c-Myc Alteration Determines the Therapeutic Response to FGFR Inhibitors

Aberration of oncogenic kinase often confers dependency of cancer cells on a particular kinase for survival and proliferation (1, 2). This phenomenon has gained increasing recognitions over the past decade and provided the rationale for targeted therapeutic strategies, in particular selective tyrosine kinase inhibitors, such as crizotinib in treating lung adenocarcinoma with ALK translocations, vemurafenib for melanoma harboring activating mutations of BRAF, and lapatinib for HER2-amplified breast cancer (3–7). Lately, emerging evidence has suggested that kinase signaling is often stringently associated with specific downstream effectors. Kinase inhibition cannot achieve therapeutic outcomes before being delivered to these downstream oncogenic effectors, particularly those regulating ultimate cellular processes, such as apoptosis or cell-cycle arrest (8). For example, it has been shown that activation of proapoptotic molecule Bim is required for achieving the therapeutic outcomes of gefitinib in EGFR-mutant cancer (9, 10). Likewise, alteration of downstream effector eIF4E and c-Myc is critical for determining the therapeutic response of rapamycin and c-Met inhibitors, respectively (11, 12). To identify kinase-associated downstream effectors may reveal “Achilles' heel” of the kinase-addicted cancer and provide important therapeutic implications such as indicating the response for kinase inhibitors (11, 13–17).

Overactivation of the tyrosine kinase FGFR occurs in a broad spectrum of solid tumors in the forms of FGFR gene amplifications, somatic mutations, or translocations. Aberrant FGFR signaling drives oncogenic growth of tumor subsets, especially those lacking effective treatments, such as 20% squamous non–small cell lung carcinoma and 4% triple-negative breast cancer, etc. FGFR has been validated as an attractive target for cancer treatment, and several selective FGFR inhibitors are undergoing clinical studies. It will be interesting to know whether there exists key downstream effector(s) that converges FGFR signaling and determines the therapeutic response of FGFR-targeted therapy.

In the present study, we have discovered that the transcription factor c-Myc functions as a key downstream effector converging FGFR signaling to drive cell-cycle progression. Mechanistically, FGFR activates MEK-ERK signaling to sustain c-Myc stability via suppressing c-Myc phosphorylation at threonine Thr58. FGFR inhibition in FGFR-addicted cancer cells increases c-Myc phosphorylation at Thr58 and, in turn, facilitates c-Myc degradation and arrests cells at G₁ phase. Moreover, monitoring c-Myc level enables us to inspect the response to FGFR inhibitors, to exclude intrinsic resistance, and to monitor the emergence of acquired resistance.

NCI-H1581, DMS-114, NCI-H520, NCI-H2444, KG1, KATOIII, SNU16, and NCI-H716 cells were obtained from the American Type Culture Collection (ATCC). UMUC14 and MFM-223 cells were obtained from European Collection of Cell Cultures (ECACC). RT112, OPM2, and BaF3 cells were obtained from Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH (DSMZ). SUM52PE was obtained from Asterand Company. All the cell lines used in this study were obtained during August 2012 to March 2013 and were maintained in appropriate medium as suppliers suggested. All the cell lines were authenticated via short tandem repeats analysis by Genesky Biopharma Technology (last tested in 2015).

FGFR inhibitors (BGJ398, AZD4547), MEK inhibitor (PD0325901), AKT inhibitor (GSK690693, MK2206), proteasome inhibitor MG132, and inhibitor of the BET family of bromodomain proteins (+)-JQ1 and (-)-JQ1 were purchased from Selleck Chemicals. For in vivo studies, BGJ398 and AZD4547 were obtained from Melone Pharmaceutical Co., Ltd. All these reagents were dissolved in DMSO for in vitro studies and in 1% Tween-80 (AZD4547) or acetic acid/acetate buffer pH 4.6/PEG300 (1:1) (NVP-BGJ398) for in vivo studies.

To generate cells resistant to FGFR inhibitors, NCI-H1581 cell was exposed to AZD4547 at concentrations increasing stepwise from 30 nmol/L to 1 μmol/L when the cells resumed growth kinetics similar to the untreated parental cells. After about 4 months, a resistant subpopulation termed as NCI-H1581AR was obtained, and the established cells were maintained in the presence of 1 μmol/L AZD4547.

