NRAS-Mutated Rhabdomyosarcoma Cells Are Vulnerable to Mitochondrial Apoptosis Induced by Coinhibition of MEK and PI3Kα Free

RAS proteins have extensively been studied due to their essential roles in normal physiology as well as in human malignancies (1). RAS proteins are small GTPases, which transmit signals from receptors on the cell surface and activate multiple downstream effector pathways, including the RAF/MEK/ERK and PI3K/AKT/mTOR signaling pathways, to regulate various cellular processes, for example, cell growth, survival, and suppression of apoptosis (1). Oncogenic mutation and activation of RAS genes frequently occur in a variety of human cancers (1).

Recently, whole-genome sequencing studies have identified a high rate of recurrent mutations in the RAS pathway in PAX gene fusion–negative rhabdomyosarcoma (RMS) samples (2–4), which correlated with intermediate- and high-risk disease (4). RMS is the most common malignant soft-tissue sarcoma in children and adolescence and comprises embryonal RMS (ERMS) and alveolar RMS (ARMS) as the two major subtypes based on histologic and genetic features (5). Although ARMS is characterized by chromosomal translocations resulting in a PAX3-FOXO1 or PAX7-FOXO1 fusion gene with few other chromosomal alterations, ERMS often acquires multiple chromosomal alterations (6). Among RAS pathway genes, NRAS was found to be most commonly mutated in RMS, followed by KRAS and HRAS (2). Besides oncogenic mutations in RAS genes, additional genetic alterations in genes directly interacting with RAS or receptor tyrosine kinases acting upstream of RAS have been identified in RMS, leading to mutational activation of the RAS pathway in at least 45% of PAX gene fusion–negative RMS (2). This stresses the clinical relevance of the RAS pathway in RMS and the high medical need for developing novel therapeutic strategies for RAS-mutated RMS.

Cancers harboring RAS mutations are often the most difficult to treat (7). Treatment resistance is often due to the evasion of programmed cell death (8), a characteristic feature of cancer cells (9). Apoptosis is one of the most intensively studied forms of programmed cell death and can be initiated via activation of the extrinsic (death receptor) or the intrinsic (mitochondrial) signaling pathway (10). These two pathways converge upon activation of caspases as effector proteins (11). The intrinsic pathway of apoptosis is characterized by mitochondrial outer membrane permeabilization (MOMP), followed by the release of proapoptotic proteins from the mitochondria into the cytosol, leading to caspase activation and cell death (12). MOMP is tightly regulated by BCL-2 family proteins, which consist of three functionally different classes, i.e., proapoptotic multidomain proteins (such as BAX and BAK), proapoptotic BH3-only proteins (e.g., BIM and BMF), and antiapoptotic proteins (e.g., BCL-2, BCL-X_L, and MCL-1). BCL-2 family proteins are regulated in multiple ways, including transcriptional or translational regulation, posttranslational modifications, and oligomerization with other BCL-2 proteins (13). Overall, the balance between pro- and antiapoptotic BCL-2 family proteins is essential for regulating sensitivity to apoptosis (13).

Because direct targeting of constitutively active RAS remains a challenge, inhibition of downstream effector pathways that are altered by oncogenic RAS offers a feasible alternative, for example, by blocking MEK (7). However, the response to MEK inhibitors varies among different RAS-mutated cancers (14), and the inefficiency of single-agent treatment has been attributed to various cross-talks and feedback loops between RAF/MEK/ERK and PI3K/AKT/mTOR pathways (15). This provides the rationale for the concurrent inhibition of both pathways. Indeed, parallel blockage of both PI3K/AKT/mTOR and RAF/MEK/ERK pathways has been shown to suppress tumor growth in some in vitro and in several in vivo models of RAS-mutated cancers (16, 17). However, the transfer of these findings into clinical application is being complicated by the fact that the RAS signaling network is regulated in a context-dependent manner in different cancer entities, for example, by alterations affecting additional pathways. This highlights the need to study RAS-controlled effector mechanisms in a given tumor.

Because RMS has recently been shown to frequently harbor mutations in RAS genes (2–4), in the current study, we aimed at dissecting the RAS effector pathways that control cell death in RMS cells in order to provide a mechanism-based rationale for therapeutic targeting of aberrant RAS signaling in RMS.

