Tissue-specific expression and regulation of sexually dimorphic genes in mice

Abstract

We report a comprehensive analysis of gene expression differences between sexes in multiple somatic tissues of 334 mice derived from an intercross between inbred mouse strains C57BL/6J and C3H/HeJ. The analysis of a large number of individuals provided the power to detect relatively small differences in expression between sexes, and the use of an intercross allowed analysis of the genetic control of sexually dimorphic gene expression. Microarray analysis of 23,574 transcripts revealed that the extent of sexual dimorphism in gene expression was much greater than previously recognized. Thus, thousands of genes showed sexual dimorphism in liver, adipose, and muscle, and hundreds of genes were sexually dimorphic in brain. These genes exhibited highly tissue-specific patterns of expression and were enriched for distinct pathways represented in the Gene Ontology database. They also showed evidence of chromosomal enrichment, not only on the sex chromosomes, but also on several autosomes. Genetic analyses provided evidence of the global regulation of subsets of the sexually dimorphic genes, as the transcript levels of a large number of these genes were controlled by several expression quantitative trait loci (eQTL) hotspots that exhibited tissue-specific control. Moreover, many tissue-specific transcription factor binding sites were found to be enriched in the sexually dimorphic genes.

Figure 1.

Two-dimensional hierarchical clustering of gene expression in livers of 334 BxH F₂ mice. A set of 1332 most transcriptionally active genes are clustered on the horizontal axis, with individual mice clustered on the vertical axis. The major groups of mice resulting from the hierarchical clustering are labeled by sex. Female mice are also labeled with the vertical blue bars on the left and right sides of the heatmap. Red represents overexpression compared to the control (a pool of RNA comprised of equal aliquots of RNA from 150 F₂ mice), and green represents underexpression. Brighter shades of red and green reflect higher degrees of over and underexpression. Black corresponds to no difference, and gray indicates missing data.

Figure 2.

Venn diagram showing tissue specificity of the active genes and the sexually dimorphic genes. (A) Overlap of active genes among liver, adipose, and muscle. (B) Overlap of sexually dimorphic genes among liver, adipose, and muscle. The gene number in each distinct area and its percentage over the combined total gene number (19,669 for A and 16,452 for B) are shown.

Figure 3.

eQTL hotspots for sexually dimorphic genes in adipose and liver. (A) Genomewise distribution of significant eQTL for all active genes (upper) and sexually dimorphic genes at >1.2-fold-change level (middle) and >2-fold-change level (lower) in adipose tissue. (B) Genomewise distribution of significant eQTL for all active genes (upper) and sexually dimorphic genes at >1.2-fold-change level (lower) in liver. Hotspots with significantly enriched eQTL counts for the sexually dimorphic genes (P < 0.05, Fisher’s exact test with Bonferroni correction) are labeled.