The retroviral constructs MSCV-MYC and empty vectors were obtained from Addgene. MSCV-MYCT58A was constructed with a site-directed mutagenesis kit (Sbsbio). pBABE-TEL-FGFR1, pBABE-TEL-FGFR3, and pBABE-TEL-FGFR4 were constructed using recombinant polymerase chain reaction and subsequently were subcloned into the pBABE-puro vector. To generate cells with stable expression, the plasmids were transfected into amphotropic phoenix 293T packaging cells with Lipofectamine 2000 (Invitrogen). After 48 hours, virus-containing medium was collected, filtered, and used to infect host cells in the presence of 6 μg/mL of polybrene. The stable transfectants were obtained by selection with 2 μg/mL puromycin (Sigma) for 2 weeks followed by immunoblotting validation.

Four- to 6-week-old nu/nu athymic BALB/cA or SCID mice were obtained from Shanghai Laboratory Animal Center, Chinese Academy of Sciences (Shanghai, China). NCI-H1581, NCI-H716, NCI-H2444, UMUC14, NCI-H1581AR, NCI-H1581 c-MycWT, or NCI-H1581 c-MycT58A cancer cells (1 × 10⁷) were suspended in 200 μL PBS and inoculated subcutaneously on the right flank of BALB/cA nude or SCID mice. When the volume of the tumor xenograft reached approximately 300 to 500 mm³, it was excised and cut into approximately 1.5 mm³ segments, which were further implanted subcutaneously via trocar needle into nude mice. When the tumor reached proper volume, mice were randomized divided into vehicle control and treated groups (n = 6 for treated group, n = 12 for vehicle group). For efficacy studies, mice were administered drugs using the indicated doses. The average tumor diameter (two perpendicular axes of the tumor were measured) was measured in control and treated groups with vernier calipers twice a week. For statistical analysis, data were analyzed by the unpaired two-tailed Student t test, and P value of <0.05 was considered statistically significant. To prepare lysates for immunoblotting, mice were sacrificed and tumor tissues were resected and homogenized in cold RIPA lysis buffer (Beyotime) supplemented with protease and phosphatase inhibitors (Roche) and then processed for immunoblotting.

Animal studies using patient-derived xenograft (PDX) models were conducted by Crown Bioscience and in strict accordance with the Guide for the Care and Use of Laboratory Animals of the NIH.

The difference between experimental groups in in vitro and in vivo studies was compared using unpaired two-tailed Student t test analysis. P < 0.05 was considered statistically significant.

The experimental procedures involving animal studies strictly adhered to the Institutional Animal Care and Use Committee guidelines and Animal Welfare policies of Shanghai Institute of Materia Medica.

We firstly selected a panel of cancer cell lines bearing FGFR aberration that covers the frequently occurring oncogenic forms of FGFR, namely gene amplifications, activating mutations, and chromosomal translocations (18, 19). BGJ398 and AZD4547, two most advanced pan-FGFR inhibitors in clinical study, were chosen to selectively inhibit FGFR signaling. Among 13 cancer cell lines, 11 were sensitive to BGJ398 and AZD4547, such as FGFR1-amplified NCI-H1581 and DMS114 cells, FGFR2-amplified KATOIII cells, and FGFR3-mutated UMUC14 cells (refs. 20, 21; Fig. 1A; Supplementary Fig. S1A; and Table 1). In contrast, NCI-H2444 cells known to harbor KRAS G12V mutation and NCI-H520 cells barely responded to FGFR inhibitors, regardless of validated FGFR1 amplification in these cells (refs. 20–22; Fig. 1A; Supplementary Fig. S1A). These two cell lines were defined as nonresponsive subset to FGFR inhibition in this study.