RMS cell lines were obtained in 2014 or 2015 from the ATCC or from DSMZ (German Collection of Microorganisms and Cell Cultures GmbH) and frozen upon arrival after authentication by short tandem repeat profiling. VJ cells were generated from a tumor specimen derived from a patient diagnosed with fusion-gene–negative ERMS. Cells were maintained up to 25 passages in RPMI 1640 or DMEM medium (Life Technologies, Inc.), supplemented with 10% FCS (Biochrom), 1 mmol/L sodium pyruvate (Invitrogen), and 1% penicillin/streptomycin (Invitrogen), and regularly tested for lack of mycoplasma contamination. NRAS-overexpressing RMS13 cells have been described previously (18). N-benzyl-oxycarbonyl-Val-Ala-Asp-fluoromethylketone (zVAD.fmk) was purchased from Bachem; BYL179, BKM120, MEK162, and ABT-199 from Selleck Chemical; A-1210477 from Active Biochem; PI103 from Merck Millipore; and all chemicals from Sigma unless indicated otherwise.

Cell viability was assessed by MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assay according to the manufacturer's instructions (Roche Diagnostics); apoptosis by analysis of DNA fragmentation of propidium iodide (PI)–stained nuclei using flow cytometry (FACSCanto II, BD Biosciences) as described previously (19); and caspase activity by CellEvent Caspase-3/7 Green Detection Reagent according to the manufacturer's instructions (ThermoFisher Scientific) using ImageXpress Micro XLS system for detection (Molecular Devices). For determination of colony formation, 200 cells were seeded in a 6-well plate, allowed to adhere overnight, and treated for 24 hours, followed by medium exchange. Colonies were stained after 12 days with crystal violet solution (0.5% crystal violet, 30% ethanol, and 3% formaldehyde). Colonies were counted, and the percentage of colonies relative to solvent-treated controls was calculated. Western blot analysis was performed as described previously (19), and antibodies are listed in Supplementary Methods. BAX/BAK activation was performed as described previously (20).

Cells were transfected with murine stem cell virus vector (pMSCV, Clontech) containing murine BCL-2 or empty vector (EV) using calcium-phosphate transfection as described (21), were transduced with pCMV-Tag3B plasmid (Genentech) containing MCL-1 or EV, followed by selection with Neomycin, or were reversely transfected with 10 nmol/L SilencerSelect siRNA (Invitrogen): Control siRNA (4390844) or two or three independent targeting siRNAs to ensure on-target effects (s54, s55, s56 for NRAS; s10520, 10521 for PIK3CA; s195011, s195012, and s223065 for BIM; s40385, s40386, and s40387 for BMF; s1880 and s1881 for BAK; s1889 and s1890 for BAX) using Opti-MEM medium (Life Technologies, Inc.) and Lipofectamine RNAi MAX reagent (Life Technologies, Inc.) according to the manufacturer's instructions.

Expression levels of the target genes were determined using quantitative real-time (qRT)-PCR method as described (20). Data were normalized on 28S-rRNA expression as reference gene. Primers are listed in Supplementary Table S3. Primers for NRAS, KRAS, and HRAS exons containing activating mutations, exons 2 and 3, are listed in Supplementary Table S1.

At day 8 of fertilization of chicken eggs, 1 × 10⁶ of RD or RH30 cells mixed 1:1 with Matrigel were implanted onto the chorioallantoic membrane (CAM) to form tumors, which were then treated for 3 consecutive days with the amount of drug per cell corresponding to 1 μmol/L BYL719 and/or 1 μmol/L MEK162 in cell culture experiments. The CAM was fixed in 4% paraformaldehyde, embedded in paraffin, cut in 3 μm sections, and then analyzed by immunohistochemistry using 1:1 hematoxylin and 0.5% eosin or rabbit polyclonal anticleaved caspase-3 (Asp175) antibody (Cell Signaling Technology) and hematoxylin counterstain. Images were digitally recorded, and tumor area was analyzed with ImageJ digital imaging software (NIH, Bethesda, MD). The number of active caspase-3–positive cells per tumor area was counted manually by two investigators.

