Division of Gastroenterology and Hepatology and Mayo Clinic Cancer Center, College of Medicine, Mayo Clinic, Rochester, MN
Address reprint requests to: Lewis R. Roberts, M.B.Ch.B., Ph.D., Division of Gastroenterology and Hepatology, College of Medicine, Mayo Clinic, 200 First St. SW, Rochester, MN 55905. E-mail: email@example.com; fax: 507-284-0762.
Potential conflict of interest: Nothing to report.
Fibroblast growth factors, or FGFs, are a large family of polypeptide cytokines exhibiting a pleiotropy of functions, from cell growth to angiogenesis, wound healing, and tissue repair. This review broadly covers the genetics and protein expression of the FGF family members and the signaling pathways involved in FGF-mediated growth regulation. We emphasize the role of FGFs in the pathogenesis of hepatocellular carcinoma (HCC), including their effects on regulation of the tumor microenvironment and angiogenesis. Finally, we present current views on FGF's potential role as a prognostic marker in clinical practice, as well as its potential as a therapeutic target in HCC. (Hepatology 2014;59:1166–1173)
Fibroblast Growth Factors: Genes, Family Members, and Structural Features
Fibroblast growth factor (FGF) was identified in the 1970s from bovine pituitary extracts exerting mitogenic effects on 3T3 fibroblasts. The 22 human FGFs are 17 to 34 kDa glycoproteins encoded by different genes. FGFs 11-14, the fibroblast homologous factors, are not secreted and act intracellularly. The remaining 18 mammalian FGFs (FGF1-FGF10 and FGF16-FGF23) are grouped into six homologous subfamilies: FGF1 (acidic FGF) and 2 (basic FGF); FGF3, 7 (keratinocyte growth factor, KGF), 10, and 22; FGF4, 5 and 6; FGF8, 17 and 18; FGF9, 16 and 20; and FGF19, 21 and 23. Human FGF19 and mouse FGF15 are orthologs with 53% amino acid identity.
Most FGFs have signal peptides and are secreted. FGF1 and FGF2 lack signal sequences and are not secreted, but cross the membrane through a process facilitated by binding cell surface and extracellular matrix (ECM) heparan sulfate (HS). FGF protein domains include FGF receptor (FGFR)-binding domains and HS-binding domains that contribute to receptor dimerization. Almost all FGFs require binding to HS for FGFR activation. Cells lacking in, or unable to synthesize, HS proteoglycans (HSPG) or those pretreated with HS-degrading agents are thus insensitive to FGFs. FGF15/19, FGF21, and FGF23 are exceptions; they have a substantially lower affinity for HS, and diffuse away from the site of secretion. They thus function as endocrine hormones, while still stimulating classical FGF pathways (Fig. 1).
FGFR Structure and Signaling
The four FGFRs, FGFR1-FGFR4, are high-affinity receptor tyrosine kinases (RTKs). FGFRs have an extracellular ligand-binding region with two or three immunoglobulin (Ig)-like domains, an extracellular HS binding domain, a single transmembrane domain, and a split intracellular tyrosine kinase domain. Alternative splicing of FGFR messenger RNA (mRNA) results in over 48 different FGFR isoforms. The binding specificities of FGFRs are determined by sequence variation and β1 strand length at the FGF amino-termini, and alternative splicing of the third Ig-like domain.
To initiate signaling, FGFs bind to the Ig-like domains II and III of FGFRs, which contain the HS-binding regions that interact with HSPGs such as syndecans, glypicans, or perlecans. Syndecans act as coreceptors for FGF signaling, while glypicans may sequester FGFs in membrane raft domains away from FGFRs. Association of FGFRs with the HS-FGF complex forms a ternary complex containing two FGFs, two FGFRs and the HSPG. This complex activates the FGFR tyrosine kinase, autophosphorylating tyrosines in the C-terminus, kinase insert, and juxtamembrane regions. Phospho-FGFR phosphorylates phospholipase Cγ (PLCγ) and the docking proteins FGFR substrate 2 (FRS2) and FGFR substrate 3 (FRS3). Phosphorylation of a C-terminal FGFR tyrosine (Y766 in FGFR1) creates a binding site for the Src homology 2 (SH2) domain of PLCγ, inducing PLCγ phosphorylation and activation. Phosphorylated FRS2 and FRS3 activate the Ras-mitogen-activated protein kinase (MAPK) and phosphoinositide 3-kinase (PI3K)-protein kinase Akt/protein kinase B (PKB) pathways. Other molecules activated by FGFRs include p90 ribosomal protein S6 kinase 2 (RSK2), signal transducers and activators of transcription (STATs), and Src. These pathways enhance cell survival, proliferation, and motility and induce secretion of additional growth factors and cytokines.
