Systematic and scalable genome-wide essentiality mapping to identify nonessential genes in phages

Abstract

Phages are one of the key ecological drivers of microbial community dynamics, function, and evolution. Despite their importance in bacterial ecology and evolutionary processes, phage genes are poorly characterized, hampering their usage in a variety of biotechnological applications. Methods to characterize such genes, even those critical to the phage life cycle, are labor intensive and are generally phage specific. Here, we develop a systematic gene essentiality mapping method scalable to new phage–host combinations that facilitate the identification of nonessential genes. As a proof of concept, we use an arrayed genome-wide CRISPR interference (CRISPRi) assay to map gene essentiality landscape in the canonical coliphages λ and P1. Results from a single panel of CRISPRi probes largely recapitulate the essential gene roster determined from decades of genetic analysis for lambda and provide new insights into essential and nonessential loci in P1. We present evidence of how CRISPRi polarity can lead to false positive gene essentiality assignments and recommend caution towards interpreting CRISPRi data on gene essentiality when applied to less studied phages. Finally, we show that we can engineer phages by inserting DNA barcodes into newly identified inessential regions, which will empower processes of identification, quantification, and tracking of phages in diverse applications.

Sequence modeling and design from molecular to genome scale with Evo

“…On a dataset of 56 whole-genome essentiality studies in bacteria from the DEG database (Zhang, 2004) and two whole-genome essentiality studies in phage from Piya et al (2023), we observed that the changes in Evo log-likelihood with 66k context are significantly predictive (Bonferroni-corrected permutation-based P < 0.05) of gene essentiality in 43 out of 58 genomes. We also observed that providing the model with additional genomic context beyond the gene sequence results in a substantial improvement in performance, especially from gene only context to 8k context.…”

“…With 8k context, the model most likely has access to enough of the genome to improve its prediction of mutational effects on organism function, whereas 66k context provides new, helpful information in only some cases. For a few genomes, the zero-shot performance with 66k context is notably strong, with an AUROC of 0.86 on lambda phage essentiality data (Piya et al, 2023) and an AUROC of 0.81 on Pseudomonas aeruginosa essentiality data (Turner et al, 2015) ( Figure 5D ).…”

“…We obtained binary genome-wide essentiality results for 56 bacterial genomes from the DEG database (Zhang, 2004) in which coding genes are labeled with “essential” or “nonessential” binary labels. We also obtained genome-wide essentiality results for two phage genomes, lambda and P1, from Piya et al (Piya et al, 2023) and used the binary labels assigned by the study authors based on the results of their CRISPRi screen.…”

Analysis and culturing of the prototypic crAssphage reveals a phage-plasmid lifestyle

“…One mechanism by which phages have evolved a non-integrative lysogenic state is through the acquisition of plasmid genes. Phage-plasmids (PPs) are large DNA elements (larger than phage or plasmid alone, >90Kb), that encode phage, plasmid, and accessory genes 22,23 .…”

“…One mechanism by which phages have evolved a non-integrative lysogenic state is through the acquisition of plasmid genes. Phage-plasmids (PPs) are large DNA elements (larger than phage or plasmid alone, >90Kb), that encode phage, plasmid, and accessory genes 22,23 . Rather than switching between lysis and integration, PPs switch between phage lysis and low-copy number plasmid replication modes 24 .…”