AU6502999A - Platelet-derived growth factor c, dna coding therefor, and uses thereof - Google Patents

Description

WO 00/18212 PCT/US99/22668 PLATELET-DERIVED GROWTH FACTOR C, DNA CODING THEREFOR, AND USES THEREOF This invention relates to growth factors for connective tissue cells, fibroblasts, myofibroblasts and glial cells and/or to growth factors for endothelial cells, and in particular to a novel platelet-derived growth factor/vascular 5 endothelial growth factor-like growth factor, a polynucleotide sequence encoding the factor, and to pharmaceutical and diagnostic compositions and methods utilizing or derived from the factor. 10 BACKGROUND OF THE INVENTION In the developing embryo, the primary vascular network is established by in situ differentiation of mesodermal cells in a process called vasculogenesis. It is believed that all subsequent processes involving the generation of new vessels 15 in the embryo and neovascularization in adults, are governed by the sprouting or splitting of new capillaries from the pre-existing vasculature in a process called angiogenesis (Pepper et al., Enzyme & Protein, 1996 49 138-162; Breier et al., Dev. Dyn. 1995 204 228-239; Risau, Nature, 1997 386 20 671-674) . Angiogenesis is not only involved in embryonic development and normal tissue growth, repair, and regeneration, but is also involved in the female reproductive cycle, establishment and maintenance of pregnancy, and in repair of wounds and fractures. In addition to angiogenesis 25 which takes place in the normal individual, angiogenic events are involved in a number of pathological processes, notably tumor growth and metastasis, and other conditions in which blood vessel proliferation, especially of the microvascular system, is increased, such as diabetic retinopathy, psoriasis 30 and arthropathies. Inhibition of angiogenesis is useful in preventing or alleviating these pathological processes. On the other hand, promotion of angiogenesis is desirable in situations where vascularization is to be established or WO 00/18212 PCT/US99/22668 extended, for example after tissue or organ transplantation, or to stimulate establishment of collateral circulation in tissue infarction or arterial stenosis, such as in coronary heart disease and thromboangitis obliterans. 5 The angiogenic process is highly complex and involves the maintenance of the endothelial cells in the cell cycle, degradation of the extracellular matrix, migration and invasion of the surrounding tissue and finally, tube formation. The molecular mechanisms underlying the complex 10 angiogenic processes are far from being understood. Because of the crucial role of angiogenesis in so many physiological and pathological processes, factors involved in the control of angiogenesis have been intensively investigated. A number of growth factors have been shown to 15 be involved in the regulation of angiogenesis; these include fibroblast growth factors (FGFs), platelet-derived growth factor (PDGF), transforming growth factor alpha (TGFa), and hepatocyte growth factor (HGF). See for example Folkman et al., J. Biol. Chem., 1992 267 10931-10934 for a review. 20 It has been suggested that a particular family of endothelial cell-specific growth factors, the vascular endothelial growth factors (VEGFs), and their corresponding receptors is primarily responsible for stimulation of endothelial cell growth and differentiation, and for certain 25 functions of the differentiated cells. These factors are members of the PDGF family, and appear to act primarily via endothelial receptor tyrosine kinases (RTKs). Nine different proteins have been identified in the PDGF family, namely two PDGFs (A and B), VEGF and six members that 30 are closely related to VEGF. The six members closely related to VEGF are: VEGF-B, described in International Patent Application PCT/US96/02957 (WO 96/26736) and in U.S. Patents 5,840,693 and 5,607,918 by Ludwig Institute for Cancer -2- WO 00/18212 PCT/US99/22668 Research and The University of Helsinki; VEGF-C, described in Joukov et al., EMBO J., 1996 15 290-298 and Lee et al., Proc. Natl. Acad. Sci. USA, 1996 93 1988-1992; VEGF-D, described in International Patent Application No. PCT/US97/14696 (WO 5 98/07832), and Achen et al., Proc. Natl. Acad. Sci. USA, 1998 95 548-553; the placenta growth factor (PlGF), described in Maglione et al., Proc. Natl. Acad. Sci. USA, 1991 88 9267 9271; VEGF2, described in International Patent Application No. PCT/US94/05291 (WO 95/24473) by Human Genome Sciences, Inc; 10 and VEGF3, described in International Patent Application No. PCT/US95/07283 (WO 96/39421) by Human Genome Sciences, Inc. Each VEGF family member has between 30% and 45% amino acid sequence identity with VEGF. The VEGF family members share a VEGF homology domain which contains the six cysteine 15 residues which form the cysteine knot motif. Functional characteristics of the VEGF family include varying degrees of mitogenicity for endothelial cells, induction of vascular permeability and angiogenic and lymphangiogenic properties. Vascular endothelial growth factor (VEGF) is a 20 homodimeric glycoprotein that has been isolated from several sources. VEGF shows highly specific mitogenic activity for endothelial cells. VEGF has important regulatory functions in the formation of new blood vessels during embryonic vasculogenesis and in angiogenesis during adult life 25 (Carmeliet et al., Nature, 1996 380 435-439; Ferrara et al., Nature, 1996 380 439-442; reviewed in Ferrara and Davis-Smyth, Endocrine Rev., 1997 18 4-25). The significance of the role played by VEGF has been demonstrated in studies showing that inactivation of a single VEGF allele results in embryonic 30 lethality due to failed development of the vasculature (Carmeliet et al., Nature, 1996 380 435-439; Ferrara et al., Nature, 1996 380 439-442) . In addition VEGF has strong chemoattractant activity towards monocytes, can induce the -3- WO 00/18212 PCTIUS99/22668 plasminogen activator and the plasminogen activator inhibitor in endothelial cells, and can also induce microvascular permeability. Because of the latter activity, it is sometimes referred to as vascular permeability factor (VPF). The 5 isolation and properties of VEGF have been reviewed; see Ferrara et al., J. Cellular Biochem., 1991 47 211-218 and Connolly, J. Cellular Biochem., 1991 47 219-223. Alterative mRNA splicing of a single VEGF gene gives rise to five isoforms of VEGF. 10 VEGF-B has similar angiogenic and other properties to those of VEGF, but is distributed and expressed in tissues differently from VEGF. In particular, VEGF-B is very strongly expressed in heart, and only weakly in lung, whereas the reverse is the case for VEGF. This suggests that VEGF and 15 VEGF-B, despite the fact that they are co-expressed in many tissues, may have functional differences. VEGF-B was isolated using a yeast co-hybrid interaction trap screening technique by screening for cellular proteins which might interact with cellular resinoid acid-binding 20 protein type I (CRABP-I). Its isolation and characteristics are described in detail in PCT/US96/02957 and in Olofsson et al., Proc. Natl. Acad. Sci. USA, 1996 93 2576-2581. VEGF-C was isolated from conditioned media of the PC-3 prostate adenocarcinoma cell line (CRL1435) by screening for 25 ability of the medium to produce tyrosine phosphorylation of the endothelial cell-specific receptor tyrosine kinase VEGFR-3 (Flt4), using cells transfected to express VEGFR-3. VEGF-C was purified using affinity chromatography with recombinant VEGFR-3, and was cloned from a PC-3 cDNA library. Its 30 isolation and characteristics are described in detail in Joukov et al., EMBO J., 1996 15 290-298. VEGF-D was isolated from a human breast cDNA library, commercially available from Clontech, by screening with an -4- WO 00/18212 PCTIUS99/22668 expressed sequence tag obtained from a human cDNA library designated "Soares Breast 3NbHBst" as a hybridization probe (Achen et al., Proc. Natl. Acad. Sci. USA, 1998 95 548-553). Its isolation and characteristics are described in detail in 5 International Patent Application No. PCT/US97/14696 (W098/07832). The VEGF-D gene is broadly expressed in the adult human, but is certainly not ubiquitously expressed. VEGF-D is strongly expressed in heart, lung and skeletal muscle. 10 Intermediate levels of VEGF-D are expressed in spleen, ovary, small intestine and colon, and a lower expression occurs in kidney, pancreas, thymus, prostate and testis. No VEGF-D mRNA was detected in RNA from brain, placenta, liver or peripheral blood leukocytes. 15 PlGF was isolated from a term placenta cDNA library. Its isolation and characteristics are described in detail in Maglione et al., Proc. Natl. Acad. Sci. USA, 1991 88 9267 9271. Presently its biological function is not well understood. 20 VEGF2 was isolated from a highly tumorgenic, oestrogen independent human breast cancer cell line. While this molecule is stated to have about 22% homology to PDGF and 30% homology to VEGF, the method of isolation of the gene encoding VEGF2 is unclear, and no characterization of the biological 25 activity is disclosed. VEGF3 was isolated from a cDNA library derived from colon tissue. VEGF3 is stated to have about 36% identity and 66% similarity to VEGF. The method of isolation of the gene encoding VEGF3 is unclear and no characterization of the 30 biological activity is disclosed. Similarity between two proteins is determined by comparing the amino acid sequence and conserved amino acid substitutions of one of the proteins to the sequence of the -5- WO 00/18212 PCT/US99/22668 second protein, whereas identity is determined without including the conserved amino acid substitutions. PDGF/VEGF family members act primarily by binding to receptor tyrosine kinases. Five endothelial cell-specific 5 receptor tyrosine kinases have been identified, namely VEGFR-1 (Flt-1), VEGFR-2 (KDR/Flk-1), VEGFR-3 (Flt4), Tie and Tek/Tie-2. All of these have the intrinsic tyrosine kinase activity which is necessary for signal transduction. The essential, specific role in vasculogenesis and angiogenesis 10 of VEGFR-1, VEGFR-2, VEGFR-3, Tie and Tek/Tie-2 has been demonstrated by targeted mutations inactivating these receptors in mouse embryos. The only receptor tyrosine kinases known to bind VEGFs are VEGFR-1, VEGFR-2 and VEGFR-3. VEGFR-1 and VEGFR-2 bind 15 VEGF with high affinity, and VEGFR-l also binds VEGF-B and PlGF. VEGF-C has been shown to be the ligand for VEGFR-3, and it also activates VEGFR-2 (Joukov et al., The EMBO Journal, 1996 15 290-298) . VEGF-D binds to both VEGFR-2 and VEGFR-3. A ligand for Tek/Tie-2 has been described in International 20 Patent Application No. PCT/US95/12935 (WO 96/11269) by Regeneron Pharmaceuticals, Inc. The ligand for Tie has not yet been identified. Recently, a novel 130-135 kDa VEGF isoform specific receptor has been purified and cloned (Soker et al., Cell, 25 1998 92 735-745) . The VEGF receptor was found to specifically bind the VEGF1 6 5 isoform via the exon 7 encoded sequence, which shows weak affinity for heparin (Soker et al., Cell, 1998 92 735-745) . Surprisingly, the receptor was shown to be identical to human neuropilin-1 (NP-1), a receptor involved 30 in early stage neuromorphogenesis. PlGF-2 also appears to interact with NP-1 (Migdal et al., J. Biol. Chem., 1998 273 22272-22278). -6- WO 00/18212 PCT/US99/22668 VEGFR-1, VEGFR-2 and VEGFR-3 are expressed differently by endothelial cells. Both VEGFR-l and VEGFR-2 are expressed in blood vessel endothelia (Oelrichs et al., Oncogene, 1992 8 11-18; Kaipainen et al., J. Exp. Med., 1993 178 2077-2088; 5 Dumont et al., Dev. Dyn., 1995 203 80-92; Fong et al., Dev. Dyn., 1996 207 1-10) and VEGFR-3 is mostly expressed in the lymphatic endothelium of adult tissues (Kaipainen et al., Proc. Natl. Acad. Sci. USA, 1995 9 3566-3570) . VEGFR-3 is also expressed in the blood vasculature surrounding tumors. 10 Disruption of the VEGFR genes results in aberrant development of the vasculature leading to embryonic lethality around midgestation. Analysis of embryos carrying a completely inactivated VEGFR-l gene suggests that this receptor is required for functional organization of the 15 endothelium (Fong et al., Nature, 1995 376 66-70). However, deletion of the intracellular tyrosine kinase domain of VEGFR-l generates viable mice with a normal vasculature (Hiratsuka et al., Proc. Natl. Acad. Sci. USA 1998 95 9349-9354). The reasons underlying these differences remain 20 to be explained but suggest that receptor signalling via the tyrosine kinase is not required for the proper function of VEGFR-l. Analysis of homozygous mice with inactivated alleles of VEGFR-2 suggests that this receptor is required for endothelial cell proliferation, hematopoesis and 25 vasculogenesis (Shalaby et al., Nature, 1995 376 62-66; Shalaby et al., Cell, 1997 8 9 981-990). Inactivation of VEGFR-3 results in cardiovascular failure due to abnormal organization of the large vessels (Dumont et al. Science, 1998 282 946-949). 30 Although VEGFR-l is mainly expressed in endothelial cells during development, it can also be found in hematopoetic precursor cells during early stages of embryogenesis (Fong et al., Nature, 1995 376 66-70) . In adults, monocytes and -7- WO 00/18212 PCT/US99/22668 macrophages also express this receptor (Barleon et al., Blood, 1996 87 3336-3343) . In embryos, VEGFR-1 is expressed by most, if not all, vessels (Breier et al., Dev. Dyn., 1995 204 228-239; Fong et al., Dev. Dyn., 1996 207 1-10). 5 The receptor VEGFR-3 is widely expressed on endothelial cells during early embryonic development but as embryogenesis proceeds becomes restricted to venous endothelium and then to the lymphatic endothelium (Kaipainen et al., Cancer Res., 1994 54 6571-6577; Kaipainen et al., Proc. Natl. Acad. Sci. 10 USA, 1995 92 3566-3570). VEGFR-3 is expressed on lymphatic endothelial cells in adult tissues. This receptor is essential for vascular development during embryogenesis. Targeted inactivation of both copies of the VEGFR-3 gene in mice resulted in defective blood vessel formation 15 characterized by abnormally organized large vessels with defective lumens, leading to fluid accumulation in the pericardial cavity and cardiovascular failure at post-coital day 9.5. On the basis of these findings it has been proposed that VEGFR-3 is required for the maturation of primary 20 vascular networks into larger blood vessels. However, the role of VEGFR-3 in the development of the lymphatic vasculature could not be studied in these mice because the embryos died before the lymphatic system emerged. Nevertheless it is assumed that VEGFR-3 plays a role in 25 development of the lymphatic vasculature and lymphangiogenesis given its specific expression in lymphatic endothelial cells during embryogenesis and adult life. This is supported by the finding that ectopic expression of VEGF-C, a ligand for VEGFR 3, in the skin of transgenic mice, resulted in lymphatic 30 endothelial cell proliferation and vessel enlargement in the dermis. Furthermore this suggests that VEGF-C may have a primary function in lymphatic endothelium, and a secondary -8- WO 00/18212 PCTIUS99/22668 function in angiogenesis and permeability regulation which is shared with VEGF (Joukov et al., EMBO J., 1996 15 290-298). Some inhibitors of the VEGF/VEGF-receptor system have been shown to prevent tumor growth via an anti-angiogenic 5 mechanism; see Kim et al., Nature, 1993 362 841-844 and Saleh et al., Cancer Res., 1996 56 393-401. As mentioned above, the VEGF family of growth factors are members of the PDGF family. PDGF plays a important role in the growth and/or motility of connective tissue cells, 10 fibroblasts, myofibroblasts and glial cells (Heldin et al., "Structure of platelet-derived growth factor: Implications for functional properties", Growth Factor, 1993 8 245-252) . In adults, PDGF stimulates wound healing (Robson et al., Lancet, 1992 339 23-25) . Structurally, PDGF isoforms are disulfide 15 bonded dimers of homologous A- and B-polypeptide chains, arranged as homodimers (PDGF-AA and PDGF-BB) or a heterodimer (PDGF-AB). PDGF isoforms exert their effects on target cells by binding to two structurally related receptor tyrosine kinases 20 (RTKs). The alpha-receptor binds both the A- and B-chains of PDGF, whereas the beta-receptor binds only the B-chain. These two receptors are expressed by many in vitro grown cell lines, and are mainly expressed by mesenchymal cells in vivo. The PDGFs regulate cell proliferation, cell survival and 25 chemotaxis of many cell types in vitro (reviewed in Heldin et al., Biochim Biophys Acta., 1998 1378 F79-113). In vivo, they exert their effects in a paracrine mode since they often are expressed in epithelial (PDGF-A) or endothelial cells (PDGF-B) in close apposition to the PDGFR expressing mesenchyme. In 30 tumor cells and in cell lines grown in vitro, coexpression of the PDGFs and the receptors generate autocrine loops which are important for cellular transformation (Betsholtz et al., Cell, -9- WO 00/18212 PCT/US99/22668 1984 39 447-57; Keating et al., J. R. Coll Surg Edinb., 1990 35 172-4) . Overexpression of the PDGFs have been observed in several pathological conditions, including maligancies, arteriosclerosis, and fibroproliferative diseases (reviewed 5 in Heldin et al., The Molecular and Cellular Biology of Wound Repair, New York: Plenum Press, 1996, 249-273). The importance of the PDGFs as regulators of cell proliferation and survival are well illustrated by recent gene targeting studies in mice that have shown distinct 10 physiological roles for the PDGFs and their receptors despite the overlapping ligand specificities of the PDGFRs. Homozygous null mutations for either of the two PDGF ligands or the receptors are lethal. Approximately 50% of the homozygous PDGF-A deficient mice have an early lethal 15 phenotype, while the surviving animals have a complex postnatal phenotype with lung emphysema due to improper alveolar septum formation because of a lack of alveolar myofibroblasts (Bostrom et al., Cell, 1996 85 863-873). The PDGF-A deficient mice also have a dermal phenotype 20 characterized by thin dermis, misshapen hair follicles and thin hair (Karlsson et al., Development, 1999 126 2611-2) . PDGF-A is also required for normal development of oligodendrocytes and subsequent myelination of the central nervous system (Fruttiger et al., Development, 1999 126 25 457-67). The phenotype of PDGFR-alpha deficient mice is more severe with early embryonic death at E10, incomplete cephalic closure, impaired neural crest development, cardiovascular defects, skeletal defects, and odemas [Soriano et al., Development, 1997 124 2691-70) . The PDGF-B and PDGFR-beta 30 deficient mice develop similar phenotypes that are characterized by renal, hematological and cardiovascular abnormalities (Lev6en et al., Genes Dev., 1994 8 1875-1887; Soriano et al., Genes Dev., 1994 8 1888-96; Lindahl et al., -10- WO 00/18212 PCTIUS99/22668 Science, 1997 277 242-5; Lindahl, Development, 1998 125 3313-2), where the renal and cardiovascular defects, at least in part, are due to the lack of proper recruitment of mural cells (vascular smooth muscle cells, pericytes or mesangial 5 cells) to blood vessels (Lev6en et al., Genes Dev., 1994 8 1875-1887; Lindahl et al., Science, 1997 277 242-5; Lindahl et al., Development, 1998 125 3313-2). 