Downstream feedback inhibition regulates the duration and extent of FGF signaling. Thus, FGF activates Sprouty (SPRY) proteins, inhibiting FGF stimulation of the MAPK pathway through interaction with growth factor receptor bound protein 2 (GRB2), SOS1, or RAF1. Therefore, the SPRY act as tumor suppressors in hepatocellular carcinoma (HCC). Likewise, the interleukin 17 receptor D (IL17RD) or SEF (similar expression to FGF) gene induced by FGF inhibits phosphorylation of FGFR1 and FGFR2. IL-17RD/SEF binds to activated MEK (mitogen-activated protein kinase), inhibiting disassembly of the MEK-ERK (extracellular signal regulated kinase) complex, thus blocking nuclear translocation of activated ERK.
FGF Signaling in Healthy Liver and Chronic Liver Disease
Substantial evidence implicates FGF signaling in normal liver function and its dysregulation in chronic liver disease. FGF19, the main source of which is the ileum, regulates hepatocyte bile acid and cholesterol metabolism and insulin sensitivity. The nuclear bile acid receptor farnesoid X receptor (FXR) induces FGF19, which signals through FGFR4, the main FGFR expressed in mature hepatocytes. Transcriptional effects of FGF19/FGFR4 include repression of CYP7A1 and acetyl CoA carboxylase 2 (ACC2). ACC2 converts acetyl CoA to malonyl CoA, a repressor of carnitine palmitoyl transferase 1-induced fatty acid oxidation. Therefore, FGF19-mediated repression of ACC2 enhances fatty acid oxidation. Another family member, FGF21, is an important regulator of liver lipid and glucose metabolism. Serum FGF21 levels are elevated in patients with nonalcoholic fatty liver disease, and both liver FGF21 mRNA and serum FGF21 show positive correlations with intrahepatic triglycerides. FGF21 appears to be more important for lipid metabolism than glucose homeostasis and insulin sensitivity.
FGF1 and FGF2 have been implicated in hepatic stellate cell activation. Serum FGF2 levels are increased in liver cirrhosis and rise with HCC development. Mice with knockout of both FGF1 and FGF2 show decreased collagen α1(I) expression and decreased liver fibrosis after chronic carbon tetrachloride (CCl4) treatment, suggesting their importance in collagen deposition during fibrogenesis. Similarly, the dual FGFR/vascular endothelial growth factor receptor (VEGFR) inhibitor brivanib inhibits liver fibrosis in mouse models (Nakamura, Roberts, et al., unpublished data). Conversely, knockout of FGFR4 enhances CCl4-induced liver fibrosis, suggesting that FGFR4 protects against chronic liver injury.
FGF Signaling in HCC
HCC usually develops in a background of liver cirrhosis. Hepatocytes in healthy liver express minimal FGF1 or FGF2. FGF1 and FGF2 are induced during chronic liver disease and increase in HCCs. FGF2 stimulates HCC proliferation by way of an autocrine mechanism, activates HCC invasion, and induces angiogenesis. The mRNA expression of FGFs and FGFRs has been studied in HCC cell lines. All cell lines except HUH6 and HUH7 expressed FGF1 and FGF2 mRNAs. Similarly, all cell lines except PLC/PRF/5 expressed FGFR mRNA. FGF1 and FGF2 stimulated proliferation of most HCC cell lines, suggesting that HCC proliferation is mediated by the FGF pathway. Ectopic expression of FGFR1 is also commonly observed in HCC cells. Mice transgenic for constitutively active human FGFR1 showed increased regeneration after partial hepatectomy, with no premalignant or malignant transformation at 18 months follow-up.
