Selective Targeting of Myoblast Fusogenic Signaling and Differentiation-Arrest Antagonizes Rhabdomyosarcoma Cells

Abstract

Rhabdomyosarcoma (RMS) is an aggressive soft tissue malignancy comprised histologically of skeletal muscle-lineage precursors that fail to exit the cell-cycle and fuse into differentiated syncytial muscle - the underlying pathogenetic mechanisms for which remain unclear. In contrast to myogenic transcription factor signaling, the molecular machinery that orchestrates the discrete process of myoblast fusion in mammals is poorly understood, and unexplored in RMS. The fusogenic machinery in Drosophila, however, is understood in much greater detail, where myoblasts are divided into two distinct pools: Founder Cells (FCs) and fusion competent myoblasts (fcms). Fusion is heterotypic and only occurs between FC and fcms. Here, we interrogated a comprehensive RNA-seq database and found that human RMS diffusely demonstrates an FC-lineage gene signature, revealing that RMS is a disease of FC-lineage rhabdomyoblasts. We next exploited our Drosophila RMS-related model to isolate druggable FC-specific fusogenic elements underlying RMS, which uncovered the Epidermal Growth Factor Receptor (EGFR) pathway. Using RMS cells, we showed that EGFR inhibitors successfully antagonized RMS RD cells, while other cell lines were resistant. EGFR inhibitor-sensitive cells exhibit decreased activation of the EGFR intracellular effector Akt, while Akt activity remained unchanged in inhibitor-resistant cells. We then demonstrate that Akt inhibition antagonizes RMS – including RMS resistant to EGFR inhibition – and sustained activity of the Akt1 isoform preferentially blocks rhabdomyoblast differentiation potential in cell culture and in vivo. These findings point towards selective targeting of fusion- and differentiation-arrest via Akt as a broad RMS therapeutic vulnerability.

PRECIS

EGFR and its downstream signaling mediator AKT1 play a role in the fusion and differentiation processes of rhabdomyosarcoma (RMS) cells, representing a therapeutic vulnerability of RMS.

INTRODUCTION

Rhabdomyosarcoma (RMS) is a well-known problem in pediatric oncology, as children with high-risk RMS endure a 3-year event-free survival rate of only 20% (1). Histologically, RMS is comprised of neoplastic skeletal muscle-lineage precursors that fail to exit the cell cycle and terminally differentiate. Rare for a somatic tissue, skeletal muscle requires that precursor cells not only undergo lineage-specific differentiation, but also fuse and form a syncytium. Though critical effort has been giving to deciphering myogenic signaling in the settings of both muscle development and RMS, the mechanisms orchestrating mammalian myoblast fusion are poorly understood.

In Drosophila, myoblast fusion is understood in greater detail (2), where myoblasts are divided into two pools: Founder Cells (FCs) and fusion competent myoblasts (fcms). FCs are seminal, establishing the position of each myofiber, while fcms seek out FCs and fuse. FC-fcm recognition is mediated by IgSF receptors – the Kirre subfamily, unique for FCs, and the fcm-specific Nephrin subfamily. Upon FC-fcm adhesion, the FC lineage-restricted adaptor molecule Rols links the transmembrane signal to the cytoskeleton and drives downstream fusion events. We and others have shown that Kirrel, Nephrin, and Rols orthologs participate in vertebrate myoblast fusion (3–5). We additionally have found that overexpression of mammalian Rols, named TANC1, influences RMS pathobiology (5).

Unknown, however, is whether misexpression of the FC program broadly underlies RMS. Additionally, identifying fusion regulators relevant to RMS for which inhibitors are available would suggest new therapeutic opportunities.

MATERIALS AND METHODS

Additional information can be found in Supplemental Methods.

Drosophila Genetics

Transgenes and screen methods were as previously described (6, 7).

