Single-cell RNA-sequencing reveals transcriptional dynamics of estrogen-induced dysplasia in the ovarian surface epithelium

Estrogen therapy increases the risk of ovarian cancer and exogenous estradiol accelerates the onset of ovarian cancer in mouse models. Both in vivo and in vitro, ovarian surface epithelial (OSE) cells exposed to estradiol develop a subpopulation that loses cell polarity, contact inhibition, and forms multi-layered foci of dysplastic cells with increased susceptibility to transformation. Here, we use single-cell RNA-sequencing to characterize this dysplastic subpopulation and identify the transcriptional dynamics involved in its emergence. Estradiol-treated cells were characterized by up-regulation of genes associated with proliferation, metabolism, and survival pathways. Pseudotemporal ordering revealed that OSE cells occupy a largely linear phenotypic spectrum that, in estradiol-treated cells, diverges towards cell state consistent with the dysplastic population. This divergence is characterized by the activation of various cancer-associated pathways including an increase in Greb1 which was validated in fallopian tube epithelium and human ovarian cancers. Taken together, this work reveals possible mechanisms by which estradiol increases epithelial cell susceptibility to tumour initiation.Author summary: Women who take estrogen replacement therapy are at higher risk of developing ovarian cancer. When ovarian epithelial cells are exposed to estrogen, there is a heterogeneous cellular response, with some cells appearing unaffected, while others become disorganized and grow at accelerated rates consistent with pre-cancerous cells. This heterogeneity confounds traditional methods for surveying gene expression, which rely on averaging the signal across a population of cells. Here, we employ single cell RNA sequencing in order to measure gene expression profiles at single-cell resolution. This allowed us to distinguish between estrogen-responsive and unresponsive populations and identify defined expression signatures for each. Also, because cellular responses are asynchronous, we were able to use the snapshot of expression profiles to infer the transcriptional changes as cells respond to estrogen and become increasingly disorganized. These techniques revealed not only the processes that may contribute to the earliest stages in the formation of estrogen-driven pre-cancerous cells, but also identified biomarkers of that transition. We have confirmed that the protein GREB1 appears in the pre-cancerous cells and is present in the majority of human ovarian cancers.