Single-Cell RNA-Seq Identifies Pathways and Genes Contributing to the Hyperandrogenemia Associated with Polycystic Ovary Syndrome

Abstract

Polycystic ovary syndrome (PCOS) is a common endocrine disorder characterized by hyperandrogenemia of ovarian thecal cell origin, resulting in anovulation/oligo-ovulation and infertility. Our previous studies established that ovarian theca cells isolated and propagated from ovaries of normal ovulatory women and women with PCOS have distinctive molecular and cellular signatures that underlie the increased androgen biosynthesis in PCOS. To evaluate differences between gene expression in single-cells from passaged cultures of theca cells from ovaries of normal ovulatory women and women with PCOS, we performed single-cell RNA sequencing (scRNA-seq). Results from these studies revealed differentially expressed pathways and genes involved in the acquisition of cholesterol, the precursor of steroid hormones, and steroidogenesis. Bulk RNA-seq and microarray studies confirmed the theca cell differential gene expression profiles. The expression profiles appear to be directed largely by increased levels or activity of the transcription factors SREBF1, which regulates genes involved in cholesterol acquisition (LDLR, LIPA, NPC1, CYP11A1, FDX1, and FDXR), and GATA6, which regulates expression of genes encoding steroidogenic enzymes (CYP17A1) in concert with other differentially expressed transcription factors (SP1, NR5A2). This study provides insights into the molecular mechanisms underlying the hyperandrogenemia associated with PCOS and highlights potential targets for molecular diagnosis and therapeutic intervention.

Keywords: androgens; cholesterol; human ovarian theca cells; polycystic ovary syndrome; single-cell RNA sequencing; steroids.

Figure 1

UMAP of affection status/treatment groups (a) and samples (b). Values from normal (control) theca cells were generated using the five theca preparations from PCOS women (MC03, Mc10, MC16, MC26, and MC27) and five theca cell preparations from normal cycling women (MC02, MC06, MC31, MC40, and MC50), that were grown until subconfluent, and treated with (F) and without (C), 20 mM forskolin for 24 h in a serum-free medium.

Figure 2

Heatmap of the Z scores for cholesterol acquisition and steroidogenesis pathways from molecular signatures database (MSigDB) C2: curated gene sets significantly enriched in at least one group. Asterisks show the significant differentially expressed group(s). scRNA-seq data were generated using theca cell preparations from normal cycling and PCOS women grown until subconfluent and either untreated (C) or treated (F) with 20 μM forskolin for 24 h in a serum-free medium.

Figure 3

Heatmap of significant differentially expressed genes likely to play a major role in cholesterol acquisition and steroidogenesis.

Figure 4

STRING protein–protein interaction network of cholesterol acquisition proteins translated from differentially expressed genes identified by scRNA-seq of normal and PCOS theca cells. The nodes represent the proteins, and the width of the edges shows the confidence of the association between connected proteins.

Figure 5

STRING protein–protein interaction network of steroidogenesis proteins translated from significant differentially expressed genes identified by scRNA-seq of normal and PCOS theca cells. The nodes represent the proteins, and the width of the edges shows the confidence of the association between connected proteins.