Identification of FGFR4-activating mutations in human rhabdomyosarcomas that promote metastasis in xenotransplanted models

Rhabdomyosarcoma (RMS) is a childhood cancer originating from skeletal muscle, and patient survival is poor in the presence of metastatic disease. Few determinants that regulate metastasis development have been identified. The receptor tyrosine kinase FGFR4 is highly expressed in RMS tissue, suggesting a role in tumorigenesis, although its functional importance has not been defined. Here, we report the identification of mutations in FGFR4 in human RMS tumors that lead to its activation and present evidence that it functions as an oncogene in RMS. Higher FGFR4 expression in RMS tumors was associated with advanced-stage cancer and poor survival, while FGFR4 knockdown in a human RMS cell line reduced tumor growth and experimental lung metastases when the cells were transplanted into mice. Moreover, 6 FGFR4 tyrosine kinase domain mutations were found among 7 of 94 (7.5%) primary human RMS tumors. The mutants K535 and E550 increased autophosphorylation, Stat3 signaling, tumor proliferation, and metastatic potential when expressed in a murine RMS cell line. These mutants also transformed NIH 3T3 cells and led to an enhanced metastatic phenotype. Finally, murine RMS cell lines expressing the K535 and E550 FGFR4 mutants were substantially more susceptible to apoptosis in the presence of a pharmacologic FGFR inhibitor than the control cell lines expressing the empty vector or wild-type FGFR4. Together, our results demonstrate that mutationally activated FGFR4 acts as an oncogene, and these are what we believe to be the first known mutations in a receptor tyrosine kinase in RMS. These findings support the potential therapeutic targeting of FGFR4 in RMS.

Rhabdomyosarcoma (RMS) is a pediatric sarcoma arising from skeletal muscle, with which the majority of patients can be subclassified as having either alveolar RMS (ARMS) or embryonal RMS (ERMS). ARMS is observed in older patients and is associated with a chromosomal translocation creating a fusion gene involving FOXO1A on chromosome 13 and members of the PAX gene family. ERMS is characterized by loss of heterozygosity and altered patterns of genomic imprinting (1). Despite marked improvement in overall prognosis during the last 4 decades, long-term survival for those with metastatic RMS remains poor (<30%) (2). Factors contributing to tumor progression and metastatic disease are not well understood.

Analysis of RMS gene expression patterns have led to improved diagnostic accuracy and new insights into possible mechanisms for metastatic regulators including SIX1 and EZRIN (3–5). We and others have reported that FGF receptor 4 (FGFR4), a receptor tyrosine kinase (RTK) member of the FGFR gene family, is highly expressed in RMS and that its mRNA expression correlates with protein levels (3, 4, 6). FGFR4 is also a key regulator of myogenic differentiation and muscle regeneration after injury, although it is not expressed in differentiated skeletal muscle (7–9). These observations raise the possibility that FGFR4 is not only a tumor-specific marker, but that it could also function as an oncogene in RMS.

The FGFRs are of considerable interest in cancer biology because they regulate essential processes including cellular survival, motility, development, and angiogenesis (10). Comparison of the complete FGFR coding regions indicates segments of high amino acid conservation in FGFR1, -2, -3, and -4. Germline mutations in these paralogs have been described for several rare Mendelian skeletal disorders, including hypochondroplasia (11, 12). RTKs may also be activated in human cancer by point mutations such as the somatic mutations within FGFR tyrosine kinase (TK) domains observed in glioblastoma multiforme, endometrial carcinoma, and lung cancer (13–17). However, FGFR4 is infrequently mutated in these and other cancers (13, 18, 19). It has also been reported that an aberrant FGFR4 isoform promotes tumorigenesis in pituitary adenomas, although the mechanism for the expression of this transcript was not due to somatic mutation but thought to be due to regulation by an alternative promoter (20). Since FGFR4 is highly expressed in RMS and during myogenesis, but not in mature skeletal muscle, we hypothesized that constitutive FGFR4 activation by either overexpression or mutation would promote an aggressive phenotype in RMS.

FGFR4 expression in RMS shows correlation with FGFR4 protein, advanced stage, ARMS histology, and poor survival. Since we previously demonstrated high FGFR4 expression in RMS tumors (3), we first investigated whether its expression was associated with aggressive clinical behavior, using available RMS expression microarray data sets (3, 5, 6). We confirmed high FGFR4 expression in primary RMS tumors compared with an independent panel of pediatric tumors and normal tissues (Figure 1A). Consistent with prior observations, mRNA overexpression was associated with high FGFR4 protein (Supplemental Figure 1; supplemental material available online with this article; doi: 10.1172/JCI39703DS1) (3). Analysis of an RMS data set for which clinical follow-up was available showed that FGFR4 messenger RNA was 2-fold higher in stage 4 metastatic tumors compared with stage 1 (P = 0.003), although there was some overlap in the range of expression between the 2 groups (Figure 1B) (6). ARMS also had higher expression compared with non-ARMS tumors (Figure 1C; P = 1.39 × 10^–9). Kaplan-Meier analysis for different FGFR4 expression quartiles showed a significant trend toward lower survival with higher expression (Supplemental Figure 2) and significantly lower overall survival for tumors with higher FGFR4 expression (Figure 1D; P = 0.03) (6). Further univariate analysis of this data set confirmed that high stage and ARMS histology were also associated with lower survival (Table 1). However a Cox proportional hazards multivariate analysis revealed that only clinical stage was associated with early death (Table 1; hazard ratio = 9.17; P = 0.002), indicating a strong association of FGFR4 expression with clinical stage and ARMS histology. To further validate the association of increased expression with advanced-stage disease, we also reexamined gene expression data from cell lines derived from a spontaneous model of murine ERMS (5, 21) and found significantly higher Fgfr4 expression in tumors of high metastatic potential (Figure 1E; P = 0.01).

