Research ArticleCancer

The HDAC3–SMARCA4–miR-27a axis promotes expression of the PAX3:FOXO1 fusion oncogene in rhabdomyosarcoma

See allHide authors and affiliations

Science Signaling  20 Nov 2018:
Vol. 11, Issue 557, eaau7632
DOI: 10.1126/scisignal.aau7632

Targeted treatment for pediatric aRMS

Rhabodomyosarcoma is a soft tissue tumor in children that is difficult to treat. A chromosomal abnormality generates a fusion protein called PAX3:FOXO1 that drives chemoresistance and aggressive progression in patients with the alveolar subtype of the disease (aRMS). Early-phase clinical trials have shown tolerability of the histone deacetylase (HDAC) inhibitor entinostat in pediatric patients. Here, using cells and animal models, Bharathy et al. found that entinostat works in aRMS by specifically blocking the activity of HDAC3, thereby preventing epigenetic suppression of a microRNA that inhibits PAX3:FOXO1 translation. Without PAX3:FOXO1 protein, aRMS growth slowed, and tumors were sensitized to the chemotherapy vincristine. These findings and ongoing clinical trials show promise for an effective therapy for some patients with aRMS.


Rhabdomyosarcoma (RMS) is the most common soft tissue sarcoma of childhood with an unmet clinical need for decades. A single oncogenic fusion gene is associated with treatment resistance and a 40 to 45% decrease in overall survival. We previously showed that expression of this PAX3:FOXO1 fusion oncogene in alveolar RMS (aRMS) mediates tolerance to chemotherapy and radiotherapy and that the class I–specific histone deacetylase (HDAC) inhibitor entinostat reduces PAX3:FOXO1 protein abundance. Here, we established the antitumor efficacy of entinostat with chemotherapy in various preclinical cell and mouse models and found that HDAC3 inhibition was the primary mechanism of entinostat-induced suppression of PAX3:FOXO1 abundance. HDAC3 inhibition by entinostat decreased the activity of the chromatin remodeling enzyme SMARCA4, which, in turn, derepressed the microRNA miR-27a. This reexpression of miR-27a led to PAX3:FOXO1 mRNA destabilization and chemotherapy sensitization in aRMS cells in culture and in vivo. Furthermore, a phase 1 clinical trial (ADVL1513) has shown that entinostat is tolerable in children with relapsed or refractory solid tumors and is planned for phase 1B cohort expansion or phase 2 clinical trials. Together, these results implicate an HDAC3–SMARCA4–miR-27a–PAX3:FOXO1 circuit as a driver of chemoresistant aRMS and suggest that targeting this pathway with entinostat may be therapeutically effective in patients.


Soft tissue sarcomas are among the most common and deadliest childhood cancers (1, 2). Despite improved survival for other childhood cancers, progress for particularly metastatic sarcoma has been minimal, and therapeutic options remain limited (2, 3). Rhabdomyosarcoma (RMS) is the most common soft tissue sarcoma in children (2). Patients with diagnosed metastatic RMS have dismal prognoses for survival that have not improved in years, if not several decades (46). Although ~85% of local lesions can be excised by surgery, undetectable circulating tumor cells and recurrence at the primary site from occult microscopic residual disease are presumed to act as a gateway to relapse and subsequent regional and distal metastasis (7).

Alveolar RMS (aRMS) is the more aggressive RMS subtype and is generally not survivable long term once metastatic (<8%) (4, 6, 8). This disease does respond to chemotherapy in most cases, but the clinical challenge is preventing or overcoming recurrence after chemotherapy and/or radiation (9). Unique to aRMS is the expression of a fusion protein called PAX3:FOXO1, which results from a chromosomal translocation and comprises the transcription factors PAX3 (paired box 3) and FOXO1 (forkhead box 1). PAX3:FOXO1 is seen in some patients with aRMS. We have previously reported that native, non-constitutive, G2 cell cycle phase-enriched expression of the PAX3:FOXO1 oncogene in aRMS facilitates checkpoint adaptation (meaning a high tolerance of DNA breaks or mitotic catastrophe) [(10) and reviewed in (11)]. In terms of clinical impact, this PAX3:FOXO1-mediated treatment resistance causes a notable 40 to 45% difference (drop) in survival over 10 years compared to PAX3:FOXO1-negative cases (9). In experimental studies, genetic knockdown of PAX3:FOXO1 abundance improves chemotherapy and radiation sensitivity and reduces tumor reestablishment (10), but drugging transcription factors such as PAX3:FOXO1 in patients is a daunting task. However, in this study, we found that PAX3:FOXO1 can be pharmacologically silenced at both the mRNA and protein levels by entinostat (ENT), a class I histone deacetylase (HDAC) inhibitor (HDACi) (with weak HDAC10/HDAC11 inhibition) that was granted a U.S. Food and Drug Administration breakthrough designation for certain breast cancers in adults.

In combination with the chemotherapy vincristine (VCR), ENT had strong antitumor activity in a genetically engineered mouse model (GEMM) of aRMS and in both orthotopic allograft and patient-derived xenograft (PDX) mouse models of aRMS. Pretreatment with ENT also improved radiation sensitivity. We further established the comparative efficacy of ENT to other HDACis at reducing the expression of PAX3:FOXO1 and that of another, related fusion protein also found in aRMS, PAX7:FOXO1. Mechanistically, we determined that, in aRMS cells, HDAC3 inhibition by ENT decreases SMARCA4, which, in turn, derepresses miR-27a, which then silences PAX3:FOXO1; by blocking HDAC3, ENT suppressed the abundance and activity of PAX3:FOXO1 and the growth of fusion-positive aRMS tumors in mice. Together, these preclinical and mechanistic studies have led to the Children’s Oncology Group pediatric phase 1 clinical trial, ADVL1513, and a planned phase 1B cohort expansion for patients with RMS for this same trial.


ENT in combination with VCR slows aRMS tumor growth in vivo

In preclinical studies using RNA interference, depletion of PAX3:FOXO1 abundance in aRMS cells is shown to be critical to overcome checkpoint adaptation, a process whereby tumor cells survive chemotherapy and radiation in late phases of the cell cycle (summarized in Fig. 1A) (10). To affirm genetically that silencing PAX3:FOXO1 increases chemosensitivity of aRMS to VCR in vitro, we first revalidated two derivative cell lines from the aRMS GEMM-derived primary tumor culture U23674 that did or did not have stable knockdown of PAX3:FOXO1 (shYFP or shCtrl, respectively) (10). PAX3:FOXO1 was then also restored by retroviral-mediated transduction into the shYFP-U23674 line. Immunoblot analysis confirmed stable knockdown of PAX3:FOXO1 and/or restored expression of PAX3:FOXO1, respectively (Fig. 1B). When treated with VCR for 24 hours, shYFP (PAX3:FOXO1 knockdown) cells alone had an absolute half maximal inhibitory concentration (IC50) of 33 nM. However, control cells did not reach an absolute IC50 (Fig. 1C). Restoring expression of PAX3:FOXO1 in the stable knockdown cells (shYFP + PAX3:FOXO1) rescued the cells from the cytotoxic effects of VCR, in that no absolute IC50 was reached (Fig. 1C). Although PAX3:FOXO1 was tumor protective against chemotherapy, PAX3:FOXO1 knockdown was only cytostatic, which is consistent with previous reports of murine and human aRMS (10, 12).

