1Universitätsklinikum Heidelberg Klinik für Innere Med. III, Kardiologie, Angiologie u. Pneumologie Heidelberg, Deutschland
The genetic background of cardiomyopathies have been widely studied and many genes could already be associated with the disease. However, for the majority of such genes linked e.g. through genome-wide association studies (GWAS), the mechanism leading to the disease is still unclear.
Chromosomal interactions on a 3D level regulate gene expression, including long-range enhancers and repressors’ control or epigenomic modifications. Despite the proven relevance of the changes in the spatial genome organization in the context of cardiac disease and development, most research has been limited to cellular (i.e. hESC, hiPSC) and mouse models.
This study aims to pinpoint the genomic 3D-structure by using HiC (High-throughput Genome-Wide Chromosome Conformation Capture) and PC-HiC (Promoter Capture HiC) and link it to genetic susceptibility loci from GWAS.
Besides the (PC-)HiC analyses, random-primer based RNA-Seq, WGS and methylation assays (Infinium EPIC) were performed in a cohort of left ventricle heart tissue samples including controls (n=9), DCM (n=13), and ICM (n=10). We generated 6.2 billion-HiC reads enabling the detection of 119.9 ± 30.8 million genomic (cis-)interactions in average. PC-HiC experiments generated 4.1 billion reads that we used to identify interacting regions of 18764 promoters.
First, we investigated the compartmentalization of the genome, with HiC contact maps that undergo an iterative correction of the matrices followed by eigenvector decomposition to derive compartment types. We found 2197 regions with changing PC1 values comparing control-DCM (adjusted p- value ≤ 0.005, using FDR’s -Benjamini Hochberg- method for adjustment). 75.6% of those regions fall on protein coding regions and 21.6% on ncRNA. In the control-ICM test, there are 726 changing regions (adjusted p-value ≤ 0.005) and ncRNAs were more present on a 28.5%. An algorithm based on a generalized linear model was used for the analysis of statistically significant changes in regulatory topologically associated domains (TAD). From a total of 6478 regions affected by a modification in regulatory domains in control vs. DCM, 31.4% of the loci reached statistical significance (adjusted p-value ≤ 0.05).
Next, using a Bayesian prioritization approach (COGS), the promoter interactome was used to rank putative DCM-associated genes across the two largest GWAS meta-analysis performed in DCM. Unlike in the nearest-exon approach with this strategy we prioritized a total of 64 genes by correlating fine mapped SNP signals from the GWAS DCM-trait to gene coding regions, using the promoter interacting regions. The prioritized genes displayed both known (i.e. FLNC, BAG3) and novel candidate genes. When evaluating such genes on the transcriptional level, we found that some of the prioritised genes that on top show a differential gene expression pattern in DCM also had chromatin rearrangements implicating regulatory elements. Further, functional validation of candidates with CRISPRi perturbation are ongoing.
To our knowledge, this is the first three-dimensional chromatin organization analysis of left ventricular heart tissue in a large cohort. We provide a comprehensive genome-wide interaction map for dilated cardiomyopathy and by colocalization of GWAS traits with the 3D-chromosomal regulation landscape we link the complex relationship between DNA activity, chromatin interactions and gene expression that will aid in the understanding the molecular etiology of cardiomyopathy.