10 SUMMARY OF THE INVENTION The invention generally provides an isolated novel growth factor which has the ability to stimulate and/or enhance proliferation or differentiation and/or growth and/or motility of cells expressing a PDGF-C receptor including, but not 15 limited to, endothelial cells, connective tissue cells, myofibroblasts and glial cells, an isolated polynucleotide sequence encoding the novel growth factor, and compositions useful for diagnostic and/or therapeutic applications. According to one aspect, the invention provides an 20 isolated and purified nucleic acid molecule which comprises a polynucleotide sequence having at least 85% identity, more preferably at least 90%, and most preferably at least 95% identity to at least nucleotides 37-1071 of the sequence set out in Figure 1 (SEQ ID NO:2), at least nucleotides 6-956 of 25 the sequence set out in Figure 3 (SEQ ID NO:3) or at least nucleotides 196 to 1233 of the sequence set out in Figure 5 (SEQ ID NO:6). The sequence of at least nucleotides 37-1071 of the sequence set out in Figure 1 (SEQ ID NO:2) or at least nucleotides 196 to 1233 of the sequence set out in Figure 5 30 (SEQ ID NO:6) encodes a novel polypeptide, designated PDGF-C (formally designated "VEGF-F") , which is structurally homologous to PDGF-A, PDGF-B, VEGF, VEGF-B, VEGF-C and VEGF-D. In a preferred embodiment, the nucleic acid molecule is a cDNA -11- WO 00/18212 PCT/US99/22668 which comprises at least nucleotides 37-1071 of the sequence set out in Figure 1 (SEQ ID NO:2), at least nucleotides 6-956 of the sequence set out in Figure 3 (SEQ ID NO:3) or at least nucleotides 196 to 1233 of the sequence set out in Figure 5 5 (SEQ ID NO:6) . This aspect of the invention also encompasses DNA molecules having a sequence such that they hybridize under stringent conditions with at least nucleotides 37-1071 of the sequence set out in Figure 1 (SEQ ID NO:2), at least nucleotides 6-956 of the sequence set out in Figure 3 (SEQ ID 10 NO:3) or at least nucleotides 196 to 1233 of the sequence set out in Figure 5 (SEQ ID NO:6) or fragments thereof. According to a second aspect, the polypeptide of the invention has the ability to stimulate and/or enhance proliferation and/or differentiation and/or growth and/or 15 motility of cells expressing a PDGF-C receptor including, but not limited to, endothelial cells, connective tissue cells, myofibroblasts and glial cells and comprises a sequence of amino acids corresponding to the amino acid sequence set out in Figure 2 (SEQ ID NO:3), Figure 4 (SEQ ID NO:5) or Figure 20 6 (SEQ ID NO:7) , or a fragment or analog thereof which has the ability to stimulate and/or enhance proliferation and/or differentiation and/or growth and/or motility of cells expressing a PDGF-C receptor including, but not limited to, endothelial cells, connective tissue cells (such as 25 fibroblasts) , myofibroblasts and glial cells. Preferably the polypeptides have at least 85% identity, more preferably at least 90%, and most preferably at least 95% identity to the amino acid sequence of in Figure 2 (SEQ ID NO:3), Figure 4 (SEQ ID NO:5) or Figure 6 (SEQ ID NO:7), or a fragment or 30 analog thereof having the biological activity of PDGF-C. A preferred fragment is a truncated form of PDGF-C comprising a portion of the PDGF/VEGF homology domain (PVHD) of PDGF-C. The minimal domain is residues 230-345. However, the domain -12- WO 00/18212 PCT/US99/22668 can extend towards the N terminus up to residue 164. Herein the PVHD is defined as truncated PDGF-C. The truncated PDGF-C is an activated form of PDGF-C. As used in this application, percent sequence identity 5 is determined by using the alignment tool of "MEGALIGN" from the Lasergene package (DNASTAR, Ltd. Abacus House, Manor Road, West Ealing, London W130AS United Kingdom) and using its preset conditions. The alignment is then refined manually, and the number of identities are estimated in the regions 10 available for a comparison. Preferably the polypeptide or the encoded polypeptide from a polynucleotide has the ability to stimulate one or more of proliferation, differentiation, motility, survival or vascular permeability of cells expressing a PDGF-C receptor 15 including, but not limited to, vascular endothelial cells, lymphatic endothelial cells, connective tissue cells (such as fibroblasts), myofibroblasts and glial cells. Preferably the polypeptide or the encoded polypeptide from a polynucleotide has the ability to stimulate wound healing. PDGF-C can also 20 have antagonistic effects on cells, but are included in the biological activities of PDGF-C. These abilities are referred to hereinafter as "biological activities of PDGF-C" and can be readily tested by methods known in the art. As used herein, the term "PDGF-C" collectively refers to 25 the polypeptides of Figure 2 (SEQ ID NO:3), Figure 4 (SEQ ID NO:5) or Figure 6 (SEQ ID NO:7), and fragments or analogs thereof which have the biological activity of PDGF-C as defined above, and to a polynucleotide which can code for PDGF-C, or a fragment or analog thereof having the biological 30 activity of PDGF-C. The polynucleotide can be naked and/or in a vector or liposome. In another preferred aspect, the invention provides a polypeptide possessing an amino acid sequence: -13- WO 00/18212 PCT/US99/22668 PXCLLVXRCGGXCXCC (SEQ ID NO:1) which is unique to PDGF-C and differs from the other members of the PDGF/VEGF family of growth factors because of the insertion of the three amino acid residues (NCA) between the 5 third and fourth cysteines (see Figure 9 - SEQ ID NOs:8-17). Polypeptides comprising conservative substitutions, insertions, or deletions, but which still retain the biological activity of PDGF-C are clearly to be understood to be within the scope of the invention. Persons skilled in the 10 art will be well aware of methods which can readily be used to generate such polypeptides, for example the use of site directed mutagenesis, or specific enzymatic cleavage and ligation. The skilled person will also be aware that peptidomimetic compounds or compounds in which one or more 15 amino acid residues are replaced by a non-naturally occurring amino acid or an amino acid analog may retain the required aspects of the biological activity of PDGF-C. Such compounds can readily be made and tested by methods known in the art, and are also within the scope of the invention. 20 In addition, possible variant forms of the PDGF-C polypeptide which may result from alternative splicing, as are known to occur with VEGF and VEGF-B, and naturally-occurring allelic variants of the nucleic acid sequence encoding PDGF-C are encompassed within the scope of the invention. Allelic 25 variants are well known in the art, and represent alternative forms or a nucleic acid sequence which comprise substitution, deletion or addition of one or more nucleotides, but which do not result in any substantial functional alteration of the encoded polypeptide. 30 Such variant forms of PDGF-C can be prepared by targeting non-essential regions of the PDGF-C polypeptide for modification. These non-essential regions are expected to fall outside the strongly-conserved regions indicated in -14- WO 00/18212 PCTIUS99/2266 8 Figure 9 (SEQ ID NOs:8- 1 7 ) In particular, the growth factors of the PDGF family, including VEGF, are dimeric, and VEGF, VEGF-B, VEGF-C, VEGF-D, PlGF, PDGF-A and PDGF-B show complete conservation of eight cysteine residues in the N-terminal 5 domains, i.e. the PDGF/VEGF-like domains (Olofsson et al., Proc. Natl. Acad. Sci. USA, 1996 93 2576-2581; Joukov et al., EMBO J., 1996 15 290-298). These cysteines are thought to be involved in intra- and inter-molecular disulfide bonding. In addition there are further strongly, but not completely, 10 conserved cysteine residues in the C-terminal domains. Loops 1, 2 and 3 of each subunit, which are formed by intra molecular disulfide bonding, are involved in binding to the receptors for the PDGF/VEGF family of growth factors (Andersson et al., Growth Factors, 1995 12 159-164). 15 Persons skilled in the art thus are well aware that these cysteine residues should be preserved in any proposed variant form, and that the active sites present in loops 1, 2 and 3 also should be preserved. However, other regions of the molecule can be expected to be of lesser importance for 20 biological function, and therefore offer suitable targets for modification. Modified polypeptides can readily be tested for their ability to show the biological activity of PDGF-C by routine activity assay procedures such as the fibroblast proliferation assay of Example 6. 25 It is contemplated that some modified PDGF-C polypeptides will have the ability to bind to PDGF-C receptors on cells including, but not limited to, endothelial cells, connective tissue cells, myofibroblasts and/or glial cells, but will be unable to stimulate cell proliferation, differentiation, 30 migration, motility or survival or to induce vascular proliferation, connective tissue development or wound healing. These modified polypeptides are expected to be able to act as competitive or non-competitive inhibitors of the PDGF-C -15- WO 00/18212 PCT/US99/22668 polypeptides and growth factors of the PDGF/VEGF family, and to be useful in situations where prevention or reduction of the PDGF-C polypeptide or PDGF/VEGF family growth factor action is desirable. Thus such receptor-binding but non 5 mitogenic, non-differentiation inducing, non-migration inducing, non-motility inducing, non-survival promoting, non connective tissue development promoting, non-wound healing or non-vascular proliferation inducing variants of the PDGF-C polypeptide are also within the scope of the invention, and 10 are referred to herein as "receptor-binding but otherwise inactive variant". Because PDGF-C forms a dimer in order to activate its only known receptor, it is contemplated that one monomer comprises the receptor-binding but otherwise inactive variant modified PDGF-C polypeptide and a second monomer 15 comprises a wild-type PDGF-C or a wild-type growth factor of the PDGF/VEGF family. These dimers can bind to its corresponding receptor but cannot induce downstream signaling. It is also contemplated that there are other modified PDGF-C polypeptides that can prevent binding of a wild-type 20 PDGF-C or a wild-type growth factor of the PDGF/VEGF family to its corresponding receptor on cells including, but not limited to, endothelial cells, connective tissue cells (such as fibroblasts), myofibroblasts and/or glial cells. Thus these dimers will be unable to stimulate endothelial cell 25 proliferation, differentiation, migration, survival, or induce vascular permeability, and/or stimulate proliferation and/or differentiation and/or motility of connective tissue cells, myofibroblasts or glial cells. These modified polypeptides are expected to be able to act as competitive or non 30 competitive inhibitors of the PDGF-C growth factor or a growth factor of the PDGF/VEGF family, and to be useful in situations where prevention or reduction of the PDGF-C growth factor or PDGF/VEGF family growth factor action is desirable. Such -16- WO 00/18212 PCTIUS99/22668 situations include the tissue remodeling that takes place during invasion of tumor cells into a normal cell population by primary or metastatic tumor formation. Thus such the PDGF C or PDGF/VEGF family growth factor-binding but non-mitogenic, 5 non-differentiation inducing, non-migration inducing, non motility inducing, non-survival promoting, non-connective tissue promoting, non-wound healing or non-vascular proliferation inducing variants of the PDGF-C growth factor are also within the scope of the invention, and are referred 10 to herein as "the PDGF-C growth factor-dimer forming but otherwise inactive or interfering variants". An example of a PDGF-C growth factor-dimer forming but otherwise inactive or interfering variant is where the PDGF-C has a mutation which prevents cleavage of CUB domain from the 15 protein. It is further contemplated that a PDGF-C growth factor-dimer forming but otherwise inactive or interfering variant could be made to comprise a monomer, preferably an activated monomer, of VEGF, VEGF-B, VEGF-C, VEGF-D, PDGF-C, PDGF-A, PDGF-B or PlGF linked to a CUB domain that has a 20 mutation which prevents cleavage of CUB domain from the protein. Dimers formed with the above mentioned PDGF-C growth factor-dimer forming but otherwise inactive or interfering variants and the monomers linked to the mutant CUB domain would be unable to bind to their corresponding receptors. 25 A variation on this contemplation would be to insert a proteolytic site between an activated monomer of VEGF, VEGF-B, VEGF-C, VEGF-D, PDGF-C, PDGF-A, PDGF-B or PlGF and the mutant CUB domain linkage which is dimerized to an activated monomer of VEGF, VEGF-B, VEGF-C, VEGF-D, PDGF-C, PDGF-A, PDGF-B or 30 PlGF. An addition of the specific protease(s) for this proteolytic site would cleave the CUB domain and thereby release an activated dimer that can then bind to its -17- WO 00/18212 PCTIUS99/22668 corresponding receptor. In this way, a controlled release of an activated dimer is made possible. According to a third aspect, the invention provides a purified and isolated nucleic acid encoding a polypeptide or 5 polypeptide fragment of the invention as defined above. The nucleic acid may be DNA, genomic DNA, cDNA or RNA, and may be single-stranded or double stranded. The nucleic acid may be isolated from a cell or tissue source, or of recombinant or synthetic origin. Because of the degeneracy of the genetic 10 code, the person skilled in the art will appreciate that many such coding sequences are possible, where each sequence encodes the amino acid sequence shown in Figure 2 (SEQ ID NO:3), Figure 4 (SEQ ID NO:5) or Figure 6 (SEQ ID NO:7), a bioactive fragment or analog thereof, a receptor-binding but 15 otherwise inactive or partially inactive variant thereof or a PDGF-C-dimer forming but otherwise inactive or interfering variants thereof. A fourth aspect of the invention provides vectors comprising the cDNA of the invention or a nucleic acid 20 molecule according to the third aspect of the invention, and host cells transformed or transfected with nucleic acids molecules or vectors of the invention. These may be eukaryotic or prokaryotic in origin. These cells are particularly suitable for expression of the polypeptide of the 25 invention, and include insect cells such as Sf9 cells, obtainable from the American Type Culture Collection (ATCC SRL-171) , transformed with a baculovirus vector, and the human embryo kidney cell line 293-EBNA transfected by a suitable expression plasmid. Preferred vectors of the 30 invention are expression vectors in which a nucleic acid according to the invention is operatively connected to one or more appropriate promoters and/or other control sequences, such that appropriate host cells transformed or transfected -18- WO 00/18212 PCTIUS99/22668 with the vectors are capable of expressing the polypeptide of the invention. Other preferred vectors are those suitable for transfection of mammalian cells, or for gene therapy, such as adenoviral-, vaccinia- or retroviral-based vectors or 5 liposomes. A variety of such vectors is known in the art. The invention also provides a method of making a vector capable of expressing a polypeptide encoded by a nucleic acid according to the invention, comprising the steps of operatively connecting the nucleic acid to one or more 10 appropriate promoters and/or other control sequences, as described above. The invention further provides a method of making a polypeptide according to the invention, comprising the steps of expressing a nucleic acid or vector of the invention in a 15 host cell, and isolating the polypeptide from the host cell or from the host cell's growth medium. In yet a further aspect, the invention provides an antibody specifically reactive with a polypeptide of the invention or a fragment of the polypeptide. This aspect of 20 the invention includes antibodies specific for the variant forms, immunoreactive fragments, analogs and recombinants of PDGF-C. Such antibodies are useful as inhibitors or agonists of PDGF-C and as diagnostic agents for detecting and quantifying PDGF-C. Polyclonal or monoclonal antibodies may 25 be