The extraordinary evolutionary history of the reticuloendotheliosis viruses

Abstract

The reticuloendotheliosis viruses (REVs) comprise several closely related amphotropic retroviruses isolated from birds. These viruses exhibit several highly unusual characteristics that have not so far been adequately explained, including their extremely close relationship to mammalian retroviruses, and their presence as endogenous sequences within the genomes of certain large DNA viruses. We present evidence for an iatrogenic origin of REVs that accounts for these phenomena. Firstly, we identify endogenous retroviral fossils in mammalian genomes that share a unique recombinant structure with REVs-unequivocally demonstrating that REVs derive directly from mammalian retroviruses. Secondly, through sequencing of archived REV isolates, we confirm that contaminated Plasmodium lophurae stocks have been the source of multiple REV outbreaks in experimentally infected birds. Finally, we show that both phylogenetic and historical evidence support a scenario wherein REVs originated as mammalian retroviruses that were accidentally introduced into avian hosts in the late 1930s, during experimental studies of P. lophurae, and subsequently integrated into the fowlpox virus (FWPV) and gallid herpesvirus type 2 (GHV-2) genomes, generating recombinant DNA viruses that now circulate in wild birds and poultry. Our findings provide a novel perspective on the origin and evolution of REV, and indicate that horizontal gene transfer between virus families can expand the impact of iatrogenic transmission events.

Figure 1. Evolutionary relationships among the RT genes of exogenous Gammaretroviruses and related ERVs.

Shaded boxes indicate taxa that are known to occur as exogenous retroviruses. Brackets to the right indicate major lineages (note: an integrated taxonomy of exogenous and ERVs has yet to be established by the International Committee on Taxonomy of Viruses, and the groupings shown here are propositional). Associations of retrovirus groups and individual retroviral taxa with avian and mammalian hosts are indicated, as shown in the key. The phylogeny shown was constructed using NJ and a multiple sequence alignment spanning 140 amino acid residues in the reverse transcriptase protein (RT), and is midpoint rooted for display purposes. To obtain putative protein sequences for ERVs, frameshifting indels were inferred and removed, and the resulting nucleotide sequence was conceptually translated. Asterisks indicate clades with bootstrap support >90% in both NJ and maximum likelihood (ML) trees, based on 1,000 bootstrap replicates. The scale bar indicates evolutionary distance in substitutions per site. Table S2 provides details of all the ERVs and exogenous retrovirus taxa shown in the phylogeny.

Figure 2. Genomic and paleovirological characteristics of REV-related retroviruses.

The schematic in panel (a) shows the genome structure of REV and SNV, a near full-length REV insertion in the FWPV genome, and the mammalian ERVs Echidna-ERV and Galidia-ERV. Percentage sequence identity to SNV, at the amino acid level, is shown for the putative Gag, Pol, and Env polyproteins of Echidna-ERV and Galidia-ERV. Proviral coding regions that disclose homology to Gammaretroviruses are shown in green, whereas those that disclose homology to Betaretroviruses are shown in blue. ORFs flanking the REV insertion in FWPV are in yellow. Panel (b) summarizes the genomic data used to estimate the minimum age of REV-related ERV insertions in Malagasy carnivore genomes. A time-scaled Carnivora phylogeny (based on Nyakatura et al. [27]) is shown on the left, with Malagasy carnivores shaded. A corresponding schematic on the right shows the genomic locus at which an orthologous ERV insertion was identified in a subset of Malagasy carnivores. Boxes represent the env gene (blue) and 3′ LTR sequences (green = U3; dark grey = R; light grey = U5). The adjacent black line represents flanking genomic DNA, spanning 238 nucleotides, obtained from the striped mongoose (Mungotictis decemlineata) and ring-tailed mongoose (Galidia elegans) genomes in our study, and aligned to a homologous genomic region (lacking a proviral insertion) in the cat (Felis catus), dog (Canis familiaris), and ferret (Mustela furo) genomes. An orthologous ERV insertion was detected in M. decemlineata and G. elegans genomes, but not in the more distantly related Fossa (Cryptoprocta ferox), indicating that germline invasion occurred between 18 and 8 Ma. Genetic data indicate that all Malagasy carnivores are derived from a single founder population that colonized Madagascar ∼19 Ma ; thus, invasion of the Malagasy carnivore germline occurred in Madagascar. The nucleotide sequence alignment on which the schematic in panel (b) is based on is shown in Figure S1. Abbreviations: RV, retrovirus; Kb, Kilobases; ORF, open reading frame; PBS, primer binding site; Pro, proline; Thr, threonine; LTR, long terminal repeat; U3, unique three prime region; R, repeat region; U5, Unique five prime region; RT, reverse transcriptase; SU, surface protein; TM, transmembrane protein; M.dec, Mungotictis decemlineata; G.ele, Galidia elegans.

Figure 3. Contrasting phylogenetic relationships of pol and env genes found in REV-related retroviruses.