Cell culture and reagents

Cells lines were handled as previously described (5). RMS lines were used up to 30 passages, C2C12 up to 20. STR profiles for all lines were obtained at UTSW’s sequencing core and verified as authentic. Mycoplasma testing regularly was negative. Lines were obtained from: C2C12, ATCC; Rh30, M. Hatley (St. Jude); RD, E. Olson (UTSW); SMS-CTR, C. Linardic (Duke). Please see Supplemental Methods for information regarding drugs and vehicles used, and shRNA (Dharmacon) sequences. MTT assays were performed using the Vibrant™ MTT Assay Kit (V-13154) (Molecular Probes/Invitrogen). TUNEL was performed using the DeadEnd™ Fluorometric TUNEL kit (Promega).

Indexes were calculated from cells cultured for six days in differentiation medium (see Supplemental Methods) from three independent experiments. For differentiation, the percentage of nuclei in MHC-positive tissue were scored. For fusion, the percentage of nuclei present in MHC-positive bi- or multi-nucleated myotubes were counted. For proliferation, the percentage of Ki67-positive cells were scored. For tumor sections, mitotic figures or Ki67-positive nuclei were scored.

Xenografts

Studies were supervised and approved by UTSW’s Institutional Animal Care and Use Committee. Drugs were administered (see Supplemental Methods for dosing) when tumor size reached ~100 mm³. An event was based on Pediatric Preclinical Testing Program criteria: tumor volume quadrupling from a base volume [here, 200 mm³ for RD (slower growing), 250 mm³ for RH30 (faster growing)].

Statistics

Type I error was evaluated by two-tailed Student’s t test. P values less than 0.05 was considered significant. Type II was evaluated by Achieved Statistical Power (post hoc) analysis, values greater than 0.80 was considered significant. Data are mean ± SEM. Software used were Excel (Microsoft), Prism 7 (GraphPad), and G*Power 3 (Heinrich-Heine-Universität).

RESULTS

To probe for RMS FC/fcm-gene expression levels, we surveyed the Oncogenomics RNA-seq database, derived from an extensive collection of human RMS specimens (8). As mentioned above, each Drosophila myofiber forms from one FC cell, with the remainder of the syncytium comprised of sequentially fused fcms. Thus, fcms dramatically outnumber FCs. However, when querying FC-marker expression levels in the panel, we found that TANC1 and KIRREL1 transcripts were broadly abundant in RMS negative and positive for the PAX-FOXO (PF) oncoprotein, while KIRREL3 [the encoding gene for which possesses a PF transcriptional activation site] is overexpressed in PF-positive RMS (Fig. 1A) (9). In contrast, expression levels of the fcm-NEPRIN orthologs, NPHS1 and NPHS2, are downregulated (Fig. 1A). These findings are consistent with our previous immunohistochemical analysis of RMS tumors, which showed diffuse positivity for TANC1 (5). These data show that RMS associates with the FC-signature.

Fig. 1. FC-genes in RMS.

(A) RMS demonstrates an FC- signature. Relative abundance levels in a human RMS tumor cohort profiled by RNA-seq. DESMIN is a muscle-specific intermediate filament, while GFAP and KRT20 are filament markers for glial and gastrointestinal adenocarcinoma neoplasms, respectively. PAX3-variant fusion (PAX3-INO80D or -NCOA1) specimens are shown separately.

(B) EGF or EGFR loss-of-function alleles suppress PAX-FOXO1. Based on Mendelian ratios, the F₁ population should be 50% control and 50% PAX7-FOXO1-expressing adults (“Expected”). PAX7-FOXO1 causes lethality, as PAX7-FOXO1 adults comprise ~20% of F₁ adults (“control”; n = 124). Chromosomal deletions Df(2L)pr-A16 (n = 47) or Df(2R)Excel⁶⁰⁷⁶ (n = 77) suppresses PAX-FOXO1 lethality, as do two EGF (named spitz in Drosophila) [spi^DG04705 (n = 55) spi^s3547 (n = 66)] or two EGFR [Egfr^f2 (n = 66) and Egfr^t1 (n = 60)] loss-of-function alleles (though the Egfr^t1 allele showed a P value of 0.067). rols^P1729 is the loss-of-function allele previously isolated as a PAX-FOXO1 suppressor. Df(2L)ed1 (n = 160) (third column) and DF(3R)23D1 (n = 74) (red column) are unrelated chromosomal deletions included as controls to demonstrate examples of a non-modifier and genetic enhancer, respectively. P values: *P < 0.05, **P < 0.01 versus Control.

