RMS is an aggressive childhood cancer arising from skeletal muscle precursors. While significant progress has been made in the overall survival of patients treated for RMS, metastatic disease remains a considerable challenge, with less than 30% survival despite aggressive multimodal therapies (2
). Therefore, there is a critical need for the development of targeted therapeutics in patients presenting with advanced-stage RMS.
We and others have previously reported FGFR4
mRNA and protein overexpression in RMS (3
), although none of these studies elucidated its functional importance in RMS pathogenesis or its potential as a molecular target for therapy. FGFR4
is also expressed in myoblasts during normal development, in regenerating muscle following injury, but not in mature skeletal muscle (6
). PAX3 and PAX7 directly induce FGFR4
expression, resulting in the progression of embryonic progenitor cells into a myogenic program (7
). Furthermore, PAX3/7-FOXO1A chimeric transcription factors are present in the majority of ARMS (33
), and they increase the expression of target genes more than wild-type PAX3 or PAX7 (34
). This predicts that these chimeric fusion products, produced as a result of chromosomal translocations, could be strong inducers of FGFR4
These reports suggest that FGFR4 pathway activation may result in a rhabdomyoblast phenotype by enhancing proliferation and blocking terminal differentiation in RMS. Therefore, we hypothesized that FGFR4 activation may be oncogenic in RMS and represent a potential novel therapeutic target. Here, we demonstrate that high FGFR4
expression was significantly associated with protein levels, ARMS histology, metastatic disease, and poor survival. However, in a multivariable regression analysis, FGFR4
mRNA expression was not independent of high stage or ARMS histology, since both of these parameters are associated with poorer prognosis and high FGFR4
). This association would also be expected if FGFR4
is a direct target of the PAX3/7-FOXO1A fusion transcription factors. Moreover, we found that suppression of wild-type FGFR4 resulted in a significant reduction in local growth and fewer early and late pulmonary metastases in xenograft models.
Oncogene activation has been described to occur through overexpression, gene amplification, or mutation (36
). Our results suggested that overexpression might result in increased activity and led us to hypothesize that activating FGFR4
mutations might also be present in RMS (5
). In this study we confirmed FGFR4
TK domain–activating mutations in 7.5% of RMS tumors, which were not present in normal populations. Additionally, all of the FGFR4
TK domain mutations were somatic in the subset of the RMS patients that had tumor DNA mutations and a paired germline DNA sample. This does not rule out the existence of germline FGFR4
mutations in addition to somatic mutations if larger populations or pedigrees were to be surveyed. However, given the dominant action of these mutations, it seems unlikely that germline mutations of this gene would result in normal development.
Computational analysis of these TK domain mutations predicted that they would likely result in FGFR4 autophosphorylation with resultant downstream pathway activation, as was confirmed by our studies. Notably, a significant increase in Stat3 activation was observed. Stat3 activation has previously been associated with cell growth and survival in RMS and other cancers and is known to occur downstream of the FGFRs (40
). Of interest, investigation of the Akt pathway revealed that both FGFR4
mutations suppressed phospho-Akt. This is consistent with previous findings that germline, activating FGFR2
mutations result in suppression of phospho-AKT and that inactive AKT can promote invasion and metastasis (44
). We speculate that the phenotypic consequences of FGFR4 mutational activation are mediated by oncogenic and metastatic effects of Stat3. Further work is required to determine which of these cellular alterations dictate the metastatic phenotype.
RTKs that are activated by point mutations have been shown to be drivers of tumorigenesis and represent ideal targets for therapy (36
). Previously identified RMS mutations that could be exploited therapeutically include PAX3/7-FOXO1A
gene translocation/fusions found only in ARMS and RAS missense mutations (NRAS
, and HRAS
), reported in a small number of ERMS (33
). However, targeting fusion transcription factors remains a significant challenge, and the mutations in RAS genes were found in studies in which only a small number of ERMS tumors were surveyed. Importantly, this is the first report of activating RTK mutations that are common to both histological types of RMS.
More generally, our study represents the highest prevalence of FGFR4
TK domain mutations reported in human cancers, while other large-scale cancer genomic screens have found infrequent missense mutations in FGFR4
). Additionally, none of the missense mutations identified in this study were found in adenocarcinoma of the lung, which has the highest prevalence of FGFR4
mutations reported to date (1.8%) (13
). In addition, 2 of the 4 mutations occurring at codons 535 and 550 are known to be mutated in FGFR paralogs (FGFR1, -2, and -3 and RET) and in FGFR4 for a single hypermutated breast cancer sample (11
Functionally these FGFR4
mutations appear to be significantly more potent than wild-type FGFR4
overexpression in promoting growth and metastasis, and they were necessary for in vivo neoplastic growth in NIH 3T3 fibroblasts. These findings are in agreement with prior work, which has shown that introduction and overexpression of wild-type FGFR4
does not transform fibroblasts or support FGF-induced growth in BaF3 cells (20
). In contrast, FGFR1–FGFR3 are all able to transform different cells through either overexpression alone or overexpression with FGF stimulation (57
), suggesting that FGFR4 has unique signaling and biological responses compared with its FGFR paralogs. Most significantly, these mutations increased invasiveness and promoted a metastasis phenotype and poor survival in our murine RMS models. Prior association studies have shown that a common variant in FGFR4, G388R, is associated with tumor progression in the absence of detectable FGFR4 activation and that this may be due to the role of FGFR4 as a tumor suppressor (59
). However, our observations demonstrate that FGFR4
mutational activation leads to an oncogenic phenotype, and this is in accord with others who have suggested that FGFR4
is an oncogene (13
Our data suggest that FGFR4 is an excellent candidate for targeted therapy in patients with advanced-stage RMS. Furthermore, we show that 7.5% of RMS tumors harbor predicted activating mutations, and we confirm that 2 of these are driver mutations that lead to enhanced sensitivity to a small molecule inhibitor. These results provide a rational basis for therapeutically targeting the FGFR4 pathway in RMS and other cancers. Overall, our findings have direct implications for rapid translation into adjuvant therapies for metastatic RMS, for which long-term prognosis remains poor.