First printed in R&D Systems' 2001 Catalog.
The name "fibroblast growth factor" (FGF) is a limiting description for this family of cytokines. The function of FGFs is not restricted to cell growth. Although some of the FGFs do, indeed, induce fibroblast proliferation, the original FGF molecule (FGF-2 or FGF basic) is now known to also induce proliferation of endothelial cells, chondrocytes, smooth muscle cells, melanocytes, as well as other cells.2 It can also promote adipocyte differentiation, induce macrophage and fibroblast IL-6 production, stimulate astrocyte migration, and prolong neuronal survival. Thus, the FGF designation is clearly limiting by its description of one target cell and one implied biological activity. To date, the FGF superfamily consists of 23 members, all of which contain a conserved 120 amino acid (aa) core region that contains six identical, interspersed amino acids. The superfamily members act extracellularly through four tyrosine kinase FGF receptors, with multiple specificities noted for almost all FGFs.10 This likely accounts for similar effects generated by many FGF molecules on common cell types.6, 11 The FGFs, partly by way of their originally recognized proliferative activities, are now considered to play substantial roles in development, angiogenesis,13 hematopoiesis, and tumorigenesis.
To date, 22 human FGFs have been identified, including FGF-1 through 14 plus 16 through 23.9 In general, their native molecular weights range from 7 kDa (FGF-1) to 38 kDa (FGF-5),16,17 and their aa lengths vary from 60 aa (an FGF-1 splice variant) to 288 aa (FGF-2). These aa lengths are generally determined by three coding exons. Over coding regions beginning with an AUG (methionine) start site, pairwise aa identity ranges from 17% to 72% (20-30% on average), with seven human subfamilies identified based on aa homology (reference 9 and Tables 1 and 2). These include FGFs-1 & 2, FGFs-3 & 5, FGFs-11, 12, 13 & 14, FGFs-4 & 6, FGFs-9, 16 & 20, FGFs-7 & 10, and FGFs-8, 17 & 18; human FGF-19 is closest to mouse FGF-15, and human FGF-21 has no counterpart. In addition to AUG start sites, many FGFs utilize CUG (leucine) start sites, and this adds a 5' aa extension to many molecules. Further complexity is introduced by the fact that select FGFs (FGF-8 and -17) exhibit a division of the first exon into three or four mini-exons, creating the possibility for multiple N-terminal splice forms. Although there is only very modest aa identity between all 22 human family members, the one characteristic apparently shared among all members is the three-dimensional, ß-trefoil motif (i.e., a fold which consists of 12 anti-parallel ß-strands). This configuration may prove to be critical for receptor recognition. There is considerable FGF aa identity across species, and species cross-reactivity is common among FGF superfamily members. Reviews on FGFs and their receptors can be found in references 2, 3, 6, 11, 15, 19, 20.
Human FGF-1 (also known as FGF acidic, FGFa, ECGF and HBGF-1) is a 17-18 kDa non- glycosylated polypeptide that is expressed by a variety of cells from all three germ layers.3, 21, 22 Synthesized as a 155 aa protein, FGF-1 has no definitive signal sequence and thus is not secreted through classical pathways. FGF-1 is released in an ATP-dependent process that involves synaptotagmin-1 in various states of cleavage. Unlike FGF-2, which is released as a monomer, FGF-1 is apparently released as a disulfide-linked dimer.24, 28 Since dimerized FGF-1 has both a reduced affinity for heparin and decreased activity, tissue resident reducing agents such as glutathione can break its disulfide linkage, creating bioactive, monomeric FGF-1 that is stored on or in the glycosaminoglycan component/heparan sulfate (HS) of the extracellular matrix. From this storage depot, FGF-1 is released as needed and interacts with both cell surface HS and any one of four known tyrosine kinase FGF receptors.2, 10 In addition to its extracellular activity, FGF-1 also functions intracellularly. Based on a nuclear localization sequence (NLS) between aa #22 and #28 (i.e., -NYKKPLK), FGF-1 may enter the nucleus via its NLS and contribute to DNA synthesis. The route that FGF-1 takes to get to the nucleus is unclear. Some sort of cell surface receptor binding is necessary, followed by its internalization and translocation to the nucleus. The binding molecule may be either an FGF receptor or a HS moiety, and limited proteolytic processing of FGF-1 after internalization might be necessary for activity.36 Human and mouse FGF-1 share 96% aa identity. 37 Human and rat FGF-1 share 95% aa identity.38 Cells known to express FGF-1 include intestinal enterochromaffin cells, renal proximal tubule cells 40 smooth muscle cells, 41 neurons, hepatocytes, skeletal muscle cells, endothelial cells and macrophages, and keratinocytes and fibroblasts.