Fig. 1 ENT treatment of aRMS in vivo.

(A) Diagrammatic representation of checkpoint adaptation, a process whereby tumor cells survive chemotherapy and radiation in late phases of the cell cycle. XRT, x-ray telescope radiotherapy; PLK1, polo-like kinase 1; AURKB, aurora kinase B; GSG2, genomic structure of haspin; CDC25B, cell division cycle 25b; CCNB1/2, cyclin B1/2; dsDNA, double-stranded DNA. (B) Basal PAX3:FOXO1 protein expression in murine U23674 aRMS cells transfected with control shRNA (shCtrl), PAX3:FOXO1-targeted shRNA (shYFP), or targeted shRNA plus a lentivirus expressing PAX3:FOXO1 (shYFP + PAX3:FOXO1). Blots are representative of n = 3 biological replicates. (C) Viability of the U23674 cells described in (B) exposed to VCR at 4 nM for 24 hours. Graph plotted using GraphPad Prism. Data are means ± SD of n = 3 independent experiments. (D) Box-and-whisker plot showing the tumoristatic efficacy of ENT or VCR, alone and in combination, in aRMS mice at day 13 (DMSO versus ENT + VCR). Treatment with ENT at a daily dose of 5 mg/kg by intraperitoneal injection or with VCR at a dose of 1 mg/kg weekly by intraperitoneal injection, or a combination of both. Data are means ± SEM (n = 5 mice per cohort). ***P < 0.001 by log-rank test. (E) Kaplan-Meier plot of the proportion of mice with tumors smaller than 1.2 cc after treatment with ENT at a daily dose of 5 mg/kg by intraperitoneal injection or with VCR at a dose of 1 mg/kg weekly by intraperitoneal injection, or a combination of both. Treatment was stopped after day 13 because body weight loss approached 10 to 15%. Data are means ± SEM (n = 5 mice per cohort). *P < 0.05 by log-rank test. In this experiment, treatment was stopped for all mice after day 13 because body weight loss approached 10 to 15%.

Our previous studies examined whether cell of origin influenced PAX3:FOXO1 expression and whether expression could be pharmacologically altered by HDACis or other epigenetic agents (13). To test whether the narrow-spectrum class I HDACi ENT could sensitize PAX3:FOXO1-positive aRMS to the chemotherapies most often used in high-risk or relapsed disease, we first tested the efficacy of ENT and VCR as single agents and in combination in orthotopic allograft mouse models of aRMS. This aRMS model was generated by injecting murine aRMS primary cell cultures into the cardiotoxin-preinjured gastrocnemius muscle of severe combined immunodeficient hairless outbred (SHO) mice. The tumor-bearing mice were treated with ENT at a daily dose of 5 mg/kg administered intraperitoneally or with VCR at 1 mg/kg administered intraperitoneally once a week, or in combination. Unlike monotherapy of either agent, the combination of ENT and VCR was effective in delaying tumor growth in the aRMS mice (Fig. 1, D and E). Histologically, no cytodifferentiation was seen (table S1).

To determine whether pretreatment with ENT potentiates the effect of radiation on RMS cells, we pretreated a murine primary tumor cell culture (U23674) with ENT or dimethyl sulfoxide (DMSO) for 24 hours followed by irradiation [10 Gy (gray)] (fig. S1A). After irradiation, 500,000 viable cells were injected into the cardiotoxin-preinjured gastrocnemius muscle of SHO mice. Tumor cells treated with DMSO alone or ENT alone were used as controls. After injection of tumor cells, mice were monitored for tumor development. Mice that received tumor cells pretreated with ENT before irradiation showed increased latency of tumor development compared to mice that received cells pretreated with DMSO only (fig. S1B), suggesting that ENT treatment sensitizes RMS cells to radiation. With respect to chemotherapy, we also confirmed that the biochemical effect of PAX3:FOXO1 silencing occurs—and with synergy—at clinically achievable concentrations of ENT and VCR in human RMS cultures (fig. S2, A to C).

To further assess the clinical feasibility of ENT in a clinical setting, we investigated whether ENT had a dose-dependent effect and/or whether ENT was unique among HDACis for its suppressive effect on PAX3:FOXO1. First, we observed a dose-dependent effect of ENT on PAX3:FOXO1 expression in Rh30 cells in response to doses from 0.1 to 2 μM, when examining the mRNA abundance at 24 hours (fig. S3A) and the protein abundance at 72 hours (fig. S3B). A PDX-derived explant aRMS culture, CF-001, also exhibited a dose-dependent response (fig. S3C), as did the murine aRMS culture, U23674 (fig. S3D).

To examine the relative efficacy of other HDACis at reducing PAX3:FOXO1 when compared to ENT (an HDAC1-3 inhibitor), we tested SAHA (suberoylanilide hydroxamic acid, a class I and class II HDACi), panobinostat (PAN), CUDC-907, and CUDC-101 in human and murine aRMS using the highest reported maximum drug serum concentration (Cmax) in humans. Where Cmax was unknown, we treated the cell cultures at the measured drug IC25 (25% inhibiting concentration). ENT at its highest achievable Cmax (1000 nM) (14) significantly reduced PAX3:FOXO1 expression in human aRMS (Rh30 and Rh41) cells (fig. S4A), which was reflected by near-complete reduction at the protein level in the human aRMS cells and in murine aRMS (U23674) cells (fig. S4B). In comparison, PAN (a broad-spectrum HDACi that targets HDAC1-11) at its Cmax (45 nM) (15) did not consistently diminish PAX3:FOXO1 at either the mRNA or protein level (fig. S4, C and D), and SAHA (an HDAC1/2/3/6/8 inhibitor) was only somewhat effective at reducing PAX3:FOXO1 abundance at its Cmax (1000 nM) (fig. S4, E and F) (16) but to a lesser degree than ENT. Near-complete cytotoxicity was observed in Rh41 cells with SAHA at the end of 72 hours of treatment, preventing inclusion of Rh41 in these biochemical studies. In addition to PAN and SAHA, CUDC-907 [a dual phosphatidylinositol 3-kinase (PI3K) and HDAC1/2/3/10 inhibitor] and CUDC-101 (a class I and class II HDACi) were tested but were found to be less efficient at reducing PAX3:FOXO1 compared to ENT at their IC25 (fig. S4, G and H). To examine whether these HDACis can have an effect on the related fusion protein, PAX7:FOXO1, the human aRMS cell line CW9019 harboring the t(1;13) translocation was treated with HDACis at their respective Cmax or IC25 concentration. ENT and all other HDACis tested were effective at reducing PAX7:FOXO1 protein abundance to an almost undetectable level (fig. S4I).