used. Monoclonal and polyclonal antibodies can be raised against polypeptides of the invention or fragment or analog thereof using standard methods in the art. In addition the polypeptide can be linked to an epitope tag, such as the FLAG® octapeptide (Sigma, St. Louis, MO), to assist in affinity 30 purification. For some purposes, for example where a monoclonal antibody is to be used to inhibit effects of PDGF-C in a clinical situation, it may be desirable to use humanized or chimeric monoclonal antibodies. Such antibodies may be -19- WO 00/18212 PCT/US99/22668 further modified by addition of cytotoxic or cytostatic drugs. Methods for producing these, including recombinant DNA methods, are also well known in the art. This aspect of the invention also includes an antibody 5 which recognizes PDGF-C and is suitably labeled. Polypeptides or antibodies according to the invention may be labeled with a detectable label, and utilized for diagnostic purposes. Similarly, the thus-labeled polypeptide of the invention may be used to identify its corresponding 10 receptor in situ. The polypeptide or antibody may be covalently or non-covalently coupled to a suitable supermagnetic, paramagnetic, electron dense, ecogenic or radioactive agent for imaging. For use in diagnostic assays, radioactive or non-radioactive labels may be used. Examples 15 of radioactive labels include a radioactive atom or group, such as 1251 or 32 P. Examples of non-radioactive labels include enzymatic labels, such as horseradish peroxidase or fluorimetric labels, such as fluorescein-5-isothiocyanate (FITC) . Labeling may be direct or indirect, covalent or non 20 covalent. Clinical applications of the invention include diagnostic applications, acceleration of angiogenesis in tissue or organ transplantation, or stimulation of wound healing, or connective tissue development, or to establish collateral 25 circulation in tissue infarction or arterial stenosis, such as coronary artery disease, and inhibition of angiogenesis in the treatment of cancer or of diabetic retinopathy and inhibition of tissue remodeling that takes place during invasion of tumor cells into a normal cell population by 30 primary or metastatic tumor formation. Quantitation of PDGF-C in cancer biopsy specimens may be useful as an indicator of future metastatic risk. PDGF-C may also be relevant to a variety of lung -20- WO 00/18212 PCTIUS99/22668 conditions. PDGF-C assays could be used in the diagnosis of various lung disorders. PDGF-C could also be used in the treatment of lung disorders to improve blood circulation in the lung and/or gaseous exchange between the lungs and the 5 blood stream. Similarly, PDGF-C could be used to improve blood circulation to the heart and 02 gas permeability in cases of cardiac insufficiency. In a like manner, PDGF-C could be used to improve blood flow and gaseous exchange in chronic obstructive airway diseases. 10 Thus the invention provides a method of stimulation of angiogenesis, lymphangiogenesis, neovascularization, connective tissue development and/or wound healing in a mammal in need of such treatment, comprising the step of administering an effective dose of PDGF-C, or a fragment or 15 an analog thereof which has the biological activity of PDGF-C to the mammal. Optionally the PDGF-C, or fragment or analog thereof may be administered together with, or in conjunction with, one or more of VEGF, VEGF-B, VEGF-C, VEGF-D, PlGF, PDGF A, PDGF-B, FGF and/or heparin. 20 Conversely, PDGF-C antagonists (e.g. antibodies and/or competitive or noncompetitive inhibitors of binding of PDGF-C in both dimer formation and receptor binding) could be used to treat conditions, such as congestive heart failure, involving accumulation of fluid in, for example, the lung 25 resulting from increases in vascular permeability, by exerting an offsetting effect on vascular permeability in order to counteract the fluid accumulation. PDGF-C can also be used to treat fibrotic conditions including those found in the lung, kidney and liver. Administrations of PDGF-C could be 30 used to treat malabsorptive syndromes in the intestinal tract, liver or kidneys as a result of its blood circulation increasing and vascular permeability increasing activities. -21- WO 00/18212 PCT/US99/22668 Thus, the invention provides a method of inhibiting angiogenesis, lymphangiogenesis, neovascularization, connective tissue development and/or wound healing in a mammal in need of such treatment, comprising the step of 5 administering an effective amount of an antagonist of PDGF-C to the mammal. The antagonist may be any agent that prevents the action of PDGF-C, either by preventing the binding of PDGF-C to its corresponding receptor on the target cell, or by preventing activation of the receptor, such as using 10 receptor-binding PDGF-C variants. Suitable antagonists include, but are not limited to, antibodies directed against PDGF-C; competitive or non-competitive inhibitors of binding of PDGF-C to the PDGF-C receptor(s), such as the receptor binding or PDGF-C dimer-forming but non-mitogenic PDGF-C 15 variants referred to above; compounds that bind to PDGF-C and/or modify or antagonize its function, and anti-sense nucleotide sequences as described below. A method is provided for determining agents that bind to an activated truncated form of PDGF-C. The method comprises 20 contacting an activated truncated form of PDGF-C with a test agent and monitoring binding by any suitable means. Agents can include both compounds and other proteins. The invention provides a screening system for discovering agents that bind an activated truncated form of PDGF-C. The 25 screening system comprises preparing an activated truncated form of PDGF-C, exposing the activated truncated form of PDGF C to a test agent, and quantifying the binding of said agent to the activated truncated form of PDGF-C by any suitable means. This screening system can also be used to identify 30 agents which inhibit the proteolytic cleavage of the full length PDGF-C protein and thereby prevent the release of the activated truncated form of PDGF-C. For this use, the full length PDGF-C must be prepared. -22- WO 00/18212 PCT/US99/22668 Use of this screen system provides a means to determine compounds that may alter the biological function of PDGF-C. This screening method may be adapted to large-scale, automated procedures such as a PANDEX* (Baxter-Dade Diagnostics) system, 5 allowing for efficient high-volume screening of potential therapeutic agents. For this screening system, an activated truncated form of PDGF-C or full length PDGF-C is prepared as described herein, preferably using recombinant DNA technology. A test 10 agent, e.g. a compound or protein, is introduced into a reaction vessel containing the activated truncated form of or full length PDGF-C. Binding of the test agent to the activated truncated form of or full length PDGF-C is determined by any suitable means which include, but is not 15 limited to, radioactively- or chemically-labeling the test agent. Binding of the activated truncated form of or full length PDGF-C may also be carried out by a method disclosed in U.S. Patent 5,585,277, which is incorporated by reference. In this method, binding of the test agent to the activated 20 truncated form of or full length PDGF-C is assessed by monitoring the ratio of folded protein to unfolded protein. Examples of this monitoring can include, but are not limited to, monitoring the sensitivity of the activated truncated form of or full length PDGF-C to a protease, or amenability to 25 binding of the protein by a specific antibody against the folded state of the protein. Those of skill in the art will recognize that IC 50 values are dependent on the selectivity of the agent tested. For example, an agent with an IC 50 which is less than 10 nM is 30 generally considered an excellent candidate for drug therapy. However, an agent which has a lower affinity, but is selective for a particular target, may be an even better candidate. Those skilled in the art will recognize that any information -23- WO 00/18212 PCT/US99/22668 regarding the binding potential, inhibitory activity or selectivity of a particular agent is useful toward the development of pharmaceutical products. Where a PDGF-C or a PDGF-C antagonist is to be used for 5 therapeutic purposes, the dose(s) and route of administration will depend upon the nature of the patient and condition to be treated, and will be at the discretion of the attending physician or veterinarian. Suitable routes include oral, subcutaneous, intramuscular, intraperitoneal or intravenous 10 injection, parenteral, topical application, implants etc. Topical application of PDGF-C may be used in a manner analogous to VEGF. For example, where used for wound healing or other use in which enhanced angiogenesis is advantageous, an effective amount of the truncated active form of PDGF-C is 15 administered to an organism in need thereof in a dose between about 0.1 and 1000 pg/kg body weight. The PDGF-C or a PDGF-C antagonist may be employed in combination with a suitable pharmaceutical carrier. The resulting compositions comprise a therapeutically effective 20 amount of PDGF-C or a PDGF-C antagonist, and a pharmaceutically acceptable non-toxic salt thereof, and a pharmaceutically acceptable solid or liquid carrier or adjuvant. Examples of such a carrier or adjuvant include, but are not limited to, saline, buffered saline, Ringer's 25 solution, mineral oil, talc, corn starch, gelatin, lactose, sucrose, microcrystalline cellulose, kaolin, mannitol, dicalcium phosphate, sodium chloride, alginic acid, dextrose, water, glycerol, ethanol, thickeners, stabilizers, suspending agents and combinations thereof. Such compositions may be in 30 the form of solutions, suspensions, tablets, capsules, creams, salves, elixirs, syrups, wafers, ointments or other conventional forms. The formulation to suit the mode of administration. Compositions which comprise PDGF-C may -24- WO 00/18212 PCT/US99/22668 optionally further comprise one or more of PDGF-A, PDGF-B, VEGF, VEGF-B, VEGF-C, VEGF-D, PlGF and/or heparin. Compositions comprising PDGF-C will contain from about 0.1% to 90% by weight of the active compound(s), and most generally 5 from about 10% to 30%. For intramuscular preparations, a sterile formulation, preferably a suitable soluble salt form of the truncated active form of PDGF-C, such as hydrochloride salt, can be dissolved and administered in a pharmaceutical diluent such 10 as pyrogen-free water (distilled), physiological saline or 5% glucose solution. A suitable insoluble form of the compound may be prepared and administered as a suspension in an aqueous base or a pharmaceutically acceptable oil base, e.g. an ester of a long chain fatty acid such as ethyl oleate. 15 According to yet a further aspect, the invention provides diagnostic/prognostic devices typically in the form of test kits. For example, in one embodiment of the invention there is provided a diagnostic/prognostic test kit comprising an antibody to PDGF-C and a means for detecting, and more 20 preferably evaluating, binding between the antibody and PDGF C. In one preferred embodiment of the diagnostic/prognostic device according to the invention, a second antibody (the secondary antibody) directed against antibodies of the same isotype and animal source of the antibody directed against 25 PDGF-C (the primary antibody) is provided. The secondary antibody is coupled to a detectable label, and then either an unlabeled primary antibody or PDGF-C is substrate-bound so that the PDGF-C/primary antibody interaction can be established by determining the amount of label bound to the 30 substrate following binding between the primary antibody and PDGF-C and the subsequent binding of the labeled secondary antibody to the primary antibody. In a particularly preferred embodiment of the invention, the diagnostic/prognostic device -25- WO 00/18212 PCT/US99/22668 may be provided as a conventional enzyme-linked immunosorbent assay (ELISA) kit. In another alternative embodiment, a diagnostic/prognostic device may comprise polymerase chain 5 reaction means for establishing sequence differences of a PDGF-C of a test individual and comparing this sequence structure with that disclosed in this application in order to detect any abnormalities, with a view to establishing whether any aberrations in PDGF-C expression are related to a given 10 disease condition. In addition, a diagnostic/prognostic device may comprise a restriction length polymorphism (RFLP)generating means utilizing restriction enzymes and genomic DNA from a test individual to generate a pattern of DNA bands on a gel and 15 comparing this pattern with that disclosed in this application in order to detect any abnormalities, with a view to establishing whether any aberrations in PDGF-C expression are related to a given disease condition. In accordance with a further aspect, the invention 20 relates to a method of detecting aberrations in PDGF-C gene in a test subject which may be associated with a disease condition in the test subject. This method comprises providing a DNA or RNA sample from said test subject; contacting the DNA sample or RNA with a set of primers 25 specific to PDGF-C DNA operatively coupled to a polymerase and selectively amplifying PDGF-C DNA from the sample by polymerase chain reaction, and comparing the nucleotide sequence of the amplified PDGF-C DNA from the sample with the nucleotide sequences shown in Figure 1 (SEQ ID NO:2) or Figure 30 3 (SEQ ID NO:5). The invention also includes the provision of a test kit comprising a pair of primers specific to PDGF-C DNA operatively coupled to a polymerase, whereby said -26- WO 00/18212 PCTIUS99/22668 polymerase is enabled to selectively amplify PDGF-C DNA from a DNA sample. The invention also provides a method of detecting PDGF-C in a biological sample, comprising the step of contacting the 5 sample with a reagent capable of binding PDGF-C, and detecting the binding. Preferably the reagent capable of binding PDGF-C is an antibody directed against PDGF-C, particularly preferably a monoclonal antibody. In a preferred embodiment the binding and/or extent of binding is detected by means of 10 a detectable label; suitable labels are discussed above. In another aspect, the invention relates to a protein dimer comprising the PDGF-C polypeptide, particularly a disulfide-linked dimer. The protein dimers of the invention include both homodimers of PDGF-C polypeptide and heterodimers 15 of PDGF-C and VEGF, VEGF-B, VEGF-C, VEGF-D, PlGF, PDGF-A or PDGF-B. According to a yet further aspect of the invention there is provided a method for isolation of PDGF-C comprising the step of exposing a cell which expresses PDGF-C to heparin to 20 facilitate release of PDGF-C from the cell, and purifying the thus-released PDGF-C. Another aspect of the invention involves providing a vector comprising an anti-sense nucleotide sequence which is complementary to at least a part of a DNA sequence which 25 encodes PDGF-C or a fragment or analog thereof that has the biological activity of PDGF-C. In addition the anti-sense nucleotide sequence can be to the promoter region of the PDGF C gene or other non-coding region of the gene which may be used to inhibit, or at least mitigate, PDGF-C expression. 30 According to a yet further aspect of the invention such a vector comprising an anti-sense sequence may be used to inhibit, or at least mitigate, PDGF-C expression. The use of a vector of this type to inhibit PDGF-C expression is favored -27- WO 00/18212 PCT/US99/22668 in instances where PDGF-C expression is associated with a disease, for example where tumors produce PDGF-C in order to provide for angiogenesis, or tissue remodeling that takes place during invasion of tumor cells into a normal cell 5 population by primary or metastatic tumor formation. Transformation of such tumor cells with a vector containing an anti-sense nucleotide sequence would inhibit or retard growth of the tumor or tissue remodeling. Another aspect of the invention relates to the discovery 10 that the full length PDGF-C protein is likely to be a latent growth factor that needs to be activated by proteolytic processing to release an active PDGF/VEGF homology domain. A putative proteolytic site is found in residues 231-234 in the full length protein, residues -RKSR-. This is a dibasic 15 motif. This site is structurally conserved in the mouse PDGF C. The -RKSR- putative proteolytic site is also found in PDGF-A, PDGF-B, VEGF-C and VEGF-D. In these four proteins, the putative proteolytic site is also found just before the minimal domain for the PDGF/VEGF homology domain. Together 20 these facts indicate that this is the proteolytic site. Preferred proteases include, but are not limited, to plasmin, Factor X and enterokinase. The N-terminal CUB domain may function as an inhibitory domain which might be used to keep PDGF-C in a latent form in some extracellular compartment 25 and which is removed by limited proteolysis when PDGF-C is needed. According to this aspect of the invention, a method is provided for producing an activated truncated form of PDGF-C or for regulating receptor-binding specificity of PDGF-C. 30 These methods comprise the steps of expressing an expression vector comprising a polynucleotide encoding a polypeptide having the biological activity of PDGF-C and supplying a proteolytic amount of at least one enzyme for processing the -28- WO 00/18212 PCT/US99/22668 expressed polypeptide to generate the activated truncated form of PDGF-C. This aspect also includes a method for selectively activating a polypeptide having a growth factor activity. 