Human FGF-2, otherwise known as FGF basic, HBGF-2, and EDGF, is an 18 kDa, non-glycosylated polypeptide that shows both intracellular and extracellular activity.19, 47-51 Like FGF-1, the 18 kDa molecule is 155 aa in length when translated from an AUG (methionine) start codon. 52 Unlike FGF-1, there are at least four alternate CUG (leucine) start codons that provide N-terminal extensions of 41, 46, 55, or 133 aa, resulting in proteins of 22 kDa (196 aa), 22.5 kDa (201 aa), 24 kDa (210 aa) and 34 kDa (288 aa), respectively.18, 49 Although these extensions could be assumed to impart special activities to the higher molecular weight (HMW) forms, all FGF-2 forms have the potential to perform the same function. The fact that they do not seems more related to their anatomical location than their structure. 51 Differences noted for 18 kDa vs. 24 kDa FGF-2 include an up-regulation of t-PA for 18 kDa FGF-2 and a down-regulation of t-PA for 24 kDa in pancreatic carcinoma cells, 53 and an increase in IL-6 expression for 24 kDa FGF-2 contrasted with a decrease in IL-6 expression for 18 kDa/FGF-2 in 3T3 fibroblasts. In general, the 155 aa/18 kDa form is considered cytoplasmic (and secretable),51, 55, 56 while the HMW forms are considered nuclear in target. Although less is known about the secretion of FGF-2 than FGF-1, release of the 18 kDa form is reported to be an energy-dependent process that is independent of the ER-Golgi. Unlike FGF-1, FGF-2 is secreted as a monomer. Following secretion, FGF-2 is sequestered on either cell surface HS or matrix glycosaminoglycans. Although FGF-2 is secreted as a monomer, cell surface HS seems to dimerize monomeric FGF-2 in a non-covalent side-to-side configuration that is subsequently capable of dimerizing and activating FGF receptors. 19, 47, 57 Notably, FGF-2 responsive cells may be able to block cell surface dimerization of basic FGF by ribosylating 18 kDa FGF-2 in the receptor-binding domain (aa 106-120).58 In addition to inducing FGF receptor signal transduction, bound 18 kDa FGF-2 can be internalized, either in association with HS or its receptor. Once internalized, FGF-2 can be degraded, possibly into 4-10 kDa bioactive fragments, 63 or translocated to the nucleus as a receptor-ligand complex. HMW forms of FGF-2 are associated with nuclear translocation. Although there is an NLS in the N-terminus of all HMW forms, it has been suggested that methylation of arginines 24-28 aa upstream of the first methionine of 18 kDa FGF-2 is responsible for nuclear entry. 65 Entry into the nucleus is associated with FGF-induced effects such as casein kinase II activation, which is necessary for cell-cycle progression and proliferation. Human 18 kDa FGF-2 is 97% aa identical to both mouse and rat FGF-2. 37, 67 Cells known to express FGF-2 include visceral and vascular smooth muscle cells, cardiac muscle cells, lining epithelium of the colon and bronchus, neurons, plus cerebellar Purkinjie cells, megakaryocytes and platelets, 69 endothelial cells,68, 70 mast cells, glomerular parietal epithelial cells and podocytes, astrocytes, CD4 + and CD8 T cells, fibroblasts (plus extracellular matrix), and numerous embryonic mesodermal and neuroectodermal tissues.
