An experimental census of retrons for DNA production and genome editing

Abstract

Retrons are bacterial immune systems that use reverse-transcribed DNA (RT-DNA) to detect phage infection. They are also deployed for genome editing, where they are modified so that the RT-DNA encodes an editing donor. Retrons are common in bacterial genomes, and thousands of unique retrons have been predicted bioinformatically. However, few have been characterized experimentally. We add to the corpus of experimentally studied retrons, finding 62 empirically determined, natural RT-DNAs that are not predictable from the retron sequence alone. We synthesize >100 previously untested retrons to identify the natural sequence of RT-DNA they produce, quantify their RT-DNA production and test the relative efficacy of editing using retron-derived donors to edit bacterial, phage and human genomes. We observe large diversity in RT-DNA production and editing rates across retrons, finding that top-performing editors are drawn from a subset of the retron phylogeny and outperform those used in previous studies, reaching precise editing rates of up to 40% in human cells.