We then intended to understand the reason accounting for this response difference. Accumulated evidence has suggested that cell-cycle arrest and apoptosis constitute the two major cellular events leading to the impaired cancer cell proliferation driven by kinase signaling (17, 23–27). We measured cell-cycle distribution and apoptosis occurrence upon FGFR inhibition to probe biological processes contributing to impaired cells proliferation. The pan-FGFR inhibitor BGJ398 and AZD4547 induced a significant G₁ phase cell-cycle arrest in all 11 tested FGFR-responsive cancer cells (Fig. 1B and C; Supplementary Fig. S1B). In contrast, both compounds barely had impacts on cell-cycle distribution in NCI-H2444 and NCI-H520 cells (Fig. 1B and C). In addition, no obvious sub-G₁ cell population or featured PARP cleavage was observed after BGJ398 or AZD4547 treatment in FGFR-dependent cancer cells (Supplementary Fig. S1C), largely excluding the occurrence of FGFR inhibition–induced apoptosis. We therefore concluded that FGFR inhibition led to G₁ phase cell-cycle arrest in FGFR-dependent cells.

We then focused on cell-cycle regulation to probe signaling molecules associated with FGFR activation. Cell-cycle regulators involved in G₁–S cell-cycle transition were probed upon FGFR inhibition in FGFR1-responsive cancer cells (NCI-H1581 and DMS-114), with NCI-H2444 and NCI-H520 cells as negative control. Phosphorylation of FGFR substrate 2α (FRS2α), a key adaptor protein in the downstream of FGFR, was examined as a surrogate for FGFR phosphorylation in FGFR1-amplified context (18, 28, 29). After treatment with BGJ398 or AZD4547 for indicated time (2, 6, 12, or 24 hours), p-FRS2α was similarly suppressed in all cell lines in spite of their diverse response in cell growth, which was also the case for the downstream molecules p-ERK and p-PLCγ (Fig. 2A; Supplementary Fig. S2A). It suggested that FGFR-related signaling inhibition could not necessarily lead to cell growth inhibition. Consistently, the alteration of G₁–S cell-cycle transition regulators, such as p21, p27, and p-RB, varied among FGFR1-addicted cell lines. Among all the tested molecules, we noticed that only the oncogenic transcription factor c-Myc unanimously showed a rapid and sustained reduction upon FGFR inhibition in NCI-H1581 and DMS-114 cells (Fig. 2A; Supplementary Fig. S2A). In contrast, c-Myc level remained intact in the nonresponsive NCI-H520 and NCI-H2444 cells in spite of obvious p-FRS2α inhibition by the treatment (Fig. 2A; Supplementary Fig. S2A). To ascertain whether FGFR inhibition caused c-Myc reduction in FGFR-addicted cancer cells is a common event in FGFR aberrant context, we also examined FGFR1-translocated KG1 cell, FGFR2-amplified cancer cell lines (KATOIII, NCI-H716, SUM52PE, MFM-223), and FGFR3-amplified or -mutated cancer cell lines (RT112, UMUC14). These cell lines consistently exhibited similar c-Myc reduction upon FGFR1, FGFR2, or FGFR3 signaling suppression (Supplementary Fig. S2B; Supplementary Fig. S3A–S3D; Supplementary Fig. S4A and S4B).

For further confirmation, we generated “gain-of-FGFR-addiction” cell lines using mouse pro-B BaF3 cells, a cell line known to intrinsically depend on IL3 for survival, and introduction of oncogenic kinase enables cell growth independent of IL3 (30, 31). We introduced TEL-FGFR1, TEL-FGFR3, or TEL-FGFR4 fusion, the oncogenic forms of FGFR gene, into BaF3 cells. As expected, TEL-FGFR–transfected cells exhibited a profound sensitivity to BGJ398 or AZD4547 (Supplementary Fig. S5A), indicating gain of FGFR dependency. Consistently, FGFR signaling inhibition caused dramatic downregulation of c-Myc in these cells (Supplementary Fig. S5B). These data indicate that c-Myc may function as a key downstream effector that dictates the blockade of G₁–S transition upon FGFR inhibition in FGFR-addicted cancer.