Human FGF-3 is the product of the int-2 gene [i.e., derived from integration region-2, a region on mouse chromosome 7 that contains a gene (int-2/FGF-3) accidentally activated following retroviral insertion]. The molecule is synthesized as a 28-32 kDa, 222 aa glycoprotein that contains a number of peptide motifs. Based on structural similarity with mouse FGF-3, the human FGF-3 precursor contains a 17 aa signal sequence, plus a 24 aa N-terminal and an 18 aa C-terminal NLS. 78, 80 Human and mouse FGF-3 share 81% aa identity over the entire AUG-initiated coding region, 78 with the mouse demonstrating an additional 274 aa extended form that results from the use of a CUG alternate start site. 81, 82 The 29 aa extension contains no special sequence motif. The N-terminal extension plus the signal sequence is treated simply as a 46 aa signal sequence. Therefore, based on its internal motifs, mouse FGF-3 can be either secreted or translocated to the nucleus, with post-translational glycosylation playing an important role in the secretion option. Secreted FGF-3 can induce DNA synthesis, whereas nuclear FGF-3 has an inhibitory effect on cell proliferation. Cells reported to express FGF-3 are limited to developmental cells and tumors; no normal adult cells are known to express FGF-3. Embryonic cells and structures expressing FGF-3 include primitive streak mesoderm, 87 pharyngeal pouch endoderm, pharyngeal arch ectoderm and hindbrain neuroepithelium, and sensory cells of the inner ear. Tumors known to express FGF-3 include breast carcinomas and colon cancer cell lines. 89, 90
Human FGF-4 is a 22 kDa, 176 aa glycoprotein that is the product of a developmentally-regulated gene.91, 92The molecule is synthesized as a 206 aa precursor that contains a large, ill-defined 30 aa signal sequence plus two heparin-binding motifs (at aa 51-55 and 140-143). The heparin-binding sites directly relate to FGF-4 activity; heparin/heparan regulate the ability of FGF-4 to activate FGFR1 and FGFR2. At low (physiological) heparin concentrations, FGF-4 only binds to FGFR2, whereas at high heparin concentrations FGF-4 can activate FGFR1. 94 There is 87% aa identity within the mature portions of human and mouse FGF-4. When secreted, FGF-4 has mitogenic activity on both fibroblasts and endothelial cells. Cells known to express FGF-4 include both tumor cells and embryonic cells. Its identification in human stomach cancer gives rise to one alternative designation (/hst-1/hst), while its isolation in Kaposi's sarcoma provides grounds for another (K-FGF). Embryonic cells known to express FGF-4 include primitive streak mesoderm,98, 99 molar enamel knot epithelium, 100 apical ridge ectoderm, 101 mature somitic myotome, 98 the ectoderm of day 4 blastocysts, 98 and cells of the notochord anterior to Hensen's node.
FGF-5 is a 32-38 kDA, secreted glycoprotein that was originally discovered in 3T3 fibroblasts following their transfection with human tumor DNA. Human FGF-5 is synthesized as a 268 aa precursor that contains a 17 aa signal sequence and a 251 aa mature peptide. FGF-5 is an unusual FGF in that it contains both O- and N-linked glycosylation, but is also quite similar to other FGFs in that it strongly binds heparin.17, 104, 105 The heavily glycosylated carbohydrate component of the molecule does not appear to be necessary for receptor activation. 106 As with other FGFs, secreted FGF-5 is mitogenic for both endothelial cells and fibroblasts.104, 106 In addition to the 268 aa precursor form, the human gene also gives rise to a glycosylated alternate splice form that is 18 kDa in size and 123 aa in length. This molecule is found in the brain and has partial agonist/antagonist activity, depending on its local concentration. 107 Human and mouse FGF-5 share 85% aa identity overall, with 88% aa identity within the mature regions. 37, 104, 108 Human and rat FGF-5 share 84% aa identity, both within the precursor molecule and the mature segment. 104, 109 Both embryonic and adult cells are known to express FGF-5. In the embryo, FGF-5 is found in the myotome, lateral splanchnic mesoderm and acoustic ganglia. In the adult, thalamic reticular and hippocampal CA3 neurons,17, 108, 110 skeletal muscle, and pancreatic ß-cells, ductal cells, fibroblasts and acinar cells all express FGF-5.
