Endocrine Abstracts

Background: Understanding the genetic causes of DSD remains challenging. Our data show that >50% of all DSD cases do not have a genetic diagnosis. Many unexplained cases may be caused by variants in non-coding regulatory regions that disrupt gene regulatory networks (GRNs) essential for gonad development. Recent single-cell omics studies of human fetal gonads show substantial differences in gene expression profiles compared with mice. These differences may explain why several genes that cause human DSD fail to produce gonadal phenotypes in mouse models. This highlights the limitations of these models to reflect the precise GRNs that active during human SD. When disrupted, these GRNs may result in DSD.

Methods: Here, we integrated GRN inference from single-cell epigenomic and transcriptomic datasets with human clinical genomics. Using published human fetal gonadal scRNA-seq and scATAC-seq datasets, and advanced GRN inference tools, we reconstructed and prioritized lineage and sex-specific GRNs. In parallel, we mapped rare/novel genomic variants from 165 individuals with unexplained DSD to candidate gene regulatory elements (GREs) predicted by our GRN models, and independently to all regions of open chromatin active during gonadal differentiation. Potentially pathogenic non-coding variants were prioritized using stringent filtering criteria and an in-house pipeline designed to identify high-confidence pathogenic variants across the entire human genome

Results: Integration of scATAC-seq with scRNA-seq analysis defined core regulons in human fetal Sertoli and pre-granulosa cells. These regulons included regulatory hubs of known DSD genes and novel genes potentially involved in DSD. We then integrated these findings with clinical genomics. Rare/novel variants from 46,XY gonadal dysgenesis cases were mapped to - (i) 37,132 candidate GREs predicted by our GRN models, and (ii) 627,158 regions of active open chromatin during gonadal differentiation. Following a strict variant prioritization protocol, we identified 14 non-coding loci in predicted GREs associated with DSD. These included 4 genes known to cause 46,XY gonadal dysgenesis (enrichment P value 10^-8) and two other genes linked to either DSD or SD.

Conclusions: This integrative approach establishes a framework of putative GRN/GREs in human Sertoli/Granulosa cell specification and development as well as defining new genetic causes of DSD due to non-coding variants.