Regulatory risk loci link disrupted androgen response to pathophysiology of Polycystic Ovary Syndrome

A major challenge in deciphering the complex genetic landscape of Polycystic Ovary Syndrome (PCOS) is the limited understanding of the molecular mechanisms driven by susceptibility loci, necessitating investigation into the regulatory pathways that contribute to the diverse phenotypic manifestations of PCOS. In this study, we integrated molecular and epigenomic annotations across proposed pathogenic cell types and employed a deep learning (DL) model to infer the cell type specific effects of risk variants. Our analysis revealed the role of these variants in brain and endocrine cell types affecting the binding sites of key transcription factors (TFs): FOXA1, FOXL1, WT1, SALL4, and CPEB1, which regulate ovarian development, folliculogenesis, and steroid hormone signaling, contributing to disease-associated transcriptomic profiles. Our DL model, which is strongly correlated with MPRA data, identified enhancer-disrupting activity in 20% of the risk variants, particularly affecting TFs involved in androgen-mediated signaling, shedding light on the molecular consequences of hyperandrogenemia. Using the FTO/IRX3 locus as a case study, we explored the potential cell-type-specific regulatory effects of risk variants in the fetal brain, pancreas, adipocytes, and an endothelial cell line, which suggest that disruptions in IRX3 regulation (previously linked to obesity) may contribute to PCOS pathogenesis through diverse mechanisms, including neuronal development, metabolic regulation, and folliculogenesis. Our findings underscore the value of integrating DL models with epigenomic annotations to identify disease relevant variants, explore the pleiotropic impact of disease risk loci, and gain novel insights into cross cell type regulatory interactions.