Integrated analysis of robust sex-biased gene signatures in human brain

Abstract

Background: Sexual dimorphism is highly prominent in mammals with many physiological and behavioral differences between male and female form of the species. Accordingly, the fundamental social and cultural stratification factors for humans is sex. The sex differences are thought to emerge from a combination of genetic and environmental factors. It distinguishes individuals most prominently on the reproductive traits, but also affects many of the other related traits and manifest in different disease susceptibilities and treatment responses across sexes. Sex differences in brain have raised a lot of controversy due to small and sometimes contradictory sex-specific effects. Many studies have been published to identify sex-biased genes in one or several brain regions, but the assessment of the robustness of these studies is missing. We therefore collected huge amount of publicly available transcriptomic data to first estimate whether consistent sex differences exist and further explore their likely origin and functional significance.

Results and conclusion: In order to systematically characterise sex-specific differences across human brain regions, we collected transcription profiles for more than 16,000 samples from 46 datasets across 11 brain regions. By systematic integration of the data from multiple studies, we identified robust transcription level differences in human brain across to identify male-biased and female-biased genes in each brain region. Firstly, both male and female-biased genes were highly conserved across primates and showed a high overlap with sex-biased genes in other species. Female-biased genes were enriched for neuron-associated processes while male-biased genes were enriched for membranes and nuclear structures. Male-biased genes were enriched on the Y chromosome while female-biased genes were enriched on the X chromosome, which included X chromosome inactivation escapees explaining the origins of some sex differences. Male-biased genes were enriched for mitotic processes while female-biased genes were enriched for synaptic membrane and lumen. Finally, sex-biased genes were enriched for drug-targets and more female-biased genes were affected by adverse drug reactions than male-biased genes. In summary, by building a comprehensive resource of sex differences across human brain regions at gene expression level, we explored their likely origin and functional significance. We have also developed a web resource to make the entire analysis available for the scientific community for further exploration, available at https://joshiapps.cbu.uib.no/SRB_app/.

Keywords: Brain disorders; Conservation; Data integration; Drug response; Gene regulation; Hormones; Human brain; Sex difference.

Fig. 1

A The number of datasets analyzed in each brain region. B A schematic map of the brain regions studied. C The number of female-biased (pink) and male-biased (blue) genes for each dataset before and after rank aggregation. D The number sex-biased genes across brain regions. E Fraction of sex-biased genes found at least in 4 primates. F Fraction of sex-biased genes with sex-biased genes expression at least in 2 species. The number of genes on the sex chromosomes and autosomes in female-biased G and male-biased H genes

Fig. 2

The top 5 enriched terms for the gene ontology and disease enrichment analysis across brain regions. A Biological process enrichment for female-biased genes. B Biological process enrichment for male-biased genes. C Cellular component enrichment for female-biased genes D Cellular component enrichment for male-biased genes. E BrainBase disease enrichment analysis for female-biased genes. F BrainBase disease enrichment analysis for male-biased genes

Fig. 3

A The number of female-biased genes by number of regions. The color grey, red and blue are shown bar graphs for proportions of genes mapped into autosome, X-chromosome and Y-chromosome, respectively. B The number of male-biased genes by number of regions. C The correlation heatmap of female-biased genes. D The correlation heatmap of male-biased genes. E DisGeNet (CURATED) enrichment of overlap female-biased genes across FC, PL, TC and OC F DisGeNet (CURATED) enrichment for overlap female-biased genes across FC, PL, TC and OC

Fig. 4

A Cell-type enrichment for female-biased genes. B Cell-type enrichment for male-biased genes. C Transcription factor (TF) enrichment for female-biased genes. D Transcription factor (TF) for male-biased genes. E The percentage of overlap between androgen receptor element (ARE) genes and sex-biased genes, brain expressed genes, and brain regionally elevated genes. The colors grey, red, yellow and orange in bar graphs represent the proportion of genes that not overlap, overlap with ARE full sites genes, overlap with ARE half sites genes and overlap both in ARE full and half sites, respectively. F The percentage of overlap between estrogen receptor element (ERE) genes and sex-biased genes, brain expressed genes and brain regionally elevated genes. The color of grey and peach are shown the proportion of genes that not overlap and overlap with ERE genes, respectively

Fig. 5

A The scatter plot of the coefficients of age and gender variables from the multiple linear regression from GSE11882. B The scatter plot of the coefficients of age and gender variables from the multiple linear regression from GSE53890 dataset. C Gene expression of XIST gene in GSE53890 dataset, labeled as red and blue for female and male samples, respectively. D Gene expression of RPS4Y1 genes in GSE53890 dataset, colored red and blue for female and male samples, respectively. E Gene expression of CALB1 genes in GSE53890 dataset, colored red and blue for female and male samples, respectively. F Gene expression of FKBP5 genes in GSE53890 dataset, colored red and blue for female and male samples, respectively. G Gene expression of FKBP5 genes in GSE53890 dataset, colored red and blue for female and male samples, respectively. H Venn diagram of overlap genes between sex-biased genes and age-biased genes from GSE53890 and GSE11882 datasets

Fig. 6

A Over-enrichment of BrainBase drug targets for female-biased genes. B Over-enrichment of BrainBase drug targets for male-biased genes. C Number of genes overlapping between female-biased genes and adverse drug reaction genes (left). Number of genes overlapping between male-biased genes and adverse drug reaction genes (right)