Figure 1

High FGFR4 expression in RMS is associated with advanced stage, ARMS histology, and poor survival. (A) Log₂ median-centered expression data showed high FGFR4 expression in RMS tumors compared with other pediatric tumors and normal tissue of human tumors. EWS, Ewings sarcoma; NB, neuroblastoma; ALL, acute lymphoblastic leukemia; other, other pediatric tumors; normal, normal tissue. (B) Stage 4 RMS was associated with significantly higher median FGFR4 expression than stage 1 RMS, as analyzed based on the data of Davicioni et al. (Mann-Whitney test) (6). (C) FGFR4 expression was significantly higher in ARMS compared with all other histologic RMS subtypes (Mann-Whitney test). (D) Kaplan-Meier analysis based on a cutoff of median FGFR4 expression showed that patients with higher expression (top 2 quartiles) had significantly higher mortality (than the bottom 2 quartiles; P = 0.03, log-rank test). (E) Murine RMS cell lines (derived from refs. 5, 21) of high metastatic potential had significantly higher median Fgfr4 mRNA than nonmetastatic cell lines (P = 0.01, Mann-Whitney test).

Table 1

Sub-analyses of FGFR4 expression on survival in RMS

Because DNA amplification or gain can lead to increased mRNA expression levels, we next examined FGFR4 DNA copy number for a set of 94 RMS tumors and measured the mRNA levels for those samples with increased copy number variation (> 2.5, n = 15). We could not demonstrate a correlation between high copy number variation and mRNA expression of FGFR4 (r² = 0.0127, P = NS; Supplemental Table 1). This suggested that high FGFR4 expression was not due to amplification at the genomic level.

FGFR4 suppression inhibits RMS tumor growth and lung metastasis. To characterize the role of FGFR4 expression in the promotion of RMS tumor progression and metastasis, we stably introduced an inducible shRNA directed against FGFR4 into the human ARMS cell line RH30. Induction of this shRNA decreased FGFR4 protein expression by nearly 92% (Figure 2A) but did not result in reduced in vitro growth for up to 4 days in tissue culture (Figure 2B). However, there was a 41% reduction (P = 0.002) in growth after prolonged culture for 13 days (Figure 2C). We then proceeded to study the phenotypic consequences of reduced FGFR4 expression in vivo. FGFR4 suppression resulted in significantly smaller tumors 31 days after intramuscular injection into SCID Beige mice compared with controls (Figure 3, A and B; P = 0.002). Quantification of early arrest of metastatic cells in the lungs 24 hours after injection by intravital video microscopy (IVVM) revealed significantly fewer RH30 cells with FGFR4 suppression (Figure 3, C and D; P = 0.008). A comparable reduction in IVVM early lung metastases was also observed in the highly metastatic murine RMS cell line RMS33 with Fgfr4 suppression (Supplemental Figure 3; P = 0.02) (5). Of note, there were also fewer pulmonary metastases with FGFR4 suppression 74 days after intravenous injection of inducible RH30 cells (Figure 3E). To confirm that this difference was not due to a reduced growth rate of the cells, we normalized for the growth rate by calculating the ratio of pulmonary to pelvic tumor signals and found a significant reduction upon FGFR4 suppression (Figure 3F; P = 0.02).

Figure 2

FGFR4 knockdown with an inducible shRNA leads to reduced in vitro growth. (A) Western blot confirmed suppression of FGFR4 with doxycycline induction of the shRNA in RH30 cells. (B) Growth curves for RH30 cell lines with an inducible shRNA directed against FGFR4 showed no measurable difference in growth rate for up to 96 hours. (C) Cell count of RH30 cell lines with an inducible shRNA against FGFR4 treated with or without doxycycline shows significantly reduced growth (41% reduction) with prolonged culture at 13 days (Mann-Whitney test).

Figure 3

FGFR4 suppression leads to inhibition of in vivo growth and lung metastasis. (A) Intramuscular injection of RH30 with inducible anti-FGFR4 shRNA resulted in significantly smaller tumors in the FGFR4-suppressed mice on day 31 using bioluminescent imaging (n = 6 mice per group, Mann-Whitney test). (B) Representative mice with intramuscular injection on day 31. The intensity of tumor cells expressing luciferase is quantified as photons/second/cm²/steridian (p/s/cm²/sr). (C) Representative IVVM images showing decreased RH30 cells in the lungs at 24 hours when FGFR4 was suppressed. (D) Quantification of IVVM early pulmonary metastases at 1 and 24 hours showed significantly fewer malignant cells remaining in the lungs with FGFR4 suppression (normalized mean values ± SEM; n = 5 mice per group; Mann-Whitney test). (E) Representative mice with intravenous injection showed decreased tumor signal in the lungs with FGFR4 suppression on day 74. (F) Intravenous injection of the same cells resulted in significantly fewer lesions in the lungs on day 74 in the FGFR4-suppressed mice (normalized for tumor growth rate by taking the ratio of pulmonary to pelvic tumor signal; n = 10 mice per group, Mann-Whitney test).

