Bacteriophage genomes rapidly evolve via mutation and horizontal gene transfer to counter evolving bacterial host defenses; such arms race dynamics should lead to divergence between phages from similar, geographically isolated ecosystems. However, near-identical phage genomes can reoccur over large geographical distances and several years apart, conversely suggesting many are stably maintained. Here, we show that phages with near-identical core genomes in distant, discrete aquatic ecosystems maintain diversity by possession of numerous flexible gene modules, where homologous genes present in the pan-genome interchange to create new phage variants. By repeatedly reconstructing the core and flexible regions of phage genomes from different metagenomes, we show a pool of homologous gene variants co-exist for each module in each location, however, the dominant variant shuffles independently in each module. These results suggest that in a natural community, recombination is the largest contributor to phage diversity, allowing a variety of host recognition receptors and genes to counter bacterial defenses to co-exist for each phage.

Microbial communities in glacial ecosystems are diverse, active, and subjected to strong viral pressures and infection rates. In this study we analyse putative virus genomes assembled from three dsDNA viromes from cryoconite hole ecosystems of Svalbard and the Greenland Ice Sheet to assess the potential hosts and functional role viruses play in these habitats. We assembled 208 million reads from the virus-size fraction and developed a procedure to select genuine virus scaffolds from cellular contamination. Our curated virus library contained 546 scaffolds up to 230 Kb in length, 54 of which were circular virus consensus genomes. Analysis of virus marker genes revealed a wide range of viruses had been assembled, including bacteriophages, cyanophages, nucleocytoplasmic large DNA viruses and a virophage, with putative hosts identified as Actinobacteria, Alphaproteobacteria, Cyanobacteria, Firmicutes, Gammaproteobacteria, eukaryotic algae and amoebae. Whole genome comparisons revealed the majority of circular genome scaffolds formed 12 novel groups, two of which contained multiple phage members with plasmid-like properties, including a group of phage-plasmids possessing plasmid-like partition genes and toxin-antitoxin addiction modules to ensure their replication and a satellite phage-plasmid group. Surprisingly we also assembled a phage that not only encoded plasmid partition genes, but a clustered regularly interspaced short palindromic repeat (CRISPR)/Cas adaptive bacterial immune system. One of the spacers was an exact match for another phage in our virome, indicating that in a novel use of the system, the lysogen was potentially capable of conferring immunity on its bacterial host against other phage. Together these results suggest that highly novel and diverse groups of viruses are present in glacial environments, some of which utilise very unusual life strategies and genes to control their replication and maintain a long-term relationship with their hosts.

Eukaryotic genomes contain a variety of endogenous viral elements (EVEs), which are mostly derived from RNA and ssDNA viruses that are no longer functional and are considered to be “genomic fossils.” Genomic surveys of EVEs, however, are strongly biased toward animals and plants, whereas protists, which represent the majority of eukaryotic diversity, remain poorly represented. Here, we show that protist genomes harbor tens to thousands of diverse, ~14 to 40 kbp long dsDNA viruses. These EVEs, composed of virophages, Polinton-like viruses, and related entities, have remained hitherto hidden owing to poor sequence conservation between virus groups and their repetitive nature that precluded accurate short-read assembly. We show that long-read sequencing technology is ideal for resolving virus insertions. Many protist EVEs appear intact, and most encode integrases, which suggests that they have actively colonized hosts across the tree of eukaryotes. We also found evidence for gene expression in host transcriptomes and that closely related virophage and Polinton-like virus genomes are abundant in viral metagenomes, indicating that many EVEs are probably functional viruses.

Background: Polintons are large mobile genetic elements found in the genomes of eukaryotic organisms that are considered the ancient ancestors of most eukaryotic dsDNA viruses. Originally considered as transposons, they have been found to encode virus capsid genes, suggesting they may actually be integrated viruses; however, an extracellular form has yet to be detected. Recently, circa 25 Polinton-like viruses have been discovered in environmental metagenomes and algal genomes, which shared distantly related genes to both Polintons and virophages (Lavidaviridae). These entities could be the first members of a major class of ancient eukaryotic viruses; however, owing to the lack of available genomes for analysis, information on their global diversity, evolutionary relationships, eukaryotic hosts, and status as free virus particles is limited. Results: Here, we analysed the metaviromes of an alpine lake to show that Polinton-like virus genome sequences are abundant in the water column. We identify major capsid protein genes belonging to 82 new Polinton-like viruses and use these to interrogate publicly available metagenomic datasets, identifying 543 genomes and a further 16 integrated into eukaryotic genomes. Using an analysis of shared gene content and major capsid protein phylogeny, we define large groups of Polinton-like viruses and link them to diverse eukaryotic hosts, including a new group of viruses, which possess all the core genes of virophages and infect oomycetes and Chrysophyceae. Conclusions: Our study increased the number of known Polinton-like viruses by 25-fold, identifying five major new groups of eukaryotic viruses, which until now have been hidden in metagenomic datasets. The large enrichment (> 100-fold) of Polinton-like virus sequences in the virus-sized fraction of this alpine lake and the fact that their viral major capsid proteins are found in eukaryotic host transcriptomes support the hypothesis that Polintons in unicellular eukaryotes are viruses. In summary, our data reveals a diverse assemblage of globally distributed viruses, associated with a wide range of unicellular eukaryotic hosts. We anticipate that the methods we have developed for Polinton-like virus detection and the database of over 20,000 genes we present will allow for continued discovery and analysis of these new viral groups.