5 This method comprises the step expressing an expression vector comprising a polynucleotide encoding a polypeptide having a growth factor activity, a CUB domain and a proteolytic site between the polypeptide and the CUB domain, and supplying a proteolytic amount of at least one enzyme for processing the 10 expressed polypeptide to generate the activated polypeptide having a growth factor activity. In addition, this aspect includes the isolation of a nucleic acid molecule which codes for a polypeptide having the biological activity of PDGF-C and a polypeptide thereof which 15 comprises a proteolytic site having the amino acid sequence RKSR or a structurally conserved amino acid sequence thereof. Also this aspect includes an isolated dimer comprising an activated monomer of PDGF-C and an activated monomer of VEGF, VEGF-B, VEGF-C, VEGF-D, PDGF-C, PDGF-A, PDGF-B or PlGF 20 linked to a CUB domain, or alternatively, an activated monomer of VEGF, VEGF-B, VEGF-C, VEGF-D, PDGF-C, PDGF-A, PDGF-B or PlGF and an activated monomer of PDGF-C linked to a CUB domain. The isolated dimer may or may not include a proteolytic site between the activator monomer and the CUB 25 domain linkage. Polynucleotides of the invention such as those described above, fragments of those polynucleotides, and variants of those polynucleotides with sufficient similarity to the non coding strand of those polynucleotides to hybridize thereto 30 under stringent conditions all are useful for identifying, purifying, and isolating polynucleotides encoding other, non human, mammalian forms of PDGF-C. Thus, such polynucleotide fragments and variants are intended as aspects of the -29- WO 00/18212 PCT/US99/22668 invention. Exemplary stringent hybridization conditions are as follows: hybridization at 420C in 5X SSC, 20 mM NaPO 4 , pH 6.8, 50% formamide; and washing at 42 0 C in 0.2X SSC. Those skilled in the art understand that it is desirable to vary 5 these conditions empirically based on the length and the GC nucleotide base content of the sequences to be hybridized, and that formulas for determining such variation exist. See for example Sambrook et al, "Molecular Cloning: A Laboratory Manual", Second Edition, pages 9.47-9.51, Cold Spring Harbor, 10 New York: Cold Spring Harbor Laboratory (1989). Moreover, purified and isolated polynucleotides encoding other, non-human, mammalian PDGF-C forms also are aspects of the invention, as are the polypeptides encoded thereby and antibodies that are specifically immunoreactive with the non 15 human PDGF-C variants. Thus, the invention includes a purified and isolated mammalian PDGF-C polypeptide and also a purified and isolated polynucleotide encoding such a polypeptide. It will be clearly understood that nucleic acids and 20 polypeptides of the invention may be prepared by synthetic means or by recombinant means, or may be purified from natural sources. It will be clearly understood that for the purposes of this specification the word "comprising" means "included but 25 not limited to". The corresponding meaning applies to the word "comprises". BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 (SEQ ID NO:2) shows the complete nucleotide 30 sequence of cDNA encoding a human PDGF-C (hPDGF-C) (2108 bp); Figure 2 (SEQ ID NO:3) shows the deduced amino acid sequence of full-length hPDGF-C which consists of 345 amino -30- WO 00/18212 PCT/US99/22668 acid residues (the translated part of the cDNA corresponds to nucleotides 37 to 1071 of Figure 1); Figure 3 (SEQ ID NO:4) shows a cDNA sequence encoding a fragment of human PDGF-C (hPDGF-C) (1536 bp); 5 Figure 4 (SEQ ID NO:5) shows a deduced amino acid sequence of a fragment of hPDGF-C(translation of nucleotides 3 to 956 of the nucleotide sequence of Figure 3); Figure 5 (SEQ ID NO:6) shows a nucleotide sequence of a murine PDGF-C (mPDGF-C) cDNA; 0 Figure 6 (SEQ ID NO:7) shows the deduced amino acid sequence of a fragment of mPDGF-C(the translated part of the cDNA corresponds to nucleotides 196 to 1233 of Figure 5); Figure 7 shows a comparative sequence alignment of the hPDGF-C amino acid sequence of Figure 2 (SEQ ID NO:3) with the 5 mPDGF-C amino acid sequence of Figure 6 (SEQ ID NO:7); Figure 8 shows a schematic structure of mPDGF-C with a signal sequence (striped box), a N-terminal Clr/Cls/embryonic sea urchin protein Uegf/bone morphogenetic protein 1 (CUB) domain and the C-terminal PDGF/VEGF-homology domain (open 0 boxes); Figure 9 shows a comparative sequence alignment of the PDGF/VEGF-homology domains in human and mouse PDGF-C with other members of the VEGF/PDGF family of growth factors (SEQ ID NOs:8-17, respectively); 5 Figure 10 shows a phylogenetic tree of several growth factors belonging to the VEGF/PDGF family; Figure 11 provides the amino acid sequence alignment of the CUB domain present in human and mouse PDGF-Cs (SEQ ID NOs:18 and 19, respectively) and other CUB domains present in 30 human bone morphogenic protein-1 (hBMP-1, 3 CUB domains CUB1 3) (SEQ ID NOs:20-22, respectively) and in human neuropilin-1 (2 CUB domains) (SEQ ID NOs:23 and 24, respectively); -31- WO 00/18212 PCT/US99/22668 Figure 12 shows a Northern blot analysis of the expression of PDGF-C transcripts in several human tissues; Figure 13 shows the regulation of PDGF-C mRNA expression by hypoxia; and 5 Figure 14 shows the expression of PDGF-C in human tumor cell lines. Figure 15 shows the results of immunoblot detection of full length human PDGF-C in transfected COS-1 cells. Figure 16 shows isolation and partial characterization 10 of full length PDGF-C. Figure 17 shows isolation and partial characterization of a truncated form of human PDGF-C containing the PDGF/VEGF homology domain only. Figure 18 provides a standard curve for the binding of 15 labeled PDGF-BB homodimers to PAE-1 cells expressing PDGF alpha receptor. Figure 19 provides a graphic representation of the inhibition of binding of labeled PDGF-BB to PAE-1 cells expressing PDGF alpha receptor by increasing amounts of 20 purified full length and truncated PDGF-CC proteins. Figure 20 shows the effects of the full length and truncated PDGF-CC homodimers on the phosphorylation of PDGF alpha-receptor. Figure 21 shows the mitogenic activities of the full 25 length and truncated PDGF-CC homodimers on fibroblasts. Figure 22 graphically presents the results of the binding assay of truncated PDGF-C to the PDGF receptors. Figure 23 shows the immunoblot of the undigested full length PDGF-C protein and the plasmin-generated 26-28 kDa 30 species. Figure 24 graphically presents the results of the competitive binding assay of full-length PDGF-C and truncated PDGF-C for PDGFR-alpha receptors. -32- WO 00/18212 PCT/US99/22668 Figure 25 shows the analyses by SDS-PAGE of the human PDGF-C CUB domain under reducing and non-reducing conditions. Figures 26A-26V show PDGF-C expression in the developing mouse embryo. 5 Figures 27A-27F show PDGF-C, PDGF-A and PDGFR-alpha expression in the developing kidney. Figures 28A-28F show histology of E 16.5 kidneys from wildtype (Figures 28A and 28C), PDGFR-alpha -/- (Figures 28B and 28F, PDGF-A -/- (Figure 28D) and PDGF-A/PDGF-B double -7 10 (Figure 28E) kidneys. DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS Figure 1 (SEQ ID NO:2) shows the complete nucleotide 15 sequence of cDNA encoding a human PDGF-C (hPDGF-C) (2108 bp), which is a new member of the VEGF/PDGF family. A clone #4 (see Figures 3 and 4 - SEQ ID NOs:4 and 5) encoding hPDGF-C was not full length and lacked approximately 80 base pairs of coding sequence when compared to the mouse protein 20 (corresponding to 27 amino acids). Additional cDNA clones were isolated from a human fetal lung cDNA library to obtain an insert which included this missing sequence. Clone #10 had a longer insert than clone #4. The insert of clone #10 was sequenced in the 5' region and it was found to contain the 25 missing sequence. Clone #10 was found to include the full sequence of human PDGF-C. Some 5'-untranslated sequence, the translated part of the cDNA encoding human PDGF-C and some 3' untranslated nucleotide sequence are shown in Figure 1 (SEQ ID NO:2). A stop codon in frame is located 21 bp upstream of 30 the initiation ATG (the initiation ATG is underlined in Figure 1). Work to isolate this new human PDGF/VEGF began after a search of the expressed sequence tag (EST) database, dbEST, -33- WO 00/18212 PCTIUS99/22668 at the National Center for Biotechnology Information (NCBI) in Washington, DC, identified a human EST sequence (W21436) which appears to encode part of the human homolog of the mouse PDGF-C. Based on the human EST sequence, two oligonucleotides 5 were designed: 5'-GAA GTT GAG GAA CCC AGT G-3' forward (SEQ ID NO:25) 5'-CTT GCC AAG AAG TTG CCA AG-3' reverse (SEQ ID NO:26) These oligonucleotides were used to amplify by polymerase chain reaction (PCR) a polynucleotide of 348 bps from a Human 10 Fetal Lung 5'-STRETCH PLUS Xgt1O cDNA library, which was obtained commercially from Clontech. The PCR product was cloned into the pCR 2.1-vector of the Original TA Cloning Kit (Invitrogen). Subsequently, the 348 bps cloned PCR product was used to construct a hPDGF-C probe according to standard 15 techniques. 106 lambda-clones of the Human Fetal Lung 5'-STRETCH PLUS xgtlO cDNA Library (Clontech) were screened with the hPDGF-C probe according to standard procedures. Among several positive clones, one, clone #4 was analyzed more carefully and 20 the nucleotide sequence of its insert was determined according to standard procedures using internal and vector oligonucleotides. The insert of clone #4 contains a partial nucleotide sequence of the cDNA encoding the full length human PDGF-C (hPDGF-C). The nucleotide sequence (1536 bp) of the 25 clone #4 insert is shown in Figure 3 (SEQ ID NO:4) . The translated portion of this cDNA includes nucleotides 6 to 956. The deduced amino acid sequence of the translated portion of the insert is illustrated in Figure 4 (SEQ ID NO:5) . A polypeptide of this deduced amino acid sequence would lack the 30 first 28 amino acid residues found in the full length hPDGF-C polypeptide. However, this polypeptide includes a proteolytic fragment which is sufficient to activate the PDGF alpha -34- WO 00/18212 PCT/US99/22668 receptors. It should be noted that the first glycine (Gly) of SEQ ID NO:5 is not found in the full length hPDGF-C. A mouse EST sequence (AI020581) was identified in a database search of the dbEST database at the NCBI in 5 Washington, DC, which appears to encode part of a new mouse PDGF, PDGF-C. Large parts of the mouse cDNA was obtained by PCR amplification using DNA from a mouse embryo AgtlO cDNA library as the template. To amplify the 3' end of the cDNA, a sense primer derived from the mouse EST sequence was used 10 (the sequence of this primer was 5'-CTT CAG TAC CTT GGA AGA G, primer 1 (SEQ ID NO:27)) To amplify the 5'end of the cDNA, an antisense primer derived from the mouse EST was used (the sequence of this primer was 5'-CGC TTG ACC AGG AGA CAA C, primer 2 (SEQ ID NO:28)) . The AgtlO vector primers were sense 15 5'-ACG TGA ATT CAG CAA GTT CAG CCT GGT TAA (primer 3 (SEQ ID NO:29)) and antisense 5'-ACG TGG ATC CTG AGT ATT TCT TCC AGG GTA (primer 4 (SEQ ID NO:30) ) . Combinations of the vector primers and the internal primers obtained from the mouse EST were used in standard PCR reactions. The sizes of the 20 amplified fragments were approx. 750 bp (3'-fragment) and 800 bp (5'-fragment), respectively. These fragments were cloned into the pCR 2.1 vector and subjected to nucleotide sequences analysis using vector primers and internal primers. Since these fragments did not contain the full length sequence of 25 mPDGF-C, a mouse liver ZAP cDNA library was screened using standard conditions. A 261 bp 32P-labeled PCR fragment was generated for use as a probe using primers 1 and 2 and using DNA from the mouse embryo AgtlO library as the template (see above). Several positive plaques were purified and the 30 nucleotide sequence of the inserts were obtained following subcloning into pBluescript. Vector specific primers and internal primers were used. By combining the nucleotide sequence information of the generated PCR clones and the -35- WO 00/18212 PCT/US99/22668 isolated clone, the full length amino acid sequence of mPDGF-C could be deduced (see Figure 6) (SEQ ID NO:7). Figure 7 shows a comparative sequence alignment of the mouse and human amino acid sequences of PDGF-C (SEQ ID NOS:6 5 and 2, respectively). The alignment shows that human and mouse PDGF-Cs display an identity of about 87% with 45 amino acid replacements found among the 345 residues of the full length proteins. Almost all of the observed amino acid replacements are conservative in nature. The predicted 10 cleavage site in mPDGF-C for the signal peptidase is between residues G19 and T20. This would generate a secreted mouse peptide of 326 amino acid residues. Figure 8 provides a schematic domain structure of mouse PDGF-C with a signal sequence (striped box), a N-terminal CUB 15 domain and the C-terminal PDGF/VEGF-homology domain (open boxes). The amino acid sequences denoted by the lines have no obvious similarities to CUB domains or to VEGF-homology domains. 20 The high sequence identity suggests that human and mouse PDGF-C have an almost identical domain structure. Amino acid sequence comparisons revealed that both mouse and human PDGF-C display a novel domain structure. Apart from the PDGF/VEGF homology domain located in the C-terminal region in both 25 proteins (residues 164 to 345), the N-terminal region in both PDGF-Cs have a domain referred to as a CUB domain (Bork and Beckmann, J. Mol. Biol., 1993 231, 539-545). This domain of about 110 amino acids (amino acid residues 50-160) was originally identified in complement factors Clr/Cls, but has 30 recently been identified in several other extracellular proteins including signaling molecules such as bone morphogenic protein 1 (BMP-1) (Wozney et al.,Science, 1988 242, 1528-1534) as well as in several receptor molecules such -36- WO 00/18212 PCT/US99/22668 as neuropilin-1 (NP-1) (Soker et al., Cell, 1998 92 735-745). The functional roles of CUB domains are not clear but it may participate in protein-protein interactions or in interactions with carbohydrates including heparin sulfate proteoglycans. 5 Figure 9 shows the amino acid sequence alignment of the C-terminal PDGF/VEGF-homology domains of human and mouse PDGF Cs with the C-terminal PDGF/VEGF-homology domains of PDGF/VEGF family members, VEGF 165 , PlGF-2, VEGF-Bi, 7 , Pox Orf VEGF, VEGF C, VEGF-D, PDGF-A and PDGF-B (SEQ ID NOs:8-17). Some of the 10 amino acid sequences in the N- and C-terminal regions in VEGF C and VEGF-D have been deleted in this figure. Gaps were introduced to optimize the alignment. This alignment was generated using the method of J. Hein, (Methods Enzymol. 1990 183 626-45) with PAM250 residue weight table. The boxed 15 residues indicate amino acids which match the PDGF-Cs within two distance units. The alignment shows that PDGF-C has the expected pattern of invariant cysteine residues, a hallmark of members of this family, with one exception. Between cysteine 3 and 4, 20 normally spaced by 2 residues there is an insertion of three extra amino acids (NCA). This feature of the sequence in PDGF-C was highly unexpected. Based on the amino acid sequence alignments in Figure 9, a phylogenetic tree was constructed and is shown in Figure 10. 25 The data show that the PDGF-C homology domain is closely related to the PDGF/VEGF-homology domains of VEGF-C and VEGF D. As shown in Figure 11, the amino acid sequences from several CUB-containing proteins were aligned (SEQ ID NOs:18 30 24) . The results show that the single CUB domain in human and mouse PDGF-C (SEQ ID NOs:18 and 19, respectively) displays a significant identify with the most closely related CUB domains. Sequences from human BMP-1, with 3 CUB domains (CUB1 -37- WO 00/18212 PCT/US99/22668 3 (SEQ ID NOs:20-22)) and human neuropilin-1 with 2 CUB domains (CUB1-2) (SEQ ID NOs:23 and 24, respectively) are shown. Gaps were introduced to optimize the alignment. This alignment was generated using the method of J. Hein, (Methods 5 Enzymol., 1990 183 626-45) with PAM250 residue weight table. Figure 12 shows a Northern blot analysis of the expression of PDGF-C transcripts in several human tissues. The analysis shows that PDGF-C is encoded by a major transcript of approximately 3.8-3.9 kb, and a minor of 2.8 kb. 10 The numbers to the right refer to the size of the mRNAs (in kb). The tissue expression of PDGF-C was determined by Northern blotting using a commercial Multiple Tissue Northern blot (MTN, Clontech) . The blots were hybridized at according to the instructions from the supplier using ExpressHyb 15 solution at 68'C for one hour (high stringency conditions), and probed with a 353 bp hPDGF-C EST probe from the fetal lung cDNA library screening as described above. The blots were subsequently washed at 50'C in 2X SSC with 0.05% SDS for 30 minutes and at 500C in 0.1X SSC with 0.1% SDS for an 20 additional 40 minutes. The blots were then put on film and exposed at -70'C. The blots show that PDGF-C transcripts are most abundant in heart, liver, kidney, pancreas and ovary while lower levels of transcripts are present in most other tissues, including placenta, skeletal muscle and prostate. 