Single-nucleus analysis of menstrual fluid highlights gene expression differences in epithelial cells of endometriosis donors

BACKGROUND: Endometriosis is a common complex gynecological condition that is difficult to efficiently diagnose and remains poorly understood. Menstrual fluid provides a non-invasive source of disease-relevant tissue and has been identified as promising for endometriosis diagnostics. Peripheral blood-based assays have yielded few clinically relevant targets, thus we hypothesize that cells specific to the uterus will exhibit more disease-related differences and may lead to insights into disease development and potential diagnostic markers. METHODS: We profiled 10 menstrual fluid samples, from 5 donors with endometriosis and 5 donors without, using both single-nucleus RNA sequencing and bulk RNA sequencing. We tested for differential abundance of cell types, differential gene expression within cell types, and differential cell communication between cell types by disease status. Finally, we compared the results of both single-nucleus and bulk RNA sequencing analyses to identify potential diagnostic targets. RESULTS: We identified endometrial and immune cell types present in menstrual fluid, with large inter-individual heterogeneity in cell type representation. While most cells were immune, the cell types with the greatest number of differentially expressed genes were endometrial epithelial cells followed by stromal cells. Epithelial cells in particular recapitulated some gene expression differences previously described in the context of endometriosis, although most identified differences were novel. Cell-cell communication in endometriosis was characterized by a loss of interactions between epithelial and stromal cells related to developmental and growth genes WNT2B, IGF2 and TGFB2, but a gain of cell-cell communications between dendritic cells and other cell types, especially within the BMP signaling pathway. Bulk RNA sequencing revealed that some of the differentially expressed genes found in epithelial cells could be replicated in the whole tissue, highlighting the following genes as biomarker candidates: TIMP2, AKR1C2, DMBT1, FERMT1, and KCNK5. CONCLUSIONS: We identified a set of promising genes that may contribute to endometriosis pathophysiology understanding and have potential as diagnosis biomarkers in whole menstrual blood. Further assessment of their stability over time within each patient and in a larger cohort will confirm whether this represents a viable non-invasive strategy to reduce diagnostic delays.