Identification of FGFR4 TK domain mutations in human RMS tumors. Our results thus far indicated that increased FGFR4 activity could contribute to tumor progression and metastatic disease in RMS. Therefore, we searched for activating FGFR4 mutations using bidirectional sequencing of all protein coding exons and their intron/exon borders in the same 94 RMS tumors (Supplemental Table 2). We found 14 missense variants, of which 6 were clustered in the TK domain (Figure 4A). Four of the TK domain mutations were localized to codons 535 and 550 (Table 2; Figure 4, A–C; and Supplemental Figure 4). None of these TK missense substitutions were present in the Human Gene Mutation database, COSMIC, or a large sequencing survey of RTK genes (18). A subset of these tumors (50 of 94) had paired germline DNA available. For these, 3 of 3 TK domain mutations were found only in the tumors and not in germline DNA and were therefore somatic (Supplemental Table 2). We did not observe mutations in the FGFR4 TK domain in 9 human RMS cell lines.

Figure 4

FGFR4 TK domain mutations in RMS. (A) Sites of 6 missense substitutions in the FGFR4 TK domain that were identified in RMS tumors (n = 94). Amino acid boundaries for protein domains were defined by the results of a search of the NCBI Conserved Domain database (NCBI CD-Search). Red, signal peptide; blue, transmembrane domain. IG, immunoglobulin-like domain; S, disulfide bond. (B) Wild-type and homozygous mutant alleles for the N535K mutation. (C) Wild-type and mutant alleles for V550E.

Table 2

FGFR4 TK domain mutations observed in primary RMSs

We confirmed that these TK domain mutations were absent in healthy populations by additional bidirectional sequencing of the 2 FGFR4 exons corresponding to codons 507 to 607 in 1,030 multiethnic controls (Supplemental Table 2). These findings suggested an overall TK domain mutation prevalence of 7.5% (7 of 94 tumors; 95% confidence interval, 4.7%–10.2%; Table 2) versus only a single R529Q single nucleotide variant allele observed among the 1,030 healthy controls (0.1%; P = 2.0 × 10^–7 for tumors vs. controls).

Predictive analysis of FGFR4 mutations. To determine the significance of these mutations, we performed multiple predictive analyses for individual amino acid substitutions (see Methods and Supplemental Table 3). These suggested that the 4 FGFR4 missense substitutions at codons 535 and 550 were most likely to result in disruption of protein function (Supplemental Table 3) (22–26). Furthermore, the codon 535 and 550 mutations mapped to adjacent sites within the hinge region on a model of the FGFR4 TK domain (Figure 5A). Codon 535 mutations are predicted to eliminate the FGFR4 R-group hydrogen bonds that inhibit receptor autophosphorylation or regulate conformational dynamics during phosphorylation in the paralog FGFR2 (Figure 5, B–D) (27, 28), while those at codon 550 are predicted to alter the ATP binding cleft (29, 30).

Figure 5

Structural modeling of the FGFR4 codon 535 and 550 mutations. (A) Codon 535 and 550 mutations (red) on a model of the FGFR4 kinase domain. Mutation sites of codons 529 and 554 are also in red. The activation loop (A loop) is yellow, the catalytic loop black, and the nucleotide binding loop blue. (B) Predicted hydrogen bonds for wild-type codon 535. (C and D) Codon 535 mutations could disrupt R-group hydrogen bonds (red dashed lines) between codon 535 and residues H530 and I533. Illustrations in B–D were created with PyMOL (http://www.pymol.org).

Mutations promote FGFR4 autophosphorylation, Stat3 phosphorylation, and activation of cell cycle and DNA replication pathways. We functionally characterized 2 of these 4 TK domain mutations (K535 and E550) to determine whether they would result in constitutive receptor activation. We transduced wild-type human FGFR4 or mutants K535 and E550 into the murine RMS cell line RMS772, which was previously derived from a spontaneous tumor occurring in a mouse model of ERMS (5, 21). RMS772 was chosen because of its low metastatic potential and undetectable Fgfr4 mRNA (5). We confirmed expression of FGFR4 in transduced cell lines at both RNA and protein levels, with FGFR4 protein levels that were comparable with 2 human RMS cell lines (Figure 6, A and B). Of note, both of these mutations resulted in significant FGFR4 receptor autophosphorylation (Figure 6C).

