There are nine identified members of the fibroblast growth factor (FGF) family and four high-affinity receptors, two with splice variant isoforms.1,2 The FGFs play major roles in development, wound healing, hematopoiesis, and tumorigenesis.1,3,5 The name FGF is misleading.3 While some FGFs do, indeed, initiate fibroblast proliferation, they induce proliferation of many other cells as well4, and their actions are more general than proliferation.4,5
There are eight known human FGFs; FGF-1 through -7 and -9.6 FGF-8 has been identified only in mice.7 They range from 17 kDa to 38 kDa. Over the entire coding sequence, homology among the eight human FGFs is only 14% (17% over a conserved internal stretch of approximately 112 aa).8,9 There is considerable species cross-reactivity for the human FGFs.
Most FGFs are translated with a signal sequence and most undergo post-translational glycosylation, characteristics typical of secreted proteins. FGF-1 and -2, have no signal sequence and are not glycosylated and FGF-9 has only a 3-residue cleaved sequence. Nevertheless, all FGFs have the potential to act extracellularly. Receptors for FGFs are expressed on cell surfaces, suggesting an extra-cellular function,2,10 and studies on single cells (where FGF release due to cell death could be ruled out) showed that FGF-2 was released, that it modified the releasing cell's migration, and that inhibitors of endo- and exocytosis blocked its release while endoplasmic reticulum-Golgi (ER/Golgi) inhibitors did not.11,12 COS-1 cells transfected with FGF-2 DNA were able to secrete FGF-2 by an energy-dependent process independent of the classic ER/Golgi pathway13, suggesting an alternative to the classical protein secretory pathway.
Some FGFs have considerable size heterogeneity. Natural FGF-2 usually has an 18 kDa mass, but larger forms result from amino-terminal extensions due to translation at non-AUG start sites.14,15 The larger forms are localized in the cell nucleus rather than the cytoplasm.16 FGF-3 is also translated from alternative start sites, leading to its localization in either the nucleus or cytosol17,18, and FGF-6 has two potential in frame ATG codons.19
Four distinct genes code for high affinity, type-I (N-terminus extracellular) transmembrane glycoprotein FGF receptors19a with intrinsic tyrosine kinase activity. There are alternative splice events for two receptors (FGF-R1 and -R2), and there are multiple ligands for three (FGF-R1, -R2 and -R3).2 Evidence suggests that cell-surface glycosaminoglycans regulate FGF binding to these receptor(s).
Nuclear localization follows either cytoplasmic synthesis or (apparently) cell-surface binding. FGF-1.19a,20, FGF-216,19a,21,22 and the long form of FGF-324 each potentially translocate to the nucleus following synthesis. In addition, cells exposed to exogenous FGF have been reported to internalize and translocate FGF-125 and FGF-224 to the nucleus (but see ref. 21). The mechanism appears to be independent of FGF receptor activation (but not necessarily independent of the FGF receptor).25,26 The mitogenic effect of FGF-1 is directly correlated with its appearance in the nucleus20,26, and FGF-2 has been proposed to activate nucleolar protein kinases and thus regulate ribosomal gene transcription.27
Activation of cells by FGF is believed to involve both high-affinity tyrosine kinase receptors (FGF-R) and low-affinity glycosaminoglycan (GAG) binding. GAG is a general term for a linear heteropolysaccharide with alternating amino sugars and uronic acids. The sugars often are sulfated, making them highly charged. GAG chains attached to a small protein core is called a proteoglycan.28 The most important GAGs for FGFs are heparin and heparan sulfate.
Heparin expression is limited essentially to the granules of mast cells and basophils. Heparin binds to, and is is physically protective of, FGF-1 and -2.29 By interaction with FGF and FGF-R30,31, heparin induces receptor dimerization and signal transduction.31 A recent report suggests that heparin alone can induce autophosphorylation of FGF-R4.32 Given the limited number of sources for its expression, however, a physiological role for heparin is unclear.
Heparan sulfate (HS), by contrast, is a normal component of the surface of most cells.33,34 HS facilitates FGF-FGF-R interactions as evidenced by high affinity binding of FGF-2 to heparan sulfate proteoglycan (HSPG)35 and by the observations that i) mutant cells deficient in HS do not show FGF-2-FGF-R binding36 and ii) HSGP-deficient cells fail to show FGF-R2 dimerization or tyrosine kinase activation.31
The form of HS on the cell surface is unclear. Interest focuses on the syndecans, a group of transmembrane proteoglycans that bind actin inside the cell23 and growth factors (FGF-2) plus extracellular matrix outside the cell.37-39 There are four syndecans, each containing some HS.40-42 There is variability in the spatial and temporal expression of syndecans, and the nature of a syndecan can vary with the physiological state of the tissue where it is expressed. It possible that a cell’s response to FGFs can be regulated by the transient form of each cell’s syndecan.
In addition to HS-FGF--FGF-R interactions, it is possible that the HS-FGF interaction itself is important. FGF-2 can be internalized following direct binding to cell-surface HS43 independently of any interaction with FGF-R. HSPGs are known to cycle from the cell surface to the nucleus, leading to speculation that HS, in an HS-FGF complex, may serve as a "shuttle" for FGF to the nucleus.
FGFs modulate cellular activity via four distinct high-affinity receptors with an intrinsic tyrosine kinase (Table 2).2,44 Each FGF-R has multiple immunoglobulin-like (Ig-like) loops in the extracytoplasmic region. In the cytoplasmic region there is a tyrosine kinase domain and a C-terminal extension that binds substrates (Figure 2). FGF receptors are reviewed in references 2 and 44-47.