FGF-6, also known as hst-2, was originally identified as the product of an FGF-4/hst homologous gene. In the human, mature FGF-6 is secreted as a 25 kDa glycosylated polypeptide. 113 FGF-6 has three potential AUG (methionine) start sites that could yield precursors of 208, 198 or 175 aa. 113, 114 Based on recent mouse studies, it would appear that the 198 aa precursor might well be the standard form. 115, 116 This results in a molecule that has a 27 aa signal sequence plus a 171 aa mature region.114, 116 Human and mouse FGF-6 share 93% aa identity. 114, 117, 118 When secreted, mouse FGF-6 shows multiple N-terminally truncated forms, varying from 171 aa to 136 aa. The 171 aa form is highly glycosylated, exhibiting both N- and O-linked glycosylation; the shorter forms lose the one potential N-linked site. 116 Although FGF-6 has been reported to be mitogenic for fibroblasts and displays marginal activity on endothelial cells, 119 fully glycosylated FGF-6 (utilizing its N-linked glycosylation site) has potent mitogenic activity on endothelial cells. 116 FGF-6 is best known for its role in skeletal muscle physiology. Thus, cells known to express FGF-6 are essentially limited to myotomes and adult skeletal muscle cells.118, 120, 121
Human FGF-7, otherwise known as keratinocyte growth factor (or KGF), is a 28 kDa, single chain, secreted glycoprotein that has a target specificity restricted to epithelium. 122-125 The molecule is synthesized as a 194 aa precursor that contains a 31 aa signal sequence and a 163 aa mature segment. The mature growth factor binds heparin with residues that lie between aa 29-38, 124 and it binds its receptor (KGFR) with residues that lie in the region of aa 101-105. Although heparin has been reported to inhibit KGF activity, 124 recent studies show that heparin is required for KGFR activation. What is unclear is whether one molecule of KGF is sufficient to dimerize two KGF receptor molecules, or whether two molecules of KGF contribute to KGFR dimerization. Although FGF-7 is glycosylated, non-glycosylated KGF shows higher mitogenic activity than native FGF-7/KGF. There is 96% aa identity in the mature segments of human and mouse FGF-7 123, 129 and 92% aa identity in the mature regions of human and rat FGF-7. Adult cells known to express FGF-7 include fibroblasts, 131γδ T cells, 132 smooth muscle cells, 133 and ovarian theca cells. 134 In the embryo, KGF is found at many stages of development throughout the mesenchyme. 135
FGF-8 was originally isolated as AIGF (or androgen-induced growth factor), a molecule secreted by testosterone-stimulated mouse Shionogi carcinoma cells. Initial studies revealed that it is a secreted, 28-32 kDa glycoprotein with two alternate splice forms. Multiple splice forms exist for both mouse and human FGF-8. In general, members of the FGF family have genes with three exons. In the case of FGF-8, the first "typical" family exon is divided into at least four small exons, and these are followed by the expected exons 2 and 3. Because of the existence of alternative donor and acceptor sites within these mini-exons, at least eight isoforms are possible in mouse 137 and four in human. 140 In mouse, the eight isoforms are labeled 8a through 8h (8b is the 28 kDa AIGF previously mentioned), and the precursor lengths range from 204 aa (8a) to 298 aa (8h). 140 All isoforms are identical within the C-terminal 142 aa as exon 2 (35 aa) and exon 3 (107 aa) are invariant. 141 In the variable N-terminus (62 to 156 aa in length), a constant 22 aa signal sequence is always present. Differences exist, however, in the number of potential N-linked glycosylation sites. 138, 140 Although the human gene parallels that for the mouse, human FGF-8 is limited to four isoforms due to the existence of a stop codon in the second mini-exon. Only the 8a (204 aa), 8b (215 aa), 8e (233 aa) and 8f (244 aa) isoforms are synthesized in man. Mouse and human 8a and 8b isoforms are 100% identical, while the 8e and 8f isoforms are 98% identical. The isoforms have different affinities for various receptor isoforms. For example, 8a activates no known receptor, 8b activates FGFR2IIIc, FGFR3IIIc and FGFR4, and 8e activates FGFR3IIIc and FGFR4; no FGF-8 isoform activates a IIIb form or FGFR1.137, 142 In the fetus, regions known to express FGF-8 include the embryonic infundibulum, the apical ectodermal ridge of the limb bud and oral epithelium of the first bronchial arch, and the pre-primitive streak embryonic ectoderm, nephrogenic cords, Bowman's capsule and developing labyrinth. In the adult, it is found in prespermatogonia and antral follicles of the ovary. 145