25 PDGF-C transcripts were below the level of detection in spleen, colon and peripheral blood leucocytes. Figure 13 shows the regulation of PDGF-C mRNA expression by hypoxia. Size markers (in kb) are indicated to the left in the lower panel. The estimated sizes of PDGF-C mRNAs is 30 indicated to the left in the upper panel (2.7 and 3.5 kbs, respectively). To explore whether PDGF-C is induced by hypoxia, cultured human skin fibroblasts were exposed to hypoxia for 0, 4, 8 and 24 hours. Poly(A)+ mRNA was isolated -38- WO 00/18212 PCT/US99/22668 from cells using oligo-dT cellulose affinity purification. Isolated mRNAs were electrophoresed through 12% agarose gels using 4 pg of mRNA per line. A Northern blot was made and hybridized with a probe for PDGF-C. The sizes of the two 5 bands were determined by hybridizing the same filter with a mixture of hVEGF, hVEGF-B and hVEGF-C probes (Enholm et al. Oncogene, 1997 14 2475-2483), and interpolating on the basis of the known sizes of these mRNAs. The results shown in Figure 13 indicate that PDGF-C is not regulated by hypoxia in human 10 skin fibroblasts. Figure 14 shows the expression of PDGF-C mRNA in human tumor cells lines. To explore whether PDGF-C was expressed in human tumor cell lines, poly(A)+ mRNA was isolated from several known tumor cell lines, the mRNAs were electrophoresed 15 through a 12% agarose gel and analyzed by Northern blotting and hybridization with the PDGF-C probe. The results shown in Figure 14 demonstrate that PDGF-C mRNA is expressed in several types of human tumor cell lines such as JEG3 (a human choriocarcinoma, ATCC #HTB-36), G401 (a Wilms tumor, ATCC 20 #CRL-1441), DAMI (a megakaryoblastic leukemia), A549 (a human lung carcinoma, ATCC #CCL-185) and HEL (a human erythroleukemia, ATCC #TID-180) . It is contemplated that further growth of these PDGF-C expressing tumors can be inhibited by inhibiting PDGF-C. As well as using PDGF-C 25 expression as a means of identifying specific types of tumors. Example 1: Generation of specific antipeptide antibodies to human PDGF-C Two synthetic peptides were generated and then used to 30 raise antibodies against human PDGF-C. The first synthetic peptide corresponds to residues 29-48 of the N-terminus of full length PDGF-C and includes an extra cysteine residue at the N- and C-terminus: CKFQFSSNKEQNGVQDPQHERC (SEQ ID NO:31) -39- WO 00/18212 PCT/US99/22668 The second synthetic peptide corresponds to residues 230-250 of the internal region of full length PDGF-C and includes an extra cysteine residue at the C-terminus: GRKSRVVDLNLLTEEVRLYSC (SEQ ID NO:32). The two peptides were 5 each conjugated to the carrier protein keyhole limpet hemocyanin (KLH, Calbiochem) using N-succinimidyl 3-(2-pyridyldithio)propionate (SPDP) (Pharmacia Inc.) according to the instructions of the supplier. 200-300 micrograms of the conjugates in phosphate buffered saline 10 (PBS) were separately emulsified in Freunds Complete Adjuvant and injected subcutaneously at multiple sites in rabbits. The rabbits were boostered subcutaneously at biweekly intervals with the same amount of the conjugates emulsified in Freunds Incomplete Adjuvant. Blood was drawn and collected from the 15 rabbits. The sera were prepared using standard procedures known to those skilled in the art. Example 2: Expression of full length human PDGF-C in mammalian cells 20 The full length cDNA encoding human PDGF-C was cloned into the mammalian expression vector, pSG5 (Stratagene, La Jolla, CA) that has the SV40 promoter. COS-1 cells were transfected with this construct and in separate transfections, with a pSG5 vector without the cDNA insert for a control, 25 using the DEAE-dextran procedure. Serum free medium was added to the transfected COS-1 cells 24 hours after the transfections and aliquots containing the secreted proteins were collected for a 24 hour period after the addition of the medium. These aliquots were subjected to precipitation using 30 ice cold 10% trichloroacetic acid for 30 minutes, and the precipitates were washed with acetone. The precipitated proteins were dissolved in SDS loading buffer under reducing conditions and separated on a SDS-PAGE gel using standard -40- WO 00/18212 PCTIUS99/22668 procedures. The separated proteins were electrotransferred onto Hybond filter and immunoblotted using a rabbit antiserum against the internal peptide of full length PDGF-C, the preparation of which is described above. Bound antibodies 5 were detected using enhanced chemiluminescence (ECL, Amersham Inc.). Figure 15 shows the results of this immunoblot. The sample was only partially reduced and the monomer of the human PDGF-C migrated as a 55 kDa species (the lower band) and the dimer migrated as a 100 kDa species (upper band). This 10 indicates that the protein is secreted intact and that no major proteolytic processing occurs during secretion of the molecule in mammalian cells. Example 3: Expression of full length and truncated human 15 PDGF-C in baculovirus infected Sf9 cells The full length coding part of the human PDGF-C cDNA (970 bp) was amplified by PCR using Deep Vent DNA polymerase (Biolabs) using standard conditions and procedures. The full length PDGF-C was amplified for 30 cycles, where each cycle 20 consisted of one minute denaturization at 94'C, one minute annealing at 560C and two minutes extension at 72'C. The forward primer used was 5'CGGGATCCCGAATCCAACCTGAGTAG3' (SEQ ID NO:33) . This primer includes a BamHI site (underlined) for in frame cloning. The reverse primer used was 25 5'GGAATTCCTAATGGTGATGGTGATGATGTTTGTCATCGTCATCTCCTCCTGTGCTC CCTCT3' (SEQ ID NO:34). This primer includes an EcoRI site (underlined) and sequences coding for a C-terminal 6X His tag preceded by an enterokinase site. In addition, residues 230 345 of the PDGF/VEGF homology domain (PVHD) of human PDGF-C 30 were amplified by PCR using Deep Vent DNA polymerase (Biolabs) using standard conditions and procedures. The residues 230 345 of the PVHD of PDGF-C were amplified for 25 cycles, where each cycle consisted of one minute denaturization at 940C, -41- WO 00/18212 PCT/US99/22668 four minutes annealing at 56 0 C and four minutes extension at 72 0 C. The forward primer used was 5'CGGATCCCGGAAGAAAATCCA GAGTGGTG3' (SEQ ID NO:35) . This primer includes a BamHI site (underlined) for in frame cloning. The reverse primer used 5 was 5'GGAATTCCTAATGGTGATGGTGATGATGTTTGTCATCGTCATCTCCTCCTGTG CTCCCTCT-3' (SEQ ID NO:36). This primer includes an EcoRI site (underlined) and sequences coding for a C-terminal 6X His tag preceded by an enterokinase site. The PCR products were digested with BamHI and EcoRI and subsequently cloned into the 10 baculovirus expression vector, pAcGP67A. Verification of the correct sequence of the PCR products cloned into the constructs was by nucleotide sequencing. The expression vectors were then co-transfected with BaculoGold linearized baculovirus DNA into Sf9 insect cells according to the 15 manufactures protocol (Pharmingen). Recombined baculovirus were amplified several times before beginning large scale protein production and protein purification according to the manual (Pharmingen). Sf9 cells, adapted to serum free medium, were infected 20 with recombinant baculovirus at a multiplicity of infection of about 7. Media containing the recombinant proteins were harvested 4 days after infection and were incubated with Ni-NTA-Agarose beads(Qiagen). The beads were collected in a column and after extensive washing with 50 mM sodium 25 phosphate buffer pH 8, containing 300 mM NaCl (the washing buffer), the bound proteins were eluted with increasing concentrations of imidazole (from 100 mM to 500 mM) in the washing buffer. The eluted proteins were analyzed by SDS PAGE using 12.5% polyacrylamide gels under reducing and 30 non-reducing conditions. For immunoblotting analyses, the proteins were electrotransferred onto Hybond filters for 45 minutes. -42- WO 00/18212 PCT/US99/22668 Figures 16A-C show the isolation and partial characterization of full length human PDGF-C protein. In Figure 16A, the recombinant full length protein was visualized on the blot using antipeptide antibodies against 5 the N-terminal peptide (described above). In Figure 16B, the recombinant full length protein was visualized on the blot using antipeptide antibodies against the internal peptide (described above). The separated proteins were visualized by staining with Coomassie Brilliant Blue 10 (Figure 16C). The numbers at the bottom of Figures 16A-C refer to the concentration of imidazole used to elute the protein from the Ni-NTA column and are expressed in molarity (M). Figures 16A-C also show that the full length protein migrates as a 90 kDa species under non-reducing 15 conditions and as a 55 kDa species under reducing conditions. This indicates that the full length protein was expressed as a disulfide-linked dimer. Figures 17A-C show the analysis of the isolation and partial characterization of a truncated form of human PDGF 20 C containing the PDGF/VEGF homology domain only. In Figure 17A, the immunoblot analysis of fractions eluted from the Ni-agarose column demonstrates that the protein could be eluted at imidazole concentrations ranging between 100-500 mM. The eluted fractions were analyzed under non-reducing 25 conditions, and the truncated human PDGF-C was visualized on the blot using antipeptide antibodies against the internal peptide (described above). Figure 17B shows the Coomassie Brilliant Blue staining of the same fractions as in Figure 17A. This shows that the procedure generates 30 highly purified material migrating as a 36 kDa species. Figure 17C shows the Coomassie Brilliant Blue staining of non-reduced (non-red.) and reduced (red.) truncated human PDGF-C protein. The data show that the protein is a -43- WO 00/18212 PCT/US99/22668 secreted dimer held together by disulfide bonds and that the monomer migrates as a 24 kDa species. Example 4: Receptor binding properties of full length and 5 truncated PDGF-C To assess the interactions between full length and truncated PDGF-C and the VEGF receptors, full length and truncated PDGF-C were tested for their capacity to bind to soluble Ig-fusion proteins containing the extracellular 10 domains of human VEGFR-1, VEGFR-2 and VEGFR-3 (Olofsson et al., Proc. Natl. Acad. Sci. USA, 1998 95 11709-11714). The fusion proteins, designated VEGFR-1-Ig, VEGFR-2-Ig and VEGFR-3-Ig, were transiently expressed in human 293 EBNA cells. All Ig fusion proteins were human VEGFRs. Cells 15 were incubated for 24 hours after transfection, washed with Dulbecco's Modified Eagle Medium (DMEM) containing 0.2% bovine serum albumin and starved for 24 hours. The fusion proteins were then precipitated from the clarified conditioned medium using protein A-Sepharose beads 20 (Pharmacia). The beads were combined with 100 microliters of 1OX binding buffer (5% bovine serum albumin, 0.2% Tween 20 and 10 pg/ml heparin) and 900 microliter of conditioned medium from 293 cells that had been transfected with mammalian expression plasmids encoding full length or 25 truncated PDGF-C or control vector, then metabolically labeled with 1 5 S-cysteine and methionine (Promix, Amersham) for 4 to 6 hours. After 2.5 hours, at room temperature, the Sepharose beads were washed 3 times with binding buffer at 40C, once with phosphate buffered saline and boiled in 30 SDS-PAGE buffer. Labeled proteins that were bound to the Ig-fusion proteins were analyzed by SDS-PAGE under reducing conditions. Radiolabeled proteins were detected using a phosphorimager analyzer. In all these analyses, -44- WO 00/18212 PCT/US99/22668 radiolabeled PDGF-C failed to show any interaction with any of the VEGF receptors. Next, full length and truncated PDGF-C were tested for their capacity to bind to human PDGF receptors alpha and 5 beta by analyzing their abilities to compete with PDGF-BB for binding to PDGF receptors. The binding experiments were performed on porcine aortic endothelial-1 (PAE-1) cells stably expressing the human PDGF receptors alpha and beta (Eriksson et al., EMBO J, 1992, 11, 543-550). Binding 10 experiments were performed essentially as in Heldin et al. (EMBO J, 1988, 7 1387-1393). Different concentrations of human full-length and truncated PDGF-C, or human PDGF-BB were mixed with 5 ng/ml of 1 25 1-PDGF-BB in binding buffer (PBS containing 1 mg/ml of bovine serum albumin). Aliquots 15 were incubated with the receptor expressing PAE-1 cells plated in 24-well culture dishes on ice for 90 minutes. After three washes with binding buffer, cell-bound 151I PDGF-BB was extracted by lysis of cells in 20 mM Tris-HCl, pH 7.5, 10% glycerol, 1% Triton X-100. The amount of cell 20 bound radioactivity was determined in a gamma-counter. A standard curve for the binding of 1 25 1-labeled PDGF BB homodimers to PAE-1 cells expressing PDGF alpha-receptor is shown in Figure 18. An increasing excess of the unlabeled protein added to the incubations competed efficiently with 25 cell association of the radiolabeled tracer. Figure 19 graphically shows that the truncated PDGF-C efficiently competed for binding to the PDGF alpha receptor, while the full length protein did not. Both the full length and truncated proteins failed to compete for 30 binding to the PDGF beta-receptor. Example 5: PDGF alpha-receptor Phosphorylation To test if PDGF-C causes increased phosphorylation of -45- WO 00/18212 PCT/US99/22668 the PDGF alpha-receptor, full length and truncated PDGF-C were tested for their capacity to bind to the PDGF alpha receptor and stimulate increased phosphorylation. Serum starved porcine aortic endothelial (PAE) cells stably 5 expressing the human PDGF alpha-receptor were incubated on ice for 90 minutes with PBS supplemented with 1 mg/ml BSA and long/ml of PDGF-AA, 100ng/ml of full length human PDGF CC homodimers (flPDGF-CC), 100ng/ml of truncated PDGF-CC homodimers (cPDGF-CC), or a mixture of lOng/ml of PDGF-AA 10 and 100ng/ml of truncated PDGF-CC. Full length and truncated PDGF-CC homodimers were produced as described above. Sixty minutes after the addition of the polypeptides, the cells were lysed in lysis buffer (20 mM tris-HCl, pH 7.5, 0.5% Triton X-100, 0.5% deoxycholic acid, 15 10 mM EDTA, 1 mM orthovanadate, 1 mM PMSF 1% Trasylol). The PDGF alpha-receptors were immunoprecipitated from cleared lysates with rabbit antisera against the human PDGF alpha-receptor (Eriksson et al., EMBO J, 1992 11 543-550). The precipitated receptors were applied to a SDS-PAGE gel. 20 After SDS gel electrophoresis, the precipitated receptors were transferred to nitrocellulose filters, and the filters were probed with anti-phosphotyrosine antibody PY-20, (Transduction Laboratories). The filters were then incubated with horseradish peroxidase-conjugated anti-mouse 25 antibodies. Bound antibodies were detected using enhanced chemiluminescence (ECL, Amersham Inc). The filters were then stripped and reprobed with the PDGF alpha-receptor rabbit antisera, and the amount of receptors was determined by incubation with horseradish peroxidase-conjugated anti 30 rabbit antibodies. Bound antibodies were detected using enhanced chemiluminescence (ECL, Amersham Inc). The probing of the filters with PDGF alpha-receptor antibodies confirmed that equal amounts of the receptor were present -46- WO 00/18212 PCT/US99/22668 in all lanes. PDGF-AA is included in the experiment as a control. Figure 20 shows that truncated, but not full length PDGF-CC, efficiently induced PDGF alpha-receptor tyrosine phosphorylation. This indicates that truncated 5 PDGF-CC is a potent PDGF alpha-receptor agonist. Example 6: Mitogenicity of PDGF-C for Fibroblasts Figure 21 shows the mitogenic activities of truncated and full length PDGF-CC on fibroblasts. The assay was 10 performed essentially as described in Mori et al., J. Biol. Chem., 1991 266 21158-21164. Serum starved human foreskin fibroblasts were incubated for 24 hours with 1 ml of serum free medium supplemented with 1 mg/ml BSA and 3ng/ml, 10ng/ml or 30ng/ml of full length PDGF-CC (flPDGF-CC), 15 truncated PDGF-CC (cPDGF-CC) or PDGF-AA in the presence of 0.2 pmCi [3H]thymidine. After trichloroacetic acid (TCA) precipitation, the incorporation of [3H]thymidine into DNA was determined using a beta-counter. The results show that truncated PDGF-CC, but not full length PDGF-CC, is a potent 20 mitogen for fibroblasts. PDGF-AA is included in the experiment as a control. PDGF-C does not bind to any of the known VEGF receptors. PDGF-C is the only VEGF family member, thus 25 far, which can bind to and increase phosphorylation of the PDGF alpha-receptor. PDGF-C is also the only VEGF family member, thus far, to be a potent mitogen of fibroblasts. These characteristics indicate that the truncated form of PDGF-C may not be a VEGF family member, but instead a novel 30 PDGF. Furthermore, the full length protein is likely to be a latent growth factor that needs to be activated by proteolytic processing to release the active PDGF/VEGF homology domain. A putative proteolytic site is the -47- WO 00/18212 PCTIUS99/22668 dibasic motif found in residues 231-234 in the full length protein, residues -R-K-S-R-. This site is structurally conserved in a comparison between mouse and human PDGF-Cs (Figure 7). Preferred proteases include, but are not 5 limited to, Factor X and enterokinase. The N-terminal CUB domain may be expressed as an inhibitory domain which might be used to localize this latent growth factor in some extracellular compartment (for example the extracellular matrix) and which is removed by limited proteolysis when 10 need, for example during embryonic development, tissue regeneration, tissue remodelling including bone remodelling, active angiogenesis, tumor progression, tumor invasion, metastasis formation and/or wound healing. 