Abstract 1516 Computational Prediction and Biochemical Validation of CVD-associated SNPs Altering NKX2-5 DNA-binding

Cardiovascular diseases (CVDs) are the leading cause of death worldwide (>1,000,000 deaths each year) and are heavily influenced by genetic factors. Genome-wide association studies (GWAS) have identified >90 % of CVD-associated mutations within non-coding regions of the genome, such as promoters and enhancers, which are essential to regulate gene expression. These non-coding regulatory regions are recognized by transcription factors (TFs) needed for gene regulation, such as NKX2-5, which is essential in regulating heart development. However, due to the overwhelming number of GWAS single nucleotide polymorphisms (SNPs) (>500,000), performing in vitro biochemical validation on all of them is not viable. To address this challenge, we prioritized CVD-associated SNPs that alter NKX2-5 DNA binding through a computational approach, that uses genomic regions bound by NKX2-5 (ChIP-seq data) as input to predict changes in TF-DNA binding. The model was successful at predicting NKX2-5 DNA binding sites within cardiac enhancers (as determined by Electromobility Shift Assay). We prioritized three CVD-associated SNPs (rs59310144, rs6715570, and rs61872084) predicted to alter NKX2-5 DNA binding and have genotype-dependent gene expression in cardiac tissue. For in vitro validation, a recombinant NKX2-5 DNA-binding domain (DBD) was expressed and purified through Ni-NTA affinity chromatography (validated by SDS-PAGE and Western Blot). NKX2-5 (DBD) DNA-binding capacity was evaluated on a known binding site within the ANF promoter EMSA. All three CVD-associated SNPs predicted to alter NKX2-5 DNA-binding were validated by EMSA and resulted in altered TF-DNA binding affinity. Variants rs59310144, and rs61872084 showed a decrease in binding affinity, while rs6715570 increased. Variants rs59310144, rs6715570, and rs61872084 are associated with genotype-dependent expression of RNASEH2B(increase) and BARD1 (increase) in the heart atrial appendage and METTL10 (decrease) in the left ventricle, respectively. Our findings suggest that NKX2-5 regulation of the RNASEH2B (repression), BARD1 (activation), and METTL10 (activation) genes are possible mechanisms that can be further explored to establish rs59310144, rs6715570, and rs61872084 as causal CVD risk-variants. Although the etiology of human diseases is complex and multifactorial, this approach can provide crucial information that can be implemented during in vivo experiments or clinical research to address genetic diseases caused by non-coding SNPs. In summary, we believe this bioinformatic approach, which considers tissue-specific eQTL analysis and SVM-based TF binding site classification, is a scalable method that can be applied to multiple types of human diseases.