In contrast, treatment of FGFR1 transgenic mice with diethylnitrosamine (DEN) promoted development of adenomas and HCC. Conversely, germline deletion of both FGF1 and FGF2 did not abrogate DEN-initiated hepatocarcinogenesis, thus overexpression of FGF1 and FGF2 are not absolutely required for liver carcinogenesis.
The FGF8 subfamily members FGF8, FGF17, and FGF18 also demonstrate oncogenic effects in HCC. At least one member of the FGF8 subfamily or their cognate receptors FGFR2, FGFR3, and FGFR4 was up-regulated in 28 of 34 (82%) HCCs studied. In serum-starved cells, addition of FGF8, FGF17, or FGF18 impaired apoptosis. In contrast, down-modulation of FGF18 by small interfering RNA reduced the viability of HCC cells and impaired their clonal proliferation. These effects appear to involve the ERK and AKT/mTOR pathways. FGF8, FGF17, and FGF18 stimulated growth of myofibroblasts cultured from HCCs and induced proliferation and tube formation of liver endothelial cells. Since myofibroblasts are a rich source of VEGF, FGF8 subfamily expression may induce neoangiogenesis by increasing the number of myofibroblasts. Interestingly, transgenic mice overexpressing FGF18 in the liver exhibited an increase in liver weight and hepatocellular proliferation. In conclusion, the FGF8 subfamily appears to stimulate proliferation in liver cells, potentially playing an important role in the formation and progression of HCC.
The FGF19 subfamily, FGF19, FGF21, and FGF23, act as endocrine factors mediating metabolic effects through interaction with FGFRs. FGF19 signaling through FGFR4 is important in liver carcinogenesis. Indeed, expression of FGF19 was observed in 48% of 281 resected HCCs. Furthermore, FGF19 expression was significantly associated with larger tumor size, higher Barcelona Clinic Liver Cancer (BCLC) stage, and early recurrence. Thus, high levels of FGF19 may predict early recurrence and poor prognosis of HCC after curative hepatectomy. Supporting in vitro studies show that FGF19 induces HCC cell proliferation and invasion and inhibits apoptosis. Conversely, suppression of FGF19 or FGFR4 expression inhibits proliferation and increases apoptosis. Ectopic expression of FGF19 in mice also promoted hepatocellular proliferation, dysplastic changes, and liver carcinogenesis. Targeting the FGF19/FGFR4 interaction using a neutralizing antibody against FGF19 inhibits hepatocarcinogenesis in mice.
FGFR4 is also overexpressed in human HCCs and HCC cell lines. Most convincingly, knockout of FGFR4 inhibited FGF19-induced hepatocarcinogenesis in both spontaneous and DEN-induced mouse models, and anti-FGFR4 neutralizing antibody inhibited growth of human HCC xenografts in nude mice. In contrast, knockout of FGFR4 enhanced DEN-induced carcinogenesis in a different mouse model. These differences may be related to the genetic context of the models. Thus, targeting the FGF19/FGFR4 pathway may be effective in the treatment of HCC.
Alternative Mechanisms of Genetic and Epigenetic Dysregulation of the FGF Pathway in Cancer
Overexpression of FGFs can drive FGF signaling in cancer. Notably, the FGF19 gene locus is amplified in some HCCs. Given the multiple splice variants of both the FGFs and FGFRs, there are many possibilities for dysregulation of FGF signaling by variations in mRNA splicing.
Chromosomal translocation resulting in fusion of the kinase domain of FGFR to the dimerization domain of another protein can constitutively activate the receptor kinase. These fusion proteins escape the normal degradation pathway, making them strongly oncogenic. Impaired down-regulation of the FGF-FGFR signaling complex also increases FGF signaling and defective degradation increases cell surface receptors, enhancing signaling. FGFR alterations, e.g., the FGFR3 G380R mutation, promoting dimerization and increasing recycling of the receptor, may prevent receptor degradation, prolonging FGF signaling.
