells in Example 8.

The viral stock from the cells that gave the strongest signal in Western blotting was amplified two rounds by infecting cells and collecting new virus from the medium. The resulting supernatant was analyzed. Uninfected cells were also analyzed as a negative control. Time course analysis showed that cells harvested between 48 and 72 hours after infection contained the greatest amount of VEGF-B. After 96 hours post infection, as a result of virus-induced cell lysis, VEGF-B could also be detected in the culture supernatant by immunoprecipitation and Western blotting. Recombinant VEGF-B could be precipitated from the lysate between 20% and 40% (NH.sub.4).sub.2 SO.sub.4.

9.2 VEGF-B with the Melittin signal peptide (pVTBac).

a) Cloning and transfection.

A polymerase chain reaction (PCR) fragment from nucleotide position 68 to 141 was used to introduce a BamHI restriction site immediately after the signal cleavage site in the plasmid pCRII-VEGF-B.sub.167 from Example 10.1. The BamHI fragment from this modified pCRII-VEGF-B.sub.167 construct was cloned into BamHI opened pVTBac ›Tessier et al., "Enhanced secretion from insect cells of a foreign protein fused to the honeybee mellitin signal peptide", Gene, Vol. 98, page 177 (1991)!. Both 3'- and 5'-junction were sequenced. Sf9 cells were cotransfected with the aforedescribed pVTBac vector which contained the human VGEF-B.sub.167 gene, and with linearized baculovirus DNA (Insectin.TM., Invitrogen Corp.). The transfected cells then were cultured in TMN-FH medium.

b) Assay for protein expression.

Forty-eight hours after transfection, the supernatant was collected and subjected to primary screening by immunoprecipitation. Four positive plaques were isolated.

The ability of VEGF-B.sub.167 to stimulate endothelial cell proliferation was established through analysis of ›.sup.3 H!thymidine incorporation in human umbilical vein endothelial cells (HUVEC) and in bovine capillary endothelial (BCE) cells.

293EBNA cells were transfected with expression vectors for VEGF-B.sub.167, VEGF.sub.165 or empty vector (mock) in the presence of 1 .mu.g/ml heparin. Conditioned media from these cells were diluted in respective media, applied to human umbilical vein endothelial cells (HUVEC) and to bovine capillary endothelial (BCE) cells and incorporation of ›.sup.3 H!thymidine was measured. As a positive control recombinant bFGF was added to BCE cells.

To elaborate, conditioned media containing human VEGF-B and human VEGF.sub.165 were collected from 293EBNA cells transfected with the appropriate expression vectors or with empty vector (mock) in the presence of heparin (1 .mu.g/ml) 48 hours posttransfection. Second passage HUVEC were plated into 96-well plates (4.times.10.sup.3 cells/well) in M-199 medium supplemented with 10% fetal bovine serum and incubated for 24 hours. Conditioned media were diluted with the growth medium and cells were stimulated for 48 hours. Fresh conditioned media containing 10 .mu.Ci/ml of ›.sup.3 H!thymidine (Amersham Inc.) were added to the cells and stimulations were continued for another 48 hours. Cells were washed with PBS and trypsinized and incorporated radioactivity was determined by liquid scintillation counting. BCE cells were seeded into 24-well plates and grown until confluence in minimal essential medium (MEM) supplemented with 10% fetal calf serum. Cells were starved in MEM supplemented with 3% fetal calf serum for 72 hours, after which conditioned media diluted into serum-free medium were added to the cells and the cells were stimulated for 24 hours. ›.sup.3 H!Thymidine was included during the last 4 hours of the stimulation (1 .mu.Ci/ml). Stimulations with bFGF were carried out as above using 6 ng/ml of recombinant bFGF (Synergen Inc.). Cells were washed with PBS, lysed with NaOH, and incorporated radioactivity was determined by liquid scintillation counting.

FIG. 20 is a bar graph showing fold of induction of ›.sup.3 H!thymidine incorporation by VEGF-B.sub.167 in human umbilical vein endothelial cells (HUVEC) and in bovine capillary endothelial (BCE) cells, as compared to basal activity induced by conditioned medium from the mock transfected cells. For comparison purposes, the induction by VEGF.sub.165 and by bFGF are also shown. The bars show the mean .+-. standard deviation of parallel samples. Similar results were obtained in several other independent experiments. The test results clearly show that VEGF-B induced ›.sup.3 H!thymidine incorporation in both HUVEC and BCE cells and stimulated proliferation of endothelial cells in vitro, thereby demonstrating that VEGF-B is an endothelial growth factor.