15 Example 7: PDGF Receptors Binding of Truncated PDGF-C To assess the interactions between truncated PDGF-C and the PDGF alpha and beta receptors, truncated PDGF-C was tested for its capacity to bind to porcine aortic endothelial-1 (PAE-1) cells expressing PDGF alpha or beta 20 receptors, respectively (Eriksson et al., EMBO J, 1992, 11 543-550). The binding experiments were performed essentially as described in Heldin et al. (EMBO J, 1988, 7 1387-1393). Five micrograms of truncated PDGF-C protein in ten microliters of sodium borate buffer was radiolabeled 25 using the Bolton-Hunter reagent (Amersham) to a specific activity of 4 x 105 cpm/ng. Different concentrations of radiolabeled truncated PDGF-C, with or without added unlabeled protein, in binding buffer (PBS containing 1 mg/ml of bovine serum albumin) was added to the receptor 30 expressing PAE-1 cells plated in 24-well culture dishes on ice for 90 minutes. After three washes with binding buffer, cell-bound 1 25 1-labeled PDGF-C was extracted by lysis of cells in 20 mM Tris-HCl, pH 7.5, 10% glycerol, 1% -48- WO 00/18212 PCTIUS99/22668 Triton X-100. The amount of cell-bound radioactivity was determined in a gamma-counter. Non-specific binding was estimated by including a 100-fold molar excess of truncated PDGF-C in some experiments. All binding data represents 5 the mean of triplicate analyses and the experimental variation in the experiment varied between 10-15%. As seen in Figure 22, truncated PDGF-C binds to cells expressing PDGF alpha receptors, but not to beta receptor expressing cells. The binding was specific as radiolabeled PDGF-C was 10 quantitatively displaced by a 100-fold molar excess of unlabeled protein. Example 8: Protease Effects on Full length PDGF-C To demonstrate that full length PDGF-C can be 15 activated by limited proteolysis to release the PDGF/VEGF homology domain from the CUB domain, the full length protein was digested with different proteases. For example, full length PDGF-C was digested with plasmin in 20 mM Tris-HCl (pH 7.5) containing 1 mM CaCl 2 , 1 mM MgCl 2 and 20 0.01% Tween 20 for 1.5 to 4.5 hours at 370C using two to three units of plasmin (Sigma) per ml. The released domain essentially corresponded in size to the truncated PDGF-C species previously produced in insect cells. Plasmin-digested PDGF-C and undigested full length PDGF-C 25 were applied to a SDS-PAGE gel under reducing conditions. After SDS-PAGE gel electrophoresis, the respective proteins were transferred to a nitrocellulose filter, and the filter was probed using a rabbit antipeptide antiserum to residues 230-250 in full length protein (residues GRKSRVVDLNLLTEEVR 30 LYSC (SEQ ID NO:37) located in just N-terminal to the PDGF/VEGF homology domain). Bound antibodies were detected using enhanced chemiluminescence (ECL, Amersham Inc). Figure 23 shows the immunoblot with a 55 kDa undigested -49- WO 00/18212 PCT/US99/22668 full length protein and the plasmin-generated 26-28 kDa species. 5 Example 9: PDGF alpha Receptors Binding of Plasmin-digested PDGF-C To assess the interactions between plasmin-digested PDGF-C and the PDGF alpha receptors, plasmin-digested PDGF C was tested for its capacity to bind to porcine aortic 10 endothelial-1 (PAE-1) cells expressing PDGF alpha receptors (Eriksson et al., EMBO J, 1992, 11 543-550). The receptor binding analyses were performed essentially as in Example 7 using 30 ng/ml of ''I-labeled truncated PDGF-C as the tracer. As seen in Figure 24, increasing concentrations of 15 plasmin-digested PDGF-C efficiently competed for binding to the PDGF alpha receptors. In contrast, undigested full length PDGF-C failed to compete for receptor binding. These data indicate that full length PDGF-C is a latent growth factor unable to interact with PDGF alpha receptors 20 and that limited proteolysis, which releases the C-terminal PDGF/VEGF homology domain, is necessary to generate an active PDGF alpha receptor ligand/agonist. Example 10: Cloning and expression of the Human PDGF-C CUB 25 domain A human PDGF-C 430 bp cDNA fragment encoding the CUB domain (amino acid residues 23 - 159 in full length PDGF-C) was amplified by PCR using Deep Vent DNA polymerase (Biolabs) using standard conditions and procedures. The 30 forward primer used was 5'cggqatcccgaatccaacctgagtag3' (SEQ ID NO:38). This primer includes a BamHI site (underlined) for in clone frame cloning. The reverse primer used was 5'ccggaattcctaatggtgatggtgatgatgtttgtcatcgtcgtcgacaatgttgta -50- WO 00/18212 PCT/US99/22668 gtg3' (SEQ ID NO:39). This primer includes an EcoRI site (underlined) and sequences coding for a C-terminal 6x His tag preceded by an enterokinase site. The amplified PCR fragment was subsequently cloned into a pACgp67A transfer 5 vector. Verification of the correct sequence of the expression construct, CUB-pACgp67A, was by automatic nucleotide sequencing. The expression vectors were then co-transfected with BaculoGold linearized baculovirus DNA into Sf9 insect cells according to the manufacture's 10 protocol (Pharmingen). Recombined baculovirus were amplified several times before beginning large scale protein production and protein purification according to the manual (Pharmingen). Sf9 cells, adapted to serum free medium, were infected 15 with recombinant baculovirus at a multiplicity of infection of about 7. Media containing the recombinant proteins were harvested 72 hours after infection and were incubated with Ni-NTA-Agarose beads(Qiagen) overnight at 4'C. The beads were collected in a column and after extensive washing with 20 50 mM sodium phosphate buffer pH 8, containing 300 mM NaCl (the washing buffer), the bound proteins were eluted with increasing concentrations of imidazole (from 100 mM to 400 mM) in the washing buffer. The eluted proteins were analyzed by SDS-PAGE using a polyacrylamide gel under 25 reducing and non-reducing conditions. Figure 25 shows the results from Coomassie blue staining of the gel. The human PDGF-C CUB domain is a disulfide-linked homodimer with a molecular weight of about 55 KD under non-reducing conditions, while two monomers of 30 about 25 and 30 KD respectively are present under reducing conditions. The heterogeneity is probably due to heterogenous glycosylation of the two putative N-linked glycosylation sites present in the CUB domain at amino acid -51- WO 00/18212 PCT/US99/22668 positions 25 and 55. A protein marker lane is shown to the left in the figure. 5 Example 11: Localization of PDGF-C transcripts in developing mouse embryos To gain insight into the biological function of PDGF C, PDGF-C expression in mouse embryos was localized by non radioactive in situ hybridization in tissue sections from 10 the head (Figures 26A-26S) and urogenital tract (Figures 26T-26V) regions. The non-radioactive in situ hybridization employed protocols and PDGF-A and PDGFR-alpha probes are described in Bostr6m et al., Cell, 1996 85 863 873, which is hereby incorporated by reference. The PDGF-C 15 probe was derived from a mouse PDGF-C cDNA. The hybridization patterns shown in Figures 26A-26V are for embryos aged E16.5, but analogous patterns are seen at E14.5, E15.5 and E17.5. Sense probes were used as controls and gave no consistent pattern of hybridization to the 20 sections. Figure 26A shows the frontal section through the mouth cavity at the.level of the tooth anlagen (t). The arrows point to sites of PDGF-C expression in the oral ectoderm. Also shown is the tongue (to). Figures 26B-26D show PDGF-C 25 expression in epithelial cells of the developing tooth canal. Individual cells are strongly labeled in this area (arrow in Figure 26D), as well as in the developing palate ectoderm (right arrow in Figure 26C). Figure 26E shows the frontal section through the eye, where PDGF-C expression is 30 seen in the hair follicles (double arrow) and in the developing eyelid. Also shown is the retina (r). In Figures 26F and 26G, the PDGF-C expression is found in the outer root sheath of the developing hair follicle -52- WO 00/18212 PCT/US99/22668 epithelium. In Figure 26H, PDGF-C expression is shown in the developing eyelid. There is an occurrence of individual strongly PDGF-C positive cells in the developing opening. Also shown is the lens (1). In Figure 261, PDGF 5 C expression in the developing lacrimal gland is shown by the arrow. In Figure 26J, PDGF-C expression in the developing external ear is shown. Expression is seen in the external auditory meatus (left arrow) and in the epidermal cleft separating the prospective auricle (e). 10 Figures 26K and 26L show PDGF-C expression in the cochlea. Expression is seen in the semi-circular canals (arrows in 26K). There is a polarized distribution of PDGF-C mRNA in epithelial cells adjacent to the developing hair cells (arrow in 26L). Figures 26M and 26N show PDGF-C expression 15 in the oral cavity. Horizontal sections show expression in buccal epithelium (arrows in 26M) and in the forming cleft between the lower lip buccal and the gingival epithelium (arrows in 26N). Also shown is the tooth anlagen (t) and the tongue (to). Figures 260 and 26P show PDGF-C 20 expression in the developing nostrils, shown on horizontal sections. PDGF-C expression appears strongest before stratification of the epithelium and the formation of the canal proper (arrows in 260 and 26P). Also shown is the developing nostrils (n). Figures 26Q-26S show PDGF-C 25 expression in developing salivary glands and ducts. Figure 26Q is the sublingual gland. Figures 26R and 26S show the maxillary glands, the salivary gland (sg) and the salivary duct (sd). Figures 26T-26V show the expression of PDGF-C in the urogenital tract. Figure 26T shows the expression 30 of PDGF-C in the developing kidney metanephric mesoderm. Figure 26U shows the expression of PDGF-C in the urethra (ua) and in epithelium surrounding the developing penis. -53- WO 00/18212 PCT/US99/22668 Figure 26V shows the PDGF-C expression in the developing ureter (u). Example 12: PDGF-C, PDGF-A and PDGFR-alpha Expression in 5 the Developing Kidney One of the strongest sites of PDGF-C expression is the developing kidney and so expression of PDGF-C, PDGF-A and PDGFR-alpha was looked at in the developing kidney. Figures 27A-27F show the results of non-radioactive in situ 10 hybridization demonstrating the expression (blue staining in unstained background visualized using DIC optics) of mRNA for PDGF-C (Figures 27A and 27B), PDGF-A (Figures 27C and 27D) and PDGFR-alpha (Figures 27E and 27F) in E16.5 kidneys. The white hatched line in Figures 27B, 27D and 15 27F outlines the cortex border. The bar in Figures 27A, 27C and 27E represents 250 pm, and in Figures 27B, 27D and 27F represents 50 pm. PDGF-C expression is seen in the metanephric mesenchyme (mm in Figure 27A), and appears to be 20 upregulated in the condensed mesenchyme (arrows in Figure 27B) undergoing epithelial conversion as a prelude to tubular development, which is situated on each side of the ureter bud (ub). PDGF-C expression remains at lower levels in the early nephronal epithelial aggregates (arrowheads in 25 B), but is absent from mature glomeruli (gl) and tubular structures. PDGF-A expression is not seen in these early aggregates, but is strong in later stages of tubular development (Figures 24C and 24D). PDGF-A is expressed in 30 early nephronal epithelial aggregates (arrowheads in Figure 27D), but once the nephron is developed further, PDGF-A expression becomes restricted to the developing Henle's loop (arrow in Figure 27D). The strongest expression is -54- WO 00/18212 PCTIUS99/22668 seen in the Henle's loops in the developing marrow (arrows in Figure 27C). The branching ureter (u) and the ureter bud (ub) is negative for PDGF-A. Thus, the PDGF-C and PDGF-A expression patterns in the 5 developing nephron are spatially and temporally distinct. PDGF-C is expressed in the earliest stages (mesenchymal aggregates) and PDGF-A in the latest stages (Henle's loop formation) of nephron development. PDGFR-alpha is expressed throughout the mesenchyme of 0 the developing kidney (Figures 27E and 27F) and may hence be targeted by both PDGF-C and PDGF-A. PDGF-B expression is also seen in the developing kidney, but occurs only in vascular endothelial cells. PDGFR-beta expression takes place in perivascular mesenchyme, and its activation by 5 PDGF-B is critical for mesangial cell recruitment into glomeruli. These results demonstrate that PDGF-C expression occurs in close spatial relationship to sites of PDGFR alpha expression, and are distinct from the expression 0 sites of PDGF-A or PDGF-B. This indicates that PDGF-C may act through PDGFR-alpha in vivo, and may have functions that are not shared with PDGF-A and PDGF-B. Since the unique expression pattern of PDGF-C in the developing kidney indicates a function as a PDGFR-alpha 5 agonist separate from that of PDGF-A or -B, a comparison was made to the histology of embryonic day 16.5 kidneys from PDGFR-alpha knockout mice (Figures 28B and 28F) with kidneys from wildtype (Figures 28A and 28C), PDGF-A knockout (Figure 28D) and PDGF-A/PDGF-B double knockout 0 (Figure 28E) mice. The bar in Figures 28A and 28B represents 250 mm, and in Figures 28C-28F represents 50 pm. Heterozygote mutants of PDGF-A, PDGF-B and PDGFR-alpha (Bostrom et al., Cell, 1996 85 863-873; Lev6en et al., -55- WO 00/18212 PCT/US99/22668 Genes Dev., 1994 8 1875-1887; Soriano et al., Development, 1997 124 2691-70) were bred as C57B16/129sv hybrids and intercrossed to produce homozygous mutant embryos. PDGF A/PDGF-B heterozygote mutants were crossed to generate 5 double PDGF-A/PDGF-B knockout embryos. Due to a high degree of lethality of PDGF-A -/- embryos before E10 (Bostrbm et al., Cell, 1996 85 863-873), the proportion of double knockout E16.5 embryos obtained in such crosses were less than 1/40. The histology of kidney phenotypes was 10 verified on at least two embryos of each genotype, except the PDGF-A/PDGF-B double knockout for which only a single embryo was obtained. It is interesting that there is lack of interstitial mesenchyme in the cortex of PDGFR-alpha -/- kidney (arrows 15 in Figure 28A and asterisk in Figure 28F) and the presence of interstitial mesenchyme in all other genotypes (asterisks in Figure 28C-E). The branching ureter (u) and the metanephric mesenchyme (mm) and its epithelial derivatives appear normal in all mutants. The abnormal 20 glomerulus in the PDGF-A/PDGF-B double knockout reflect failure of mesangial cell recruitment into the glomerular tuft due to the absence of PDGF-B. These results indicate that PDGFR-alpha knockouts have a kidney phenotype, which is not seen in PDGF-A or PDGF 25 A/PDGF-B knockouts, hence potentially reflecting loss of signaling by PDGF-C. The phenotype consists of the marked loss of interstitial mesenchyme in the developing kidney cortex. The cells lost in PDGFR-alpha -/- kidneys are thus normally PDGFR-alpha positive cells adjacent to the site of 30 expression of PDGF-C. BIOASSAYS TO DETERMINE THE FUNCTION OF PDGF-C Assays are conducted to evaluate whether PDGF-C has -56- WO 00/18212 PCT/US99/22668 similar activities to PDGF-A, PDGF-B, VEGF, VEGF-B, VEGF-C and/or VEGF-D in relation to growth and/or motility of connective tissue cells, fibroblasts, myofibroblasts and glial cells; to endothelial cell function; to angiogenesis; 5 and to wound healing. Further assays may also be performed, depending on the results of receptor binding distribution studies. 10 I. Mitogenicity of PDGF-C for Endothelial Cells To test the mitogenic capacity of PDGF-C for endothelial cells, the PDGF-C polypeptide is introduced into cell culture medium containing 5% serum and applied to bovine aortic endothelial cells (BAEs) propagated in medium 15 containing 10% serum. The BAEs are previously seeded in 24-well dishes at a density of 10,000 cells per well the day before addition of the PDGF-C. Three days after addition of this polypeptide the cells were dissociated with trypsin and counted. Purified VEGF is included in the 20 experiment as positive control. II. Assays of Endothelial Cell Function a) Endothelial cell proliferation Endothelial cell growth assays are performed by 25 methods well known in the art, e.g. those of Ferrara & Henzel, Nature, 1989 380 439-443, Gospodarowicz et al., Proc. Natl. Acad. Sci. USA, 1989 86 7311-7315, and/or Claffey et al., Biochem. Biophys. Acta, 1995 1246 1-9. 30 b) Cell adhesion assay The effect of PDGF-C on adhesion of polymorphonuclear granulocytes to endothelial cells is tested. -57- WO 00/18212 PCT/US99/22668 c) Chemotaxis The standard Boyden chamber chemotaxis assay is used to test the effect of PDGF-C on chemotaxis. 5 d) Plasminogen activator assay Endothelial cells are tested for the effect of PDGF-C on plasminogen activator and plasminogen activator inhibitor production, using the method of Pepper et al., Biochem. Biophys. Res. Commun., 1991 181 902-906. 10 e) Endothelial cell Migration assay The ability of PDGF-C to stimulate endothelial cells to migrate and form tubes is assayed as described in Montesano et al., Proc. Natl. Acad. Sci. USA, 1986 83 7297 15 7301. Alternatively, the three-dimensional collagen gel assay described in Joukov et al., EMBO J., 1996 15 290-298 or a gelatinized membrane in a modified Boyden chamber (Glaser et al., Nature, 1980 288 483-484) may be used. 20 III. Angiogenesis Assay The ability of PDGF-C to induce an angiogenic response in chick chorioallantoic membrane is tested as described in Leung et al., Science, 1989 246 1306-1309. Alternatively the rat cornea assay of Rastinejad et al., Cell, 1989 56 25 345-355 may be used; this is an accepted method for assay of in vivo angiogenesis, and the results are readily transferrable to other in vivo systems. 30 IV. Wound Healing The ability of PDGF-C to stimulate wound healing is tested in the most clinically relevant model available, as -58- WO 00/18212 PCT/US99/22668 described in Schilling et al., Surgery, 1959 46 702-710 and utilized by Hunt et al., Surgery, 1967 114 302-307. V. The Haemopoietic System 5 A variety of in vitro and in vivo assays using specific cell populations of the haemopoietic system are known in the art, and are outlined below. In particular a variety of in vitro murine stem cell assays using fluorescence-activated cell sorter to purified cells are 10 particularly convenient: a) Repopulating Stem Cells These are cells capable of repopulating the bone marrow of lethally irradiated mice, and have the Lin~, Rh, 15 Ly-6A/E+, c-kit+ phenotype. PDGF-C is tested on these cells either alone, or by co-incubation with other factors, followed by measurement of cellular proliferation by 3

