About You

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. This possible mechanism could explain how despite significant neurogenesis over development the effects of in utero testosterone exposure are maintained and able to alter adult behavior. While the active effects of testosterone on the transcriptome are robust in both sexes, it appears that these differences revert back to near baseline if TP is removed from downstream cellular lineages. The effects of gonadal hormones on these cell types are just beginning to be uncovered, as such; this is the first documentation of global gene changes in eNSCs as a result of testosterone exposure during a developmentally relevant time in gestation31. Therefore, we provide evidence here that basal gene expression differences between XX and XY cells may contribute to differences in proliferation of NSCs in the absence of testosterone.
Hence, this cluster (termed "male SHBG cluster", Methods) represents a genetic instrument with primary SHBG-increasing effects, and secondary divergent effects on total (higher) and bioavailable testosterone (lower) that are consistent with the known hormone-regulatory role of SHBG. To validate these findings, we performed replication using three available datasets (Methods) - a previously published GWAS meta-analysis of SHBG levels in 21,791 individuals19, 9,138 individuals with testosterone measurements from the EPIC-Norfolk study and published data on 2,913 individuals from the Twins UK study with nine sex hormones measured20. After extensive quality control (Methods), serum levels of SHBG, total testosterone and estradiol were available in up to 425,097 individuals with genetic data in UK Biobank (UKBB) (Table S1). This study substantially advances our understanding of the genetic regulation of sex hormone levels, increasing the number of known genetic determinants by two orders of magnitude. Using 2,571 genome-wide significant associations, we demonstrate the genetic determinants of testosterone levels are substantially different between sexes, and that genetically higher testosterone is harmful for metabolic diseases in women but beneficial in men. Sex-biased gene expression in the liver is best described in rodent models and is often driven by sex-specific patterns of GH secretion from the pituitary. In addition to reducing our power to detect DEGs, fewer replicate transcriptomes per group should increase error in the estimation of mean sex, age, and treatment effects (i.e., log2 FC values) for individual genes.
These findings demonstrate the ability of gonadal hormones to modulate effects of sex chromosome composition, resulting in diminished XX and XY basal gene expression differences, making such expression more similar in nature. To further analyze possible functional groupings of sexually dimorphic patterns of gene expression we subjected the top 200 most down regulated genes from each genetic background to pathway analysis using the recently available Broad Institute’s GeNets software. (C) Distribution ratios comparing the effects of chromosomal sex and testosterone on global gene expression of XX in TP/baseline gene expression (Y-axis), versus global gene expression changes due to testosterone of XY in TP/baseline gene expression (X-axis).
Lastly, we evaluated causal relationships and genetic correlations between the studied T traits and complex traits, leveraging publicly available GWAS summary statistics (Fig. 1). In brief, we conducted sex-stratified GWAS for T, SHBG and free T based on immunoassay measurements from the UK Biobank, and constructed sex-specific polygenic scores (PGS) for these traits. We utilized the rich biochemical and health information available in two population-scale genetic datasets and analysis methods building on GWAS discovery. A high GCP value and a statistically significant effect support partial genetic causality between the traits, and suggest that interventions targeting trait 1 are likely to affect trait 2.
The number of samples per tissue in the final gene expression files ranged from 78 (brain spinal cord cervical c-1) to 469 (muscle skeletal) in males and 33 (brain substantia nigra) to 237 (skeletal muscle) in females. For males, BioT was inverse normal transformed and adjusted for fasting time, age at baseline, assessment centre, genotyping chip, genetic data release, and 10 PCs. In this study, we utilized publicly available sex-stratified GWAS summary statistics for total6 and bioavailable testosterone levels3 to construct sex-specific polygenic scores (PGS) for these traits in the Genotype-Tissue Expression (GTEx) samples19. Indeed, autosomal genetic correlations between males and females are close to one across human quantitative traits and disorders3. The relationship between testosterone and gene expression levels is complex, showing variation across tissues and between the sexes. No transcript or gene showed a significant association with predicted bioavailable testosterone across all study tissues in both males and females. We quantified the proportion of variance in the genomically predicted testosterone levels that was captured by gene expression measures within each sex/tissue pair.
Consistent with potentially divergent biological effects for the SHBG bound and unbound T, we however observed distinct association profiles for total and free T fractions in men. Especially for many metabolic traits, SHBG—either directly or through its metabolic network—appeared to modify some of total T’s associations and causality estimates. Although highly valuable, clinical trials to investigate the relationship of T levels to many complex phenotypes would be often infeasible or unethical to conduct. Reflecting the results from FinnGen, for most traits there was no evidence for substantial causality, but the few suggested causal relationships involved traits with clear biological links to T function. Notably, the behavioral traits showing significant correlations were clearly linked to metabolism (smoking, sleep duration and exercise).
In females, the gene expression in the mammary breast tissue captured the largest proportion of variance in PGST, consistent with other studies showing that this tissue has relatively many sex-biased genes29. T-treatment affected largely different genes in males and female, suggesting that T-treatment may bring about similar behavioral and physiological effects on the sexes by different transcriptional mechanisms, potentially opening a route to the reduction of sexual conflict over optimal levels of T. In other species, many of the genes and functions identified as differentially expressed between males and females relate to translation and suggest large downstream effects that cannot be identified by gene expression analysis . We predicted that control males and control females would differ in the expression of key genes related to known sex differences, such as sexual and social behavior. Collectively, despite being based on immunoassays which may have limited use in clinical settings especially in females61,62, these findings illustrate that the UK Biobank data permits construction of robust genetic instruments to study how post-pubertal T and SHBG levels relate to adult health. Recent efforts have used genetics to address the potential causal contribution of adult T to selected complex traits in both sexes, including T2D, body composition and hormonal cancers, or have examined the effects of free T more broadly in males based on the UK Biobank data23,24,26,27,28. To better understand this, we identified genetic determinants of testosterone levels and related sex hormone traits in 425,097 UK Biobank study participants.
Notably, the sex-specific PGSs for T and free T had no predictive value in the opposite sex, as expected based on serum T being determined by distinct genetic variants in both sexes (Supplementary Data 6 and Supplementary Fig. 5). The PGSs capture the genetic effects on T and SHBG levels, and therefore serve as a proxy for cumulative post-pubertal T exposure. For the traits from public GWAS included in genetic correlation analyses, in 16 out 44 instances we could perform two sample MR (phenotype data not including UK Biobank samples) and in 19/44 instances the studies included X-chromosomal data (Supplementary Data 11). For all causality analyses X-chromosomal effects in males are presented for (0,2) allele dosage coding. For 23 traits we performed the analyses using sex-specific GWAS results and compared these to data from sex-combined GWAS (Supplementary Data 13).

Profile Info