ENT plus VCR has efficacy in aRMS PDXs

We next investigated the antitumor efficacy of ENT and VCR as single agents or in combination for various biologically independent PDX models of aRMS. The dosing details are given in table S2, and PDX model characteristics are given in tables S3 and S4. Seven of the eight models were from recurrent and/or metastatic tumors taken from biopsy or autopsy. All these contemporary models were established after 2010. Seven of eight models carried the PAX3:FOXO1 oncogene, whereas one model (J99873/CF-2) was established from an autopsy tumor harboring PAX7:FOXO1. In each case, tumor growth inhibition by combined ENT plus VCR was superior to either drug alone (Fig. 2, A to G) except for one model (CTG-1008), in which tumor growth inhibition by ENT alone was better than combined treatment (Fig. 2H). Waterfall plots summarizing the tumor growth inhibition (%) for the eight models (Fig. 2I and fig. S5, A to C) showed that, in five of eight cases, ENT had single-agent activity relative to the control. No difference was seen between different treatment groups in terms of rhabdomyoblast differentiation (fig. S5, D and E, and table S5). Biochemical analysis of tumor lysates from J77636 PDXs treated with ENT showed pharmacodynamic down-regulation of PAX3:FOXO1 protein abundance, which was consistent with results seen in two-dimensional cultures of human and murine aRMS cell lines (Fig. 2J). Statistical summaries of the eight aRMS PDX models are provided in tables S6 to S13.

Fig. 2 In vivo evaluation of ENT and VCR in aRMS PDX.

(A to H) Graphical analysis of tumor volume inhibition by either ENT alone or in combination with VCR in eight different PDX aRMS mouse models (Champions Oncology and The Jackson Laboratory) established from clinical biopsies, recurrent aRMS, or autopsies. Demographic features of these models are given in tables S3 and S4; treatment schedules are given in table S2. Statistical analyses are given in tables S6 to S13. (I) Waterfall plot showing tumor growth inhibition (%) for combination treatment (ENT + VCR) in the eight models shown in (A) to (H), as labeled. (J) Pharmacodynamic assessment of tumor lysates from J77636 treated with ENT for PAX3:FOXO1 protein expression. Blots show n = 3 biological replicates (n represents tumor lysates from three mice per cohort). Ctrl, control.

Mechanisms underlying ENT treatment converge on known downstream genes

ENT targets class I HDACs (HDAC1, HDAC2, and HDAC3) but that at higher concentrations also inhibits the class II member HDAC10. To determine further the mechanism by which ENT silences PAX3:FOXO1 expression, we reduced the endogenous expression of HDAC1, HDAC2, and/or HDAC3 in Rh30 cells using small interfering RNA (siRNA) (siHDAC1/2/3) alongside a scrambled siRNA control (Fig. 3A). PAX3:FOXO1 abundance was decreased to some extent by each individual HDAC siRNA, but depletion of HDAC2 or HDAC3 was the most potent (Fig. 3A). Depleting Rh30 cells of HDAC10, albeit somewhat less efficient than HDAC1/2/3 depletion, had no detectable impact on the protein abundance of PAX3:FOXO1 (Fig. 3B). These results indicated the importance of class I HDACs, specifically HDAC2 and HDAC3, for PAX3:FOXO1 expression and suggest that inhibition of specifically these HDACs is the mechanism through which ENT represses the fusion protein.

Fig. 3 siRNA-mediated knockdown of HDAC and gene expression and PAX3:FOXO1 binding data for key aRMS gene targets.

(A and B) PAX3:FOXO1 expression in Rh30 cells transfected with siRNA at 100 nM for 72 hours, targeting HDAC1, HDAC2, and/or HDAC3 (A), as well as HDAC10 (B). Blots are representative of n = 3 independent experiments. (C to F) Three aRMS samples [U23674 in two replicates (U23674A and U23674B), Rh30, and Rh41] were sequenced after treatment with ENT at 2 μM for 72 hours or DMSO. Featured genes were decreased (log2 ratio of ENT-induced expression divided by control expression > 1) in all samples. Additional expression data for these key targets were curated from previous publications (10, 33), and PAX3:FOXO1 binding data were curated from the literature as indicated (31, 32) or were generated through chromatin immunoprecipitation sequencing (ChIP-seq) experiments. Four subclasses of ENT-induced gene expression were identified as (C) PAX3:FOXO1 binding with or without HDAC binding; (D) HDAC1, HDAC2, HDAC3, or HDAC11 binding only; (E) indirect targets of PAX3:FOXO1; and (F) otherwise-regulated genes.

Our RNA sequencing (RNA-seq) and chromatin immunoprecipitation–exonuclease (ChIP-exo) data identified four classes of cell-autonomous gene expression changes related to ENT treatment: These subclasses were PAX3:FOXO1 binding with or without HDAC binding (Fig. 3C); HDAC1, HDAC2, HDAC3, or HDAC11 binding only (Fig. 3D); indirect targets of PAX3:FOXO1 (Fig. 3E); and otherwise-regulated genes (Fig. 3F) (13). Although HDAC-related and HDAC-unrelated changes in expression of the known PAX3:FOXO1 direct and indirect targets were expected, neither these nor those of the otherwise-regulated genes could explain altered transcription of PAX3:FOXO1 with respect to PAX3 regulatory elements.

To understand the mechanisms of action for ENT and PAN, we also analyzed the expression of wild-type PAX3 and FOXO1 in U23674 and Rh30 aRMS cells, as well as C2C12 mouse myoblasts, human skeletal muscle myoblasts (HSMMs), and CureFast-1 (CF-1) PDX explant cell culture each treated separately with ENT and PAN. Consistent with previous results (13), PAX3:FOXO1 was significantly reduced by ENT but not by PAN in U23674 cells (Fig. 4, A and B). These cells do not express detectably sufficient wild-type PAX3 for analysis (Fig. 4A); however, in C2C12 cells, which do so, ENT significantly reduced wild-type PAX3 but not wild-type FOXO1 abundance (Fig. 4, A and C), and in both cell lines, neither ENT nor PAN altered FOXO1 (Fig. 4, A, D, and E). Similar results were observed in Rh30, CF-1, and HSMM cells (Fig. 4, F to I). These results suggest that, for PAX3:FOXO1 in aRMS tumor cells, the PAX3 cis-regulatory elements are affected differently by ENT treatment versus PAN treatment, perhaps attributable to secondary effects on PAX3 cis-elements or PAX3 mRNA stability factors (Fig. 4J).

Fig. 4 Wild-type PAX3 expression altered by ENT and not by PAN.

