Heart failure encompasses a diverse group of cardiomyopathies, including myocardial infarction, hypertrophic, dilated, and arrhythmogenic forms, each defined by distinct etiologies. By integrating single-nuclei transcriptomic data from human heart tissue across these conditions, we constructed a unified atlas containing 1.8 million nuclei from 195 individuals. The atlas reveals disease-specific cellular clusters, transcriptional changes and altered ligand-receptor interactions. We found cell states specific to the ischemic zone of myocardial infarction, explored the influence of cytokines on fibroblast cell-states, and identified etiological pathways in different cell types. The integration of summary statistics of 52 genome-wide association studies with atlas-wide gene expression highlighted genetically associated pathways involving metabolic dysregulation and ion channel dysfunction. Implementation of an AI agent led to the identification of ZLN005, a small molecule that boosts mitochondrial biogenesis via PGC-1α, whose cardioprotective effect we validated experimentally, underscoring the utility of the atlas in early therapeutic target discovery for heart failure.
Bleckwehl et al. (Mon,) studied this question.