Removing testosterone in male rats was linked to 5, 104 differently expressed genes in the PVN relating to hormone signaling. In the female SHR, testosterone replacement in oophorectomized animals induced the regulation of 1,727 genes, sharing many biological functions with those in the high T males. Further investigation is warranted to determine whether these differences in gene expression contributed to the previously identified phenotypic effects of T-treatment and to assess the mechanisms relating T-treatment, gene expression, and phenotypic effects. Sexual dimorphism in the genes that are affected by T-treatment may be a key step in resolving sexual conflict over optimal circulating T levels – by modifying the phenotypic effect of T separately in each sex. Female insensitivity to T-treatment with respect to some phenotypes suggests that, in some species, the sexes may process or interpret a hormonal signal differently, which is consistent with our finding that T-treatment affects different genes in each sex. However, it is also known that females are not sensitive to all of the same behavioral and physiological effects of hormonal treatment as males (reviewed in ). It appears that the similar phenotypic outcomes of T-treatment described in previous studies may be caused by expression changes in different genes in each sex.
When these same sets of testosterone-responsive genes were tested for age-biased expression in males, we found that DEGs inferred to be upregulated by testosterone were significantly subadult-biased in all cases (Figures 4I–L; Supplementary Table S4), whereas DEGs inferred to be downregulated by testosterone were significantly juvenile-biased in all cases (Figures 4I–L; Supplementary Table S4). Volcano plots in the top row illustrate effects of testosterone on gene expression based on comparisons of (A) juvenile females with empty vs. testosterone implants (JF vs. JFT), (B) juvenile males with empty vs. testosterone implants (JM vs. JMT), (C) castrated vs. intact subadult males (SMC vs. SM), and (D) castrated subadult males with empty vs. testosterone implants (SMC vs. SMCT). Likewise, when genes that were significantly age-biased in males (Figure 2C) were tested for age-biased expression in females (Supplementary Table S2), juvenile-biased DEGs were significantly juvenile-biased (Figure 2I) and subadult-biased genes were significantly subadult-biased (Figure 2I). When genes that were significantly age-biased in females (Figure 2B) were tested for age-biased expression in males (Supplementary Table S2), juvenile-biased DEGs were significantly juvenile-biased (Figure 2H) and subadult-biased DEGs were significantly subadult-biased (Figure 2H).
To avoid any effects which may be attributed to collider bias33, we compared BMI adjusted estimates to BMI unadjusted estimates across all identified genome-wide significant SHBG signals. After application of QC criteria, a maximum of 425,097 UK Biobank participants were available for analysis with genotype and phenotype data. In addition to the quality control metrics performed centrally by UK Biobank, we defined a subset of "white European" ancestry samples using a K-means clustering approach applied to the first four principal components calculated from genome-wide SNP genotypes. At baseline a panel of 34 biomarkers were measured across the full ~500,000 study participants. All UK Biobank participants provided written informed consent, the study was approved by the National Research Ethics Service Committee North West–Haydock and all study procedures were performed in accordance with the World Medical Association Declaration of Helsinki ethical principles for medical research. Discovery analyses were performed in the full UK Biobank study which has been described extensively elsewhere29.
Briefly, this Nimblegen 12-plex microarray (Roche Nimblegen, Inc., Madison, WI) contained 100,635 features representing 33,545 contigs (assembled sequencing reads) in triplicate covering 22,765 isogroups (putative genes). These direct and indirect effects reflect the natural response of the organism to elevated T levels and to the T-implants utilized in previous studies. Furthermore, the medial amygdala plays a key role in relaying and mediating social signals between brain areas , and the hypothalamus is a major control center of hormones and behavior with projections extending throughout the brain . Importantly, these brain areas are also major sites of estrogenic action, and many of the effects of sex steroids in these regions may occur after local conversion of T to estradiol , .
One can inherit genetic polymorphisms which make muscle hypertrophy easier than others who do not possess those polymorphisms. Muscle CSA can be affected by numerous environmental factors, but it is also highly determined by genetic factors. In young healthy, physically active women (20–35 years) treated with testosterone cream for 10 weeks, muscle hypertrophy was primarily driven by increases in CSA of type II fibers (Horwath et al. 2020). Indeed, female athletes with higher free testosterone performed better in 400 to 800-m sprinting events, hammer throw and pole vault compared with female competitors with lower free testosterone (Bermon and Garnier 2017). It is possible that individuals who have higher levels of endogenous testosterone are more predisposed to certain power sports.
However, the statistically significant correlation estimates in our study suggests some shared variance in testosterone genetics across sexes, despite the distinct genetic architectures – consistent with previous studies3,6. The proportion of variance () for PGST or PGSBioT captured by the genome-wide gene expression transcripts was estimated using restricted maximum likelihood (OREML) based on OSCA (OmicS-data-based Complex trait Analysis) software24 for each sex and tissue, separately. The expression values were adjusted for possible covariates, including PEER probabilistic estimation of expression residuals factors to account for hidden batch effects, the top 5 genetic PCs, the genotyping platform and protocol, and sex as described in19. The relationship between testosterone and disease becomes more complex as testosterone can affect gene expression in a tissue-specific manner.
The genotypes have been imputed with using the SISu v3 population-specific reference panel developed from high-quality data for 3775 high-coverage (25–30×) whole-genome sequencing in Finns. In males, allele dosage of 2 was used for X-chromosomal haploid variants. SHBG quantification for males was done in 2009 and for females in 2011 with Spectria SHBG IRMA kit (Orion Diagnostica, Espoo, Finland). For testosterone and SHBG, we used data on 2001 follow-up (Subjects aged 24–39 years). Including up to 127 covariates (based on ref. 34), e.g., assay center, dilution factors, blood draw time, and socioeconomic status indicators, or excluding related individuals from the analysis all showed negligible effects on the genetic findings (Supplementary Fig. 10).