MicroRNAs (miRNAs) can also enhance FGF signaling. Recently, FGF2-induced CREB-mediated induction of the lysine specific demethylase 2B (KDM2B) was shown to synergize with the histone H3K27 methyltransferase EZH2 to silence miR-101, an inhibitor of EZH2, enhancing FGF2-mediated cell proliferation, migration, and angiogenesis. Repression of miR-15 and miR-16 by the HBV X protein has also been implicated in FGF signal activation and HCC pathogenesis.
FGF and Tumor Angiogenesis
The most widely recognized angiogenic factors are VEGFs, FGFs, angiopoietin, hepatocyte growth factor (HGF), and platelet-derived growth factor (PDGF). These activate RTKs and the Ras-p42/44 MAPK pathway in endothelial cells. While VEGF is the main driver of tumor angiogenesis, there is crosstalk between VEGF and FGF signaling in angiogenesis. FGF1, 2, and 4 stimulate arteriogenic collateral vessel growth in vivo. FGF2 released from tumor and stromal cells is stored in an inactive form in the subendothelial ECM of blood vessels through HSPG binding, with the FGF binding protein 1 releasing FGF2 from the ECM and serving as an extracellular chaperone for FGFs. Binding of FGF2 to HS in the ECM is HS sulfation-dependent. The HS desulfatase, sulfatase 2 (SULF2), releases FGF2 from storage sites in the ECM, enhancing FGF signaling.
Angiogenic signaling pathways involving FGFs show substantial redundancy. While FGF1 and FGF2 induce angiogenesis in vivo, FGF2 knockout mice demonstrate only mild delays in wound repair. Simultaneous knockout of both FGF1 and FGF2 did not impact angiogenesis, suggesting redundancy in the ligand-receptor system. Thus, the precise role of FGF1 and 2 in angiogenesis is yet to be defined. In HCCs, expression of FGF1 and FGF2 was associated with an increase in capillarized sinusoids, suggesting a role for FGF signaling in tumor angiogenesis. The clinical effects of anti-FGF agents on tumor angiogenesis suggest that at least some tumors have a requirement for FGF. However, functional redundancy is an important consideration in targeting FGF signaling for cancer therapy.
Effects of FGF on Nonvascular Elements of the Tumor Microenvironment
The interaction between cancer cells and their microenvironment is critical in tumor development, progression, and metastasis. FGFs and FGFRs regulate several pathways in the microenvironment. FGF stimulation modulates integrin and ECM expression, altering the composition of the microenvironment and cellular parameters critical to angiogenesis and tumor progression. Integrin-mediated interactions modulate angiogenic signaling pathways; antagonists of integrin αvβ3 inhibit FGF-induced angiogenesis, while antagonists of integrin αvβ5 inhibit VEGF-induced angiogenesis. Studies show differences between FGF and VEGF in regulating endothelial cells from different microenvironments.
FGFs as HCC Tumor and Prognostic Markers
FGF2 has been evaluated as a marker for HCC diagnosis. An initial study found progressively more elevated serum FGF2 in liver cirrhosis and HCC patients than those with chronic hepatitis. Conversely, a similar study showed a marked increase in FGF2 in cirrhosis and HCC patients, as compared to those with chronic hepatitis or healthy individuals, but with no significant difference between the cirrhosis and HCC groups. FGF2 was increased with more advanced tumor stage, but was also elevated in the presence of acute illness; thus, it had no utility as a diagnostic marker or for follow-up of HCC patients. Another study found higher FGF2 in patients with chronic hepatitis C and liver cirrhosis than in those with HCC. High preoperative serum FGF2 predicted an invasive HCC phenotype with poorer survival after resection. Therefore, it was suggested that serum FGF2 levels could be used as a prognostic marker in HCC. The available results for the clinical utility of serum FGF2 measurements are therefore conflicting and require further study and validation. Serum FGF19 was also shown to decrease after curative resection, suggesting that FGF19 may have utility as a marker for HCC tumor burden.