Statistical significance was assessed by the Student t-test (two-tailed distribution, two-sample, unequal variance) or ANOVA and Tukey multiple comparison test. Drug interactions were analyzed by calculation of CI using CalcuSyn software (Biosoft) based on the methods described by Chou (22). CI < 0.9 indicates synergism, 0.9–1.1 additivity, and >1.1 antagonism.

To investigate the role of oncogenic RAS in RMS, we initially assessed the RAS mutational status in a panel of RMS cell lines. We found NRAS mutation (codon Q61H) in four of nine established RMS cell lines and in one patient-derived RMS culture as well as HRAS mutation (codon Q61K) in one RMS cell line, whereas four RMS cell lines were wild-type for NRAS, KRAS, and HRAS (Supplementary Table S2).

Next, we investigated the effects of NRAS knockdown on cell survival and cell death in three NRAS-mutated RMS cell lines. The efficacy of siRNA-imposed silencing of NRAS was controlled at both 48 hours (Fig. 1A) and 144 hours (Supplementary Fig. S1A). NRAS knockdown significantly decreased cell viability as assessed by MTT and crystal violet assays (Fig. 1B; Supplementary Fig. S1B) as well as long-term clonogenic survival (Fig. 1C). In contrast, NRAS silencing failed to cause spontaneous cell death in the absence of a cytotoxic stimulus (Fig. 1D). This shows that cell viability and long-term clonogenic growth of NRAS-mutated RMS depend on the presence of oncogenic RAS; however, NRAS depletion is not sufficient to induce spontaneous cell death. Thus, NRAS-mutated RMS cells can compensate for the loss of oncogenic NRAS to prevent cell death.

To explore how NRAS-mutated RMS cells evade cell death upon depletion of oncogenic NRAS, we tested their dependency on active PI3K/AKT/mTOR signaling. To this end, we used three inhibitors that block distinct pathway elements, i.e., the PI3Kα-specific inhibitor BYL719, the pan-PI3K inhibitor BKM120, and the dual pan-PI3K/mTOR inhibitor PI103. Importantly, all three PI3K inhibitors significantly increased cell death in NRAS knockdown cells compared with control cells (Fig. 2A–C). Interestingly, the PI3Kα-specific inhibitor BYL719 turned out to be similarly effective as pan-PI3K inhibitors to induce cell death in NRAS knockdown cells with no or little single-agent cytotoxicity in both tested RMS cell lines, whereas BKM120 and PI103 already as single agents induced cell death in a dose-dependent manner in one or both RMS cell lines (Fig. 2A–C). In sharp contrast to PI3K inhibitors, addition of the MEK inhibitor MEK162 failed to preferentially trigger cell death in NRAS knockdown compared with control cells (Fig. 2D).

Next, we used a dual pharmacological approach to block effector pathways of oncogenic RAS using the MEK1/2 inhibitor MEK162 in combination with the three PI3K inhibitors. Importantly, BYL719 turned out to be the most potent inhibitor to induce cell death together with MEK162 in all tested RMS cell lines as compared with BKM120 or PI103 (Fig. 2E; Supplementary Fig. S2A–S2C). Calculation of CI revealed that the interaction in particular of MEK162 and BYL719 is highly synergistic (Supplementary Table S3).

To further investigate the specific relevance of PI3Kα to compensate for inhibition of RAF/MEK/ERK signaling, we silenced PI3Kα via siRNA. Western blotting confirmed that two distinct siRNAs against p110α caused marked, although not complete downregulation of p110α protein (Fig. 2F). Importantly, p110α knockdown significantly enhanced the sensitivity of all NRAS-mutated RMS cells to MEK162-induced cell death (Fig. 2G).

Together, this set of experiments shows that parallel inhibition of both, the RAF/MEK/ERK pathway and PI3Kα, is necessary to elicit cell death in NRAS-mutated RMS cells, whereas vertical inhibition of the RAF/MEK/ERK cascade is not sufficient. This indicates that RAS-mutated RMS cell lines rely on PI3Kα to prevent cell death upon NRAS depletion or MEK inhibition.