H

thymidine incorporation. 20 b) Late Stage Stem Cells These are cells that have comparatively little bone marrow repopulating ability, but can generate D13 CFU-S. These cells have the Lin-, Rh h, Ly-6A/E+, c-kit+ phenotype. PDGF-C is incubated with these cells for a period of time, 25 injected into lethally irradiated recipients, and the number of D13 spleen colonies enumerated. c) Progenitor-Enriched Cells These are cells that respond in vitro to single growth 30 factors and have the Lin-, Rh h, Ly-6A/E+, c-kit+ phenotype. This assay will show if PDGF-C can act directly on haemopoietic progenitor cells. PDGF-C is incubated with -59- WO 00/18212 PCT/US99/22668 these cells in agar cultures, and the number of colonies present after 7-14 days is counted. VI. Atherosclerosis 5 Smooth muscle cells play a crucial role in the development or initiation of atherosclerosis, requiring a change of their phenotype from a contractile to a synthetic state. Macrophages, endothelial cells, T lymphocytes and platelets all play a role in the development of 10 atherosclerotic plaques by influencing the growth and phenotypic modulations of smooth muscle cell. An in vitro assay using a modified Rose chamber in which different cell types are seeded on to opposite cover slips measures the proliferative rate and phenotypic modulations of smooth 15 muscle cells in a multicellular environment, and is used to assess the effect of PDGF-C on smooth muscle cells. VII. Metastasis The ability of PDGF-C to inhibit metastasis is assayed 20 using the Lewis lung carcinoma model, for example using the method of Cao et al., J. Exp. Med., 1995 182 2069-2077. VIII. Migration of Smooth Muscle Cells The effects of the PDGF-C on the migration of smooth 25 muscle cells and other cells types can be assayed using the method of Koyama et al., J. Biol. Chem., 1992 267 22806 22812. IX. Chemotaxis 30 The effects of the PDGF-C on chemotaxis of fibroblast, monocytes, granulocytes and other cells can be assayed -60- WO 00/18212 PCTIUS99/22668 using the method of Siegbahn et al., J. Clin. Invest., 1990 85 916-920. X. PDGF-C in Other Cell Types 5 The effects of PDGF-C on proliferation, differentiation and function of other cell types, such as liver cells, cardiac muscle and other cells, endocrine cells and osteoblasts can readily be assayed by methods known in the art, such as 3 H-thymidine uptake by in vitro 10 cultures. Expression of PDGF-C in these and other tissues can be measured by techniques such as Northern blotting and hybridization or by in situ hybridization. 15 XI. Construction of PDGF-C Variants and Analogues PDGF-C is a member of the PDGF family of growth factors which exhibits a high degree of homology to the other members of the PDGF family. PDGF-C contains eight conserved cysteine residues which are characteristic of 20 this family of growth factors. These conserved cysteine residues form intra-chain disulfide bonds which produce the cysteine knot structure, and inter-chain disulfide bonds that form the protein dimers which are characteristic of members of the PDGF family of growth factors. PDGF-C 25 interacts with a protein tyrosine kinase growth factor receptor. In contrast to proteins where little or nothing is known about the protein structure and active sites needed for receptor binding and consequent activity, the design of 30 active mutants of PDGF-C is greatly facilitated by the fact that a great deal is known about the active sites and important amino acids of the members of the PDGF family of growth factors. -61- WO 00/18212 PCT/US99/22668 Published articles elucidating the structure/activity relationships of members of the PDGF family of growth factors include for PDGF: Oestman et al., J. Biol. Chem., 1991 266 10073-10077; Andersson et al., J. Biol. Chem., 5 1992 267 11260-1266; Oefner et al., EMBO J., 1992 11 3921 3926; Flemming et al., Molecular and Cell Biol., 1993 13 4066-4076 and Andersson et al., Growth Factors, 1995 12 159-164; and for VEGF: Kim et al., Growth Factors, 1992 7 53-64; Pbtgens et al., J. Biol. Chem., 1994 269 32879-32885 10 and Claffey et al., Biochem. Biophys. Acta, 1995 1246 1-9. From these publications it is apparent that because of the eight conserved cysteine residues, the members of the PDGF family of growth factors exhibit a characteristic knotted folding structure and dimerization, which result in 15 formation of three exposed loop regions at each end of the dimerized molecule, at which the active receptor binding sites can be expected to be located. Based on this information, a person skilled in the biotechnology arts can design PDGF-C mutants with a very 20 high probability of retaining PDGF-C activity by conserving the eight cysteine residues responsible for the knotted folding arrangement and for dimerization, and also by conserving, or making only conservative amino acid substitutions in the likely receptor sequences in the 25 loop 1, loop 2 and loop 3 region of the protein structure. The formation of desired mutations at specifically targeted sites in a protein structure is considered to be a standard technique in the arsenal of the protein chemist (Kunkel et al., Methods in Enzymol., 1987 154 367-382). 30 Examples of such site-directed mutagenesis with VEGF can be found in Potgens et al., J. Biol. Chem., 1994 269 32879 32885 and Claffey et al., Biochem. Biophys. Acta, 1995 1246 -62- WO 00/18212 PCT/US99/22668 1-9. Indeed, site-directed mutagenesis is so common that kits are commercially available to facilitate such procedures (e.g. Promega 1994-1995 Catalog., Pages 142 145). 5 The connective tissue cell, fibroblast, myofibroblast and glial cell growth and/or motility activity, the endothelial cell proliferation activity, the angiogenesis activity and/or the wound healing activity of PDGF-C mutants can be readily confirmed by well established 10 screening procedures. For example, a procedure analogous to the endothelial cell mitotic assay described by Claffey et al., (Biochem. Biophys. Acta., 1995 1246 1-9) can be used. Similarly the effects of PDGF-C on proliferation of other cell types, on cellular differentiation and on human 15 metastasis can be tested using methods which are well known in the art. 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 20 disclosed embodiments incorporating the spirit and substance of the invention may occur to persons skilled in the art, the invention should be construed broadly to include all variations falling within the scope of the appended claims and equivalents thereof. -63-

