Abstract 18835: Chromatin Gene Cohorts, Not Individual Cardiac Genes, Modulate Cardiac Phenotype in a Genetically Diverse Population

Abstract

Re-expression of fetal genes is a hallmark of transcriptome change in models of cardiac hypertrophy, but how this remodeling is affected by the common genetic variation present in populations is unknown. We examined gene expression in 86 strains of control and isoproterenol (ISO)-treated mice, which exhibited varying degrees of disease. Correlations between individual fetal genes (Anf, Serca2a, alpha-MHC, beta-MHC) and hypertrophic phenotypes were not significant. Expanded analysis of 37 fetal genes showed sparse correlation (at p<0.05) between expression of individual genes with one or more of 52 systemic phenotypes. We next tested the hypothesis that groups of genes exhibit emergent association with phenotype not observed with individual transcripts. Examining the fetal genes as a group unmasked a correlation with left atrial mass (enriched 3.04-fold vs transcriptome, p<0.001). Grouping cardiac transcription factors (n=31 genes) or genes shown by gain/loss of function studies to contribute to hypertrophic signaling (n=142) revealed no correlation to cardiac phenotypes, whereas chromatin regulators (n=124) were associated with right ventricular mass (1.55-fold, p<0.01), fractional shortening (1.52-fold, p<0.05) and ejection fraction (1.50-fold, p<0.05). We next analyzed histone genes. HOPACH clustering generated 39 clusters of strains based on the basal transcriptome that collapsed to 13 clusters after ISO, suggesting that adrenergic stress entrains gene expression. We observed transcriptional co-regulation of H2A with H2B, and H3 with H4 (the other pairings did not show correlation), mirroring the protein level assembly, which proceeds via pairs of dimers according to the same logic. Comparison of strain grouping based on partitioning clustering using histone stoichiometry or phenotype revealed that hypertrophic mice were enriched in histone cluster 4, marked by minimal change to histone variant expression (enriched 2.27-fold, p<0.01). The two novel conclusions are: fetal gene regulation is buffered within populations by common genetic variation and is not a good predictor of disease; and cardiac phenotype is influenced in the basal and disease states through global changes in chromatin.