FGF Signaling as a Target for HCC Therapy
With improved understanding of HCC pathogenesis, there is growing interest in targeted therapy for HCC. The crucial roles played by the FGF signaling cascades in carcinogenesis make this pathway an important target in cancer therapy. Acyclic retinoid inhibited growth of HCC cells through down-regulation of FGFR3 expression, suppressing Rho activity and serum response factor-mediated transcription. Silencing the FGFR3 gene by RNAi also inhibited cell growth. Thus, in oncogene-addicted HCCs with coamplification of FGF19 and CCND1, inhibiting FGF19 with RNAi or a therapeutic monoclonal antibody could block HCC tumorigenicity.[35, 55] Currently, the only systemic agent improving survival in advanced HCC is sorafenib, which inhibits VEGFR, PDGFR, and Raf kinases, increasing survival in advanced HCC by 3 months. FGF activation appears to be a common resistance mechanism to sorafenib. Concomitant blockade of FGF and VEGF signaling could limit the FGF-mediated reinduction of angiogenesis and tumor regrowth seen with anti-VEGF agents. FGF3/FGF4 amplification may also predict response to sorafenib in patients with HCC. FGF3/FGF4 was amplified in 30% of HCCs from sorafenib responders. In in vivo xenograft experiments, sorafenib was partially effective for FGF4-expressing tumors but less effective in FGF3 tumors.
Brivanib is a selective dual inhibitor of VEGFR2 and FGFR1. In phase III HCC trials brivanib showed similar antitumor activity to sorafenib, but had a less well-tolerated safety profile. Dovitinib, a potent inhibitor of the VEGFR, PDGFR, and FGFR RTKs, showed significantly reduced basal phosphorylation of FGFR-1, FRS2α, and ERK1/2 but not Akt in HCC cells in vitro, and blocked FGF2-induced migration of HCC cells. In vivo mouse studies showed potent inhibition of liver tumor growth. A randomized phase II study comparing dovitinib versus sorafenib as first-line treatment in advanced HCC is ongoing (NCT01232296). TSU-68 (orantinib), a dual VEGF/FGFR inhibitor, showed promising efficacy with a good safety profile in phase I/II HCC studies (NCT00784290). E7080 (lenvatinib) is a multikinase inhibitor with higher affinity for VEGFR than FGFR. A phase I/II trial is ongoing in patients with advanced HCC (NCT00946153). Toxicity may be higher with nonselective multitargeted RTK inhibitors; therefore, more selective FGFR tyrosine kinase inhibitors (Fig. 2) are being developed, including AZD4547 and NVP-BGJ398.[62, 63] Therapeutic antibodies (Fig. 3) against FGFRs are also under development, including IMC-A1 (targeting FGFR1 (IIIc)), and R3Mab and PRO-001 (targeting FGFR3).[64-66]
FGF Activation as a Mechanism of HCC Resistance to Antiangiogenic Therapy
Since most antiangiogenic therapeutic strategies are only transiently effective, it is important to determine the tumor resistance mechanisms that they induce. A key mechanism is up-regulation of alternative proangiogenic factors such as FGFs. For example, treatment with an anti-VEGFR2 mAb showed decreased vascular density after 10 days only to manifest an angiogenic rebound after 4 weeks. Subsequently, the levels of the proangiogenic ligands FGF1, FGF2, FGF7, and FGF8 were found to be up-regulated in tumor cells, with a partially overlapping set (FGF1 and FGF2) up-regulated in tumor endothelial cells. Blocking FGF signaling with an FGF-trap minimized the acquired resistance to VEGF-targeted therapy, decreasing the tumor burden. Dual targeting of VEGF/FGF is therefore an important strategy to consider against therapy resistance.
Future Directions and Opportunities for Targeting FGF Signaling in Cancer
The role of FGF signaling in the molecular pathogenesis of HCC is well established. FGF signaling is also implicated in the development of resistance to anti-VEGF therapeutics. The role of anti-FGFR agents in management of advanced HCC is still evolving and the results of additional clinical trials evaluating small molecule FGFR inhibitors and antibodies targeting FGF signaling in HCC are awaited. If these trials show promising results with acceptable side effect profiles, the next step would be ascertaining the profiles of tumors resistant to anti-FGF therapy. This will help design therapeutic regimens targeted at parallel signaling pathways, with the goal of diminishing drug resistance.
Table 1. Overview of Clinical Trials With FGF and FGFR-Targeting Agents