Figure 6

FGFR4 mutations promote FGFR4 autophosphorylation, STAT3 phosphorylation, and activation of cell cycle and DNA replication pathways. (A and B) High levels of RNA (A) and protein (B) for wild-type human FGFR4, FGFR4 K535, and FGFR4 E550 after stable introduction into murine RMS cell line RMS772. Protein levels are comparable to those of human RMS cell lines RH41 and RH28. (C) Immunoblot of FGFR4 and phospho-FGFR4 after immunoprecipitation showed that the mutant forms of FGFR4 were constitutively autophosphorylated. (D) Increased total Stat3 and phospho-Stat3 were observed in FGFR4 mutant cell lines by immunoblot, while mutants also showed decreased phospho-Akt. All transductants expressing human FGFR4 also had less phospho-Erk1/2. (E) Representative GSEA showing enrichment of cell cycle genes in murine RMS cells expressing FGFR4 mutations. Gene names listed at the bottom are at the leading edge of genes ranked by expression enrichment score that also belong to the Cell Cycle KEGG gene set.

Western blot analysis of known FGFR downstream signaling molecules found significant differences in the Stat, Akt, and Mapk/Erk pathways with increased total Stat3 and phospho-Stat3 in both mutant lines (Figure 6D). Interestingly, the mutant cell lines also had a decrease in phospho-Akt and a discernible difference in the Mapk pathway with decreased phospho-Erk1/2 in both wild-type FGFR4 and mutant transductants compared with the vector control (Figure 6D). Other signaling molecules including mTor, S6k, 4Ebp1, and Gsk3β showed no significant differences between transductants (Supplemental Figure 5).

We then examined the global downstream effects of expressing these human mutations in RMS772 cells using gene expression profiling. Gene set enrichment analysis showed that the mutants had significant upregulation of cell cycle and DNA replication gene pathways (false discovery rate [FDR] < 0.01), while cell adhesion pathways and markers of muscle differentiation were diminished (Table 3 and Figure 6E).

Table 3

Altered expression of 6 pathways in cells expressing FGFR4 TK domain mutations by GSEA

FGFR4 mutants increase proliferation, invasion, and metastatic potential. The association of high FGFR4 expression with advanced stage and poor survival and the upregulation of cell cycle genes led us to hypothesize that constitutive FGFR4 activation would be associated with increased growth and result in a metastatic phenotype in RMS. Both the K535 and E550 mutations caused significantly higher growth rates in RMS772 cell lines grown in vitro when compared with wild-type FGFR4 at 72 hours (Figure 7A; P = 0.0071 and 0.0090, respectively). Consistent with these data, subcutaneous injection of the RMS772 transductants into nude mice demonstrated rapid increases in tumor volume for the mutants at 18 days (Figure 7B; both P = 0.0079), providing evidence for increased in vivo growth.

Figure 7

FGFR4 mutations accelerate growth and promote a metastasis phenotype. (A) Growth was significantly higher for cultured mutant RMS772 cell lines (*P < 0.01 versus wild type, Mann-Whitney test). (B) FGFR4 mutants in RMS772 cells significantly enhanced in vivo growth after subcutaneous injection in nude mice (*P < 0.001 versus wild-type, Mann-Whitney test). (C) FGFR4 RMS772 mutants showed 3-fold higher invasion (normalized to the vector control) by a modified Boyden chamber assay (t test with Welch correction). (D) Left: Representative IVVM images show the presence of RMS cells in the lung (1 and 24 hour time points shown). Right: Quantification of pulmonary lesions at 1 and 24 hours showed the persistence of RMS cells transduced with the 2 mutants (normalized to the vector control at each time point; n = 5 mice per group; Mann-Whitney test). (E) Gross pulmonary metastases from RMS772 cells at 21 days (comparison using Student’s t test). (F) Left: Kaplan-Meier survival analysis demonstrated a significantly poorer survival in mice injected intravenously with RMS772 cell lines transduced with the 2 mutants. Right: Pulmonary lesions in representative animals injected with mutants are shown.

Using a modified Boyden chamber invasion assay, we found 3-fold enhanced invasiveness associated with FGFR4 mutant cell lines compared with the vector control or wild-type FGFR4 in RMS772 (Figure 7C; P = 0.002 for vector versus K535; P = 0.005 for E550). Cellular arrest in the lungs and early metastasis was assayed by IVVM in mice after intravenous injection of fluorescently labeled RMS772 transductants. IVVM at 1 hour after injection demonstrated an equal number of cells for all 4 transductants. However, only the mutations significantly enhanced the presence of foci in the lungs after 24 hours (46-fold and 22-fold increase for K535 and E550, respectively, compared with vector control), although there was a small (2.6-fold) but significant increase in the number of foci caused by the wild-type human FGFR4 (Figure 7D). To determine whether these differences in growth and invasion influence in vivo metastatic potential, RMS772 cells expressing wild-type human FGFR4 or mutations were introduced into nude mice intravenously. At 3 weeks, mutant cell lines produced significantly more gross pulmonary metastases compared with those expressing wild-type FGFR4 (Figure 7E). In a follow-up experiment, Kaplan-Meier survival analysis demonstrated earlier mortality for mice injected with mutant cell lines (Figure 7F; P < 0.0001 by log-rank test for trend; median survival: E550, 19 days; K535, 29 days; wild type, 59 days; vector control, 78 days). Necroscopy confirmed large tumor burden in the lungs due to metastatic disease (Figure 7F).