Further, we determined whether c-Myc played a critical role in FGFR-driven cell proliferation using c-Myc–specific siRNAs. Reduction of c-Myc level in FGFR1-amplified NCI-H1581 cell suppressed cell survival and induced G₁ phase arrest (Fig. 2B; Supplementary Fig. S2C and S2D). Likewise, we treated DMS114 cells with (+)-JQ1, a chemical compound that inhibits c-Myc expression by downregulating c-Myc transcription via inhibition of BET family of bromodomain proteins. (-)-JQ1, the enantiomer of (+)-JQ1, that lost ability to downregulate c-Myc transcription was used as a negative control (32). As shown in Supplementary Fig. S2E, (+)-JQ1 dramatically inhibited the survival of DMS114 cells. Consistently, c-Myc knockdown significantly suppressed cell survival and induced G₁ cell-cycle arrest in FGFR2-amplified KATOIII cell and cell survival in FGFR3-mutated UMUC14 cell (Fig. 2C; Supplementary Fig. S2C and S2D; Supplementary Fig. S4C). Moreover, AZD4547 treatment did not further inhibit cell survival in c-Myc knockdown cells compared with c-Myc knockdown alone (Fig. 2B and C, left). These data suggested that c-Myc inhibition was sufficient to recapitulate the proliferation inhibition caused by FGFR inhibitors. We speculated that c-Myc functioned as a dominant downstream effector required for FGFR signaling in FGFR-dependent cells.

Downregulation of c-Myc could result from protein stability change or transcriptional reduction upon FGFR inhibition. The rapid reduction of c-Myc suggested that the downregulation of c-Myc by FGFR inhibition may be mainly due to protein degradation. We then used the proteasome inhibitor MG-132 to examine its impact on FGFR inhibition caused c-Myc downregulation. The results showed that the elimination of c-Myc by FGFR inhibition was reversed by MG-132 (Fig. 3A). In addition, we measured the mRNA level of c-Myc and did not observe a significant reduction of c-Myc mRNA within 24 hours of FGFR inhibitor treatment (Supplementary Fig. S6A, data not shown). These results indicated that FGFR inhibition–induced c-Myc downregulation was mainly due to c-Myc degradation. We hence examined c-Myc phosphorylation at Thr58, a key event for ubiquitin-mediated degradation of c-Myc (33). The results showed that FGFR inhibition resulted in immediate elevation of c-Myc Thr58 phosphorylation, which may trigger c-Myc degradation. Following instant increase of c-Myc Thr58 phosphorylation, we observed the decline of c-Myc Thr58 phosphorylation which was presumably resulted from the reduction of total c-Myc (Fig. 3B). Mutation of Thr 58 to alanine (T58A) results in a stable c-Myc protein and is no longer a substrate for ubiquitination (33). For further conformation, wild-type c-Myc or undegradable c-Myc T58A mutant was ectopically expressed in NCI-H1581 and SUM52PE cells. As expected, FGFR inhibitor reduced ectopically expressed c-Myc wild-type but not c-Myc T58A mutant (Fig. 3C; Supplementary Fig. S6B and S6C). Accordingly, the c-Myc T58A mutant significantly reversed antiproliferative effect of FGFR inhibitors, whereas cells transfected with wild-type c-Myc showed similar sensitivity as parental cells (Fig. 3C; Supplementary Fig. S6B and S6C). Consistent with this in vitro observation, xenograft model of NCI-H1581 cells that expressed wild-type c-Myc showed the comparable response to AZD4547 as that of NCI-H1581 parental xenograft model. In contrast, stable expression of nondegradable c-MycT58A mutant significantly reversed antitumor effect of AZD4547 in vivo (Figs. 3D and 4A). Thus, FGFR activation is essential for maintaining c-Myc stability to drive cell growth, and blockage of FGFR signaling causes rapid c-Myc degradation and impedes cell growth.

