Distinct genetic architecture of gene and isoform level QTL in the Diversity Outbred (DO) mouse population
Genetic association studies of mRNA abundance phenotypes link regulatory gene loci to mRNA abundance (quantitative trait loci; QTL). Here, we used liver RNA-seq data from a large cohort of the Diversity Outbred (DO) mouse population to map QTL for gene abundance (eQTL), isoform abundance (isoQTL), isoform ratios (irQTL), and splicing (sQTL). QTL studies using this heterogeneous mouse population have been limited to gene-level abundance of mRNAs and have not focused on mRNA isoforms or splicing. We show for the first time that genetic effects on expression are distinct from splicing in the DO mouse population. Using allele-effect patterns from local QTL for protein-coding gene-isoform pairs, we show that genetic variation drives allele-specific isoform usage, generating isoforms whose genetic signals diverge from their aggregated gene-level effects. We conducted pathway enrichment on distal eQTL and isoQTL hotspots and uncovered pathways not detected with eQTL. We then applied a composite mediation approach at these distal hotspots that compares gene-gene, isoform-isoform, and isoform-gene mediator models. By contrasting these causal models of transcriptional regulation, we identified unique associations between mRNA isoforms. We identified candidate genes at irQTL and sQTL hotspots that regulate mRNAs primarily through isoform usage and splicing, including Alkbh1 and Dicer1. For these driver genes, we nominated novel candidate targets through mediation analysis and pathway associations with known RNA-binding targets. We integrated our QTL data with human genetic data, prioritizing effector genes in loci associated with metabolically relevant traits. Our data also suggest that sex and diet influence eQTL and isoQTL distinctly and primarily through distal-acting gene loci. Overall, our findings highlight distinctive genetic effects on transcriptional and post-transcriptional mechanisms of gene regulation.