Broad-scale phylogenomics provides insights into retrovirus–host evolution

Abstract

Genomic data provide an excellent resource to improve understanding of retrovirus evolution and the complex relationships among viruses and their hosts. In conjunction with broad-scale in silico screening of vertebrate genomes, this resource offers an opportunity to complement data on the evolution and frequency of past retroviral spread and so evaluate future risks and limitations for horizontal transmission between different host species. Here, we develop a methodology for extracting phylogenetic signal from large endogenous retrovirus (ERV) datasets by collapsing information to facilitate broad-scale phylogenomics across a wide sample of hosts. Starting with nearly 90,000 ERVs from 60 vertebrate host genomes, we construct phylogenetic hypotheses and draw inferences regarding the designation, host distribution, origin, and transmission of the Gammaretrovirus genus and associated class I ERVs. Our results uncover remarkable depths in retroviral sequence diversity, supported within a phylogenetic context. This finding suggests that current infectious exogenous retrovirus diversity may be underestimated, adding credence to the possibility that many additional exogenous retroviruses may remain to be discovered in vertebrate taxa. We demonstrate a history of frequent horizontal interorder transmissions from a rodent reservoir and suggest that rats may have acted as important overlooked facilitators of gammaretrovirus spread across diverse mammalian hosts. Together, these results demonstrate the promise of the methodology used here to analyze large ERV datasets and improve understanding of retroviral evolution and diversity for utilization in wider applications.

Pan-vertebrate comparative genomics unmasks retrovirus macroevolution

“…More specifically, host switching is incredibly common from the level of host superorder and below but is relatively infrequent between different vertebrate classes. These results confirm that retroviral interclass host switches are rare (11,18), implying that key constraints act to restrict retroviruses to focal host classes. Although host taxon sampling remains limited by currently sequenced vertebrate genomes, sufficient host taxonomic overlap demonstrates that observed patterns are distinct from sampling features.…”

“…Mouse ERVs are distributed throughout the phylogenetic tree (Fig. S1) and frequently are associated with rat and other rodent ERVs, as previously discussed for Gamma retroviruses (11). However, peaks in other ERV clades result largely from a small number of host-specific radiations of closely related ERVs, and the peaks largely disappear if these radiations are discounted from totals for each host taxon.…”

“…These analyses will help place other research on retroviral biology in context and highlight key priorities for more targeted studies of host-retrovirus interactions and host-pathogen biology more widely. (Tables S1 and S2), using the RetroTector software (10) as previously implemented (11). These genomes consist of assemblies and scaffolds available at the start of our analysis (hgdownload.…”

“…determined by cross-referencing the motif names provided in Hayward et al (11) with the RetroTector documentation (10). Retroviral regions subject to rapid rates of evolution and increased instances of recombination, specifically those of the env gene, were excluded to maintain a high phylogenetic signal-tonoise ratio.…”

“…Of ∼94,000 ERVs detected from the 65 genomes spanning vertebrate diversity, we use ∼36,000 high-quality ERVs, which are relatively complete sequences with little to moderate mutational degradation and that receive a score of at least 300 from the RetroTector software (Tables S1 and S2) (10). To combat high sequence divergence and permit homology estimation among sequences across retroviral groups, we use a strategy to analyze conserved regions sampled from multiple locations across the retroviral genome (10,11). In total, 28 conserved amino acid sequences were sampled, including six from gag [2 in matrix (MA); 2 in capsid (CA); 2 in nucleocapsid (NC), 2 from pro (protease, PR), and 20 from pol (11 in reverse transcriptase, RT, and 9 in integrase, IN)].…”