Effect of testosterone on within-sex gene expression across 40 human tissues

Abstract Background Variations in testosterone levels is associated with pronounced health risks, often in a discordant manner between males and females. While studies have demonstrated a sex-specific genetic architecture for testosterone, the biological basis for the differential impact on diseases between the sexes is largely unknown. In this study, we correlated predicted testosterone and within-sex gene expression measures across 40 human tissues to identify genes that show sex-differential control of gene expression and examine how this varies across tissues. Methods Gene expression measures were obtained from the Genotype-Tissue Expression project (v8 GTEx release), with sex-specific genome-wide summary statistics from the UK Biobank used to construct polygenic scores as proxies for total testosterone and bioavailable testosterone. We quantified the proportion of variance in the genomically predicted testosterone levels that was captured by gene expression measures within each sex/tissue pair. The association between predicted testosterone measures was tested within each sex/tissue pair for over 20,000 gene transcripts. Results Gene expression was able to capture a consistently higher proportion of predicted testosterone variance in females than in males across the majority of study tissues. Gene expression levels in the mammary breast and adipose (visceral omentum) tissue explained the largest proportion of total testosterone variance in females, while various brain tissues (anterior cingulate cortex BA24, putamen basal ganglia and cerebellum) captured the largest proportion of total testosterone variance in males. Association analyses identified significant associations between total testosterone and gene expression across four transcripts (NUPR1L, PTPRD, PSPHP1, and RP11-208G20.3) in the skeletal muscle, tibial artery, and pancreas tissue for females but were not associated in males. No transcript/gene showed a significant association with predicted bioavailable testosterone across all study tissues in males and females. Conclusions The relationship between testosterone and gene expression levels is complex, showing variation across tissues and between the sexes. While we found limited numbers of associations between individual gene expression levels and predicted testosterone levels, our whole transcriptome approach found that a substantial proportion of testosterone levels could be captured by gene expression levels, indicating that more associations could be identified with larger sample sizes and directly measured hormone levels.