New approaches to characterise viral pathogens in aquaculture

Aquaculture is the fastest growing food-producing sector in the world and of considerable economic, cultural and environmental relevance. This sector will be vital to achieving future food security demands, but its continued sustainable expansion is severely threatened by infectious diseases, with viral diseases amongst the most problematic to control. Unlike farmed livestock, fish are generally reared in open systems with constant circulation between farms and the natural aquatic environment. This routinely exposes the animals to naturally occurring viruses in the water, both pathogenic and non-pathogenic, which are generally uptaken through mucosal surfaces (i.e. gills and gut surfaces). However, with the increase in globalisation, aquatic species are frequently farmed in non-native habitats, thus exposing them not only to the pathogens present in wild relatives of the same species, but to pathogens of other species in their introduced habitat. Moreover, wild fish are threatened by viral disease outbreaks on fish farms due to the high density of individuals available to carry and transmit the pathogen. Characterising viral infections is therefore important to support the prevention and control of disease outbreaks, as understanding the disease agent enables both fish farmers and regulating agencies to tailor appropriate mitigation strategies. The routine use of whole genome sequencing to screen infected animals is not yet commonplace in the aquaculture industry, where genetic screening of viruses is largely done using PCR for 1 or 2 marker genes. However, the ‘genomic surveillance’ approach has been used to great effect in cases of disease outbreaks relevant to human health, and could be applied in aquaculture to enhance the resolution of molecular epidemiology investigations and diagnostic tests. Moreover, with the under-researched genetic diversity of aquatic viruses, significant advances in the understanding of host-pathogen interactions could be achieved with a denser and better curated genomic database of viruses. To address these knowledge gaps, I have developed and optimised several approaches to characterise aquatic viruses up-taking various sequencing methods depending on the resolution required for the specific study, using salmonid alphavirus (SAV) as a primary study system. To rapidly and accurately generate consensus-level genomes of specific pathogenic viruses, I developed a targeted PCR approach using overlapping long amplicons tiled across the SAV genome for full coverage. These amplicons are sequenced on the Oxford Nanopore Technologies MinION long-read platform. An analysis workflow was then optimised to generate consensus genomes while maintaining capability to discover SAV subtype-level co-infections by simultaneously mapping to multiple reference sequences. This approach can generate highly accurate consensus sequences (as judged by independent Sanger sequencing) and detect co-infections of strains with ≥ 95% pairwise identity over a 2kb region, even when minor infecting strains are present at just 5% frequency. This approach was used to investigate the population dynamics and phylogeography of the SAV3 epidemic in Norwegian aquaculture, revealing repeated seedings of SAV3 from ‘source’ to ‘sink’ counties. To characterise viral genetic diversity within a host, I applied a targeted sequence capture strategy to obtain SAV genomes at high coverage (using Illumina technology) from infected fish using both pooled and individual tissue samples. This approach utilises RNA baits to capture and enrich for specific DNA (or cDNA) strands in a sample, and allows for greater sequencing efficiency. These baits, while designed from specific templates, are less specific than PCR primers and can tolerate a certain amount of template mismatches, thus capturing all genetic variation of a specific viral species within a sample. This approach was used to compare the genetic diversity of SAV in farmed Atlantic salmon and rainbow trout, in addition to two wild flatfish species, sampled from multiple regions in Scottish and Irish waters. In the same study, I provided evidence of complex infections on single fish farms, and for co-infections within single wild fish. Finally, I developed a pipeline to detect viral infections in metagenomics samples, which can be applied even when the infectious agent is unknown. This involves an optional step of mapping to the host reference genome to increase efficiency of later steps, assembly of the remaining reads with a transcriptome assembler, and identifying viral transcripts using homology-based tools. Before implementation, this pipeline was benchmarked against several datasets, including a simulated virome and a simulated co-infection of two strains of the same virus. It was also tested against datasets with known pathogens, resulting in similar efficiencies of detection as a mapping-based approach. Finally the pipeline was used on datasets with unknown viromes to demonstrate its applicability to detect novel viral species. Overall, my research has led to the development of several cutting-edge approaches for the genomic analysis of aquatic viruses and other pathogens, and helps clarify which approach is most useful in different epidemiological settings. I also demonstrate that genome-wide analyses of viral pathogens impacting salmonid aquaculture generates valuable additional information on viral diversity compared to standard surveillance methods using particular marker genes, advocating for route use of genomic approaches in this sector.