UCSF

A central goal of organismal biology is to understand how the physiological functions of organs and tissues arise from the coordinated activity of thousands of genes in diverse principal and support cell types. Recent large-scale atlasing efforts have characterized the molecular composition of thousands of distinct cell types across the organs; ongoing work is mapping the spatial organization of these cell types across tissues. Although such efforts provide a foundation resource, a major challenge for the field is to causally understand how cell types, cell states, and multicellular structures and motifs are produced and regulated by the actions of specific genes and molecular pathways. This thesis describes an approach that could enable a systematic and comprehensive understanding of how specific genes control diverse cellular and tissue phenotypes in living animals, under different physiological states. Chapter 1 describes the use of Perturb-seq to create a comprehensive portrait of the transcriptional impact of genetic perturbations at genome-scale in cell culture, enabling the prediction of function for thousands of genes, the investigation of complex cellular phenotypes. Chapter 2 describes two new technologies to take this many genotypes/complex phenotypes approach into animal tissue. The first is a new method for fixed cell Perturb-seq, which drastically simplifies the processing of animal tissues and enables large-scale experiments. The second is an imaging-based method that captures both spatial gene expression and multiplexed immunofluorescence. Through an integrated analysis of imaging and transcriptional phenotypes, we identify novel regulators of hepatocyte zonation, reveal how proteostatic stress pathway activation can broadly affect the expression of secreted proteins, and show how diverse cellular states can produce convergent effects on lipid accumulation. In total, we provide a new path to characterizing multidimensional genotype-phenotype relationships in living tissues, in a scalable manner that will be broadly applicable across diverse tissues as well as physiological and disease state.