Claims (74)

1. An isolated nucleic acid molecule comprising a polynucleotide sequence having at least 85% identity with the sequence of Figure 1, 3 or 5 (SEQ ID NOs: 2, 4 and 6, respectively).

2. An isolated nucleic acid molecule according to claim 1, wherein the sequence identity is at least 90%.

3. An isolated nucleic acid molecule according to claim 1, wherein the sequence identity is at least 95%.

4. An isolated nucleic acid molecule according to claim 1, wherein said nucleic acid is a cDNA.

5. An isolated nucleic acid molecule according to claim 1, wherein said nucleic acid is a mammalian polynucleotide.

6. An isolated nucleic acid molecule according to claim 5, wherein said nucleic acid is a murine polynucleotide.

7. An isolated nucleic acid molecule according to claim 6, comprising the sequence shown in Figure 5 (SEQ ID NO:6).

8. An isolated nucleic acid molecule according to claim 5, wherein said nucleic acid is a human polynucleotide. -64- WO 00/18212 PCT/US99/22668

9. An isolated nucleic acid molecule according to claim 8, comprising the sequence shown in Figure 1 or Figure 3 (SEQ ID NOs: 2 and 4, respectively).

10. An isolated nucleic acid molecule which encodes a polypeptide molecule comprising an amino acid sequence having at least 85% identity with the amino acid sequence of Figure 2 (SEQ ID NO:3) or the amino acid sequence of Figure 4 (SEQ ID NO:5), or a fragment or analog thereof having the biological activity of PDGF-C.

11. An isolated nucleic acid molecule according to claim 10, wherein the amino acid sequence identity is at least 90%.

12. An isolated nucleic acid molecule according to claim 10, wherein the amino acid sequence identity is at least 95%.

13. An isolated nucleic acid molecule which codes for a polypeptide which comprises the amino acid sequence PXCXXVXRCGGXXXCC (SEQ ID NO:1).

14. A vector comprising a nucleic acid according to claim 1, which nucleic acid is operably linked with a promoter sequence.

15. A vector according to claim 14, wherein said vector is a eukaryotic vector.

16. A vector according to claim 14, wherein said vector is a prokaryotic vector. -65- WO 00/18212 PCT/US99/22668

17. A vector according to claim 14, wherein said vector is a plasmid.

18. A vector according to claim 14, wherein said vector is a baculovirus vector.

19. A method of making a vector which expresses a polypeptide comprising an amino acid sequence having at least 85% identity with the amino acid sequence of Figure 2 (SEQ ID NO:3) or the amino acid sequence of Figure 6 (SEQ ID NO:7), or fragment or analog thereof having the biological activity of PDGF-C, said method comprising incorporating an isolated nucleic acid according to claim 1, claim 10 or claim 13, into said vector in operatively linked relation with a promoter.

20. A host cell transformed or transfected with a vector according to claim 14.

21. A host cell according to claim 20, wherein said host cell is a eukaryotic cell.

22. A host cell according to claim 20, wherein said host cell is a COS cell.

23. A host cell according to claim 20, wherein said host cell is a prokaryotic cell.

24. A host cell according to claim 20, wherein said host cell is a 293EBNA cell.

25. A host cell according to claim 20, wherein said host cell is an insect cell. -66- WO 00/18212 PCT/US99/22668

26. A host cell transformed or transfected with a vector comprising a nucleic acid sequence according to claim 1, operatively linked to a promoter, such that said host cell expresses a polypeptide comprising an amino acid sequence having at least 85% identity with the amino acid sequence of Figure 2 (SEQ ID NO:3) or the amino acid sequence of Figure 6 (SEQ ID NO:7), or a fragment or analog thereof having the biological activity of PDGF-C.

27. An isolated polypeptide comprising at least 85% identity with the amino acid sequence of Figure 2 (SEQ ID NO:3) or the amino acid sequence of Figure 6 (SEQ ID NO:7), or a fragment or analog thereof having the biological activity of PDGF-C.

28. An isolated polypeptide according to claim 27, wherein said polypeptide is a murine polypeptide.

29. An isolated polypeptide according to claim 27, wherein said polypeptide is a human polypeptide.

30. An isolated polypeptide according to claim 27, wherein said polypeptide has the ability to stimulate and/or enhance proliferation and/or differentiation and/or growth and/or motility of cells expressing a PDGF-C receptor.

31. An isolated polypeptide produced by expression of a polynucleotide comprising the polynucleotide sequence having at least 85% identity with the Figure 1, 3 or 5 (SEQ ID NOs: 2, 4 or 6, respectively), or a polynucleotide which hybridizes under stringent conditions with at least one of the foregoing DNA sequences. -67- WO 00/18212 PCT/US99/22668

32. An isolated polypeptide which comprises the characteristic sequence PXCXXVXRCGGXXXRCC (SEQ ID NO:1).

33. An isolated polypeptide dimer comprising a polypeptide according to claim 27.

34. An isolated polypeptide dimer according to claim 33, wherein said polypeptide dimer is a homodimer of said polypeptide.

35. An isolated polypeptide dimer according to claim 33, wherein said polypeptide dimer is a heterodimer of said polypeptide and VEGF, VEGF-B, VEGF-C, VEGF-D, PDGF-A, PDGF B or PlGF.

36. An isolated polypeptide dimer according to claim 33, wherein said polypeptide dimer is a disulfide-linked dimer.

37. A pharmaceutical composition comprising an effective cell proliferation promoting amount of a polypeptide according to claim 27, claim 31 or claim 32, and at least one further growth factor selected from the group consisting of VEGF, VEGF-B, VEGF-C, VEGF-D, PDGF-A, PDGF-B or PlGF.

38. A pharmaceutical composition according to claim 37, further comprising heparin.

39. A pharmaceutical composition comprising an effective cell proliferation promoting amount of an isolated polypeptide according to claim 27, claim 31 or -68- WO 00/18212 PCT/US99/22668 claim 32, and at least one pharmaceutical carrier or diluent.

40. A pharmaceutical composition comprising a PDGF receptor stimulating amount of an isolated polypeptide according to claim 27, claim 31 or claim 32, and at least one pharmaceutical carrier or diluent.

41. A pharmaceutical composition comprising an effective connective tissue or wound healing stimulating amount of an isolated polypeptide according to claim 27, claim 31 or claim 32, and at least one pharmaceutical carrier or diluent.

42. A means for amplifying a polynucleotide according to claim 1 in a test sample, said means comprising at least one pair of primers complementary to a nucleic acid according to claim 1.

43. A means for amplifying a polynucleotide as according to claim 1 in a test sample, said means comprising a polymerase and at least one pair of primers complementary to a nucleic acid according to claim 1, for amplifying the polynucleotide by polymerase chain reaction in order to facilitate a sequence comparison of the polynucleotide with the nucleic acid according to claim 1.

44. An antibody specifically reactive with a polypeptide according to claim 27, claim 31 or claim 32.

45. An antibody according to claim 44, wherein said antibody is a polyclonal antibody. -69- WO 00/18212 PCTIUS99/22668

46. An antibody according to claim 44, wherein said antibody is a monoclonal antibody or a F(ab') 2 , F(ab'), F(ab) fragment or chimeric antibody.

47. An antibody according to claim 45 or 46, wherein said antibody is labeled with a detectable label.

48. An antibody according to claim 47, wherein said detectable label is radioactive isotope.

49. A method of making a polypeptide according to claim 27, claim 31 or claim 32, said method comprising the steps of: culturing a host cell transformed or transfected with a vector comprising a polynucleotide encoding said polypeptide operably associated with a promoter sequence such that the nucleic acid sequence encoding said polypeptide is expressed; and isolating said polypeptide from said host cell or from a growth medium in which said host cell is cultured.

50. A method of stimulating growth of connective tissue or wound healing in a mammal, comprising administering to said mammal an effective amount of the polypeptide according to claim 27, claim 31 or 32.

51. A method for producing an active truncated form of PDGF-C, comprising the steps of expressing an expression vector comprising a polynucleotide encoding a polypeptide as claimed in claim 69.

52. A method for regulating receptor-binding specificity of PDGF-C, comprising the steps of expressing -70- WO 00/18212 PCT/US99/22668 an expression vector comprising a polynucleotide encoding a polypeptide as claimed in claim 27. claim 31 or claim 32 and supplying a proteolytic amount of at least one enzyme for processing the expressed polypeptide to generate the active truncated form of PDGF-C.

53. A method for selectively activating a polypeptide having a growth factor activity comprising the step expressing an expression vector comprising a polynucleotide encoding a polypeptide having a growth factor activity, a CUB domain and a proteolytic site between the polypeptide and the CUB domain, and supplying a proteolytic amount of at least one enzyme for processing the expressed polypeptide to generate the active polypeptide having a growth factor activity.

54. An isolated polypeptide according to claim 27, claim 31 or claim 32 which comprises a proteolytic site having the amino acid sequence RKSR or a structurally conserved amino acid sequence thereof.

55. An isolated nucleic acid molecule according to claim 10 which codes for a polypeptide which comprises a proteolytic site having the amino acid sequence RKSR or a structurally conserved amino acid sequence thereof.

56. An isolated heterodimer comprising an active monomer of VEGF, VEGF-B, VEGF-C, VEGF-D, PDGF-C, PDGF-A, PDGF-B or PlGF and an active monomer of PDGF-C linked to a CUB domain.

57. An isolated heterodimer comprising an active monomer of PDGF-C and an activated monomer of VEGF, VEGF-B, -71- WO 00/18212 PCT/US99/22668 VEGF-C, VEGF-D, PDGF-C, PDGF-A, PDGF-B or PlGF linked to a CUB domain.

58. An isolated heterodimer according to claim 56, further comprising a proteolytic site between the active monomer and the CUB domain linkage.

59. An isolated heterodimer according to claim 57, further comprising a proteolytic site between the active monomer and the CUB domain linkage.

60. An isolated polynucleotide, comprising a polynucleotide sequence having at least 85% identity with the sequence of Figure 1, 3 or 5 (SEQ ID NOs: 2, 4 or 6, respectively), or a polynucleotide which hybridize under stringent conditions with at least one of said DNA sequences.

61. A method of promoting fibroblast mitogenesis in a mammal, comprising the step of administering to said mammal an effective fibroblast mitogenesis promoting amount of the polypeptide according to claim 27, claim 31 or claim 32.

62. A method of inducing PDGF alpha receptor activation, comprising the step of adding a PDGF alpha receptor stimulating amount of the polypeptide according to claim 27, claim 31 or claim 32.

63. A method of inhibiting tumor growth of a tumor expressing PDGF-C in a mammal, comprising administering to said mammal a PDGF-C inhibiting amount of a PDGF-C antagonist. -72- WO 00/18212 PCT/US99/22668

64. A method of identifying specific types of human tumors, comprising the step of taking a sample of the tumor and testing for the expression of PDGF-C.

65. The method of claim 64, wherein the specific types of tumors are selected from the group selected from choriocarcinoma, Wilms tumor, megakaryoblastic leukemia, lung carcinoma and erythroleukemia.

66. A method for identifying an PDGF-C antagonist comprising: admixing a substantially purified preparation of an activated truncated form of PDGF-C with a test agent; and monitoring, by any suitable means, an inhibition in the biological activity of PDGF-C.

67. A method for identifying an PDGF-C antagonist comprising: admixing a substantially purified preparation of an full-length PDGF-C with a test agent; and monitoring, by any suitable means, an inhibition in the cleavage of the CUB domain from PDGF-C.

68. An isolated polypeptide according to claim 27, wherein the cells are selected from the group consisting of endothelial cells, connective tissue cells, myofibroblasts and glial cells.

69. A method of making a vector which expresses a polypeptide comprising an amino acid sequence having at least 85% identity with the amino acid residues 230 to 345 of Figure 2 (SEQ ID NO:3) or of Figure 6 (SEQ ID NO:7), -73- WO 00/18212 PCT/US99/22668 said method comprising incorporating an isolated nucleic acid molecule encoding said amino acid residues into said vector in operatively linked relation with a promoter.

70. An antibody according to claim 46, wherein said monoclonal antibody is a humanized antibody.

71. A method for producing an activated truncated form of PDGF-C, comprising the steps of expressing an expression vector comprising a polynucleotide encoding a polypeptide as claimed in claim 27, claim 31 or claim 32 and supplying a proteolytic amount of at least one enzyme for processing the expressed polypeptide to generate the activated truncated form of PDGF-C.

72. A method of inhibiting tissue remodeling during invasion of tumor cells into a normal population of cells, comprising administering to said mammal a PDGF-C inhibiting amount of a PDGF-C antagonist.

73. A method of treating fibrotic conditions in a mammal in need a such treatment, comprising administering to said mammal a PDGF-C inhibiting amount of a PDGF-C antagonist.

74. A method of claim 73, wherein the fibrotic conditions are found in the lung, kidney or liver. -74-