(A) Immunoblots of PAX3:FOXO1, wild-type PAX3, and wild-type FOXO1 in murine C2C12 myoblasts and U23674 aRMS cells upon 72 hours of treatment with ENT (1 μM) and PAN (45 nM). Blots are representative of n = 3 independent experiments. Note that, in the murine aRMS GEM culture U23674, both wild-type PAX3 alleles are homozygously replaced by PAX3:FOXO1, but both FOXO1 alleles are intact (homozygous wild type). (B to E) Densitometric analysis of (A) PAX3:FOXO1, wild-type PAX3, and wild-type FOXO1 expression upon treatment with ENT and PAN. Data are means ± SD (n = 3 independent experiments). ***P < 0.001, two-sided Student’s t test. (F) Western blots of PAX3:FOXO1, wild-type PAX3, and wild-type FOXO1 in HSMMs, Rh30, and CF-1 cells upon 72 hours of treatment with ENT (1μM) or PAN (45 nM). Blots are representative of n = 3 biological replicates. In contrast to murine aRMS cultures, PAX3 is heterozygous (wild type/PAX3:FOXO1) and FOXO1 is heterozygous (wild type/null) in human aRMS cultures, and both are homozygous wild type in HSMMs. (G to I) Densitometric analysis of PAX3:FOXO1, wild-type PAX3, and wild-type FOXO1 protein expression upon treatment with ENT. Data are means ± SD (n = 3 independent experiments). *P < 0.05 and ***P < 0.001, two-sided Student’s t tests. (J) A diagrammatic representation attributing the effect of ENT on PAX3:FOXO1 expression to PAX3 cis-regulatory elements.

Loss of SMARCA4 expression or activity increases, or derepresses, the expression of the microRNA miR-27a

To further understand how ENT silences the PAX3:FOXO1 gene, whereas PAN does not, RNA-seq analysis of ENT-treated versus PAN-treated human and murine aRMS cells was carried out with an emphasis on assessing the expression of transcription factors and chromatin remodeling complexes. Differential expression across samples after ENT treatment versus after PAN treatment identified key up-regulated and down-regulated genes in the ENT-specific effects (Fig. 5A). Many epigenetic targets, including chromatin-modifying enzymes and remodeling complexes, were indeed decreased by ENT treatment and increased by PAN treatment in Rh30 and U23674 cells (Fig. 5A). Specifically, ENT treatment had a strong effect on the expression of many of the SWI/SNF (switch/sucrose non-fermentable) complexes, which included that of SMARCA4/BRG1, which was markedly reduced by ENT but not by PAN treatment (Fig. 5A). At the protein level, only ENT reduced SMARCA4 protein abundance in both cells, notably to an undetected level (Fig. 5B). Consistent with ENT-mediated pharmacological inhibition of class I HDACs, siRNA-mediated knockdown of HDAC3 decreased SMARCA4 protein abundance to an almost undetectable level; knockdown of HDAC2 had markedly less of an effect (Fig. 5C). This finding concurs with a previous report indicating a noncanonical role of HDAC3 as a transcriptional activator (17). Furthermore, the abundance of the PAX3:FOXO1 fusion protein was decreased in Rh30 cells upon reduction of SMARCA4 by siRNA (siSMARCA4) (Fig. 5D), altogether suggesting an HDAC3-SMARCA4 circuit in promoting PAX3:FOXO1 fusion protein abundance in aRMS cells.

Fig. 5 Loss of SMARCA4 expression or activity derepresses miR-27a.

(A) Differential expression analysis by RNA-seq for genes associated with chromatin-modifying complexes in Rh30 and U23674 cells after treatment for 72 hours with ENT (1 μM) or PAN (45 nM) relative to cells treated with DMSO. SMARCA4 average transcripts per million for DMSO-treated cells, ENT-treated cells, and PAN-treated cells were 86.86, 56.15, and 72.19, respectively. (B) SMARCA4 protein abundance in Rh30 and U23674 aRMS cultures treated with ENT or PAN. Blots are representative of n = 3 independent experiments. (C) SMARCA4 protein abundance in Rh30 cells transfected with HDAC2- or HDAC3-targeted siRNA (100 nM for 72 hours). Blots are representative of n = 3 independent experiments. (D) PAX3:FOXO1 protein abundance in Rh30 cells transfected with SMARCA4-targeted siRNA (100 nM for 72 hours). Blots are representative of n = 3 independent experiments. (E) Quantitative polymerase chain reaction (qPCR) analysis of miR-27a expression in Rh30 cells treated with the SMARCA4 bromodomain inhibitor PFI-3 (10 μM for 24 hours). (F) qPCR analysis of miR-485 in Rh30 cells treated with PFI-3. **P < 0.01, two-sided Student’s t test. (G and H) qPCR analysis of miR-27a expression in Rh30 and CF-1 cells transfected with SMARCA4-targeted siRNA (100 nM for 72 hours). Data were normalized to the expression of U6snRNA. Gene expression was quantified using the 2−ΔCt method. Data are means ± SD (n = 3 independent experiments each in triplicate). *P < 0.05, two-sided Student’s t test. (I) Diagrammatic representation of how ENT and PFI-3 disrupt SMARCA4-mediated interference of miR-27a transcription.

Upon searching the miR search engine miRBase for microRNA (miRNA) that bound the PAX3 promoter or cis-elements, we observed that miR-27a was unique in that, in both mouse and human cells, it bound PAX3 upstream (5′) of the PAX3:FOXO1 breakpoint (in intron 4). miR-27a-5p reportedly targets PAX3, as well as PAX7, in developing and adult muscle cells (18). For comparison, we chose miR-485-5p as a conserved miRNA binding the PAX3 3′ untranslated region (3′UTR). Referencing the University of California, Santa Cruz (UCSC) genome browser visualization of ChIP-seq data from the Encyclopedia of DNA Elements (ENCODE) predicted that SMARCA4 binds to the 3′UTR of miR-27a. We then measured miR-27a expression upon pharmacological inhibition of SMARCA4 bromodomain activity using the small-molecule inhibitor PFI-3 at a concentration of 10 μM. PFI-3 inhibits 95% of the SMARCA4 bromodomain activity at this concentration according to analysis by time-resolved fluorescence energy transfer (TR-FRET; fig. S6). Inhibition of SMARCA4 bromodomain activity increased miR-27a expression (Fig. 5E) but not that of miR-485 (Fig. 5F). We then carried out siRNA-mediated knockdown of SMARCA4 in Rh30 cells and CF-1 cells, which showed an increase in miR-27a compared to scrambled siRNA control, consistent with the effect seen on miR-27a upon pharmacological inhibition of SMARCA4 (Fig. 5, G and H). Together, these results suggested that a SMARCA4–miR-27a regulatory circuitry controls the expression of PAX3:FOXO1 (Fig. 5I).