Further, we explored which signaling pathway was involved in regulating the c-Myc protein stability in FGFR aberrant cellular context. MAPK/ERK, PI3K–AKT, STAT3, and PLCγ are known as major downstream pathways proceeding activated FGFR signaling (18). Notably, upon BGJ398 or AZD4547 treatment, the phosphorylation of ERK and PLCγ was unanimously inhibited in all tested FGFR-addicted cancer cells (Fig. 2a; Supplementary Fig. S2A, S3, S4A–4B). Suppression of p-AKT and p-STAT3 was only observed in FGFR2- or FGFR3 aberrant cancer cells. To determine which signaling pathway was involved in FGFR inhibition that caused c-Myc degradation, we treated the cells with inhibitors of the MAPK/ERK and AKT or siRNAs against STAT3 and PLCγ. p-ERK suppression induced c-Myc decrease in FGFR1-amplified H1581, DMS114 cells, and FGFR1OP2-FGFR1–translocated KG1 cell (Fig. 3E). Similar results were also noted in FGFR2- and FGFR3-addicted cancer cells treated with MAPK/ERK inhibitor, PD0325901 (Fig. 3F). However, the c-Myc protein level was not affected by AKT inhibitors GSK690693 and MK2206 (Fig. 3F), or siRNAs against STAT3 or PLCγ (Supplementary Fig. S6D and S6E). These results suggest that the c-Myc stability is predominantly regulated by FGFR-MEK-ERK signaling in FGFR aberrant cancer. Consistent with the known notion that MEK/ERK activation promotes c-Myc stability (34–36), FGFR inhibition and resultant MEK/ERK signal silencing causes c-Myc degradation.

Our findings above suggest a fundamental role of c-Myc in mediating growth inhibition by FGFR inhibitors in FGFR-dependent cells. Hence, inspection of c-Myc level may serve to indicate therapy response to FGFR inhibition in FGFR-addicted cancer. This capacity has important clinical implications, including assessing therapy response and thereby stratifying nonresponders and monitoring the emergence of acquired resistance. To test this possibility, we utilized a series of xenograft tumor models that harbor FGFR aberration but exhibit differential sensitivity to FGFR inhibition (sensitive, FGFR1-amplified NCI-H1581, FGFR2-amplified NCI-H716, FGFR3-mutated UMUC14; nonresponsive, NCI-H2444). Mice bearing NCI-H1581 xenograft were treated with AZD4547 at 6.25, 12.5, or 25 mg/kg once a day for 21 days. Tumor volume was examined twice a week, and intratumoral expression of c-Myc was determined. Along with the strikingly inhibited tumor growth at the dose of 12.5 and 25 mg/kg, intratumoral c-Myc level was profoundly decreased in AZD4547-treated mice. AZD4547 at the dose of 6.25 mg/kg had marginal effect on tumor growth, in which the intratumoral level of c-Myc was almost intact, despite the suppression of FGFR signaling (Fig. 4A and B). Moreover, c-Myc downregulation was detectable as early as on day 3 (Supplementary Fig. S7A), indicating that inspecting c-Myc level could enable to predict the therapy response of FGFR inhibitors at the early stage of the treatment. Similar results were recapitulated in FGFR2-driven H716 (Supplementary Fig. S7B and S7C) and FGFR3-driven UMUC14 model (Fig. 4C and D). We also extended our study to FGFR-nonresponsive model. In NCI-H2444 xenograft model, mice barely responded to AZD4547 treatment even at the high dose of 25 mg/kg. Consistently, the intratumoral level of c-Myc following AZD4547 treatment remained constant, regardless of the abolishment of FGFR signaling (Fig. 4E and F).

Given the accessibility to clinical test of FGFR inhibitors is strictly limited, we alternatively used PDX models to test the potential translational value of c-Myc. The models were established from the fresh tumor tissue of cancer patients, and believed to recapitulate the heterogeneity and histologic characteristics of primary tumor (37). We interrogated a collection of 30 human primary gastric tumor. Consistent with the report that FGFR2 amplification occurs in around 10% gastric cancer, we identified 3 FGFR2-amplified models with FGFR2 copy number > 10, along with FGFR2 transcript overexpression (Supplementary Table S1). We used these three models for further study. Mice were treated with AZD4547 at 12.5 mg/kg or BGJ398 at 10 mg/kg once a day for 21 consecutive days, and tumor volume was examined twice a week. Given dramatic tumor regression to FGFR inhibitors, GA3055 model was treated only for 11 days. As shown in (Fig. 4G, I, and K), all three models showed response to FGFR inhibitor treatment (GA3055, GA1224, and GA0033). Then, we detected intratumoral c-Myc expression and found that c-Myc levels in responders were reduced upon treatment, and downregulation was detectable as early as on day 3 (Fig. 4H, J, and L; Supplementary Fig. S7D and S7E).

