Dissecting Regulatory Mechanisms in Human Metabolic Tissues Using Genetic and Multiomic Approaches

Complex human traits and metabolic diseases are shaped by genetic and regulatory mechanisms that act across tissues, cell types, and molecular layers. Although genome-wide association studies have identified thousands of disease- and trait-associated loci, most associated variants lie in noncoding regions, making it challenging to identify the causal regulatory mechanisms through which they act. Molecular QTL mapping and single-nucleus multiomic profiling provide complementary approaches for addressing this problem. Genetic variation can be used as a natural perturbation to identify regulatory elements and chromatin features associated with disease-relevant loci, while experimental perturbation can test how disruption of regulatory factors affects chromatin state, gene expression, and cellular function. This dissertation uses genetic and multiomic approaches to dissect regulatory mechanisms in human metabolic tissues. In the first project, single-nucleus ATAC-seq data from 281 human skeletal muscle biopsies were used to decompose chromatin accessibility into nucleosome-free regions (NFR) and nucleosome (nuc)-associated regions across major skeletal muscle cell types. By mapping genetic effects on these partitioned chromatin features, this work identified 76,027 nfrQTLs and 24,623 nucQTLs. Fine mapping and colocalization analyses showed that most nucQTLs share genetic signals with nearby nfrQTLs, while causal inference supported a predominant model in which genetic effects on nucleosome-free region accessibility shape adjacent nucleosome positioning. Integration with GWAS further demonstrated that nfrQTLs prioritize disease-relevant regulatory variants and molecular mechanisms for muscle and metabolic traits, including grip strength, atrial fibrillation, type 2 diabetes, and fasting insulin. In the second project, perturbational single-nucleus multiome profiling was applied to study the function of NKX2.2 in adult human pseudoislets. NKX2.2 is an islet-enriched homeodomain transcription factor required for pancreatic endocrine cell specification, beta-cell differentiation, and maintenance of mature beta-cell function. Functional experiments showed that NKX2-2 perturbation increased insulin secretion, with weaker or absent effects on glucagon secretion. Computational analysis of integrated single-nucleus RNA and ATAC profiles showed limited gene-level differential expression but clearer chromatin accessibility changes after NKX2-2-targeted perturbation. Differential accessibility and motif enrichment analyses suggested that NKX2-2 perturbation remodels endocrine regulatory chromatin landscapes, consistent with chromatin-level effects that are not fully captured by steady-state nuclear RNA abundance. Together, these projects demonstrate that chromatin-centered molecular phenotypes can improve regulatory mechanism discovery in human metabolic tissues. In skeletal muscle, partitioning chromatin accessibility into NFR and nucleosome components improves interpretation of regulatory genetic effects and GWAS loci. In pseudoislets, perturbational multiome profiling reveals chromatin accessibility responses to transcription factor disruption that complement functional hormone secretion phenotypes. This dissertation supports a framework in which genetic variation, transcription factor perturbation, chromatin accessibility, gene expression, and cellular phenotypes are integrated to prioritize causal regulatory mechanisms relevant to metabolic disease.