Understanding Cellular Specialization Through Functional Genomics

Abstract

The human body is composed of hundreds of specialized cell types, each fulfilling distinct functions that are together essential for normal tissue homeostasis. This thesis is aimed at identifying genes that contribute to to cell type-specific functions, with major projects focused on (1) a specialized epithelial transport pathway called transcytosis and (2) the challenge of measuring cell type-specific gene expression. In both projects, we applied high-throughput methods to narrow down from the ~25,000 protein coding genes to distinguish the subset that contribute to specialized cellular functions. Common themes include the development of enabling technology and the value of integrating diverse genomic datasets. The results described here implicate new genes in cell type-specific processes and provide a starting place for subsequent investigation into the individual genes and pathways. In the first project, we performed an RNA interference (RNAi) screen to identify genes necessary for receptor-mediated transcytosis, a specialized endosomal pathway in epithelial cells. We developed high-throughput assays to measure the transcytosis of immunoglobulin G (IgG) across cultured epithelial cells in conjunction with gene knockdown. Then we selected a set of 582 candidate genes to screen using a combination of literature review and integrated high-throughput evidence, including expression data, proteomics, and domain annotation. We knocked-down each of these candidates in parallel and identified many reagents that interfered with transcytosis. In small-scale validation assays, we confirmed a reproducible decrease in transcytosis after knocking down 7 genes with multiple independent reagents (7 confirmed out of 8 genes tested). The validated hits included genes with an established role in related pathways, such as EXOC2 and PARD6B, and genes that have not been implicated in epithelial trafficking before, such as LEPROT, VPS13C, and ARMT. In the second project, we developed an approach to identify genes expressed selectively in specific cell types, using a computational algorithm that searches thousands of microarrays for genes with a similar expression profile to known cell type-specific markers. Our method, CellMapper, is accurate without the need for cell isolation and can be applied to any cell type where at least one cell-specific marker gene is known. We demonstrated the approach for 30 diverse cell types, many of which have not been isolated for expression analysis in humans before. Furthermore, we explored the applicability of our method to infer causal relationships in genome-wide association studies (GWAS) and to investigate the transcriptional identity of a poorly understood cell type, enteric glia. We provided a user-friendly R implementation that will enable researchers from systems biology, molecular biology of disease, and population genetics to identify cellular localization of genes of interest or to expand the catalog of known marker genes for difficult-to-isolate cell types.