The foregoing results indicate that that VEGF-B is a novel growth factor for endothelial cells which plays a role in vascularization, in particular of muscle. The cell-association of VEGF-B may have several implications for regulation of vascularization and endothelial cell growth. In developing embryos and in contractile tissues, cell-associated VEGF-B may provide spatial cues to outgrowing endothelial cells during establishment and maintenance of the vascular tree. It could also, through its cell-association, support the regeneration of damaged endothelium in adult vessels. The ability of VEGF-B to modulate the secretion of VEGF by heterodimer formation suggests an indirect role of VEGF-B in VEGF signalling. The formation of multiple heterodimeric complexes of these growth factors could provide a basis for a diverse array of regulatory signals for endothelial cells.

VEGF-B can be used as a growth factor for populations of endothelial cells in vitro. VEGF-B may be used to promote desirable angiogenesis, i.e. the formation of new blood vessels and capillaries. For example, it may be useful in promoting the development of the corpus luteum and endometrium as an aid to initiating and/or maintaining pregnancy. It would also be useful in bone repair by virtue of its action on endothelial cells. Administration of VEGF-B may also be useful in supporting embryogenesis, as well as somatic growth and vascular development and differentiation. Topical application of VEGF-B to wounds may be useful in promoting wound healing, and oral administration of VEGF-B may be useful to accelerate the healing of gastric and/or duodenal ulcers. The ability of VEGF-B to modulate the secretion of VEGF by heterodimer formation could provide a therapeutic role for VEGF-B in diseases where VEGF agonists would be useful.

VEGF-B may exert proliferative effects on mesodermal cells either directly or via improvements in the blood supply.

Tumor assays for VEGF-B may be useful as indicators of metastatic risk. Assays of VEGF-B in body fluids or the tumor itself by histochemistry may be useful as a tumor prognostic factor. Furthermore, because tumor growth requires angiogenesis, administration of VEGF-B may also be useful in promoting tumor growth in laboratory animals in order to test anti-tumorigenic drugs. VEGF-B may also be useful to increase the microvascularity of hypoxic areas of tumors and make them more sensitive to radiation, radiation sensitizing drugs, etc.

The angiogenic action of VEGF-B may be useful in treating ischemic conditions. VEGF-B or agonists could be used to stimulate the development of collateral circulation in cases of arterial and/or venous obstruction, e.g. myocardial infarcts, ischaemic limbs, deep venous thrombisis, and/or postpartum vascular problems.

A VEGF-B/VEGF-B receptor system may be used as an assay system to detect small molecules as agonists/antagonists for development as new drugs. Examples of small molecules which could be detected include, but are not limited to, organic chemicals, peptides, and RNA molecules.

Pharmaceutical compositions may be produced by admixing a pharmaceutically effective amount of VEGF-B protein with one or more suitable carriers or adjuvants such as water, mineral oil, polyethylene glycol, starch, talcum, lactose, thickeners, stabilizers, suspending agents, etc. Such compositions may be in the form of solutions, suspensions, tablets, capsules, creams, salves, ointments, or other conventional forms.

As demonstrated in Example 7, VEGF-B protein also can be used to produce antibodies. In general, conventional antibody production techniques may be used to produce VEGF-B antibodies. For example, specific monoclonal antibodies may be produced via immunization of fusion proteins obtained by recombinant DNA expression.

Labelled monoclonal antibodies, in particular, should be useful in screening for conditions associated with abnormal levels of VEGF-B in the body. For example, assays of VEGF-B levels in blood or urine may be useful as a tumor marker. These monoclonal antibodies to VEGF-B also may be useful in inhibiting angiogenesis associated with high levels of VEGF-B in the body, e.g. in rapidly proliferating, angiogenesis-dependent tumors in mammals, and thereby may retard the growth of such tumors. Treatment may be effected, e.g., by twice weekly intraperitoneal injection of 10 to 500 .mu.g, preferably 50-100 .mu.g of monoclonal antibody. For the therapy of humans, chiaserization or humanization of such monoclonal antibodies is to be preferred.

VEGF-B antagonists such as antibodies may be useful to inhibit new blood vessels in diabetic retinopathy, psoriasis, arthopathies and/or vascular tumors such as haemangiomas.

The foregoing description and examples have been set forth merely to illustrate the invention and are not intended to be limiting. Since modifications of the disclosed embodiments incorporating the spirit and substance of the invention may occur to persons skilled in the art, the invention should be construed to include everything within the scope of the appended claims and equivalents thereof.