Next, we asked whether the sensitivity of RMS cells toward MEK162/BYL719 correlates with their RAS mutational status. To address this question, we tested subtoxic concentrations of BYL719 and MEK162 that cause less than 20% cell death as single agents against a broad panel of RMS cell lines. Strikingly, MEK162/BYL719 cotreatment significantly increased apoptosis in the six RAS-mutated RMS cell lines, whereas it failed in the four RAS wild-type cell lines (Fig. 3A). Similarly, ectopic expression of mutant NRAS rendered the ARMS cell line RMS13 responsive to the MEK162/BYL719 cotreatment, whereas no cooperative induction of cell death by MEK162 and BYL719 was found in RMS13 cells expressing EV control (Fig. 3B and C). Also, PI3Kα silencing failed to sensitize RAS wild-type RH30 cells to MEK162-mediated cell death (Fig. 3D and E), in contrast to NRAS-mutated RMS cell lines (Fig. 2G). To test whether the RAS mutational status predicts sensitivity of RMS to MEK162/BYL719 cotreatment in vivo, we used the CAM model, an established in vivo model for anticancer drug testing (21, 23). Of note, MEK162/BYL719 cotreatment was more effective in vivo to trigger caspase-3 activation as a marker of apoptosis in NRAS-mutated than in RAS wild-type RMS tumors (Fig. 3F). Beyond RMS, MEK162 and BYL719 synergized to induce cell death in NRAS-mutated SK-N-AS neuroblastoma and HL60 leukemia cells (Supplementary Fig. S3). These findings indicate that the RAS mutational status predicts the responsiveness of RMS and other malignancies toward combined MEK and PI3Kα inhibition.

Next, we aimed at unraveling the molecular mechanisms underlying the synergy of combined MEK and PI3Kα inhibition. To this end, we assessed the activation of caspases known to mediate apoptotic cell death. Treatment with MEK162/BYL719 significantly increased caspase-3/7 activation in all three RMS cell lines (Fig. 4A). To examine whether caspases are required for the execution of cell death, we used the broad-range caspase inhibitor zVAD.fmk. Of note, the addition of zVAD.fmk significantly decreased MEK162/BYL719-induced apoptosis (Fig. 4B). Kinetic analysis showed that cotreatment with BYL719 and MEK162 induced apoptosis in a time-dependent manner (Fig. 4C) and that zVAD.fmk-conferred protection from cell death was more pronounced at earlier time points (Supplementary Fig. S4A and S4B). Monitoring of long-term clonogenic survival revealed that BYL719 and MEK162 cooperated to significantly suppress colony formation compared with each agent alone or to control cells (Fig. 4D).

This set of experiments emphasizes that MEK162/BYL719 cotreatment triggers caspase-dependent apoptosis.

To investigate how treatment with MEK162 alone or in combination with BYL719 alters RAF/MEK/ERK and PI3K/AKT/mTOR signaling, we assessed phosphorylation of AKT as surrogate readout for PI3K activity, phosphorylation of ERK for MEK activity, and phosphorylation of S6 ribosomal protein and 4E-BP1 as readouts for mTORC1 activity and cap-dependent translation. Constitutively, all three RMS cell lines exhibited phosphorylation of AKT, ERK, S6 ribosomal protein, and 4E-BP1 (Fig. 5A). As expected, BYL719 as single agent reduced phosphorylation of AKT, whereas MEK162 alone suppressed phosphorylation of ERK (Fig. 5A). We also observed that BYL719 slightly reduced ERK1/2 phosphorylation in RD cells (Fig. 5A), in line with a previous study reporting that PI3Kα might regulate ERK phosphorylation via RAF/MEK signaling (24). By comparison, knockdown of NRAS had a minor effect on ERK phosphorylation (Supplementary Fig. S5), which might be due to differences between acute pharmacological inhibition and siRNA-imposed knockdown of gene expression. Importantly, MEK162 and BYL719 cooperated to reduce phosphorylation of S6 ribosomal protein and 4E-BP1 in all three NRAS-mutated RMS cell lines (Fig. 5A), whereas it failed to reduce 4E-BP1 phosphorylation in RAS wild-type RH30 cells (Supplementary Fig. S6A). This indicates that combined MEK and PI3Kα inhibition is necessary to potently suppress cap-dependent translation, a common downstream process regulated by both PI3K/AKT/mTOR and RAF/MEK/ERK signaling.

