Analyses of allele-specific gene expression in highly divergent mouse crosses identifies pervasive allelic imbalance

Abstract

Complex human traits are influenced by variation in regulatory DNA through mechanisms that are not fully understood. Because regulatory elements are conserved between humans and mice, a thorough annotation of cis regulatory variants in mice could aid in further characterizing these mechanisms. Here we provide a detailed portrait of mouse gene expression across multiple tissues in a three-way diallel. Greater than 80% of mouse genes have cis regulatory variation. Effects from these variants influence complex traits and usually extend to the human ortholog. Further, we estimate that at least one in every thousand SNPs creates a cis regulatory effect. We also observe two types of parent-of-origin effects, including classical imprinting and a new global allelic imbalance in expression favoring the paternal allele. We conclude that, as with humans, pervasive regulatory variation influences complex genetic traits in mice and provide a new resource toward understanding the genetic control of transcription in mammals.

Figure 1

Diallel crossing scheme and sample sizes. We selected three divergent inbred strains representative of three subspecies within the Mus musculus species group. We generated offspring from all possible pairwise crosses to form a 3×3 diallel, including age- and sex-matched biological replicates for each of the nine possible genotypic combinations. Mice were aged to 23 days, sacrificed, and total RNA extracted from whole brain, liver, kidney, and lung. Sample size shown is for RNAseq (52 female, 39 male). RNAseq was performed on RNA extracted from brain and microarrays were run on RNA extracted from brain, liver, kidney, and lung.

Figure 2

Principal components of brain RNAseq and microarray expression levels across four tissues. Each point represents one animal with shape indicating sex (circle = female, square = male) and color indicating genotype. For the F₁ animals, the outer color indicates maternal strain and the inner color paternal strain. (a) PC1 versus PC2 of the brain RNAseq total read counts for all autosomal genes. The three inbred strains form a near-perfect triangle with the F₁ samples located between their corresponding parental strains. PC1 and PC2 account for 31% of the variance in TReC, indicating that genetic background is the overwhelming driver of gene expression difference, greatly exceeding the effects of parent-of-origin and sex. (b) PC1 versus PC2 of microarray expression values for all autosomal genes across four tissues. The pattern seen in the brain extends to multiple diverse tissues.

Figure 3

Balanced contribution of different subspecies to the identification of cis regulated genes. (a) Venn diagram showing the number of genes with allelic imbalance (FDR < 0.05) in each cross and the relationship to other crosses. (b) Distribution of the allelic imbalance effect size for the 11,287 autosomal genes that showed allelic imbalance in at least one cross. In each cross, the proportion is the fraction of allele-specific reads from the strain listed second in the legend (i.e., PWK or WSB). The inset magnifies the distribution of effect sizes in the vicinity of 0.5 and provides, in the background, the distribution of effect size for genes that did not reach statistical significance for a strain effect (filled distributions).

Figure 4

Differential gene expression is positively correlated to sequence diversity at multiple evolutionary scales. Each square indicates the relationship between the local level of sequence diversity (SNPs/kb) and the fraction of genes that show differential gene expression (proportion of genes with additive, consistent strain effects), for regions of the genome with the same or different subspecific origin (indicated by dendrograms). Colored circles represent strain (magenta: PWK, blue: WSB, green: CAST), while colored text represents the subspecific origin in the regions of the genome considered (magenta: musculus, blue: domesticus, green: castaneous). For each of the six pairwise comparisons, only expressed genes with allele-specific information were considered and only SNPs within the entire gene body (±10 kb) were included. The portion of the genome considered for each of these six comparisons was approximately, from left to right in the figure: 50 Mb, 150 Mb, 175 Mb and 2.25 Gb for the final three comparisons.

Figure 5

Imprinted genes in mouse brain. (a) Paternal expression ratio for 95 genes with a significant parent-of-origin effect. Each dot corresponds to a reciprocal cross (e.g., CASTxPWK vs PWKxCAST) and dot size is proportional to the parent-of-origin effect P-value. Genes known from the literature to be maternally expressed are shown in red, those known to be paternally expressed in blue, and novel imprinted genes in black (n = 54 novel genes). Genes with a strain by parent-of-origin effect are underlined (n = 47 genes). (b) Distribution of the parental expression proportion in the vicinity of 0.5 for genes that are imprinted (lines) and, in the background, genes that did not reach statistical significance for parent-of-origin-dependent expression (filled distributions).

Figure 6

Global allelic imbalance in favor of the paternal allele. (a) Distribution of the proportion of paternal expression for all genes, except the 95 imprinted genes described in Figure 5. The distribution reflects aggregate data for ~10,000 genes × 3 crosses × 2 sexes. The dashed red line represents a reflection of the values to the left of 0.5 (the expectation if no paternal skew was present). (b) Genes with consistent allelic imbalance (found in all three crosses) are clustered in most autosomes. The red lines denote the expected proportion of clustering based on the number of genes with consistent paternal or maternal expression in every autosome. (c) Genes with consistent paternal expression in all three crosses and both sexes (N = 467) tend to be closer to CpG islands, while those with consistent maternal expression (N = 116) tend to be farther away, relative to inconsistent genes (N = 9,540). Plotted is the cumulative proportion of genes with a given distance between transcriptional start site (TSS) and the nearest CpG island. (d) Expanded analysis including genes not fully consistent in both sexes, but still consistent in all three crosses. Genes with consistent paternal expression (N = 3,338) retain enrichment for CpG islands, while those with consistent maternal expression (N = 1,631) are not different from inconsistent genes (N = 5,154).