Improving marine viral metagenomics through long read sequencing

Abstract

As arch manipulators of their host communities’, marine phages (viruses of bacteria) are key players in global biogeochemical cycles. Phage-mediated host lysis is a critical component of nutrient cycles and structures host communities. Horizontal gene transfer and co-evolution alter host-substrate interactions. Phages even ‘hijack’ their hosts, modifying host function to maximise virus production, or to improve host fitness for survival of the infected cell. Unfortunately investigation of marine viruses is hampered by the low proportion of hosts in culture and the resulting dearth of model host-virus systems. As phages lack a universal marker gene for PCR-based survey, metagenomic approaches have become the standard approach for assessments of viral community structure. However, viral ‘populations’ (equivalent: species) which are highly microdiverse (i.e. diverse at single nucleotide level) and/or replete with hypervariable regions (i.e. extremely variable regions within viral populations) fail to assemble from short-read metagenomic datasets. The central aim of this thesis was to leverage long-read sequencing technology for better characterisation of marine viral community diversity and structure. Here, I present ‘VirION’ (Viral, long-read metagenomics via MinION sequencing), a hybrid, short- and long-read sequencing and informatics workflow. After validation with mock viral communities, VirION was applied to both coastal (Western English Channel) and Open Ocean (Sargasso Sea) samples. The results here show that VirION captured: I). Many globally abundant and ubiquitous viruses missing from short-read metagenomic assemblies; II) Viral genomes that were more microdiverse and hypervariable-region rich than those in short-read assemblies; III) Hypervariable regions containing niche-defining genes of host origin, plus genes likely involved in host-virus interactions and host-hijacking; IV) Single-virus resolution in variations of viral hypervariable regions. Although this approach did not improve host-prediction, the identification of many previously ‘invisible’ viruses, including their host-virus interaction hotspots, improved characterisation of marine dsDNA phage communities, and should be used to guide future investigations of viral ecosystem function. Further development of VirION for the capture of cell-associated-, ssDNA and RNA viruses could provide a powerful tool for environmental monitoring whose scope goes beyond marine systems.