Together, all these data suggested that the inspection of c-Myc level will enable excluding the nonresponsive subset at early stage of the treatment of FGFR inhibitors.

From the above results, we found that c-Myc degradation was essential for the efficacy of FGFR inhibition. It raises a possibility that lack of c-Myc degradation may associate with loss of response to FGFR inhibition. To this end, we generated AZD4547 acquired resistance cells using H1581 cells, designated as NCI-H1581AR. As shown in Fig. 5A, NCI-H1581AR was resistant to AZD4547, and the IC₅₀ values were more than 300-fold less potent than that of the parent cells. Meanwhile, c-Myc protein level remained intact, although the p-FRS2α was effectively inhibited by the FGFR inhibitor in this cell line (Fig. 5B), suggesting that dissociation of c-Myc and FGFR is closely related to the loss of sensitivity to FGFR inhibition. This observation was confirmed in vivo using xenograft models. NCI-HA581AR mice with acquired resistance to FGFR inhibition barely responded to 12.5 mg/kg of AZD4547 treatment (Fig. 5C and D), further confirming that c-Myc level was closely associated with response to FGFR inhibition.

Further, we proceeded to test whether c-Myc is dynamically restored in the process of acquired resistance in FGFR-addicted cells. Because the resistant cell NCI-H1581AR was generated by gradiently increasing concentrations of the indicated FGFR inhibitor for 4 months, the sensitivity of cells and the status of phospho-FRS2α and c-Myc were monitored on days 0, 30, 60, 90, 120, respectively. As expected, along with the development of acquired resistance, the diminished c-Myc expression by FGFR inhibitors was gradually restored in spite of the FGFR signaling inhibition (Fig. 5E and F). These findings strongly support that a serial and dynamic inspection of the response of c-Myc may allow us to monitor the acquired resistance developed during the treatment of FGFR inhibitors.

A profound mechanistic illumination of the growth control of kinase addiction may pave the way for a better understanding of drug response in molecularly targeted therapy of kinase inhibitors. This study was initially intrigued by the emerging evidence that kinase signaling is often stringently associated with specific downstream effectors. Responsive alteration of these downstream effectors to upstream kinase inhibition seems essential to determine the therapeutic outcomes of kinase inhibitors (8, 17, 23, 26, 27). This observation may partially explain the clinical observations that kinase inhibitors often obtained a limited response in patients similarly bearing kinase aberrations. For example, up to 30% of patients with EGFR-mutant cancers failed to show responses to gefitinib. A series of preclinical studies later on discovered that alteration of Bim, a proapoptotic molecule downstream of EGFR signaling, was sufficient to ensure the therapeutic outcomes of gefitinib in EGFR-mutant cancer (9). It appears to us that to identify kinase associated downstream effectors may reveal “Achilles' heel” to the kinase addicted cancer, and provide important therapeutic application for kinase inhibitors (11, 13–17). In this study, we chose to focus on FGFR inhibitors, which have gained tremendous attention in the field of anticancer drug discovery worldwide. We have discovered that in FGFR-dependent cancer, growth inhibition upon FGFR signaling deprival is mediated by G₁ cell-cycle arrest and requires degradation of c-Myc. Mechanistically, FGFR inhibition is proceeded by MEK-ERK suppression to facilitate c-Myc degradation via phosphorylation of c-Myc at Thr 58, the key event for degradation of c-Myc through the ubiquitin pathway (33). This model is consistent with the previous notion that MEK/ERK inhibition promoted c-Myc degradation (34–36).