Next, we analyzed expression levels of BCL-2 family proteins as key regulators of apoptosis. Interestingly, we found that MEK162 and BYL719 cooperated to upregulate BMF mRNA as well as protein levels (Fig. 5B and C). Also, MEK inhibition caused dephosphorylation and increased expression of BIM protein, and the addition of BYL719 further increased mRNA levels of BIM and, to a lesser extent, its protein expression (Fig. 5B and C). Furthermore, MEK162 especially in combination with BYL719 suppressed MCL-1 mRNA and protein levels in all three NRAS-mutated RMS cell lines (Fig. 5B and C), in line with the cooperative inhibition of 4E-BP1 phosphorylation by MEK162/BYL719 cotreatment (Fig. 5A). By comparison, MEK162/BYL719 cotreatment had little effect on MCL-1 levels in RAS wild-type RH30 cells (Supplementary Fig. S6B), in line with its failure to inhibit 4E-BP1 phosphorylation in these cells (Supplementary Fig. S6A), whereas it increased BMF and BIM expression (Supplementary Fig. S6B). Consistently, genetic silencing of NRAS resulted in upregulation of BMF and BIM mRNA and protein expression, whereas it had little effect on MCL-1 protein levels (Fig. 5D; Supplementary Fig. S7). Therefore, we investigated whether NRAS silencing is synthetic lethal with MCL-1 inhibition. Intriguingly, the MCL-1 inhibitor A-1210477 cooperated with NRAS silencing to trigger cell death in NRAS-mutated RMS cells (Supplementary Fig. S8).

To further investigate whether oncogenic NRAS regulates expression of BCL-2 family proteins, we used the ARMS cell line RMS13 with ectopic expression of mutant NRAS (18). Consistently, ectopic expression of mutant NRAS resulted in marked suppression of BMF and BIM protein levels compared with RMS13 cells expressing EV control, whereas MCL-1 protein expression remained largely unchanged (Fig. 5E). By comparison, screening of a panel of RMS cell lines showed no obvious association between RAS mutational status and expression levels of BMF or BIM (Supplementary Fig. S9), indicating that additional factors besides RAS mutation regulate BIM and BMF expression. This set of experiments shows that MEK162 and BYL719 cooperate to upregulate BIM and BMF and to suppress MCL-1 levels, whereas oncogenic NRAS suppresses BMF and BIM with little alterations of MCL-1 levels.

To test if the elevated expression levels of BIM and BMF upon MEK/PI3K coinhibition result in their enhanced interaction with antiapoptotic BCL-2 proteins, we immunoprecipitated BCL-2, BCL-X_L, and MCL-1 and analyzed their binding to BIM and BMF. Interestingly, treatment with MEK162 alone and in combination with BYL719 increased the binding of BIM to BCL-2, BCL-X_L, and MCL-1, and BMF mainly to BCL-2 (Supplementary Fig. S10). This shows that elevated expression levels of BIM and BMF result in their enhanced interaction with antiapoptotic BCL-2 proteins and thus increase the priming of these cells, rendering them more susceptible to undergo apoptosis.

To investigate the functional relevance of BMF and BIM for MEK162/BYL719-induced apoptosis, we genetically silenced these proteins using three distinct siRNA sequences for each gene. Importantly, silencing of BMF or BIM significantly reduced MEK162/BYL719-induced apoptosis in all three RMS cell lines (Fig. 6A–D). To test the relevance of MCL-1, we generated RMS cells with overexpression of MCL-1. Indeed, MCL-1 overexpression significantly decreased MEK162/BYL719-induced apoptosis (Fig. 6E and F). These results indicate that MEK162/BYL719-stimulated changes in expression levels of BIM, BMF, and MCL-1 contribute to the induction of apoptosis by this combination.