To isolate potentially targetable FC-signaling elements, we turned to our Drosophila PAX-FOXO1 model (6, 7), which we exploit to uncover new influential RMS genetic elements (5, 10)]. We identified two chromosomal deletions, Df(2L)pr-A16 and Df(2R)Excel⁶⁰⁷⁶, that suppress PAX-FOXO1-based lethality (Fig. 1B). Df(2L)pr-A16 deletes EGFR, while Df(2R)Excel⁶⁰⁷⁶ deletes EGF (named spitz in flies). As EGFR signaling is known to drive naïve Drosophila myoblasts to the FC-differentiation program we tested individual loss-of-function alleles in EGFR or spitz, which similarly suppressed PAX-FOXO1 lethality (Fig. 1B). These data identify EGFR signaling – a druggable pathway – as a potential FC-based RMS target.

To extrapolate this finding to mammals, we established that EGFR signaling is active and regulated in wild-type cultured myoblasts (Fig. S1A). We questioned whether the EGFR inhibitors Erlotinib (EGFR tyrosine kinase inhibitor) or Cetuximab (humanized monoclonal interfering antibody) antagonize RMS, utilizing the PF-negative RD and RH36 cell lines, and the RH30 PF-positive line (Fig. S1B). We profiled viability for each line against each agent (Fig. S1C) [IC₅₀ values were similar to human carcinoma cells (11, 12)] and demonstrated that each inhibitor antagonized EGFR. Both inhibitors interfered with RD cell proliferation, increased Myosin Heavy Chain (MHC)-positive terminal differentiation (Fig. 2A & B) (Fig. S1D and E), and blocked anchorage independent growth (Fig. S1F and G). TUNEL assay for apoptotic cell death was negative (Fig. S1H). In vivo, tumorigenesis was inhibited, event-free survival increased, mitotic activity decreased, and MHC-expression enhanced (Fig. 2C & D) (Fig. S1I–L). RH36 and RH30 cells, however, were not antagonized (Fig. S1M).

Fig. 2. Erlotinib and Cetuximab block tumorigenicity in RD RMS cells.

(A,B) Erlotinib- or Cetuximab-treated RD cells show decreased proliferation and enhanced differentiation. Cells were stained with Ki67 or MHC antibody, and DAPI. Erlotinib concentration = 10 μM, Cetuximab concentration = 1 μg/mL.

(C,D) Erlotinib or Cetuximab antagonizes tumorigenesis. Shown are tumor growth and event-free survival (see Methods) for “Control” (6% Captisol) (n = 3) versus Erlotinib-treated (n = 4) (Panel C), and “Control” (PBS) (n = 3) versus Cetuximab-treated (n = 3) tumors (Panel D). Achieved Statistical Power for “tumor volume” and “event-free analyses were 0.99 and 0.82, respectively, for the Erlotinib study; for Cetuximab, 0.97 and 0.95, respectively. Myosin Heavy Chain IHC shows enhanced differentiation within Erlotinib- or Cetuximab-treated xenografts.

Scale bar = 100 μm. P values: *P < 0.05 versus Control.

As the RD line carries an oncogenic N-RAS mutation (Q61H), we hypothesized that an intracellular signaling pathway other than RAS must be antagonized upon Erlotinib/Cetuximab treatment. We analyzed inhibitor-treated cells and observed downregulated Akt activation, whereas MEK/MAPK or STAT3 activation levels showed no decrease (Fig. S2A). We next found that Akt activation levels remained unaltered in EGFR-inhibitor resistant RH36 and RH30 cells (Fig. S2B). These results infer that Akt is a critical RMS effector.

