Abstract
Characterizing virus-host pairs and the infection state of individual cells is the major technical challenge in microbial ecology. We addressed these challenges using state-of-the-art single-cell genome technology (SAG-gel) combined with extensive metagenomic datasets targeting the bacterial and viral communities in Lake Biwa. From two water layers and two seasons, we obtained 862 single-cell amplified genomes (SAGs), including 176 viral (double-stranded DNA phage) contigs, which identified novel virus-host pairs involving dominant freshwater lineages. The viral infection rate, estimated by mapping the individual SAG's raw reads to viral contigs, showed little variation among samples (12.1%-18.1%) but significant variation in host taxonomy (4.2%-65.3%), with copiotrophs showing higher values than oligotrophs. The high infection rates of copiotrophs were attributed to collective infection by diverse viruses, suggesting weak density-dependent virus-host selection, presumably due to their nonpersistent interactions with viruses resulting from fluctuating abundance. In contrast, the low infection rates of oligotrophs supported the idea that their codominance with viruses is achieved by genomic microdiversification, which diversifies the virus-host specificity, sustained by their large population size and persistent density-dependent fluctuating selection. Notably, we discovered viruses infecting CL500-11, the dominant bacterioplankton lineage in deep freshwater lakes worldwide. These viruses showed extremely high read coverages in cellular and virion metagenomes but were detected in <1% of host cells, suggesting a low infection rate and high burst size. Overall, we revealed highly diverse virus-host interactions within and between host lineages that were overlooked at the metagenomic resolution.