Pan-vertebrate comparative genomics unmasks retrovirus macroevolution

Abstract

Although extensive research has demonstrated host-retrovirus microevolutionary dynamics, it has been difficult to gain a deeper understanding of the macroevolutionary patterns of host-retrovirus interactions. Here we use recent technological advances to infer broad patterns in retroviral diversity, evolution, and host-virus relationships by using a large-scale phylogenomic approach using endogenous retroviruses (ERVs). Retroviruses insert a proviral DNA copy into the host cell genome to produce new viruses. ERVs are provirus insertions in germline cells that are inherited down the host lineage and consequently present a record of past host-viral associations. By mining ERVs from 65 host genomes sampled across vertebrate diversity, we uncover a great diversity of ERVs, indicating that retroviral sequences are much more prevalent and widespread across vertebrates than previously appreciated. The majority of ERV clades that we recover do not contain known retroviruses, implying either that retroviral lineages are highly transient over evolutionary time or that a considerable number of retroviruses remain to be identified. By characterizing the distribution of ERVs, we show that no major vertebrate lineage has escaped retroviral activity and that retroviruses are extreme host generalists, having an unprecedented ability for rampant host switching among distantly related vertebrates. In addition, we examine whether the distribution of ERVs can be explained by host factors predicted to influence viral transmission and find that internal fertilization has a pronounced effect on retroviral colonization of host genomes. By capturing the mode and pattern of retroviral evolution and contrasting ERV diversity with known retroviral diversity, our study provides a cohesive framework to understand host-virus coevolution better.

Keywords: endogenous retrovirus; evolution; phylogenetics; retrovirus; transmission.

Fig. 1.

Retroviral evolution, host switching, and composition. (A) Retroviral tree schematic derived from reference retroviruses and ERVs sampled from 65 vertebrate host genomes (Tables S1 and S2). Reference retroviruses for the seven genera of Retroviridae (1), to which we apply the retrovirus “-like” ERV nomenclature (ref. and Fig. S1), include, from the top, Gamma: murine leukemia virus (MLV), Epsilon: walleye dermal sarcoma virus (WDSV), Beta: mouse mammary tumor virus (MMTV), Alpha: avian leukosis virus (ALV), Lenti: HIV 1 (HIV1), Delta: human T-lymphotrophic virus 1 (HTLV1), and Spuma: simian foamy virus (SFV). The tree was rooted using C. elegans retrotransposon Cer1 (GenBank accession no. U15406) and additional gypsy/Ty3 sequences identified from the 65 analyzed vertebrate genomes. (B) Host switching estimated from the full retroviral phylogeny (Fig. S1). Bar graphs, numbers, and dashed lines indicate the frequency of switches and trends along the retrovirus phylogeny, with reference to switches between retroviral lineages and host classes, superorders, orders, and families, respectively, as shown in the key. (C) Abundance of classified ERVs in major retroviral lineages determined by phylogenetic analysis.

Fig. 2.

Distribution of major retroviral lineages present as ERVs within individual host genomes. Bar graphs indicate proportions of ERV groups (0–100%) within host genomes (Tables S1 and S2). The trend line in the upper panel indicates the number of classified ERVs for each genome along the host phylogeny.