To test this notion, we treated the RMS cell lines with an allosteric Akt inhibitor, MK-2206 [IC₅₀ values were similar to human carcinoma cells (Fig. S3A) (13)]. We additionally included a fourth line, SMS-CTR (PF-negative), found to be EGFR inhibitor resistant (Fig. S3B). We observed potent blockage of Akt activation in each cell line (Fig. S3C), with all lines now exhibiting decreased proliferative activity and anchorage independent growth (Fig. 3A and B) (Fig. S3D) (note - RH36 cells did not form colonies in soft agar). TUNEL assays performed on RD and RH30 MK-2206-treated cells were negative (Fig. S3E). In vivo, we observed inhibited tumorigenesis, increased event-free survival, and decreased proliferation (Fig. 3C) (Fig. S3F and G). No difference in MHC-positivity was observed, however (Fig. S3H). Together, these findings point towards Akt as a broadly targetable RMS vulnerability.

Fig. 3. MK-2206 blocks tumorigenicity in PF-negative and -positive RMS.

(A) MK-2206 antagonizes RMS proliferation in culture. RD, RH36, RH30, and SMS-CTR cells were each cultured with MK-2206 (0.5 uM) and stained with Ki67 antibody and DAPI.

(B) MK-2206-treated RMS cells show decreased colony formation in soft agar. Shown are average number of colonies per 20×-objective field.

(C) MK-2206 antagonizes tumorigenesis. Shown are tumor growth and event-free survival (see Methods) plots for “Control” (15% Captisol) (n = 4) and MK-2206-treated (n = 4) RD or RH30 tumors. Achieved Statistical Power for RD and RH30 tumor volume and event-free survival analyses were 1.00 and 0.95 (RD), and 1.00 and 0.93 (RH30), respectively.

P values: **P < 0.01, ***P < 0.001 versus Control.

Though transcripts for Akt½/3 are detectable in human skeletal muscle, only Akt1 and 2 protein are detected (14). As Akt1 has been shown to function early in myogenesis and promote myoblast proliferation, while Akt2 downstream directs myoblast fusion and differentiation (15), we hypothesized that the MK-2206 Akt inhibitor, though promoting RMS cell cycle exit, failed to induce RMS rhabdomyoblast fusion and differentiation due to dual blockage of Akt½. As differing roles for Akt½ in RMS are unexplored, we silenced Akt1 or Akt2 (Fig. S4A) and found that Akt1-silenced RD cells exhibited a marked rescue of fusion and differentiation potential when compared to control or Akt2-silenced cells (Fig. 4A and B). In vivo, tumorigenesis was inhibited, event-free survival increased, mitotic activity decreased, and MHC-expression enhanced (Fig. 4C) (Fig. S4C). These results reveal that sustained Akt1 activity preferentially influences the failure of RMS cells to complete the myogenic developmental program, and that targeting of Akt1 is sufficient to rescue RMS cell differentiation-arrest.

Fig. 4. Akt1 silencing rescues RMS cell differentiation-arrest in vitro and in vivo.

(A,B) Akt1 silencing rescues RD cell fusion- and differentiation-arrest. RD cells expressing shRNA (transient transfection) against GFP (Control), Akt1, or Akt2 are shown, stained with MHC antibody and DAPI (panel A). Fusion and differentiation indexes (panel B) show that Akt1-silenced RD cells exhibited rescue of fusion and differentiation when compared to control or Akt2-silenced cells.