The role of c-Myc in FGFR signaling may enable c-Myc as a probe to distinguish response and resistance for FGFR-targeted therapies. Our results have tested this possibility by monitoring c-Myc protein alteration both in the FGFR-addicted xenografts and PDX models, in which the occurrence of c-Myc degradation can differentiate ultimate responders and the nonresponders. On the third day in a treatment cycle of up to 11 or 21 days, the alteration of c-Myc protein level was clearly detected in responsive models. In the clinical practice, this capacity of c-Myc may allow stratification of the nonresponders at the beginning of treatment, thereby avoiding delay in starting alternative treatment and unnecessary expenses of continuous treatment. More importantly, along with gain of FGFR resistance, c-Myc reduction upon FGFR inhibition was gradually restored, indicating that monitoring the dynamic change of c-Myc in the process of treatment could predict the occurrence of acquired resistance. These together suggested that c-Myc may serve as a functional biomarker for effectiveness and resistance/sensitivity loss to FGFR-targeted therapy. In the meanwhile, we also noticed that in cells with aberrant FGFR, the oncogenic activity of FGFR was enhanced by coexpression of c-Myc. Tumor cells coexpressing c-Myc were more sensitive to FGFR inhibition, suggesting c-Myc may have implications for patient selection for treatment with FGFR inhibitors besides FGFR aberration (22, 38). All these findings by us and others further emphasized the therapeutic value of c-Myc in FGFR axis.

While our study reveals the translational value of FGFR-associated downstream effector c-Myc in FGFR-targeted therapy, it may also suggest a paradigm that can be extended to a broad range of kinase inhibitors. In our recently published study, we have shown that c-Myc is also essential for determining the response to c-Met inhibition in c-Met–amplified cancer. Different from the current study, in c-Met–amplified cancer, c-Myc level was closely associated with c-Met signaling at both transcriptional and protein levels, suggesting more sophisticated molecular basis linking the c-Met and c-Myc. Nevertheless, in both cases, c-Myc maintains a tumorigenic state via regulating G₁–S cell-cycle progression. We hence speculated that downstream effectors of oncogenic kinases may be classified according to the ultimate cellular processes they are driving, such as prevention of apoptosis or cell-cycle progression. In cases where kinases inhibition causes cell-cycle arrest, cell-cycle regulators, in particular those with oncogenic power like c-Myc, are more likely critical to determine ultimate outcomes of the cells. In the meanwhile, in circumstances that kinases inhibition leads to apoptosis, proapoptotic molecules could critically associate with upstream kinases to maintain the tumorigenic state. In support of this hypothesis, EGFR inhibition is known to induce apoptosis in EGFR-mutant cancer, and proapoptotic protein Bim is identified as a downstream effector to promote apoptosis upon EGFR inhibition (9, 10, 26); elevation of Bim indicates the therapeutic outcomes of the target inhibition, whereas its suppression results in loss of sensitivity (9). Our findings together with others present a “Kinase-downstream effector” paradigm that oncogenic kinases are generally coupled to the downstream effector to govern tumor growth. A broad exploration of kinases and their associated downstream effectors may help further test this paradigm.

In conclusion, we have revealed an association between tyrosine kinase FGFR and downstream effector c-Myc in FGFR-dependent cancer. Our results support a paradigm that oncogenic kinase is coupled to the downstream effector to govern tumor growth, and this association determines drug responses. This conceptual progress may help understand growth addiction and resistance acquisition in kinase-addicted cancer, and potentially benefits the patients by optimizing treatment design and increasing the success of kinase inhibitors in clinical practice.

No potential conflicts of interest were disclosed.

Conception and design: J. Ding, M. Geng

Development of methodology: X. Peng

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H. Liu, Y. Chen, X. Wang, X. Peng, H. Chen, Y. Shen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Liu, J. Ai, A. Shen, Y. Chen, Y. Shen, M. Geng

Writing, review, and/or revision of the manuscript: H. Liu, J. Ai, M. Huang, M. Geng

Study supervision: J. Ai, A. Shen, M. Huang, J. Ding, M. Geng

This research was supported by grants from the National Program on Key Basic Research Project of China (No. 2012CB910704 for M. Geng), National Key Sci-Tech Project (No. 2012ZX09301001-007 for M. Geng), the Natural Science Foundation of China (No. 81321092 for J. Ding; No. 81473243 for J. Ai; No. 81222049 for M. Huang; No. 81402966 for A. Shen), “Personalized Medicines-Molecular Signature-based Drug Discovery and Development”, Strategic Priority Research Program of the Chinese Academy of Sciences (No. XDA12020101 for J. Ding), and Youth Innovation Promotion Association (for J. Ai).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.