SRA

Microbes and their associated viruses are key drivers of biogeochemical processes in both marine and soil biomes. After a decade dominated by the use of metagenomics to evaluate functional and taxonomic diversity in natural microbial communities, there has been a recent resurgence of successful cultivation efforts to isolate model bacterial organisms for hypothesis testing and synthetic community ecology. However, the success in isolating important bacterial representatives has not been matched by similar efforts to cultivate their associated bacteriophages, despite the importance of the latter in nutrient recycling and shaping of community composition. Efforts in viral isolation via plaque assay have been restricted to the minority of bacterial hosts that grow on solid medium. Isolation of phages for ecologically important, but fastidious bacteria such as SAR11 in liquid culture is technically challenging, time-intensive and thus practiced only by a handful of research groups. This paucity of representative viral isolates presents a challenge to current attempts to predict hosts of metagenomic viral contigs through machine learning, limiting the capacity for incorporating phage predation and viral metabolic reprogramming into ecological models. Here, we describe a novel high-throughput system for isolating ecologically important host-virus model systems that couples recent advances in dilution-to-extinction culturing of hosts with sequential enrichment and isolation of associated phages. Applied to 5 monthly samples from the Western English Channel, and samples from the Sargasso Sea, this method successfully isolated a novel member of the SAR11 clade and three new isolates of the marine methylotrophic OM43 clade from the Western English Channel. Sequential enrichment and isolation of phages for SAR11 and OM43 clades yielded 109 new phage isolates, including the first known siphovirus of SAR11 and the first isolated phages for the OM43 clade. 13 isolates were selected for genome sequencing, and organised into 7 viral populations. A total of 109 out of 204 isolation attempts using 6 different host strains of the SAR11 and OM43 clades successfully yielded isolates, indicating that our method has the potential for rapidly increasing the number of viral isolates available for inclusion in ecological models and for better training of machine learning algorithms for host prediction from metagenomically derived viral contigs.