Alveolar rhabdomyosarcoma (aRMS) is an aggressive sarcoma of skeletal muscle characterized by expression of the paired box 3-forkhead box protein O1 (PAX3-FOXO1) fusion oncogene. Despite its discovery nearly two decades ago, the mechanisms by which PAX3-FOXO1 drives tumor development are not well characterized. Previously, we reported that PAX3-FOXO1 supports aRMS initiation by enabling bypass of cellular senescence checkpoints. We have now found that this bypass occurs in part through PAX3-FOXO1-mediated upregulation of RASSF4, a Ras-association domain family (RASSF) member. RASSF4 expression was upregulated in PAX3-FOXO1-positive aRMS cell lines and tumors. Enhanced RASSF4 expression promoted cell cycle progression, senescence evasion, and tumorigenesis through inhibition of the Hippo pathway tumor suppressor MST1. We also found that the downstream Hippo pathway target Yes-associated protein 1 (YAP), which is ordinarily restrained by Hippo signaling, was upregulated in RMS tumors. These data suggest that Hippo pathway dysfunction promotes RMS. This work provides evidence for Hippo pathway suppression in aRMS and demonstrates a progrowth role for RASSF4. Additionally, we identify a mechanism used by PAX3-FOXO1 to inhibit MST1 signaling and promote tumorigenesis in aRMS. Introduction Rhabdomyosarcoma (RMS) is the most common soft tissue sarcoma of childhood and adolescence. Over the past 30 years, clinical trials in North America, Europe, and Australia have identified superior treatment strategies leading to the improved survival of discrete groups of RMS patients. A distinctly worse outcome is encountered for patients with the alveolar histologic variant of RMS (aRMS), who have a 5-year survival rate of less than 50% (1). Even more dismal is the survival for those whose tumors express the signature paired box 3-forkhead box protein O1 (PAX3-FOXO1) fusion gene; in the metastatic setting, their survival rate at 4 years is less than 10% (2). Although PAX3-FOXO1 (and the related fusion protein PAX7-FOXO1) was identified in the 1990s (3-5) and intensely studied in terms of its regulation, downstream targets, and cellular phenotypic effects, a unified understanding of how the fusion gene and its resulting oncoprotein contribute so profoundly to aRMS tumorigenesis remains obscure.

Rhabdomyosarcoma (RMS) is a mesenchymal malignancy composed of neoplastic primitive precursor cells that exhibit histological features of myogenic differentiation. Despite intensive conventional multimodal therapy, patients with high-risk RMS typically suffer from aggressive disease. The lack of directed therapies against RMS emphasizes the need to further uncover the molecular underpinnings of the disease. In this Review, we discuss the notable advances in the model systems now available to probe for new RMS-targetable pathogenetic mechanisms, and the possibilities for enhanced RMS therapeutics and improved clinical outcomes.

SUMMARY Rhabdomyosarcoma (RMS) is an aggressive skeletal muscle-lineage tumor composed of malignant myoblasts that fail to exit the cell cycle and are blocked from fusing into syncytial muscle. Rhabdomyosarcoma includes two histolopathologic subtypes: alveolar rhabdomyosarcoma, driven by the fusion protein PAX3-FOXO1 or PAX7-FOXO1, and embryonal rhabdomyosarcoma (ERMS), which is genetically heterogeneous. Here, we show that adipocyte-restricted activation of Sonic Hedgehog signaling through expression of a constitutively active Smoothened allele in mice gives rise to aggressive skeletal muscle tumors that display the histologic and molecular characteristics of human ERMS with high penetrance. Our findings suggest that adipocyte progenitors can be a cell of origin for Sonic Hedgehog-driven ERMS, showing that RMS can originate from non-skeletal muscle precursors.

Dorsoventral polarity in the Drosophila embryo is established by a signaling pathway active on the ventral and ventrolateral surfaces of the embryo. Signal transduction via the protein kinase Pelle frees the Rel-related protein Dorsal from its cytoplasmic inhibitor Cactus, allowing Dorsal to translocate into ventral and ventrolateral nuclei and direct gene expression. Here, we show by immunochemical analyses that Pelle-mediated signaling induces the spatially graded degradation of Cactus. Using a tissue culture system which reconstitutes Pelle-dependent Cactus degradation, we show that a motif in Cactus resembling the sites of signal-dependent phosphorylation in the vertebrate homologs IkappaB-alpha and IkappaB-beta is essential for Pelle-induced Cactus degradation. Substitution of four serines within this motif with nonphosphorylatable alanine residues generated a mutant Cactus that still functions as a Dorsal inhibitor but is resistant to induced degradation. Injection of RNA encoding this altered form of Cactus has a dominant negative effect on establishment of dorsoventral polarity in the embryo. We conclude that dorsoventral signaling results in a Cactus concentration gradient and propose that signal-dependent phosphorylation directs the spatially regulated proteolysis of Cactus protein.

Pediatric rhabdomyosarcoma occurs as two biologically distinct histological variants, embryonal (ERMS) and alveolar (ARMS). To identify genomic changes that drive ERMS pathogenesis, we used a new array comparative genomic hybridization (aCGH) platform to examine a specific subset of ERMS tumors, those occurring in children with clinically defined intermediate-risk disease. The aCGH platform used has an average probe spacing ∼1 kb, and can identify genomic changes with single gene resolution. Our data suggest that these tumors share a common genomic program that includes inactivation of a master regulator of the p53 and Rb pathways, CDKN2A/B, and activation of FGFR4, Ras, and Hedgehog (Hh) signaling. The CDKN2A/B tumor suppressor is deleted in most patient samples. FGFR4, which encodes a receptor tyrosine kinase, is activated in 20% of tumors, predominantly by amplification of mutant, activating FGFR4 alleles. Over 50% of patients had low-level gains of a region containing the Hh-pathway transcription factor GLI1, and a gene expression pattern consistent with Hh-pathway activation. We also identified intragenic deletions affecting NF1, a tumor suppressor and inhibitor of Ras, in 15% of tumor samples. Deletion of NF1 and the presence of activating Ras mutations (in 42% of patients) were mutually exclusive, suggesting NF1 loss is an alternative and potentially common mechanism of Ras activation in ERMS. Our data suggest that intermediate-risk ERMS is driven by a common set of genomic defects, a finding that has important implications for the application of targeted therapies to improve the treatment of children diagnosed with this disease.