There are multiple alternatively spliced variants for the first two FGF receptors44-48, leading to a complex descriptive nomenclature (Figure 2). With three Ig-like domains, the receptor is an alpha-type; with only the second and third loops, it is a beta-type; without a secretory signal, it is a gamma-type. The binding site for HS-FGF, which spans the second loop to the third loop, has no variant forms45 in contrast to three variants of loop III (IIIa, IIIb, and IIIc). The IIIa variant leads to an abbreviated, secreted receptor49, while the IIIb and IIIc forms differ in FGF specificity. Within the cytoplasmic segment, the presence or absence of a threonine-valine insert in the juxtamembrane region defines a or b forms. A full catalytic kinase domain is form 1, while a truncated catalytic domain is form 2.44 Thus an FGF-R1 alpha a2(IIIc) is a product of the FGF-R1 gene with a three Ig-loop extracellular region, a Thr-Val insert, a truncated tyrosine kinase domain and the "c" variant of the third Ig-loop.
There apparently is physiological significance to this complexity. Expression of an isoform is tissue specific, but a single cell can express more than one isoform. This is more significant when the a- and b-forms are considered with variants in the cytoplasmic domain. In theory, there can be 1-1, 1-2, and 2-2 dimers. 1-1 dimers are fully functional, 2-2 are silent and 1-2, are relatively inactive (if not silent).46,50 Since specificity for the various FGFs varies with isoform, since the signalling by the isoforms may differ and since the ratio of occupied receptor to functional dimer will vary, a cell’s response to a mixture of FGFs may vary with the relative amounts of different FGFs and with relative amounts of different receptor isoforms expressed on a cell. The beta-forms of receptors are associated with malignant development51, while IIIa (secreted) and 2 (kinase truncated) forms are suggested to act as FGF antagonists.44
There is roughly 50-75% homology among human FGF receptors, but homology between human and mouse FGF-R1 is 98%, undoubtedly accounting for the marked species cross-reactivity of FGFs. Soluble (or IIIa) forms of FGFR1 beta have been found in blood53, cerebral spinal fluid54 and connective tissue basement membranes.55 These soluble forms bind both FGF-1 and 2 at reduced affinities.51 Soluble forms of FGF-R2 differ, and may be the result of proteolysis rather than differences in RNA splicing.56
There are an enormous number of functions attributed to FGFs. Only a few are mentioned here.
FGFs have been implicated in two aspects of hematopoiesis: development of specific lineages of blood cells and development of bone-marrow stroma. FGF-2 stimulates myeloid progenitors. FGFs also alter stromal cells, which contribute to hematopoiesis by releasing essential growth factors.57 FGF-2 stimulates stromal growth, either directly or by inducing differentiation of a common stem cell (CD34+CD38-HLA DR-) into stromal-type cells.58,59
FGF-6 appears to play a major role in skeletal muscle development. FGF-6 and its receptor, FGF-R4, induce myoblast proliferation and suppress myocyte differentiation, suggested as a way to insure a supply of self-renewing myocytes.60,61 The branching of bronchi to form alveoli is believed to be a function of an FGF-7:FGF-R2/IIIb interaction.62 FGF-5 in the embryo is notable for its highly specific pattern of expression63, first in pre-gastrulation embryonic ectoderm and later in a small patch of mesoderm through which the hepatic bud will penetrate.63 Ectodermally-derived FGF-8 is suggested to be a principal mitogen driving the proliferation of the underlying mesenchyme, thus adding length to the limb.64-66
There are three stages in soft tissue (or skin) repair: 1) hemostasis and inflammation; 2) granulation tissue formation and re-epithelialization; and 3) remodeling. FGFs play minor roles in stages 1 and 3 but major roles in stage 2, characterized by macrophage, fibroblast, and blood vessel migration into the damaged tissue, with proliferation and migration of nearby epithelium to form new epidermis. FGF-1 and 2 induce endothelial cell proliferation67, and FGF-1 is chemotactic for both endothelial cells and fibroblasts. FGF-2 promotes the release of endothelium from its normal connective tissue anchor, thus encouraging the entry of new vascular endothelium.68 FGF-7 expression is markedly increased in fibroblasts underlying the epidermal layer.69
Tumor growth beyond a very small size requires the development of new blood vessels into the tumor.70,71 The development of new vessels, or angiogeneisis, requires specific signals. An important signal is FGF-2. The risk of metastasis has been correlated with the degree of angiogenesis and, in some cases72, with an elevation of FGF in blood or urine. In other cases, there is no apparent increase in FGF-2, raising the possibility that while a tumor requires an angiogenic factor for growth, it need not always be the same angiogenic factor.
There is a large variety within the FGF family of proteins and even a larger variety of specific high-affinity receptors due to the many alternative exons and splice variants. The different FGFs bind with different affinities to the different receptors and their variants. This leads to an exceptionally complex pattern of specific interactions and specific effects on cells. Additional complexity arises from the requirement for a heparan sulfate-like binding in the presence of many possible variations in the heparan sulfate on cell surfaces. Finally, some FGF is translocated in the cell where it is synthesized to the nucleus, where it presumably serves some function. In total, these facts suggest that there is a nearly unimaginable variety and subtlety in the responses to FGFs.