Panels (a) and (b) show ML phylogenies constructed from alignments of Gamma- and Betaretrovirus protein sequences. The phylogeny in panel (a) was constructed from an alignment spanning 157 residues of the RT protein encoded by pol, whereas the phylogeny in panel (b) was constructed from an alignment spanning 153 residues of the TM domain in the polypeptide encoded by env. Asterisks on internal nodes indicate ML bootstrap support >95%, (based on 1,000 bootstrap replicates). Asterisks beside taxa names indicate ERV families identified in this study. Open triangles indicate ERV lineages for which env genes were not identified. Scale bars indicate evolutionary distance in substitutions per site. Brackets to the right indicate genus designations, and viruses previously identified as Gamma- and Betaretrovirus (γ-β) recombinants. Table S2 provides details of all the ERVs and exogenous retrovirus taxa shown in the phylogeny. Abbreviations: RV, retrovirus; MoMLV, Moloney murine leukemia virus; FeLV, feline leukemia virus; GaLV, gibbon ape leukemia virus; KoRV, koala retrovirus; BAEV, baboon endogenous virus; SMRV, squirrel monkey retrovirus; TvERV, Trichosurus vulpecula endogenous retrovirus; JSRV, Jaagsiekte sheep retrovirus; SRV, simian retrovirus; MMTV, mouse mammary tumor virus.

Figure 4. Phylogenetic relationships of REV coding regions.

ML phylogenies constructed using (a) an alignment spanning residues 183–481 of the Pol polyprotein (DIAV coordinates) and containing REV and mammalian gammaretroviruses sequences and (b) a nucleotide alignment of the entire internal coding region of full-length avian isolates. The tree in panel (b) indicates the number of strain-specific, nonsynonymous mutations estimated to have occurred in the nucleocapsid (NC), capsid (CA), matrix (MA), protease (PR), RT, RNase-H (RH), integrase (IN), surface (SU), and TM genes of the exogenous isolate HA9901. Asterisks on internal nodes indicate ML bootstrap support >95%. All trees are midpoint rooted for display purposes. Scale bars indicate evolutionary distance in substitutions per site. Taxa labels include sequence accession numbers, and in panel (b) two-letter ISO country codes enclosed by brackets indicating the country of sampling. Further details of REV sequences included in these trees can be found in Table S4.

Figure 5. Phylogenetic relationships of REV LTR sequences.

ML phylogenies constructed using an alignment of REV LTR sequences. Asterisks on internal nodes indicate ML bootstrap support >95%. The phylogeny is midpoint rooted for display purposes. Scale bars indicate evolutionary distance in substitutions per site. Taxa labels include two-letter ISO country codes indicating the country of sampling.. Taxa labels include accession numbers and two-letter ISO country codes enclosed by brackets indicating country of sampling. Where appropriate, FWPV and GHV-2 strain designations are shown in bold. Further details of REV sequences used in the tree, and an alignment figure highlighting lineage-specific LTR indels, can be found in Table S4 and Figure S2, respectively.

Figure 6. Interclass transmission and the origin of REV.

A schematic showing the three possible scenarios via which the ancestor of REV could have crossed from birds (class Aves) into mammals (class Mammalia), assuming a maximum of one inter-class transmission (ICT) event in total. For each of the three scenarios shown, the phylogenetic relationships between REV isolates that would be expected to arise as a result are indicated (all phylogenies are rooted on the mammalian ancestor of avian REVs). A REV founder strain could conceivably have been transmitted from mammals to birds after first inserting into the genome of FWPV (panel a) or GHV-2 (panel b). However, only a scenario in which the SNV and DIAV lineage were established first (panel c)—as would be expected to occur if P. lophurae contamination enabled the iatrogenic emergence of virus—is compatible with the relationships observed in rooted phylogenetic trees (see Figure 4a). Abbreviations: FWPV, fowlpox virus; GHV-2, gallid herpesvirus 2.

Figure 7. A hypothesis of REV origin and evolution.

A schematic representation of REV evolutionary history is shown, summarizing our hypothesis regarding the origin and evolution of the three major avian REV lineages (SNV/DIAV, REV/FWPV-REV, and HA9901) from a mammalian retrovirus ancestor that originated in the Cenozoic Era. REV-associated events (i.e., outbreaks of REV-associated disease, isolation of new REV strains, or identification of REV-containing DNA virus strains) reported in the literature have been mapped onto this schematic, as indicated in the key. Numbers shown above key symbols refer to Table 2, where details of the associated publication or report can be found. The broken scale bar shows time in years A.D. to the right of the break and Ma to the left of the break. A shaded background region indicates the time window for invasion of FWPV genome following iatrogenic introduction into poultry (assuming that reports of REV sequences in FWPV vaccine strains lyophilized in 1949 are accurate). Abbreviations: REV, reticuloendotheliosis virus; SNV, spleen necrosis virus; DIAV, duck infectious anemia virus; FWPV, fowlpox virus; GHV-2, gallid herpesvirus 2; FWPV-REV, Fowlpox virus with REV insertion.