ENT-induced reexpression of miR-27a silences PAX3:FOXO1

To understand whether ENT silences PAX3:FOXO1 transcript via miR-27a–mediated transcription interference, we reexpressed miR-27a in Rh30 and CF-1 cells by transient transfection of miR-27a mimics and examined PAX3:FOXO1 protein expression. Compared to a control oligonucleotide (a “negative mimic”), miR-27a reexpression decreased PAX3:FOXO1 abundance in both Rh30 and CF-1 cells (Fig. 6, A and B), which also exhibited altered cell morphology (Fig. 6, C and D). qPCR analysis then showed a more potent increase in the expression of miR-27a in ENT-treated than in PAN-treated Rh30 and CF-1 cells (Fig. 6, E and F). However, PAN increased miR-485 expression more potently than did ENT (Fig. 6, G and H). Because our earlier studies showed that ENT reduced PAX3:FOXO1 expression by inhibiting HDAC2/3 (Fig. 3A), we silenced HDAC2/3 by siRNA in Rh30 and CF-1 cells, which resulted in increased miR-27a expression relative to that in scrambled siRNA control cells (Fig. 6, I and J). These results altogether suggest that, by inhibiting HDAC3 and maybe in part HDAC2, ENT blocks epigenetic suppression of an miRNA, thereby enabling its restriction of PAX3:FOXO1 expression and reducing the growth of aRMS tumors (Fig. 6K).

Fig. 6 miR-27a overexpression silences PAX3:FOXO1.

(A and B) Western blot of PAX3:FOXO1 protein abundance in Rh30 and CF-1 cells transfected with mimics of miR-27a (10 μM for 72 hours) or a negative control (Ctrl). Also shown is blotting in lysates from untransfected control cells (Unt). Blots are representative of n = 3 independent experiments. (C and D) Light microscopy images of Rh30 and CF-1 cells transfected with mimics of miR-27a. Images are representative of n = 3 independent experiments. Scale bars, 50 μM. (E to H) qPCR analysis of miR-27a (E and F) or miR-485p (G and H) expression in Rh30 and CF-1 cells treated for 72 hours with ENT (1 μM) or PAN (45 nM). (I and J) qPCR of miR-27a expression in Rh30 and CF-1 cells transfected with HDAC2- or HDAC3-targeted siRNA (100 nM). Data were normalized to U6snRNA expression. Gene expression was quantified using the 2−ΔCt method. Data are means ± SD (n = 3 independent experiments each in triplicate). *P < 0.05, **P < 0.01, and ***P < 0.001, two-sided Student’s t tests. (K) Summary of the HDAC3–SMARCA4–miR-27a–PAX3:FOXO1 regulatory circuit targeted by ENT.


Our previous studies uncovered that cell of origin conveys an epigenetic memory to tumor cells, leading to differences in histological phenotype and drug sensitivity, and that cell of origin also epigenetically influences the transcription of the PAX3:FOXO1 oncogene in tumor cells (13). Apropos to those findings, mechanistic studies that we performed here suggest that HDAC2/HDAC3 inhibition is the primary mechanism of ENT-mediated repression of PAX3:FOXO1 in aRMS. We also found that ENT and PAN, despite both being HDACis, are very different drugs with different global actions and specifically different activity with respect to SMARCA4, the inhibition of which we have now shown is critical to miR-27a expression and PAX3:FOXO1 mRNA destabilization.

Deregulated expression and function of miRNAs has been reported in many cancers, including RMS. HDACis and ENT specifically have been previously reported to alter miRNA expression (19, 20). In this study, we found an miRNA that inhibited PAX3:FOXO1 mRNA translation and/or its stability. Our focus was on miRNAs that could bind either the PAX3 promoter and exons 1 to 8 or the FOXO1 exons 1 to 3 and its 3′UTR, as well as those that were conserved across humans and mice, given that ENT reduced both human and mouse PAX3:FOXO1. miR-27a-5p targets both PAX3 and PAX7 (18); miR-27a’s own promoter is regulated in muscle by the myogenic transcription factors MYF6, MYOD, and MEF2C (which aligns with RMS features) (21); and SMARCA4 is predicted by ENCODE to bind the 3′UTR of miR-27a. Knockdown of SMARCA4 or pharmacological inhibition of its bromodomain activity increased the expression of miR-27a. Although SMARCA4 is typically characterized as a transcriptional activator, reports suggest that SMARCA4 can also act as a repressor (22, 23). How precisely SMARCA4 represses miR-27a requires further investigation and may reveal insight into its function in other contexts.

From a clinical point of view, ENT in combination with the chemotherapy agent VCR showed strong antitumor activity in aRMS orthotopic mouse models at clinically achievable adult drug concentrations. For the purposes of clinical trial planning, we have also established the comparative efficacy of ENT to other HDACis at reducing PAX3:FOXO1 and PAX7:FOXO1 expression. ENT not only reduced PAX3:FOXO1 abundance across species in both murine and human aRMS cell lines/cultures and PDX explant cell cultures but also inhibited PAX3:FOXO1 levels more effectively than the other HDACis tested. We note that, in the time since our studies were performed, Malempati et al. (24) have reported the phase 1 clinical trial results for ENT in pediatric patients, showing higher drug exposures and decreased clearance without any additional toxicity for children versus adults (Cmax of 140.8 nM at a dose of 4 mg/m2 and a half-life of 45 hours in children). Thus, our preclinical studies may be overly stringent with respect to clinical modeling.

In summary, our preclinical data with strong mechanical evidence suggest that targeting HDAC3–SMARCA4–miR-27a–PAX3:FOXO1 regulatory circuitry in aRMS may provide real therapeutic benefits for patients with aRMS. The preclinical efficacy in vivo and overall biochemical performance at PAX3:FOXO1 suppression make ENT a promising drug candidate for combining with chemotherapy and/or radiation in clinical trials for aRMS. Our findings directly support the pediatric phase 1B clinical trial ADVL1513 concept under review for the use of ENT as a single agent in RMS and possible phase 2 chemotherapy-ENT trials to follow.


Cell culture

Murine primary tumor cell cultures (U23674) were generated as described previously (25). Human aRMS cell lines Rh30 and Rh41 were cultured in growth medium (GM) RPMI 1640 (Thermo Fisher Scientific, catalog no. 11875-093) supplemented with 10% fetal bovine serum (FBS) (Thermo Fisher Scientific, catalog no. 26140079) and 1% penicillin/streptomycin (Thermo Fisher Scientific, catalog no. 15140-122). CW9019 was cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Sigma, catalog no. 11965-084) supplemented with 10% FBS and 1% penicillin/streptomycin. The mouse myoblast cell line C2C12 was obtained from the American Type Culture Collection (ATCC) and was cultured in DMEM supplemented with 20% FBS and 1% penicillin/streptomycin. Primary HSMMs (Lonza Inc., catalog no. CC-2580) were cultured in GM (Cell Applications, catalog no. 151-500). Phoenix-Amphotropic cells (a second-generation retrovirus producer cell line) (ATCC, catalog no. CRL-3213) were cultured in DMEM supplemented with 10% FBS and 1% antibiotics. PDX explant cell culture CF-1 was cultivated in RPMI 1640 supplemented with 10% FBS and 1% antibiotics. All cells were incubated at 37°C and 5% CO2. Rh30 and Rh41 were obtained from the Children’s Oncology Group ( and were authenticated by Short Tandem Repeat validation assay through Biosynthesis. The murine cell culture U23674 was authenticated by PCR validation by Transnetyx. CF-001 was authenticated by analyzing the expression of PAX3:FOXO1.