To validate these results in an independent cell line, we transduced NIH 3T3 cells with the same constructs and repeated the in vivo subcutaneous growth and intravenous experimental metastasis assays. We again observed rapid growth in mice receiving subcutaneous NIH 3T3 cells transduced with FGFR4 K535 or E550, compared with no growth among the controls at 18 days (Figure 8A). Intravenous injection of 3T3 cell lines similarly resulted in earlier mortality due to metastatic disease in animals receiving mutant FGFR4 cells (Figure 8B; P < 0.0001 by log-rank test for trend).

Figure 8

FGFR4 mutations transform 3T3 cells. (A) FGFR4 mutants transformed NIH 3T3 cells and promoted in vivo tumor growth after subcutaneous injection in nude mice (*P < 0.001 versus wild type, Mann-Whitney test). Representative animals injected with 3T3 cells transduced with the respective constructs are shown at day 18. In these animals, tumors were observable in 4 of 5 mice (K535) and 5 of 5 mice (E550). (B) Kaplan-Meier survival analysis in mice after intravenous injection of NIH 3T3 cells with wild-type and mutant FGFR4 (n = 5 mice per group).

Effect of mutations on cell cycle, apoptosis, and FGFR inhibition. Our results predicted mutational activation of an oncogenic pathway. Consequently, we tested whether these mutations would result in increased survival under adverse conditions and whether RMS tumor cells become dependent upon FGFR4 activation for survival, potentially making them more sensitive to a FGFR inhibitor. Despite the increased proliferation rate of mutant cell lines grown in 10% serum (Figure 7A), there were no differences in the distribution in the cell cycle compared to the empty vector or wild-type FGFR4 (Figure 9A). However, both vector and wild-type controls had significantly higher proportions of apoptotic cells under conditions of serum starvation compared with mutant cell lines, as demonstrated by an increased subG1 fraction (Figure 9B).

Figure 9

Oncogene dependence and inhibition of FGFR4. (A) Cell cycle analysis of RMS772 in 10% serum. (B) Cell cycle analysis in the absence of serum showed increased apoptotic cells (subG1 fraction) for the vector and wild-type FGFR4 RMS772 cell lines. (C) Inhibition of FGFR4 phosphorylation after PD173074 treatment at 3 hours. (D) RMS772 survival after 48 hour treatment with PD173074. IC₅₀ values for RMS772 vector control, wild-type FGFR4, K535, and E550 are presented. (E) Increased subG1 fraction of mutant RMS772 cells lines after 24 hours of treatment with 20 μM PD173074. (F) Increased caspase-3 activation of FGFR4 mutant RMS772 cell lines after 24 hours of treatment with 20 μM PD173074.

To demonstrate oncogene dependence upon mutational activation of FGFR4, the 4 cell lines were treated with the FGFR inhibitor PD173074 (29, 31). We confirmed that PD173074 treatment reduced phospho-FGFR4 (normalized to total FGFR4) for both K535 and E550 mutants (Figure 9C). The IC₅₀ after 48 hours of treatment decreased from 12.7 μM and 11.8 μM in the vector control and wild-type FGFR4 cell lines to 8.2 μM and 5.9 μM for K535 and E550, respectively (Figure 9D). Increased apoptosis with PD173074 was apparent in both mutant cell lines, as evidenced by an increased SubG1 fraction (Figure 9E) and increased activated caspase-3 (Figure 9F).

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, 4, 6), 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 (6–8, 32). 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 in ARMS.

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 expression (35). 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 (36–39). 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, 6). 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 (40–43). 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 (44–46). 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 (36–39, 47). 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, KRAS, and HRAS), reported in a small number of ERMS (33, 48–50). 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 (13, 18, 19, 51–54). 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, 51). 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, 12, 14, 16, 17, 19, 55, 56).

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, 41, 57). In contrast, FGFR1–FGFR3 are all able to transform different cells through either overexpression alone or overexpression with FGF stimulation (57, 58), 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 (59–61). 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, 20).

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.

View Supplemental data

The authors thank Bryan Clary for the pMSCVzeo retroviral luciferase gene construct and Kent Hunter (NCI) and Mauro Tiso (University of Pittsburgh, Pittsburgh, Pennsylvania, USA) for useful discussions. This work was supported in part by the Cooperative Human Tissue Network, which is funded by the NCI, and by the intramural programs of the NIH (NHLBI and NCI).

Address correspondence to: Javed Khan, Advanced Technology Center, 8717 Grovemont Circle, Bethesda, Maryland 20892-4605, USA. Phone: (301) 435-2937; Fax: (301) 480-0314; E-mail: khanjav@mail.nih.gov.

Conflict of interest: J. Khan, A.T. Cheuk, and J.G. Taylor VI are named as inventors with the NIH on a provisional patent application for the therapeutic treatment of cancers overexpressing FGFR4 or harboring FGFR4 mutations.

Reference information: J. Clin. Invest.119:3395–3407 (2009). doi:10.1172/JCI39703.

Stephen J. Qualman is deceased.