Human FGF-9 (also known as GAF or glia-activating factor) is a 30 kDa, single chain, heparin-binding secreted glycoprotein that has principal activity in, and on, nervous tissue.146, 147 The molecule is synthesized as a 208 aa precursor that has no definitive (leader) signal sequence, a structure reminiscent of those for FGFs-1 and 2. Unlike FGF-1 and 2, however, FGF-9 is abundantly secreted. This is attributed to an interaction between the molecule's first 33 N-terminal aa and a 26 aa stretch between aa 95-120 that serves as an uncleavable internal signal sequence.146, 148 Secreted FGF-9 may exist in multiple N-terminally truncated forms. These range from 30 kDa (205 aa) to 29 kDa (197 aa) to 25 kDa (175 aa), and it is unknown if they are an artifact of isolation or naturally-occurring, proteolytically processed forms.92, 147 Human FGF-9 shares 98% aa identity with mouse and rat FGF-9 (which are 100% identical). Cells known to express FGF-9 include prostatic stromal cells, 151 neurons, astrocytes and oligodendrocytes, 152, 153 embryonic oral epithelium of the first bronchial arch and primary enamel knot and fetal skeletal myoblasts, ventricular myocardium, gut lumenal epithelium and olfactory bulb.
Based on studies in the mouse, 156 FGF-10 in the human is presumably a glycosylated, 30 kDa heparin binding protein that has unusual secretory properties. Also known as KGF-2, human FGF-10 is synthesized as a 209 aa precursor that contains an unusually long signal sequence (35 aa) and an N-terminal serine repeat motif. Although the molecule is presumed to be secreted, it does not appear free unless exposed to heparin, suggesting it is sequestered on cell surfaces or within the extracellular matrix. There is 94 % aa identity within the mature segments of human and mouse FGF-10. 156, 158 There is 100% aa identity within the mature segments of human and rat FGF-10, with the rat exhibiting an extended repeat of serines in the N-terminus (13 for rat vs. 5 for human). 159 In addition to the full-length molecule, rat FGF-10 is suggested to hold the potential for alternate splicing, with translation initiated at a methionine at position 42. 160 Cells known to express FGF-10 include fibroblasts and pre-adipocytes in the adult,156, 161 plus lung mesenchyme, posterior limb mesoderm and mesenchyme associated with the development of the seminal vesicle and prostate in the fetus.
Human FGF-11 was originally cloned from a human retinal cDNA library and found to encode a protein 225 aa in length with no signal sequence and internal basic aa clusters that may constitute an NLS. Mouse FGF-11 has also been cloned, and found to consist of two alternate splice forms; one 225 aa in length and another 165 aa in length. Their significance is unknown. 165, 166 The long form of mouse and human FGF-11 share 97% aa identity. FHF-3 (originally named as a fibroblast growth factor homologous factor) is found throughout the embryo at various stages.165, 166
FGF-12 is a 243 aa, heparin-binding protein associated with a variety of developmental structures. The molecule has no definitive signal sequence, but does possess two N-terminal aa sequences at aa 11-18 and 28-38 that qualify as a bipartite NLS, and a secondary NLS at aa 56-68. Between human and mouse, there is almost 100% aa identity in the 243 aa long form (FGF-12A). 165, 167 In both mouse and human, there is at least one other alternatively spliced short form of 181 aa (FGF-12B). 166-168 This form does not possess an NLS and its function is unknown. In the embryo, cells known to express FGF-12 include mesenchyme surrounding sites of skeletal formation, glomeruli, and atrial myocardium; in the adult, FGF-12 message is found in various regions.