VCR chemotherapy, HDAC inhibition, and SMARCA4 inhibition

The chemotherapy agent VCR sulfate, used here in the treatment of Rh30 and U23674 cells, was obtained from Sigma (product no. V8879). HDACis used in this study were purchased from Selleckchem: ENT (also known as MS-275, a class I HDACi; catalog no. S1053), PAN (a broad-spectrum HDACi; catalog no. S1030), SAHA (catalog no. S1047), CUDC-907 (a PI3K inhibitor and a class I and class II HDACi; catalog no. S2759), and CUDC-101 (a class I and class II HDACi and an inhibitor of epidermal growth factor receptor family members EGFR and HER2; catalog no. S1194). aRMS cell lines and primary tumor cell cultures were treated with these drugs at their clinically relevant Cmax (maximum plasma concentrations) or, where Cmax is not yet reported, their determined IC25.

TR-FRET assay of the treatment of cells with the SMARCA4 inhibitor

The SMARCA4 bromodomain activity inhibitor, PFI-3 (S7315), was purchased from Selleckchem. Rh30 cells were treated with PFI-3 at a concentration of 10 μM. PFI-3 inhibits 95% of the SMARCA4 bromodomain activity at 10 μM based on the TR-FRET assay carried out by a commercial vendor, BPS Bioscience. Briefly, the assay was performed by TR-FRET technology using recombinant bromodomain and BET ligand. The TR-FRET signal from the assay is correlated with the amount of ligand binding to the bromodomain. Binding experiments were performed in duplicate at each concentration. The TR-FRET data were analyzed using the computer software GraphPad Prism (GraphPad Software Inc.). In the absence of the compound in wells containing BET ligand, the TR-FRET signal (Ft) in each dataset was defined as 100% activity. Select wells where control inhibitors were more than 100-fold the IC50 were used to define the TR-FRET signal (Fb) as 0% activity. The percent activity in the presence of each compound was calculated according to the following equation: % activity = [(F − Fb)/(Ft − Fb)] × 100, where F = the TR-FRET signal in the presence of the compound. The percent inhibition was calculated according to the following equation: % inhibition = 100 − % activity.

RNA-extraction and RT-PCR

Rh30 (human aRMS) cell line and murine aRMS primary tumor cell cultures (U23674) were treated with 0.1, 0.2, 0.4, 0.8, 1, and 2 μM ENT for 24 hours. DMSO treatment was used as a control. After treatment with ENT, total RNA was extracted and complementary DNA (cDNA) was synthesized as previously described (13). Expression of PAX3:FOXO1 was determined by real-time PCR (RT-PCR) using custom Taqman primers and probe (catalog nos. 4304970 and 4316034) on a StepOnePlus RT-PCR machine (Applied Biosystems). Rh30 and Rh41 cell lines were treated with 1 μM ENT, 45 nM PAN, and 1 μM SAHA for 24 hours. DMSO-treated cells were used as control. Total RNA was extracted from cells using TRIzol reagent (Invitrogen) according to the manufacturer’s protocol. cDNA was prepared from RNA using a high-capacity cDNA reverse transcription kit (Applied Biosystems) with a ribonuclease inhibitor. qPCR was performed using TaqMan Universal Master Mix, no AmpErase Uracil-N glycosylase on the ABI 7900HT Fast Real-Time PCR System (Applied Biosystems). Primers used were Gapdh-Hs02758991_g1 and PAX3:FOXO1-Hs03024825_ft. Gene expression was quantified using the 2−ΔCt method.


Human Rh30 and Rh41 cell lines and murine aRMS primary tumor cell cultures (U23674) were treated with 1 μM ENT, 45 nM PAN, 1 μM SAHA, 150 nM CUDC907, or 150 nM CUDC-101 for 72 hours. DMSO (vehicle)–treated cells were used as control. After 72 hours, lysates were collected using radioimmunoprecipitation assay lysis buffer with Halt Protease and Phosphatase Inhibitor Cocktail (Thermo Fisher Scientific, catalog no. 78440) and analyzed for PAX3:FOXO1 expression using an anti-PAX3 antibody (1:400; R&D Systems, catalog no. MAB2457), PAX7:FOXO1 expression using an anti-FKHR antibody (H-128, 1:300; Santa Cruz Biotechnology), and SMARCA4 expression using an anti-SMARCA4 antibody (G-7, 1:300; Santa Cruz Biotechnology) and matched for protein expression using an anti–β-actin antibody (1:10,000; Abcam, catalog no. ab8227). All blots were visualized using a FluorChem Q system (ProteinSimple). To analyze endogenous wild-type FOXO1 expression, recombinant FOXO1 (NM_002015) (Origene) was used. FOXO3 (NM_001455) was used as negative control to choose the appropriate antibody (fig. S7). Anti-FOXO1 antibody (1:500; Millipore, clone 2H8.2) detects recombinant FOXO1 without background compared to other antibodies and subsequently used for all immunoblots (fig. S7).

Orthotopic allograft studies

All animal studies were conducted with Institutional Animal Care and Use Committee (IACUC) approval at the Oregon Health & Science University. An orthotopic allograft mouse model of aRMS (U23674, genotype Myf6Cre, PAX3:FOXO1, p53) was generated as previously described (13). Mice were treated with ENT at a daily dose of 5 mg/kg by intraperitoneal injection or with VCR sulfate at a dose of 1 mg/kg weekly by intraperitoneal injection, or a combination of both. Treatment was started once the tumors reached 0.25 cc and ended when the tumors reached 1.5 cc. During treatment, mice that experienced body weight loss (10 to 15%) were euthanized early. For the radiation studies, murine aRMS primary tumor cell cultures (U23674) were pretreated with either DMSO or 2 μM ENT and then subjected to 10 Gy radiation before injecting 500,000 viable cells into each mouse. Failure was defined as an event for tumor size greater than or equal to 1.2 cc.

PDX models at Champions Oncology

The Champions Personalized Tumorgraft chemosensitivity tests were conducted using a TumorGraft model established from two independent RMS biopsy specimens. The explants were received and immediately implanted into immunodeficient mice. The antitumor activity of ENT and VCR was tested in a low-passage immune-compromised female mice (Harlan; nu/nu) between 5 and 8 weeks of age housed on irradiated papertwist-enriched 1/8-inch corncob bedding (Shepherd) in individual HEPA (high-efficiency particulate air) ventilated cages (Innocage IVC, Innovive, USA) on a 12-hour light/12-hour dark cycle at 68° to 74°F (20° to 23°C) and 30 to 70% humidity. Animals were fed water ad libitum (reverse osmosis, 2 parts per million Cl2) and an irradiated test rodent diet (Teklad 2919) consisting of 19% protein, 9% fat, and 4% fiber. All compounds were formulated according to the manufacturer’s specifications. Beginning day 0, tumor dimensions were measured twice weekly by a digital caliper, and data including individual and mean estimated tumor volumes (mean TV ± SEM) were recorded for each group; tumor volume was calculated using the following formula (1): TV = width2 × length × 0.52. All studies were done with the approval of Champions Oncology IACUC.

