Characterizing Heterogeneity of Estrogen Receptor Isoform Expression in Endometriosis: Rethinking the ERβ Dominance Hypothesis

Introduction: Endometriosis is an estrogen-dependent disease where endometrial tissue grows outside of the uterus on or around organs throughout the pelvic region. Individuals with endometriosis often experience pelvic pain, digestive issues, and painful periods ranging in severity as well as infertility due to the lesion growth and development. However, despite about 10% of all reproductive age women estimated to experience endometriosis, the diagnostic and treatment processes are incomplete and often ineffective, establishing the need for a further understanding of the disease state as well as potential causes. Previous studies have honed in on the potential role of estrogen’s two nuclear receptors – ERɑ and ERβ – in the establishment, progression, and symptoms of endometriosis, specifically the hypothesis that aberrant dominance of ERβ is a driver of disease. Here, we conducted a secondary analysis of single cell RNA-seq data from endometriosis patients and disease-free patients across multiple tissue origins using data sourced from the Gene Expression Omnibus (GEO). There is lack of knowledge on how ERɑ and ERβ are expressed on specific cell types and how changes in nuclear receptor expression might shift membrane estrogen receptor expression (GPER), particularly across different tissue origins. We stratified our analysis of estrogen receptor isoform-specific expression by cell type and tissue origins, examining expression of genes encoding these receptors – ESR1 for ERɑ, ESR2 for ERβ, and GPER1 for GPER respectively – in each cell as well as the ratios of nuclear estrogen receptor expression in both a tissue-origin and cell-type dependent manner.

Materials and

Methods: Data was sourced from GEO under accession code GSE179640. Data included five different tissue origins including paired samples taken from eutopic endometrial tissue in a patient with no disease (Control) and a patient with endometriosis (EuE) as well as ectopic ovarian lesions (EcO), ectopic peritoneal lesions (EcP), and areas adjacent to the peritoneal lesions (EcPA). Samples were processed for scRNA-seq, and the filtered feature barcode matrices were supplied to GEO. Using the R package Seurat, we processed the data and ran dimensionality reduction analysis to create UMAP visualizations, allowing us to classify clusters by cell type (stromal, endothelial, epithelial, myeloid, lymphocyte) for each tissue origin. Each cell in the data was labeled with its respective cell type, allowing for examination of the expression and distribution of ESR1, ESR2, and GPER1 on each cell asking the following questions: (1) How does expression of each estrogen receptor isoform differentiate by cell type and tissue origin, (2) Do the nuclear estrogen receptors tend to be uniquely expressed or coexpressed by cell type and tissue origin, and (3) How do single cell ratios of ESR1/ESR2 expression vary by cell type and tissue origin. We employed non-parametric statistical testing, including Wilcoxon Rank Sum and Fisher’s Exact Test, to examine these questions. Using this information, differential gene expression analysis will be performed within the different cell types and tissue origins comparing cells with ERɑ dominance to those with ERβ dominance further elucidating the consequences of expression of estrogen receptor isoforms in endometriosis.

Results, Conclusions, and Discussions: The expression of each estrogen receptor significantly differed across most tissue origins and cell types, but certain tissue and cell types showed indistinguishable expression of the isoforms. Of these, endothelial cells in the Control samples showed similar expression of ESR1 and GPER1. Endothelial cells in the EuE and EcPA samples also showed similar expression of ESR1 and ESR2. Epithelial cells in the EcPA samples showed similar expression of ESR2 and GPER1. Myeloid cells in the EcO samples showed similar expression of ESR1 and GPER1. Takeaway point being ERβ expression was most pronounced in the endothelial cells of endometriotic tissues.

We examined if there were any subgroups where ERɑ or ERβ were significantly uniquely expressed or coexpressed. Stromal cells of the EcO origin yielded significant results, with the majority of cells (83.54%) expressing neither estrogen receptor. Myeloid, endothelial, and lymphocyte cells of the EuE origin also yielded significant results – 81.92% of myeloid cells, 73.53% of endothelial cells, and 76.08% of lymphocyte cells expressed neither estrogen receptor isoform. Of these subgroups’ uniquely expressing cells, cells more often uniquely expressed ESR1 over ESR2.

We assessed the ratio of ESR1 to ESR2 expression in each tissue and cell type to identify where the ERɑ/ERβ ratio differs significantly. Stromal cells were the only cell type with significant differences in distribution across every tissue origin pairing. Endothelial cells’ distributions were significantly changed in control/EcP, control/EcPA, and all EcO origin pairings. Myeloid and lymphocyte cells were significantly changed in EcP/EcO, EcP/EcPA, and all EuE origin pairings. Epithelial cells’ distributions were significantly changed across every tissue origin pairing except control/EcPA

These results, alongside considerable heterogeneity in the ESR1/ESR2 ratio, suggests ERβ dominance in endometriosis is overstated. Past research focused on ERβ dominance only in tissues of EcO origin; however, our results indicate no definitive shift to ERβ dominance in any subgroup. Instead, the ESR1/ESR2 ratios and aberrant expression of both ERɑ and ERβ should be considered as an indicator of endometriosis development, particularly within the eutopic endometrium.

Acknowledgements (Optional):