Despite being the second most-frequent gynecological cancer, the mortality rate and incidence for endometrial cancer increased over the past decade. These data underline an urgent need of patient care improvement for this cancer type. Around 85% of endometrial cancer express favorable prognostic factor Estrogen Receptor alpha (ERα), but little is known about its transcriptional regulation in this tumor type.

We identified robust and reproducible ERα reprogramming in endometrial cancer development, associated with a gain of tumor-specific enhancer activity and alterations in 3D genome organization, as determined by ChIP-seq and Hi-C in human endometrial tissue. Integration of these findings with WGS data from endometrial cancer metastases revealed a striking enrichment of non-coding somatic mutations at tumor-gained ERα binding sites. Enhancer regions proximal to the ESR1 locus that were specifically mutated in metastatic disease, showed tumor-specific ERα binding and acquired H3K27ac, with increased promoter-enhancer interactions in tumorigenesis. Using a machine learning approach, we predicted one of the mutations occurring at the ESR1 enhancer in metastatic tumors to be potentially responsible for the 3D genome organization perturbation of the ESR1 locus. In vitro locus-specific mass spectrometry analyses revealed that this mutation, in contrast to the wild-type sequence, leads to the loss of interactions with the transcriptional repressor EHMT2/G9a. Analyses of publicly available data revealed that high expression of EHMT2/G9a is associated with lower survival probability of endometrial cancer patients, in contrast to high survival associated to ESR1 expression.

In summary, we identified a complex genomic-epigenomic interplay with consequences on 3D genome regulation in endometrial cancer, driving expression of the critical driver ERα, along with the epigenetic plasticity thereof in endometrial tumor carcinogenesis.