Next, we investigated the question as to whether or not the observed changes in the ratio of pro- and antiapoptotic BCL-2 proteins promote activation of BAX and BAK. To this end, we immunoprecipitated BAX and BAK using conformation-specific antibodies, which specifically bind to their activated forms. BYL719 and MEK162 cooperated to trigger activation of BAX and BAK (Supplementary Fig. S11). Notably, BAK and BAX double knockdown significantly reduced cell death upon MEK162/BYL719 cotreatment (Fig. 7A and B), demonstrating that BAX and BAK activation contributes to MEK162/BYL719-induced apoptosis.

To further test the requirement of mitochondrial apoptosis, we overexpressed BCL-2, which is known to interfere with mitochondrial apoptosis. Notably, overexpression of BCL-2 prevented BAX/BAK activation and significantly rescued cells from MEK162/BYL719-induced apoptosis as well as loss of clonogenic growth (Fig. 7C–F). To test whether BCL-2 might impede MEK162/BYL719 cotreatment, we used the BCL-2-selective inhibitor ABT-199 to neutralize BCL-2. Intriguingly, addition of ABT-199 completely reversed the BCL-2-imposed resistance to MEK162/BYL719 cotreatment (Fig. 7D). Together, this set of experiments underscores the relevance of an intact mitochondrial signaling pathway for MEK162/BYL719-induced apoptosis.

Recent sequencing studies have revealed recurrent mutations in the RAS pathway in primary RMS samples with NRAS as the most commonly mutated RAS gene (2–4). The signaling network of oncogenic RAS is regulated in a context-dependent manner. In RMS, however, the RAS effector pathways are still poorly understood.

Here, we report that NRAS-mutated RMS cells can compensate for the depletion of oncogenic NRAS or pharmacological inhibition of MEK to prevent cell death. Importantly, subtoxic concentrations of the MEK inhibitor MEK162 and the PI3Kα-specific inhibitor BYL719 synergize to trigger apoptosis in NRAS-mutated RMS cells, demonstrating that parallel inhibition of both the RAF/MEK/ERK and the PI3K pathways is necessary to elicit cell death in NRAS-mutated RMS cells. This suggests that oncogenic NRAS signaling occurs via these two pathways in RMS cells. In addition, aberrant receptor tyrosine kinase signaling, a frequent event in RMS (2, 3, 25), likely contributes to PI3K/AKT/mTOR pathway activation. In contrast to parallel blockade of RAF/MEK/ERK and PI3K signaling, vertical inhibition of NRAS and MEK fails to elicit cell death in NRAS-mutated RMS cells.

Interestingly, we discover that NRAS-mutated RMS cells rely in particular on PI3Kα to prevent cell death upon NRAS silencing or MEK inhibition. This conclusion is supported by our data showing that specific knockdown of PI3Kα is sufficient to cooperatively trigger cell death together with pharmacological MEK inhibition. In addition, pharmacological inhibitors of MEK or NRAS knockdown synergize in particular with the PI3Kα-specific inhibitor BYL719 to trigger cell death in NRAS-mutated RMS cells. Our study is the first to demonstrate that PI3Kα plays a critical role in activating the PI3K/AKT/mTOR pathway in NRAS-mutated RMS cells. This finding has important implications, because it provides a rationale for the combined use of PI3Kα-specific inhibitors together with MEK inhibitors in RMS with NRAS mutation. Although the four isoforms of PI3K (p110α, p110β, p110γ, and p110δ) might exhibit functional redundancy in retaining cell survival, one of the isoforms preferentially mediates signal transmission depending on the tumor type and genetic aberration of the cell (26). Some RAS-mutated cancers, for example, myeloid leukemia, non–small cell lung cancer, and pancreatic cancer, have previously been reported to depend in particular on PI3Kα in addition to RAF/MEK/ERK signaling (27–30). This dependency has been attributed to the direct interaction of oncogenic RAS with PI3Kα (26, 27, 31). However, KRAS depletion on its own has also been reported to induce apoptosis in some KRAS-mutated cancer cell lines (32), emphasizing the context dependency of RAS signaling.

