Systematic genome-wide querying of coding and non-coding functional elements in E. coli using CRISPRi

Abstract

Genome-wide repression screens using CRISPR interference (CRISPRi) have enabled the high-throughput identification of essential genes in bacteria. However, there is a lack of functional studies leveraging CRISPRi to systematically explore targeting of both the coding and non-coding genome in bacteria. Here we perform CRISPRi screens in Escherichia coli MG1655 K-12 targeting ~13,000 genomic features, including nearly all protein-coding genes, non-coding RNAs, promoters, and transcription factor binding sites (TFBSs) using a ~33,000-member sgRNA library, which represents the most compact and comprehensive genome-wide CRISPRi library in E. coli to date. Our data reveal insights into the conditional essentiality of the genome with key refinements to screen design and profiling. First, we demonstrate that strong fitness defects associated with essential cellular processes can be resolved using inducible time-series measurements. We show that knockdowns of different classes of genes exhibit distinct, transient responses that are correlated to gene function with genes involved in translation exhibiting the strongest responses. We also query feature essentiality across several biochemical conditions and show that several genes, sRNAs, and operons exhibit conditional phenotypes not reported by previous high-throughput efforts. Second, we evaluate systematically targeting non-genic features (promoters and TFBSs) in the E. coli genome. We show that promoter-targeting guides can be used to add phenotypic confidence to promoter annotations and verify computationally predicted promoters. In contrast to prior studies, we find that promoter knockdowns exhibit a strong targeting orientation dependency where targeting the non-template strand of the promoter closest to the target gene is more effective in knocking down gene expression than other promoter targeting orientations. Unlike eukaryotic genomes, we note that interpreting the effects of TFBS targeting is particularly challenging due to the small size of such features and their proximity to and overlap with other genomic features. Together, this work reveals novel conditionally essential gene phenotypes, provides a characterized set of sgRNAs for future E. coli CRISPRi screens, and highlights considerations for CRISPRi library design and screening for microbial genome characterization.