Abstract Body: Background: Genetics substantially contribute to heart failure (HF), yet the genetically driven changes in cardiac cellular composition or interactions within and between cell types are not fully understood. To better understand this aspect of HF, we modeled heart failure in a large genetically diverse mouse panel, the Collaborative Cross (CC). We then integrated genomic, phenotypic and transcriptomic data with bioinformatics approaches to highlight candidate genes with potential druggability for HF.

Methods and Results: HF was modeled in the CC panel with the beta-agonist isoproterenol (ISO, 30 mg/kg/day for 21 days via osmotic minipumps). We collected data on HF traits (echocardiography, organ weights) and the left ventricle’s transcriptome via RNA-seq.
We first employed the MuSiC pipeline for cell-type deconvolution to re-estimate the RNA-seq data based on inferred cell type contribution and determine differentially expressed genes in response to ISO. We also clustered genes based on their expression pattern using Weighted Gene Correlation Network Analysis, and analyzed the clusters for correlation with CC HF traits, enriched functions, and hub genes. We notice that the ISO network is enriched for immune response or matrix remodeling, whereas the control network shows more general cellular functions like protein control or cell growth. Additionally, we performed quantitative trait loci (QTL) mapping, the traits of interest being cell type proportions (from MuSiC) and module eigengenes (from WGCNA, representing a mouse’s overall cluster pattern). Finally, to infer directions of these associations, we conducted mediation analysis with bmediatR. Combining these with our previously published HF trait QTL results allows us to detect genomic loci and genes that may regulate other genes in one or many cell types, as well as cardiac function, which might be attractive drug targets for HF. Notably, our candidate list consist of both known HF genes and new ones including Ccn2, Ltbp2, Adamts2 and Cyfip2.

Conclusion: Harnessing the genetic and trait data of the CC and performing QTL analysis has revealed multiple genes with previously unknown roles in HF. Future works will focus on in vivo validation and mechanistic studies of these genes.