1. Parham, D.M., Ellison, D.A. 2006. Rhabdomyosarcomas in adults and children: an update. Arch. Pathol. Lab. Med. 130:1454-1465.
   View this article via: PubMed Google Scholar
2. Raney, R.B., et al. 2001. Rhabdomyosarcoma and undifferentiated sarcoma in the first two decades of life: a selective review of intergroup rhabdomyosarcoma study group experience and rationale for Intergroup Rhabdomyosarcoma Study V. J. Pediatr. Hematol. Oncol. 23:215-220.
   View this article via: CrossRef PubMed Google Scholar
3. Khan, J., et al. 2001. Classification and diagnostic prediction of cancers using gene expression profiling and artificial neural networks. Nat. Med. 7:673-679.
   View this article via: CrossRef PubMed Google Scholar
4. Baird, K., et al. 2005. Gene expression profiling of human sarcomas: insights into sarcoma biology. Cancer Res. 65:9226-9235.
   View this article via: CrossRef PubMed Google Scholar
5. Yu, Y., et al. 2004. Expression profiling identifies the cytoskeletal organizer ezrin and the developmental homeoprotein Six-1 as key metastatic regulators. Nat. Med. 10:175-181.
   View this article via: CrossRef PubMed Google Scholar
6. Davicioni, E., et al. 2006. Identification of a PAX-FKHR gene expression signature that defines molecular classes and determines the prognosis of alveolar rhabdomyosarcomas. Cancer Res. 66:6936-6946.
   View this article via: CrossRef PubMed Google Scholar
7. Lagha, M., et al. 2008. Pax3 regulation of FGF signaling affects the progression of embryonic progenitor cells into the myogenic program. Genes Dev. 22:1828-1837.
   View this article via: CrossRef PubMed Google Scholar
8. Zhao, P., et al. 2006. Fgfr4 is required for effective muscle regeneration in vivo. Delineation of a MyoD-Tead2-Fgfr4 transcriptional pathway. J. Biol. Chem. 281:429-438.
   View this article via: CrossRef PubMed Google Scholar
9. Zhao, P., Hoffman, E.P. 2004. Embryonic myogenesis pathways in muscle regeneration. Dev. Dyn. 229:380-392.
   View this article via: CrossRef PubMed Google Scholar
10. Eswarakumar, V.P., Lax, I., Schlessinger, J. 2005. Cellular signaling by fibroblast growth factor receptors. Cytokine Growth Factor Rev. 16:139-149.
    View this article via: CrossRef PubMed Google Scholar
11. Bellus, G.A., et al. 1995. A recurrent mutation in the tyrosine kinase domain of fibroblast growth factor receptor 3 causes hypochondroplasia. Nat. Genet. 10:357-359.
    View this article via: CrossRef PubMed Google Scholar
12. Kan, S.H., et al. 2002. Genomic screening of fibroblast growth-factor receptor 2 reveals a wide spectrum of mutations in patients with syndromic craniosynostosis. Am. J. Hum. Genet. 70:472-486.
    View this article via: CrossRef PubMed Google Scholar
13. Ding, L., et al. 2008. Somatic mutations affect key pathways in lung adenocarcinoma. Nature. 455:1069-1075.
    View this article via: CrossRef PubMed Google Scholar
14. Rand, V., et al. 2005. Sequence survey of receptor tyrosine kinases reveals mutations in glioblastomas. Proc. Natl. Acad. Sci. U. S. A. 102:14344-14349.
    View this article via: CrossRef PubMed Google Scholar
15. Pollock, P.M., et al. 2007. Frequent activating FGFR2 mutations in endometrial carcinomas parallel germline mutations associated with craniosynostosis and skeletal dysplasia syndromes. Oncogene. 26:7158-7162.
    View this article via: CrossRef PubMed Google Scholar
16. Byron, S.A., et al. 2008. Inhibition of activated fibroblast growth factor receptor 2 in endometrial cancer cells induces cell death despite PTEN abrogation. Cancer Res. 68:6902-6907.
    View this article via: CrossRef PubMed Google Scholar
17. Dutt, A., et al. 2008. Drug-sensitive FGFR2 mutations in endometrial carcinoma. Proc. Natl. Acad. Sci. U. S. A. 105:8713-8717.
    View this article via: CrossRef PubMed Google Scholar
18. Greenman, C., et al. 2007. Patterns of somatic mutation in human cancer genomes. Nature. 446:153-158.
    View this article via: CrossRef PubMed Google Scholar
19. Stephens, P., et al. 2005. A screen of the complete protein kinase gene family identifies diverse patterns of somatic mutations in human breast cancer. Nat. Genet. 37:590-592.
    View this article via: CrossRef PubMed Google Scholar
20. Ezzat, S., Zheng, L., Zhu, X.F., Wu, G.E., Asa, S.L. 2002. Targeted expression of a human pituitary tumor-derived isoform of FGF receptor-4 recapitulates pituitary tumorigenesis. J. Clin. Invest. 109:69-78.
    View this article via: JCI PubMed Google Scholar
21. Sharp, R., et al. 2002. Synergism between INK4a/ARF inactivation and aberrant HGF/SF signaling in rhabdomyosarcomagenesis. Nat. Med. 8:1276-1280.
    View this article via: CrossRef PubMed Google Scholar