PDX models at The Jackson Laboratory

To establish each PDX model, NSG (NOD.Cg-Prkdcscid IL2rgtm1Wjl/SzJ) mice were obtained by The Jackson Laboratory. Tumor explants were obtained from the patients and immediately implanted into the rear flanks of recipient female NSG (JAX, stock no. 005557) mice using a trochar. Once tumors reached about 2000 mm3, they were collected and passaged for serial transplantation in NSG mice to create low-passage fragments or cohort for future studies. The tumor volume range for enrollment was 150 to 250 mm3. All studies were done with the approval of The Jackson Laboratory IACUC. Mice were treated with vehicle or ENT/VCR as a single agent or a combination at doses and routes of administration provided in table S2 until tumors reached 2000 mm3 or reached study day 28. The antitumor activity of ENT and VCR was tested. All compounds were formulated according to the manufacturer’s specifications. Beginning day 0, tumor dimensions were measured twice weekly by a digital caliper, and data including individual and mean estimated tumor volumes (mean TV ± SEM) were recorded for each group; tumor volume was calculated using the following formula: TV = width2 × length/2.

siRNA-mediated silencing of class I and class II HDACs

For silencing of class I HDACs (13), Rh30 cells were transfected with 100 nM HDAC1 (L-003493-00-0005, 5 nmol), HDAC2 (L-003495-02-0005, 5 nmol), HDAC3 (L-003496-00-0005, 5 nmol), or HDAC10 (L-004072-00-0005, 5 nmol) SMARTpool siRNA reagent (a pool of four siRNA duplexes all designed to target distinct sites within the specific gene of interest) (Dharmacon) versus scrambled siRNA (Dharmacon) using Lipofectamine RNAimax (Invitrogen, catalog no. 13778030). siRNA cell lysates were subjected to Western blotting using anti-HDAC1 (H-51, 1:400; Santa Cruz Biotechnology), anti-HDAC2 (C-8, 1:400; Santa Cruz Biotechnology), anti-HDAC3 (B-12, 1:400; Santa Cruz Biotechnology), and anti-HDAC10 (E-2, 1:400; Santa Cruz Biotechnology) antibodies and PAX3:FOXO1 expression using anti-PAX3 antibody (R&D Systems).

Retroviral infection

To restore expression of PAX3:FOXO1 in U23674 with stable knockdown of PAX3:FOXO1 (U23674 shYFP), retroviral vector pk1-PAX3:FOXO1 was transfected into Phoenix-AMPHO packaging cells using Lipofecatmine Plus reagent (Invitrogen, catalog no. 15338100). Medium was harvested after 24, 48, and 72 hours and filtered with a 0.45-μm syringe filter. Retroviral supernatant with polybrene was used for transduction of U23674 shYFP culture.

siRNA-mediated silencing of SMARCA4

For silencing of SMARCA4, Rh30 cells were transfected with 100 nM SMARCA4 (L-010431-00-0005, 5 nmol) SMARTpool siRNA reagent (Dharmacon) versus scrambled siRNA (Dharmacon) using Lipofectamine RNAimax (Invitrogen). Cell lysates were subjected to Western blotting using anti-SMARCA4 (G-7, 1:300; Santa Cruz Biotechnology).

miRNA overexpression

For overexpression of miR-27a-5p, Rh30 and CF-1 cells were transfected with 10 μM miR-27-a-5p mimics (assay name, hsa-miR-27a-5p; accession no., MI0000085) or miR-27a-5p negative control (5 nmol; Thermo Fisher Scientific, catalog no. 4464058) using Lipofectamine RNAimax (Invitrogen). Cell lysates were collected at the end of 72 hours and subjected to Western blotting for analysis of PAX3:FOXO1 expression using an antibody against PAX3 (R&D Systems).

miRNA isolation and RT-PCR

miRNA isolation was performed as per the manufacturer’s protocol (Invitrogen, catalog no. K157001). Briefly, Rh30 and CF-1 cells were seeded at a density of 0.3 million cells and, the next day, were treated with ENT (1 μM) or DMSO (vehicle). At the end of 24 hours of treatment, miRNA was isolated as per the manufacturer’s standard protocol. The recovery tube with RNA was then quantified using a BioTek microvolume plate. Complementary DNA was prepared from 5 ng of miRNA using a TaqMan miRNA reverse transcriptase kit (Thermo Fisher Scientific, catalog no. 4366596). qPCR was performed using TaqMan Universal Master Mix, no AmpErase UNG (Thermo Fisher Scientific, catalog no. 4440040) on the Bio-Rad Thermocycler Real-Time PCR System. Primers used were hsa-miR-27a and hsa-miR-485 (TaqMan miRNA assays). U6snRNA (TaqMan miRNA assays) was used as loading control. miRNA expression was quantified using the 2−ΔCt method.

RNA sequencing

For the identification of transcriptional changes (Fig. 3, C to F) in aRMS cells after ENT treatment, each of the three aRMS cultures (human cell lines Rh30 and Rh41 and mouse cell culture U23674) was treated with ENT alongside a paired untreated sample for a fixed time period. All samples were treated with ENT for 72 hours, except for Rh41, which was treated for 24 hours because of higher sensitivity to ENT. All cells were cultured on 10-cm dishes, and treatment began when plates were 60% confluent. Passages lower than 7 were used for all mouse cultures. RNA isolation and sequencing were performed by the commercial service provider, Beijing Genomics Institute (BGI). Differential expression for a single sample was defined as post-ENT treatment expression divided by vehicle-treated expression. These criteria identified 348 overexpressed and 358 underexpressed genes in aRMS. Differential expression of genes associated with chromatin modification and remodeling after ENT and PAN treatment (72 hours) (Fig. 5A) was analyzed by one more set of RNA-seq with RNA isolation and sequencing performed by BGI. Log2 scaled ratios of ENT-treated versus vehicle-treated RH30 and U23674 and PAN-treated versus vehicle-treated RH30 and U23674 were organized by change in regulation.

Bioinformatic analysis of RNA-seq

The paired-end raw reads for all RNA-seq data samples were aligned using TopHat version 2.0.9. Up to two mismatches in the alignment were permitted before a read alignment was discarded. The reads for human samples were aligned to the UCSC hg36 human reference genome. The reads for mouse samples were aligned to the UCSC mm19 mouse reference genome. The aligned reads were assembled into transcripts using Cufflinks version 2.2.1. Differential comparisons were performed by the Cuffdiff function of Cufflinks version 2.2.1. Differential comparisons were made between samples treated with ENT and untreated control samples. For the differential analysis of treated versus untreated samples, the standard Cuffdiff parameters were used. Reported in all figures are log2 scaled differential values. Genes for which treated samples had nonzero expression and untreated samples had zero expression fixed were set to the minimum of the treated expression and 1024, before log2 scaling, to prevent divide-by-zero errors during downstream analysis. Only genes with quantified expression across all samples (total of 14,575 genes) were considered. Log2 scaled mean differential expression across all samples of a common subtype was used to identify overexpression and underexpression.

