Deciphering the Enhancer Landscape and Regulatory Code of Heart Development at Single-Cell Resolution - SUMMARY Heart disease is the most common cause of death world-wide and congenital heart disease (CHD) is the most common class of life-threatening human birth defects. However, the gene regulatory mechanisms underlying heart development and their contribution to heart disease remain poorly understood. Enhancers, non-coding regulatory elements that orchestrate gene expression, are critical for cardiac development but their cell-type- specific activity, regulatory grammar, and the impact of mutations on their function remain largely unexplored. This project aims to map and decode the enhancer landscape of human heart development through single-cell multiomic data analysis, extensive in vivo validation in mice, and creation of deep learning (DL) models of cardiac enhancer function. In Aim 1, we will create a detailed single-cell enhancer atlas of the developing human heart. We will integrate genome-wide single-cell multiome data from critical stages of human heart development from preliminary studies with additional, publicly available data from human and mouse embryonic hearts. Our preliminary studies indicate that tens of thousands of enhancers are active during heart development, most with highly cell type-specific activities. The fully integrated atlas will provide a prediction of cell type specificity, activity states, developmental profile, and target gene for each enhancer, providing a foundational resource for understanding heart development and studying CHD, and enabling the discovery of enhancers active in rare cardiac cell populations. To assess the quality of this atlas rigorously and demonstrate its utility, in Aim 2, we will functionally validate ~100 candidate enhancers identified in Aim 1 in transgenic mice, using in vivo single- cell resolution assays. Given that the concept of using cell type-resolved multiome data for defining cell type specificity of developmental in vivo enhancers has not been extensively used to date, this validation will be critical to verify the reliability of the predictions. This aim will also provide molecular reagents for experimental targeting of specific cardiac cell populations. In Aim 3, we will train a DL framework on cell-type-specific data from Aim 1 to build sequence models of enhancers active in different cardiac cell types. We will then use these models to classify human CHD-associated variants. We will test 70 predicted causative enhancer mutations and control alleles in transgenic mouse assays to study the predicted activity changes in vivo. Finally, to fully assess the extent to which the DL models capture the cell type regulatory code of cardiac enhancers, will design synthetic, cell type-specific enhancers using DL-driven in silico evolution and will validate 30 synthetic enhancers in vivo. Selected CHD-associated and synthetic enhancers will be further characterized in knockin mice. Taken together, this project will elucidate the enhancer landscape of human heart development at single-cell resolution, enable cell type-resolved analysis of genetic findings from CHD patients, and establish a framework for engineering regulatory elements, transforming our ability to interpret non-coding variation and develop precision tools for cardiac research and therapeutics.