Human and mouse FGF-13 are 245 aa proteins that arise from genes that show N-terminal alternate splicing. In human and mouse, transcripts for 245 aa, 199 aa, 226 aa, 192 aa and 255 aa have been identified, with almost complete cross-species aa identity among all splice forms (> 98%). FGF-13 has been identified in fetal ependyma, dorsal root and cranial ganglia, both atrial and ventricular myocardium, and in renal collecting duct-associated mesenchyme.
Human FGF-14 is a 247 aa polypeptide that contains no signal sequence, but possesses a bipartite NLS with a secondary signal motif. This 245 aa molecule, termed FGF-14A, is 98% aa identical to mouse 14A, which is one of three alternate splice forms known for the mouse.165, 169, 170 In both human and mouse, there is also a second 252 aa splice form that has high cross-species aa orthology.165, 171 This isoform, termed FGF-14B, has a potential signal sequence with no definitive NLS thus suggesting it could be secreted. 165, 172 In mouse, there is a third known alternate splice form termed 14C that contains only the C-terminal 163 aa common to splice forms 14A and 14B. This makes it simply an N-terminally truncated form of FGF-14. Embryonic regions known to express FGF-14 include central nervous system zones that contain migrating neurons and glia, the aortic arch and its immediate branches, granule cells of the cerebellum, and the mammillary nuclei whereas adult cells known to express FGF-14(C) are limited to spermatocytes.165, 170, 172
This molecule is only known to exist in mouse and consists of a 218 aa protein with a 23 aa signal sequence and an unusual cysteine pattern. 173
Human FGF-16 is a 26 kDa, 207 aa polypeptide that shows no identifiable signal sequence. Human and rat FGF-16 share 98% aa identity, with the rat cDNA coding for a secreted protein when transfected into Sf9 cells. This suggests a non-classical secretion pathway, perhaps similar to FGF-9. FGF-16 is expressed by brown adipocytes, where it has marked activity.174, 175
FGF-17 is a molecule analogous to FGF-8 as the gene encoding the factor has a series of mini-exons substituting for a standard FGF family exon 1. This creates the potential for a number of different isoforms varying within the N-terminus. In human, the standard isoform (and only one known to date) is 216 aa in length and termed FGF17B. It has a 22 aa signal sequence and, based on rat studies, would be expected to be secreted by insect cells.176, 177 The size of the secreted molecule is 25 kDa, but given the nature of insect cells, it is not known what native molecular weight would result from mammalian expression. Human FGF-17B shares 93% aa identity with mouse and rat FGF-17B (which are identical). 178 In mouse, two additional isoforms have been reported, an A and C form. FGF-17 has been found in the embryo at the midbrain-hindbrain junction, in the telencephalon-diencephalon transition, in the smooth muscle of major artery walls, in chondrocytes, osteoblast precursors, and in mesenchymal cells. 176
Human FGF-18 is a 31 kDa, secreted, heparin-binding glycoprotein that is expressed in both fetal and adult tissues. The molecule is 207 aa in length and contains a 26 aa signal sequence with a 181 aa mature segment. Human FGF-18 shares 99% aa identity with both mouse and rat FGF-18. In light of its homology to FGF-8 and FGF-17, the FGF-18 gene may code for alternate splice forms. FGF-18 appears to be involved in lung physiology as it is present in both adult and fetal lung respiratory system. 180
FGF-19 is only known in human, and is 216 aa in length with a 22 aa signal sequence. The molecule is secreted from insect cells. Embryonic regions known to express FGF19 include skin, cartilage, retina, gall bladder and small intestine.
Human FGF-20 is a 23 kDa, 211 aa polypeptide that contains no signal sequence. FGF-20 demonstrates 72% homology with FGF-9, suggesting that it may be secreted in a similar manner to FGF-9. 8
FGF-21 is a 209 aa polypeptide that presumably has a signal sequence and is secreted. Human FGF-21 demonstrates only 75% aa identity with mouse FGF-21. In mouse, it is expressed preferentially in liver.