Abstract 1489 Congenital Heart Disease-associated Mutations Alter TBX5-DNA Binding Affinity

Congenital heart diseases (CHDs) are the most common birth defect, affecting 1 in every 100 liveborn infants. CHDs are characterized by structural abnormalities that occur during heart development. Among the possible causes of CHDs are mutations that alter cardiac transcription factor (TF) function, such as TBX5. TBX5 is a cardiac TF crucial to regulating heart development in embryonic stages and mutations in the TBX5 gene have been associated with several cardiovascular diseases, including CHDs. However, genome-wide association studies (GWAS) have mapped over 97% of CHD-associated mutations within non-coding regions of the genome, such as promoters and enhancers. In this work, we evaluated the potential of CHD-associated non-coding single nucleotide polymorphisms (SNPs) to alter TBX5-DNA binding. First, we trained a machine learning-based predictive model (LS-GKM SVM) using TBX5 ChIP-seq data from human cardiomyocytes to identify TF binding sites (TFBS). The model was tested by scoring known human heart footprints and enhancers and validating them in vitro through electrophoretic mobility shift assay (EMSA). A recombinant TBX5 DNA-binding domain, also known as a T-box domain, was expressed and purified through Ni-NTA affinity chromatography with >90 % purity (determined by SDS-PAGE). Afterward, we identified 121 CHD GWAS SNPs and included variants in linkage disequilibrium (LD, R2 = 0.80) from 4 populations (EUR, AFR, EAS, and SAS), resulting in 3,232 SNPs. We then used the predictive model to score the CHD-associated SNPs potential to alter TBX5-DNA binding affinity. We tested the model by validating three variants that are predicted to increase (rs11904437and rs2438368) and decrease (rs262828) in vitro through EMSA. After biochemical validation, the model was successful in predicting changes in TBX5-DNA binding for all three CHD-associated SNPs, suggesting that we can successfully predict changes in binding affinity for most variants. Future steps include a high-throughput approach to evaluate TBX5-DNA binding for all 3,232 CHD-associated SNPs and determine changes in gene expression. Completion of this project will provide valuable information needed to understand genetic diseases that are caused by non-coding mutations.