22. Yue, P., Moult, J. 2006. Identification and analysis of deleterious human SNPs. J. Mol. Biol. 356:1263-1274.
    View this article via: CrossRef PubMed Google Scholar
23. Yue, P., Melamud, E., Moult, J. 2006. SNPs3D: candidate gene and SNP selection for association studies. BMC Bioinformatics. 7:166.
    View this article via: PubMed Google Scholar
24. Sunyaev, S., et al. 2001. Prediction of deleterious human alleles. Hum. Mol. Genet. 10:591-597.
    View this article via: CrossRef PubMed Google Scholar
25. Ramensky, V., Bork, P., Sunyaev, S. 2002. Human non-synonymous SNPs: server and survey. Nucleic Acids Res. 30:3894-3900. .
    View this article via: PubMed Google Scholar
26. Ng, P.C., Henikoff, S. 2003. SIFT: Predicting amino acid changes that affect protein function. Nucleic Acids Res. 31:3812-3814.
    View this article via: CrossRef PubMed Google Scholar
27. Chen, H., et al. 2007. A molecular brake in the kinase hinge region regulates the activity of receptor tyrosine kinases. Mol. Cell. 27:717-730.
    View this article via: CrossRef PubMed Google Scholar
28. Lew, E.D., Bae, J.H., Rohmann, E., Wollnik, B., Schlessinger, J. 2007. Structural basis for reduced FGFR2 activity in LADD syndrome: Implications for FGFR autoinhibition and activation. Proc. Natl. Acad. Sci. U. S. A. 104:19802-19807.
    View this article via: CrossRef PubMed Google Scholar
29. Mohammadi, M., et al. 1998. Crystal structure of an angiogenesis inhibitor bound to the FGF receptor tyrosine kinase domain. EMBO J. 17:5896-5904.
    View this article via: CrossRef PubMed Google Scholar
30. Torkamani, A., Schork, N.J. 2008. Prediction of cancer driver mutations in protein kinases. Cancer Res. 68:1675-1682.
    View this article via: CrossRef PubMed Google Scholar
31. Grand, E.K., Chase, A.J., Heath, C., Rahemtulla, A., Cross, N.C. 2004. Targeting FGFR3 in multiple myeloma: inhibition of t(4;14)-positive cells by SU5402 and PD173074. Leukemia. 18:962-966.
    View this article via: CrossRef PubMed Google Scholar
32. Marics, I., Padilla, F., Guillemot, J.F., Scaal, M., Marcelle, C. 2002. FGFR4 signaling is a necessary step in limb muscle differentiation. Development. 129:4559-4569.
    View this article via: PubMed Google Scholar
33. Galili, N., et al. 1993. Fusion of a fork head domain gene to PAX3 in the solid tumour alveolar rhabdomyosarcoma. Nat. Genet. 5:230-235.
    View this article via: CrossRef PubMed Google Scholar
34. Fredericks, W.J., et al. 1995. The PAX3-FKHR fusion protein created by the t(2;13) translocation in alveolar rhabdomyosarcomas is a more potent transcriptional activator than PAX3. Mol. Cell. Biol. 15:1522-1535.
    View this article via: PubMed Google Scholar
35. Sorensen, P.H., et al. 2002. PAX3-FKHR and PAX7-FKHR gene fusions are prognostic indicators in alveolar rhabdomyosarcoma: a report from the children’s oncology group. J. Clin. Oncol. 20:2672-2679.
    View this article via: CrossRef PubMed Google Scholar
36. Chen, Y., et al. 2008. Oncogenic mutations of ALK kinase in neuroblastoma. Nature. 455:971-974.
    View this article via: CrossRef PubMed Google Scholar
37. George, R.E., et al. 2008. Activating mutations in ALK provide a therapeutic target in neuroblastoma. Nature. 455:975-978.
    View this article via: CrossRef PubMed Google Scholar
38. Janoueix-Lerosey, I., et al. 2008. Somatic and germline activating mutations of the ALK kinase receptor in neuroblastoma. Nature. 455:967-970.
    View this article via: CrossRef PubMed Google Scholar
39. Mosse, Y.P., et al. 2008. Identification of ALK as a major familial neuroblastoma predisposition gene. Nature. 455:930-935.
    View this article via: CrossRef PubMed Google Scholar
40. Bromberg, J.F., et al. 1999. Stat3 as an oncogene. Cell. 98:295-303.
    View this article via: CrossRef PubMed Google Scholar
41. Hart, K.C., et al. 2000. Transformation and Stat activation by derivatives of FGFR1, FGFR3, and FGFR4. Oncogene. 19:3309-3320.
    View this article via: CrossRef PubMed Google Scholar
42. Chen, C.L., et al. 2007. Signal transducer and activator of transcription 3 is involved in cell growth and survival of human rhabdomyosarcoma and osteosarcoma cells. BMC Cancer. 7:111.
    View this article via: PubMed Google Scholar
43. Huang, S. 2007. Regulation of metastases by signal transducer and activator of transcription 3 signaling pathway: clinical implications. Clin. Cancer Res. 13:1362-1366.
    View this article via: CrossRef PubMed Google Scholar
44. Yoeli-Lerner, M., et al. 2005. Akt blocks breast cancer cell motility and invasion through the transcription factor NFAT. Mol. Cell. 20:539-550.
    View this article via: CrossRef PubMed Google Scholar