We identify the RAS mutational status as a marker predicting the sensitivity of RMS toward combined inhibition of MEK and PI3Kα by screening a panel of RMS cell lines including a patient-derived RMS culture. This conclusion is underscored by genetic evidence showing that (i) PI3Kα silencing cooperates with MEK inhibition to trigger cell death in NRAS-mutated but not in RAS wild-type RMS cells and that (ii) ectopic expression of NRAS in RAS wild-type ARMS confers sensitivity to MEK162/BYL719 cotreatment. In addition, MEK162/BYL719 cotreatment proved to be more effective to trigger apoptosis in NRAS-mutated than in RAS wild-type RMS tumors in vivo. RAS mutations were reported to predominantly occur in ERMS, and only some rare cases were found in fusion-negative ARMS (2, 3). This implies that the RAS mutational status may help to select RMS patients that are particularly susceptible to MEK162/BYL719 cotreatment. Nevertheless, RMS cell lines harboring wild-type RAS genes were also found to respond to parallel inhibition of RAF/MEK/ERK and PI3K/AKT/mTOR pathways under certain circumstances, i.e., under cellular stress caused by serum starvation or when MEK inhibition was combined with inhibition of downstream elements of the PI3K/AKT/mTOR pathway using a catalytical mTOR inhibitor or a dual pan-PI3K/mTOR inhibitor (33, 34).

Our study provides novel insights into the molecular mechanisms underlying the synergy of combined MEK and PI3Kα inhibition in NRAS-mutated RMS cells, i.e., (i) changes in the ratio of pro- and antiapoptotic BCL-2 proteins leading to apoptosis via the mitochondrial pathway and (ii) suppression of cap-dependent translation (Fig. 7G).

Importantly, we identify BMF as a target that is suppressed by oncogenic NRAS and required for MEK162/BYL719-induced apoptosis. So far, BMF has not yet been implied in mediating apoptosis upon combined inhibition of RAF/MEK/ERK and PI3K/AKT/mTOR in RAS-mutated cancer cells. We show that NRAS depletion or MEK inhibition upregulates mRNA and protein levels of BMF, which are further increased upon the addition of BYL719. This increase in BMF expression is critical for MEK162/BYL719-stimulated apoptosis, because knockdown of BMF significantly rescues cell death. BMF has recently been identified as a direct transcriptional target of FOXO3a (35), indicating that dephosphorylation of FOXO3a upon MEK and PI3K inhibition might contribute to transcriptional upregulation of BMF. In addition, BMF has been reported to be upregulated via internal ribosome entry sites (IRES)–mediated translation upon inhibition of cap-dependent translation (36). This could explain why we detected upregulation of BMF protein even under conditions when general translation was blocked. Upregulation of BMF has previously been observed in BRAF-mutated melanoma cells treated with the BRAF inhibitor vemurafenib or the MEK inhibitor CI-1040 (37, 38), and phosphorylation of BMF by ERK2 has been reported to reduce its proapoptotic activity without changing its protein turnover (39). In mammary epithelial cells, oncogenic transformation by RAS^V12 has been described to suppress upregulation of BMF during anoikis (40).

Besides BMF, we show that also other BCL-2 family proteins contribute to MEK162/BYL719-stimulated apoptosis, as (i) MEK162/BYL719 cotreatment cooperates to upregulate BIM and to downregulate MCL-1 and as (ii) BIM silencing or MCL-1 overexpression significantly protects RMS cells from MEK162/BYL719-induced cell death. Both AKT and ERK have been reported to phosphorylate and inactivate the transcription factor FOXO3a that regulates BIM expression (41). In addition, ERK directly phosphorylates BIM, which enhances its proteasomal degradation (42). MCL-1 expression is tightly controlled by both RAF/MEK/ERK and PI3K/AKT/mTOR pathways (13, 43), involving transcription factors such as ELK-1 (44) or CREB (45), mTORC1-stimulated translation of MCL-1 (46), or posttranslational modifications, such as phosphorylation, that control the stability of MCL-1 (47).

