Abstract
In most cancers, including endometrial cancer, tumor suppressor genes harboring inactivating mutations have been systematically cataloged. However, locus-specific epigenetic alterations contributing to cancer initiation and progression remain only partly described, creating knowledge gaps about functionally significant tumor suppressors and underlying mechanisms associated with their inactivation. Here, we show that PAX2 is an endometrial tumor suppressor recurrently inactivated by a distinct epigenetic reprogramming event not associated with promoter hypermethylation. PAX2 is expressed throughout normal endometrial glands; however, microscopic clones with PAX2 protein loss arise spontaneously in adulthood and progress to endometrial precancers and cancers. Here we show that PAX2 protein loss occurs via transcriptional silencing in 80% of endometrial cancers. Mechanistically, transcriptomic, epigenomic, 3D genomic, and machine learning analyses showed that silencing is associated with replacement of open/active chromatin features (H3K27ac/H3K4me3) with inaccessible/repressive features (H3K27me3). The spread of the H3K27me3 signal is restrained by cohesin loops, preventing transcriptional dysregulation of neighboring genes. Functionally, PAX2 loss promoted endometrial carcinogenesis by rewiring the transcriptional landscape via global enhancer reprogramming. Genetically engineered mouse models together with organoid and human cell line studies established Pax2 as an in vivo endometrial tumor suppressor that cooperates with Pten to produce lethal cancers. Our discovery of a specific and recurring epigenetic alteration that transcriptionally silences PAX2 to initiate most endometrial cancers opens new lines of investigation into the origins of endometrial cancer with diverse implications for its diagnosis and treatment. ONE SENTENCE SUMMARY: We demonstrate that PAX2 is silenced by a highly recurrent epigenetic mechanism to initiate most endometrial cancers, establishing a novel mechanism of tumor suppressor inactivation.