Fibroblast growth factors (FGFs) constitute a family of structurally related but genetically distinct polypeptides that share strong affinity for heparin or heparan sulfates (HS), act as intrinsic regulators of cell to cell communication within all tissues including the developing embryo, have impact on practically all known cellular functions and therefore are important in associated tissue pathologies. Among them, FGF-1 (acidic FGF) and FGF-2 (basic FGF) are widely distributed in most adult tissues and show strong angiogenic activity in vivo. Both FGF-1 and FGF-2 stimulate the proliferation of stromal cells such as fibroblasts, vascular endothelial and smooth muscle cells in vitro. In contrast, the expression of FGF-7(keratinocyte growth factor, KGF) is limited to the stromal compartment of parenchymal tissues containing a well-defined stroma and epithelium. Of all the FGF ligands, FGF-7 exhibits the strictest range of specificity for receptor isotype. It acts specifically on epithelial cells which express a specific splice variant of receptor in epithelium makes FGF-7 a directionally-specific paracrine factor underlying communication of epithelial cells with their stroma.
FGF action is mediated through binding and activation of receptors with tyrosine kinase activity. FGF receptors (FGFR) consist of four different gene products that have two or three immunoglobulin (Ig)-like loop structures in the extracellular domain and a tyrosine kinase in the intracellular domain. Regulated combinatorial alternate splicing resulting in coding sequence for cassettes of receptor subdomains generates a large degree of functional diversity of the monomeric product from one FGFR gene. The FGFR type 1 gene in general, and its splice variant FGFR1IIIc in particular, appears to be the member of the four gene family whose expression is limited to stromal cells. The IIIb splice variant of FGFR2, which arises by mutually exclusive alternate splicing of exon cassettes coding for the second half of the third Ig loop of the receptor ectodomain, appears limited to epithelial-like cells.
In addition to ligand and thyrosine kinase, our results along with others show that heparan sulfate proteoglycans (HSPG) on the cell surface or within the extracellular matrix play an integral role in the FGF signal transduction complex. In addition to their role as a reservoir or sequestering agent for FGF ligands, HS chains interact with the ectodomain of FGFR through a specific amino acid sequence in Ig loop II which requires divalent cations and potentially anchors the ectodomain in a ligand-dependent conformation for dimerization and activation of receptor kinase. Through the combined requirement for a single bivalent HS chain that will react with both an FGF ligand and FGFR ectodomain plus the length sufficient to the two together, HS chains may exhibit an element of ligand- and receptor-related specificity. In fibroblasts and endothelial cells, exogenous heparin potentiates and, in some cases, is absolutely required for the activity of FGF-1, but not FGF-2. This and our results suggest that endothelial cells express an HSPG which is specific for FGF-2 and FGFR1. In contrast, nonmalignant immortalized and tumor-derived epithelial cells appear to express HS which support the interaction of both FGF-1 and FGF-2 under the same conditions. These results raise the possibility that specific HS chains of specific classes of HSPG may complement different combinations of FGF ligands and FGFR ectodomains in assembly and activation of the FGF signal transduction complex.
Transplantable rat prostate tumors exhibit characteristics of human tumors as they progress from the slow-growing, androgen-responsive, well-differentiated state to the malignant, androgen-independent, undifferentiated state. Dunning tumors exhibit well-differentiated stromal and epithelial compartments. Cloned epithelial cell lines (DT-E) derived from the R3327PAP tumors express exclusively FGFR2IIIb and respond to stromal-derived FGF-7 (or FGF-1). The derived stromal cells (DT-S) express FGFR2IIIc and respond to FGF-2 (and FGF-1). Cells derived from malignant R3327AT3 tumors which arise from tumors similar to the R3327PAP tumor after castration of male hosts and prolonged passage in castrated hosts or females or tumors that eventually emerge from the DT-E cells in absence of stroma kill hosts after several weeks, are completely insensitive to androgen, are undifferentiated and express FGFR2IIIc and FGFR1IIIc. The AT3 cells are completely insensitive to FGF-7 and stromal cells. Increasing evidence suggests that the control of epithelial cell functions by androgen is partitioned between the stroma and epithelium. Androgenic regulation of the communication between stroma and epithelium underlies the indirect control of epithelial cell functions (growth and differentiation ) by androgen. The partitioning of the expression of FGF-7 in stromal cells and its receptor FGFR2IIIb on epithelial cells appears to be a directionally-specific paracrine communication system in which signal arises in stroma and reception lies in the epithelium. In prostate and other steroid-responsive tissues, the FGF-7 signal is under control of androgen and, although not proven to date, it is predicted that the reception end in epithelium may be regulated by androgen. Subversion of the system from multiple directions may underlie the diverse properties of prostate tumors as they progress eventually to a state of independence on androgen and stroma, to a state of growth autonomy and to the undifferentiated state. Androgen-independence may occur by loss of dependence of expression of FGF-7 on androgen without loss of stromal independence and differentiation. Exon switching from FGFR2IIIb to FGFR2IIIc and activation of FGFR1 or loss of the FGFR2 gene entirely which has been observed to occur during progression of model prostate tumors potentially confers both androgen-independence and lack of differentiation concurrent with independence on stroma. Activation of abnormally-expressed FGF ligands (e.g. FGF-2, FGF-3, FGF-5) may confer growth autonomy by the autocrine loop formed between the tumor-associated ligands and the FGFR2IIIc and FGFR1IIIc isoforms. Indeed, introduction of FGFR1 by tranfection in nonmalignant DT-E cells accelerates their progression in vivo and introduction of FGFR2 in malignant AT3 cells inhibits their growth in vitro and in vivo. Most recently, increasing evidence suggests that FGF action occurs through much more tightly regulated steps than simply binding of ligands to the ectodomain of tyrosine kinase receptors. HS chains attached to the cores of HSPG appear to be bivalent and potentially specific integral components of the FGFR signal transduction complex. The HSPG component plays a critical role in both restriction and regulated activation of the oligomeric FGFR signal transduction complex. In this integral role, HSPG are expected to play an equally important role to ligand and receptor kinase isotype in normal prostate tissues and prostate tumors.