Together, these MEK162/BYL719-stimulated changes in the ratio of pro- and antiapoptotic BCL-2 proteins toward a proapoptotic state act in concert to promote activation of BAX and BAK and caspase-dependent mitochondrial apoptosis, as individual silencing of one of these proteins provides partial protection from MEK162/BYL719-triggered apoptosis (Fig. 7G). Our findings showing that MEK162/BYL719 cotreatment triggers upregulation of BMF and BIM as well as MCL-1 downregulation, whereas NRAS knockdown upregulates BMF and BIM but has little effect on MCL-1 levels may explain why NRAS knockdown on its own is not sufficient to induce apoptosis. The crucial role of the mitochondrial pathway in mediating MEK162/BYL719-induced apoptosis is emphasized by our data showing that BAX/BAK silencing, overexpression of MCL-1 or BCL-2 as well as caspase inhibition rescue cell death. Intriguingly, the BCL-2–specific inhibitor ABT-199 completely reversed the BCL-2–conferred resistance to MEK162/BYL719-induced apoptosis. This implies that therapeutic modulation of mitochondrial apoptosis may offer new opportunities to enhance the antitumor activity of combined RAF/MEK/ERK and PI3K/AKT/mTOR inhibition. It also implicates that markers of an intact mitochondrial apoptotic pathway, for example using BH3 profiling, may help to predict the sensitivity of RMS cells toward MEK162/BYL719 cotreatment, e.g., for patient stratification.

Besides changes in BCL-2 family proteins, we show that combined inhibition of RAF/MEK/ERK and PI3K/AKT/mTOR pathways is necessary in NRAS-mutated RMS cells to potently suppress cap-dependent translation. Both 4E-BP1 and S6 are important regulators of cap-dependent translation (48), and 4E-BP1 has been identified as a key common downstream effector of active RAF/MEK/ERK and PI3K/AKT/mTOR signaling in RAS- and PIK3CA-mutated cancer cells (49). Interestingly, we show that MEK162/BYL719-conferred dephosphorylation of 4E-BP1 and S6 is accompanied by downregulation of MCL-1 and upregulation of BMF levels. MCL-1 is a short-lived protein that for its translation depends on active mTORC1 (46), and IRES-mediated upregulation of BMF has been described as a cellular stress response to inhibition of cap-dependent translation (36). Thus, MEK162/BYL719-imposed suppression of cap-dependent translation may well lead to downregulation of MCL-1 and upregulation of BMF levels. In addition, reduced proliferation upon blockage of cap-dependent translation is expected to facilitate the induction of programmed cell death.

Our study has important implications. First, it might inspire future clinical trials in RMS by providing a rationale for the combined use of MEK- and PI3Kα-specific inhibitors in NRAS-mutated RMS. Because sequencing of RMS tumor samples prior to treatment is being more and more introduced into clinical practice and because several therapeutics targeting RAS effector networks have already been approved or are in late-stage development, it is in principle feasible to translate our approach into clinical application in the future. The relevance of concomitant inhibition of RAS and PI3K signaling in RMS is further emphasized by documented simultaneous mutations of RAS and PIK3CA in some PAX gene fusion–negative RMS (2–4).

Second, specific targeting of PI3Kα may offer therapeutic advantages, as it reduces the side effects of pan-PI3K inhibitors. BYL719 has been reported to have an improved safety profile with respect to glucose metabolism and proved to be well-tolerated with manageable side effects (50). Currently, the combination of BYL719 and MEK162 is being evaluated in an ongoing clinical trial for adult solid cancers with documented RAS of BRAF mutations (NCT01449058).

Third, our findings are not restricted to NRAS-mutated RMS but are likely of broader relevance also for RMS-harboring mutations in other RAS genes, because we show that HRAS-mutated RMS cells similarly respond to MEK162/BYL719 cotreatment. In summary, by elucidating the RAS effector pathways in RMS cells, our study provides a rationale for the combined use of MEK- and PI3Kα-specific inhibitors in RAS-mutated RMS that warrants further investigation.

No potential conflicts of interest were disclosed.

Conception and design: N. Dolgikh, S. Fulda

Development of methodology: N. Dolgikh, M. Hugle, M. Vogler

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): N. Dolgikh

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): N. Dolgikh, M. Hugle, M. Vogler, S. Fulda

Writing, review, and/or revision of the manuscript: N. Dolgikh, M. Hugle, M. Vogler, S. Fulda

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Fulda

Study supervision: S. Fulda

This work has partially been supported by grants from the BMBF (to S. Fulda). We thank D. Brücher for expert technical assistance and C. Hugenberg for expert secretarial assistance.

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.