Comparative analysis of poxvirus genome evolution

Abstract

In this thesis I examine the molecular evolution of poxviruses using various comparative genomics approaches. Poxviruses are large DNA viruses that infect vertebrates and insects. Understanding the evolutionary processes that have shaped them can help us explain their infectious cycle and host interactions. I lay the groundwork for the later analyses presented here by reconstruct-ing a robust phylogenetic tree of the 104 poxvirus genomes whose complete genome sequences were available. This tree is based on concatenated multiple alignments of core families found in all poxviruses, and in order to identify these families, I used homology searches (BLAST) and clustering algorithms (MCL). The resulting tree agreed with previously published phylogenies of smaller groups of poxvirus genomes. The first study discussed in this thesis aims to identify cases of horizontal gene transfer (HGT) from a host or host-related organism into poxviruses. The main genomic feature used here is conserved synteny, the genomic neighbourhood of the gene in question, which, for most poxviruses, is extremely well conserved and helps to determine whether one or multiple independent HGT events have taken place. The main finding from this study is that taking synteny into account can provide very different conclusions about HGT history than merely looking at sequence homology between host and viral genes. In particular, for thymidine kinase and interleukin-10, multiple HGT events are implied by different, but conserved, syntenic environments, where phylogenetic relationships favoured single origins. In the second study, an attempt is made to predict interactions between proteins, based on their phylogenetic profiles, i.e. presence or absence of genes encoding them across all poxviruses. This method assumes that where two proteins are present in a non-monophyletic group of genomes, and therefore have experienced co-gains or co-losses, this implies a biological link. While some predictions can be made, some with plausible functional links, biological relevance is hard to ascertain. When examining all known protein-protein interactions between proteins encoded by vaccinia virus, it becomes very clear that such interactions exist mainly between well-conserved proteins present since the early poxviral lineage, which tend to be located near the genomic centre. Using these characteristics, some tentative predictions of further interactions can be made, although there are too many that are too inter-connected to offer a useful shortlist of candidates for further investigation. The last study presented here takes a close look at each gene in a representative vaccinia virus genome with the aim to identify its evolutionary point and mechanism of origin. Evidence for gene fusion and fission, HGT and de novo origins is sought but the origins of a large proportion of genes (38%) remain undiscovered. A poxviral origin via horizontal transfer from another organism is by far the most common, at 32%.