Chromatin immunoprecipitation–exonuclease

For identification of ENT-related gene expression changes for the known targets of PAX3:FOXO1, ChIP-exo studies were performed by Peconic Genomics using anti-PAX3 antibody (R&D Systems, catalog no. MAB2457) with the murine U23674 aRMS culture as previously described (26). Paired-end, 40–base pair (bp) reads were generated. Putative binding sites (peaks) were identified using the Model-based Analysis of ChIP-seq (MACS) program version, with default parameters without a control sample (in which case MACS uses regional read levels for background) (27). Overall data quality and peak quality were verified by a combination of methods, including cross-correlation of positive- and negative-strand reads (which showed a peak expected from punctate binding at a plausible offset of about 15 bp) (26) and the R package ChIPQC (which showed good numbers of reads in peaks) (28). ChIP-seq data for additional ENT targets HDAC1, HDAC2, HDAC3, and HDAC11 were taken from published sources (29, 30). HDAC11 was used as a surrogate for HDAC10, because HDAC10 data were not available. Additional data were gathered from published sources for PAX3:FOXO1 binding identified by cyclic amplification and selection of targets (CASTing) (31) and by previously published PAX3:FOXO1 ChIP-seq data (32). Published expression data (33) and cell cycle–dependent expression data (10) were also integrated to infer indirect targets of PAX3:FOXO1. Differential expression across samples after ENT treatment was used to identify key up-regulated and down-regulated genes in aRMS. Bioinformatic analysis of PAX3:FOXO1 and HDACs binding sites was then cross-referenced for RNA-seq data for ENT-treated cells.

Statistical analysis

Bioinformatics and computational methods are described above. Continuously distributed outcomes were summarized with the mean ± 1 SD. Treatment groups were contrasted on the mean with analysis of variance in log units. Time-to-event distributions were summarized with Kaplan-Meier curves, and significance of variation with treatment group was assessed with log-rank tests. Corrections for multiple comparisons were made with the Dunnett method for analysis of variance (ANOVA) and the Bonferroni method for log-rank testing. Statistical testing on means and time to event was two sided with a nominal significance level of 5% and was carried out with R. For PDX mouse models, the significance of variation in tumor volume with treatment was assessed with a repeated-measures linear model with an autoregressive order 1 autocorrelation matrix and a Tukey correction for multiple comparisons in terms of treatment, day, and the treatment × day interaction. All analyses were carried out in log10 units, and all statistical testing was two sided with a 5% experiment-wise significance level. SAS version 9.4 for Windows (SAS Institute) was used throughout. GraphPad Prism was used for statistical analysis of cell viability assay and TR-FRET assay. For densitometric analysis (% change), significance was determined by a two-tailed Student’s t test, and P values of <0.05 were considered to be statistically significant. Statistical significance was set at *P < 0.05, **P < 0.01, and ***P < 0.001. Error bars indicate mean ± SD or SEM.


Fig. S1. ENT treatment of aRMS in vivo.

Fig. S2. ENT reduces PAX3:FOXO1 abundance at a clinically relevant dose and time of exposure and reduces cell viability synergistically with VCR.

Fig. S3. PAX3:FOXO1 mRNA expression is reduced by ENT in aRMS cell lines.

Fig. S4. Better than other HDACis, ENT reduces PAX3:FOXO1 abundance in murine and human aRMS cells.

Fig. S5. Representative histology of PDX mouse aRMS tissue.

Fig. S6. PFI-3–mediated inhibition of SMARCA4 bromodomain activity.

Fig. S7. Validation of commercially available antibodies for detecting endogenous FOXO1 abundance in aRMS cell lines.

Table S1. Histological markers of differentiation in aRMS orthotopic allograft PDX mice.

Table S2. Treatment schedules for the CF-4/PCB-513 PDX mice.

Table S3. Patient history of the Champions Oncology PDX aRMS models.

Table S4. Patient history of The Jackson Laboratory PDX aRMS models.

Table S5. Histological scoring of markers of differentiation in PDX aRMS mice.

Table S6. Statistical analysis of CTG-1604/POS-14175 data.

Table S7. Statistical analysis of J101220/CF-4 data.

Table S8. Statistical analysis of J77636/PCB-481 data.

Table S9. Statistical analysis of J0103366/CF-13A data.

Table S10. Statistical analysis of J099761/CF-1 data.

Table S11. Statistical analysis of CTG-1409/POS 14107 data.

Table S12. Statistical analysis of J099873/CF-2 data.

Table S13. Statistical analysis of CTG-1008 data.


Acknowledgments: We are grateful to the families who shared the PDX model data. We thank K. Kikuchi for assistance. Funding: This work was supported by NIH grants 5R01CA189299,1R01CA143082 P30CA006973, UL1TR001079, and UL1TR001079; the St. Baldrick’s Foundation; the Braver, Stronger, Smarter Foundation; the Michelle Paternoster Foundation for Sarcoma Research;; the Christina Renna Foundation; the Friends of Doernbecher Foundation; the Friends of T.J. Foundation; the Clarke Gilles Foundation; and a private anonymous gift in memory of Nanette. Additional funding was provided to F.G.B. by the Intramural Research Program of the National Cancer Institute. Author contributions: C.K., C.R.V., D.S.H., E.W., J.A., J.E.M., J.K., L.X., M.A.R., N.B., N.E.B., N.M.A., P.O., and T.P.S. participated in the design or interpretation of the experimental results. A.A., A.P.H., A.M., A.M.D., B.A.G., B.S.H., C.K., C.N., C.R.V., D.B., E.W., E.R.R., H.P.B., J.A., J.E.M., J.K., J.L., K.Z., L.X., M.A.R., M.M., M.N.S., M.W.G., N.B., N.C.G., N.E.B., N.M.A., R.P., T.J.P., T.P.S., Y.I., and Z.B. participated in the acquisition or analysis of data. B.C., F.G.B., J.E.H., and J.E.W. contributed resources to these studies. C.K., N.B., and T.P.S. participated in writing the manuscript. C.K. directed the studies. Competing interests: C.K. received an unrestricted grant from Syndax Pharmaceuticals that supported third-party testing of ENT in PDXs. P.O. is an employee of Syndax Pharmaceuticals. J.E.W. is an employee of AbbVie Pharmaceuticals. Unrelated to this study, C.K. has had sponsored research agreements with Eli Lilly and Roche-Genentech, and N.E.B. is a scientific officer at First Ascent Biomedical Corp. C.R.V. is an advisor to KSQ Therapeutics and receives research funding from Boehringer-Ingelheim. D.S.H. has received reimbursement for travel to Medical Advisory Board meetings for Loxo Oncology, Bristol-Myers Squibb, Celgene, and Bayer but no compensation otherwise. All other authors declare that they have no competing interests. Data and materials availability: RNA-seq data have been deposited in the NCBI Gene Expression Omnibus Database (GEO accession number GSE115698). All other data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.

Stay Connected to Science Signaling

Navigate This Article