EZH2 Cooperates with BRD4-NUT to Drive NUT Carcinoma Growth by Silencing Key Tumor Suppressor Genes.

NUT carcinoma is an aggressive carcinoma driven by the BRD4-NUT fusion oncoprotein, which activates chromatin to promote expression of progrowth genes. BET bromodomain inhibitors (BETi) are a promising treatment for NUT carcinoma that can impede BRD4-NUT's ability to activate genes, but the efficacy of BETi as monotherapy is limited. Here, we demonstrated that enhancer of zeste homolog 2 (EZH2), which silences genes through establishment of repressive chromatin, is a dependency in NUT carcinoma. Inhibition of EZH2 with the clinical compound tazemetostat potently blocked growth of NUT carcinoma cells. Epigenetic and transcriptomic analysis revealed that tazemetostat reversed the EZH2-specific H3K27me3 silencing mark and restored expression of multiple tumor suppressor genes while having no effect on key oncogenic BRD4-NUT-regulated genes. Indeed, H3K27me3 and H3K27ac domains were found to be mutually exclusive in NUT carcinoma cells. CDKN2A was identified as the only gene among all tazemetostat-derepressed genes to confer resistance to tazemetostat in a CRISPR-Cas9 screen. Combined inhibition of EZH2 and BET synergized to downregulate cell proliferation genes, resulting in more pronounced growth arrest and differentiation than either inhibitor alone. In preclinical models, combined tazemetostat and BETi synergistically blocked tumor growth and prolonged survival of NUT carcinoma-xenografted mice, with complete remission without relapse in one cohort. Identification of EZH2 as a dependency in NUT carcinoma substantiates the reliance of NUT carcinoma tumor cells on epigenetic dysregulation of functionally opposite, yet highly complementary, chromatin regulatory pathways to maintain NUT carcinoma growth. SIGNIFICANCE: Repression of tumor suppressor genes, including CDKN2A, by EZH2 provides a mechanistic rationale for combining EZH2 and BET inhibitors for the clinical treatment of NUT carcinoma. See related commentary by Kazansky and Kentsis, p. 3827.

EZH2 synergizes with BRD4-NUT to drive NUT carcinoma growth through silencing of key tumor suppressor genes.

NUT carcinoma (NC) is an aggressive carcinoma driven by the BRD4-NUT fusion oncoprotein, which activates chromatin to promote expression of pro-growth genes. BET bromodomain inhibitors (BETi) impede BRD4-NUT's ability to activate genes and are thus a promising treatment but limited as monotherapy. The role of gene repression in NC is unknown. Here, we demonstrate that EZH2, which silences genes through establishment of repressive chromatin, is a dependency in NC. Inhibition of EZH2 with the clinical compound tazemetostat (taz) potently blocked growth of NC cells. Epigenetic and transcriptomic analysis revealed that taz reversed the EZH2-specific H3K27me3 silencing mark, and restored expression of multiple tumor suppressor genes while having no effect on key oncogenic BRD4- NUT-regulated genes. CDKN2A was identified as the only gene amongst all taz-derepressed genes to confer resistance to taz in a CRISPR-Cas9 screen. Combined EZH2 inhibition and BET inhibition synergized to downregulate cell proliferation genes resulting in more pronounced growth arrest and differentiation than either inhibitor alone. In pre-clinical models, combined taz and BETi synergistically blocked growth and prolonged survival of NC-xenografted mice, with all mice cured in one cohort. STATEMENT OF SIGNIFICANCE: Identification of EZH2 as a dependency in NC substantiates the reliance of NC tumor cells on epigenetic dysregulation of functionally opposite, yet highly complementary chromatin regulatory pathways to maintain NC growth. In particular, repression of CDKN2A expression by EZH2 provides a mechanistic rationale for combining EZH2i with BETi for the clinical treatment of NC.

NUTM1 -rearranged Carcinoma of the Thyroid : A Distinct Subset of NUT Carcinoma Characterized by Frequent NSD3 - NUTM1 Fusions.

NUT carcinoma (NC) is a rare subtype of squamous cell carcinoma defined by NUTM1 rearrangements encoding NUT fusion oncoproteins (the most frequent fusion partner being BRD4 ) that carries a very poor prognosis, with most patients dying in under 1 year. Only rare primary thyroid NCs have been reported. Here, we evaluated a series of 14 cases. The median patient age at diagnosis was 38 years (range: 17 to 72 y). Eight of 13 cases with slides available for review (62%) showed a morphology typical of NC, whereas 5 (38%) had a non-NC-like morphology, some of which had areas of cribriform or fused follicular architecture resembling a follicular cell-derived thyroid carcinoma. For cases with immunohistochemistry results, 85% (11/13) were positive for NUT on biopsy or resection, though staining was significantly decreased on resection specimens due to fixation; 55% (6/11) were positive for PAX8, and 54% (7/13) for TTF-1. Tumors with a non-NC-like morphology were all positive for PAX8 and TTF-1. The fusion partner was known in 12 cases: 9 (75%) cases had a NSD3-NUTM1 fusion, and 3 (25%) had a BRD4-NUTM1 fusion. For our cohort, the 2-year overall survival (OS) was 69%, and the 5-year OS was 58%. Patients with NC-like tumors had a significantly worse OS compared with that of patients with tumors with a non-NC-like morphology ( P =0.0462). Our study shows that NC of the thyroid can mimic other thyroid primaries, has a high rate of NSD3 - NUTM1 fusions, and an overall more protracted clinical course compared with nonthyroid primary NC.