45. Liu, H., et al. 2006. Mechanism of Akt1 inhibition of breast cancer cell invasion reveals a protumorigenic role for TSC2. Proc. Natl. Acad. Sci. U. S. A. 103:4134-4139.
    View this article via: CrossRef PubMed Google Scholar
46. Dufour, C., et al. 2008. FGFR2-Cbl interaction in lipid rafts triggers attenuation of PI3K/Akt signaling and osteoblast survival. Bone. 42:1032-1039.
    View this article via: CrossRef PubMed Google Scholar
47. Demetri, G.D., et al. 2002. Efficacy and safety of imatinib mesylate in advanced gastrointestinal stromal tumors. N. Engl. J. Med. 347:472-480.
    View this article via: CrossRef PubMed Google Scholar
48. Chen, Y., et al. 2006. Mutations of the PTPN11 and RAS genes in rhabdomyosarcoma and pediatric hematological malignancies. Genes Chromosomes Cancer. 45:583-591.
    View this article via: CrossRef PubMed Google Scholar
49. Stratton, M.R., Fisher, C., Gusterson, B.A., Cooper, C.S. 1989. Detection of point mutations in N-ras and K-ras genes of human embryonal rhabdomyosarcomas using oligonucleotide probes and the polymerase chain reaction. Cancer Res. 49:6324-6327.
    View this article via: PubMed Google Scholar
50. Yoo, J., Robinson, R.A. 1999. H-ras and K-ras mutations in soft tissue sarcoma: comparative studies of sarcomas from Korean and American patients. Cancer. 86:58-63.
    View this article via: CrossRef PubMed Google Scholar
51. Marks, J.L., et al. 2007. Mutational analysis of EGFR and related signaling pathway genes in lung Adenocarcinomas identifies a novel somatic kinase domain mutation in FGFR4. PLoS ONE. 2:e426.
    View this article via: PubMed Google Scholar
52. Wood, L.D., et al. 2007. The genomic landscapes of human breast and colorectal cancers. Science. 318:1108-1113.
    View this article via: CrossRef PubMed Google Scholar
53. Parsons, D.W., et al. 2008. An integrated genomic analysis of human glioblastoma multiforme. Science. 321:1807-1812.
    View this article via: CrossRef PubMed Google Scholar
54. Jones, S., et al. 2008. Core signaling pathways in human pancreatic cancers revealed by global genomic analyses. Science. 321:1801-1806.
    View this article via: CrossRef PubMed Google Scholar
55. Berndt, I., et al. 1998. A new hot spot for mutations in the ret protooncogene causing familial medullary thyroid carcinoma and multiple endocrine neoplasia type 2A. J. Clin. Endocrinol. Metab. 83:770-774.
    View this article via: CrossRef PubMed Google Scholar
56. Bolino, A., et al. 1995. RET mutations in exons 13 and 14 of FMTC patients. Oncogene. 10:2415-2419.
    View this article via: PubMed Google Scholar
57. Wang, J.K., Gao, G., Goldfarb, M. 1994. Fibroblast growth factor receptors have different signaling and mitogenic potentials. Mol. Cell. Biol. 14:181-188.
    View this article via: PubMed Google Scholar
58. Li, Z., et al. 2001. The myeloma-associated oncogene fibroblast growth factor receptor 3 is transforming in hematopoietic cells. Blood. 97:2413-2419.
    View this article via: CrossRef PubMed Google Scholar
59. Bange, J., et al. 2002. Cancer progression and tumor cell motility are associated with the FGFR4 Arg(388) allele. Cancer Res. 62:840-847.
    View this article via: PubMed Google Scholar
60. Morimoto, Y., et al. 2003. Single nucleotide polymorphism in fibroblast growth factor receptor 4 at codon 388 is associated with prognosis in high-grade soft tissue sarcoma. Cancer. 98:2245-2250.
    View this article via: CrossRef PubMed Google Scholar
61. Stadler, C.R., Knyazev, P., Bange, J., Ullrich, A. 2006. FGFR4 GLY388 isotype suppresses motility of MDA-MB-231 breast cancer cells by EDG-2 gene repression. Cell Signal. 18:783-794.
    View this article via: CrossRef PubMed Google Scholar
62. Barr, F.G., Xiong, Q.B., Kelly, K. 1995. A consensus polymerase chain reaction-oligonucleotide hybridization approach for the detection of chromosomal translocations in pediatric bone and soft tissue sarcomas. Am. J. Clin. Pathol. 104:627-633.
    View this article via: PubMed Google Scholar
63. Stone, E.A., Sidow, A. 2005. Physicochemical constraint violation by missense substitutions mediates impairment of protein function and disease severity. Genome Res. 15:978-986.
    View this article via: CrossRef PubMed Google Scholar
64. Ngo, V.N., et al. 2006. A loss-of-function RNA interference screen for molecular targets in cancer. Nature. 441:106-110.
    View this article via: CrossRef PubMed Google Scholar
65. Subramanian, A., et al. 2005. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. U. S. A. 102:15545-15550.
    View this article via: CrossRef PubMed Google Scholar
66. Khanna, C., et al. 2004. The membrane-cytoskeleton linker ezrin is necessary for osteosarcoma metastasis. Nat. Med. 10:182-186.
    View this article via: PubMed Google Scholar