LONG-RANGE GENE REGULATION IN HUMAN DEVELOPMENT AND ALZHEIMER’S DISEASE

A wide variety of specialized cell types all contain the exact same genomic DNA sequences. Through the process of gene regulation, different cell types will express different genes that allow them to exhibit specific phenotypes and perform specific functions. Less than 1% of the genome encodes protein products and the rest is thought to be regulatory in nature. Different cell types also express different epigenetic profiles that describe their regulatory capacities. While regulatory elements such as enhancers often affect transcription of their closest gene, they may also affect genes up to a million base pairs away in linear space through the formation of chromatin loops. Common genetic variation associated with various human diseases is also enriched in the noncoding regulatory regions of the genome. Understanding specifically which genetic variants are driving the association with a phenotype is complicated due to (1) linkage disequilibrium (LD) and (2) their location within the noncoding genome makes it difficult to discern which genes are impacted. In the first part of this work, I describe a time course study mapping chromatin loops over cell differentiation of lymphoblastoid K562 cells into megakaryocytes with deeply sequenced Hi-C. This was coupled with matched RNA-seq, ATAC-seq, and CUT&RUN for transcription factors (TFs) and Histone H3 Lysine 27 acetylation (H3K27ac). Differential loop analysis identified over 1,500 loops that are gained or lost over differentiation, and linear models revealed that changes in chromatin features at loop anchors and interiors were predictive of changes in looping. I go on to identify putative causal variants in Alzheimer’s Disease (AD) by performing a Massively Parallel Reporter Assay (MPRA) in resting and proinflammatory macrophages. Our MPRA identified 47 expression modulating variants (emVars) that are likely causal in nature. Coupling previously published macrophage multi-omics and microglia eQTL data connected emVars to 76 target genes enriched for AD pathology. Overall, these studies show how complex multi-omic data sets can identify genes and mechanisms of gene regulation over cellular transitions and within human disease. In the case of AD, these insights have increased our understanding of the genetic basis of AD and may identify targets for therapeutic development.