Profiling Essential Functions and Translation Control via Bacterial Optical Pooled Screening

Abstract

Early cytology established the principle that cells of a given type, from a particular organism, are highly stereotyped in terms of both form and composition. The replicative process, therefore, is one characterized by precision, resulting in an orderly partition of contents and maintenance of subcellular organization. Single-celled model organisms represent an experimentally tractable prism through which replication can be understood. Escherichia coli stands out as the best-studied model for rapidly growing bacteria, which organize division, DNA replication and protein synthesis at the fastest rates known in nature. Moreover, dysfunction in proliferation is increasingly associated with antibiotic persistence and the acquisition of resistance, making a fundamental understanding of growth a health priority. Our ability to disentangle the myriad components of growth largely depends on timelapse microscopy, especially multigenerational imaging enabled by microfluidic cultivation. On the other hand, high-throughput genetic perturbation techniques like TNseq and CRISPRi are powerful tools to screen for the molecular basis of phenotypes. Few methods combine these techniques to perform systematic genetic screening of imaging phenotypes. In this dissertation, I describe my development of multiplex timelapse imaging of E. coli in a microfluidic device and its application to measure cell cycle defects in a CRISPRi library targeting essential genes. In chapter 1, I describe MARLIN, a novel technique for bacterial optical pooled screening (OPS) in the mother machine microfluidic device. I show that MARLIN achieves multiplexing of 100k member genetic libraries, while measuring tens of millions of cell cycles per experiment. In chapter 2, I detail the use of MARLIN to collect imaging data for a ~30k sgRNA pooled CRISPRi library targeting all essential genes in E. coli. I demonstrate that the morphological and growth phenotypes of these knockdowns are reproducible and consistent with known gene function. In chapter 3, I use these data to quantify growth-size scaling among different translation perturbations. Based on these data, I demonstrate that slow translation elongation leads to a SpoT-dependent reduction in the size regulator (p)ppGpp. I use modeling to show that this mechanism accounts for many features of previously proposed E. coli growth laws. In chapter 4, I perform clustering analysis on our MARLIN-CRISPRi data to annotate genes of unknown/unrelated function as involved in division or replication, and validate a role for RNase E in promoting replication initiation.