(C) Akt1 silencing antagonizes tumorigenesis. Stable RD cell lines that conditionally express shRNA against eGFP, Akt1, or Akt2 upon doxycycline administration were generated. Shown are tumor growth and event-free survival plots for “Control” (shRNA>>eGFP) (n = 4), shAkt1 (shRNA>>Akt1) (n = 3), and shAkt2 (shRNA>>Akt2) expressing (n = 3) RD tumors. Achieved Statistical Power for RD Control versus Akt1 tumor volume and event-free survival analyses were 0.97 and 0.97, respectively.

P values: *P < 0.05, **P < 0.01, *** P < 0.001 versus Control.

DISCUSSION

We previously reported that correcting FC-lineage TANC1 overexpression induces RMS cells to terminally differentiate (5). Here, utilizing the Oncogenomics database, we now reveal that RMS broadly demonstrate the FC-program gene signature. Collectively, these findings argue that misexpression of FC-programming is a common RMS mechanism. Since it remains unclear the extent to which the lineage-restricted process of Drosophila myoblast fusion is precisely conserved in mammals, unknown is whether human RMS tumor initiation occurs in FC-lineage cells, or whether misexpression of the FC-program occurs downstream during tumor progression.

Utilizing our Drosophila model, we probed for FC-elements that possess druggable human orthologs and uncovered EGFR, which [though studied in cultured RMS cells (16, 17)] has not been functionally probed in vivo. Though EGFR inhibitors demonstrated efficacy against RD cells, EGFR inhibition was ineffective against the remaining lines tested. As numerous receptor kinases (e.g., FGFR4, c-MET) have been shown to influence RMS (1), we speculate that the RMS cells resistant to EGFR inhibition do not rely upon EGFR for growth-promoting signaling. Interestingly, we note that a subset of fly myoblasts utilizes an FGFR4 ortholog for FC program activation. We thus speculate that PF-positive RMS – FGFR4 is a direct target of the PF transcription factor – and the subset of PF-negative RMS possessing activating FGFR4 mutations [~10% (8, 18)] instead rely upon FGFR4 for FC-program dysregulation. Whether EGFR inhibitor sensitivity is common or limited in PF-negative RMS remains an open question.

MK-2206, however, broadly antagonized RMS, including RMS driven by oncogenic N-RAS (RD cells), mutationally activated FGFR4 (RH36), PAX3-FOXO1 (RH30), and oncogenic H-RAS (SMS-CTR) (19). Though Akt-mediated phosphorylation has been shown to inhibit the activity of PF (20), other studies have demonstrated that synthetic lethality in RMS can be induced by dual inhibition of the PI3K pathway (PI3K/mTOR or TORC½ inhibition) and either RAS or Hedgehog signaling (21–24), and that PI3K/mTOR inhibitors demonstrate efficacy against FGFR4-driven RMS (25). Here we newly reveal that mono-targeting of Akt is both effective and sufficient to antagonize RMS and point towards Akt as a critical RMS nodal point. We additionally found that sustained Akt1 activity preferentially incites RMS differentiation-arrest, suggesting that inhibitors specific for Akt1 would be similarly effective against RMS, and presumably with less overall toxicity than pan-Akt inhibitors.

Focusing on the mechanisms that underlie RMS cell differentiation-arrest, the read-outs surveyed in these studies focus on myoblast maturation, and not cytotoxicity. Thus, we suggest that scoring for tumor regression (which requires cytotoxicity) is not the most appropriate metric, and that event-free-survival is a better preclinical gauge. This notion differs in part from the Pediatric Preclinical Testing Program (PPTP), which tests for cytotoxicity and tumor regression, and thus reported MK-2206 as not inducing greater than 50% tumor volume regression. Neither Erlotinib or Cetuximab has been tested by the PTPP. We next anticipate testing EGFR and Akt inhibitors in the context of conventional chemotherapy agents, as an emphasis has been placed on identifying agents that enhance outcomes in combination with current therapeutic protocols. As both MK-2206 and Erlotinib have been successfully Phase I tested in pediatric patients (26, 27), we speculate that these agents, when combined with established protocols, will improve RMS outcomes.