Ancient papillomavirus-host co-speciation in Felidae

Abstract

Background: Estimating evolutionary rates for slowly evolving viruses such as papillomaviruses (PVs) is not possible using fossil calibrations directly or sequences sampled over a time-scale of decades. An ability to correlate their divergence with a host species, however, can provide a means to estimate evolutionary rates for these viruses accurately. To determine whether such an approach is feasible, we sequenced complete feline PV genomes, previously available only for the domestic cat (Felis domesticus, FdPV1), from four additional, globally distributed feline species: Lynx rufus PV type 1, Puma concolor PV type 1, Panthera leo persica PV type 1, and Uncia uncia PV type 1.

Results: The feline PVs all belong to the Lambdapapillomavirus genus, and contain an unusual second noncoding region between the early and late protein region, which is only present in members of this genus. Our maximum likelihood and Bayesian phylogenetic analyses demonstrate that the evolutionary relationships between feline PVs perfectly mirror those of their feline hosts, despite a complex and dynamic phylogeographic history. By applying host species divergence times, we provide the first precise estimates for the rate of evolution for each PV gene, with an overall evolutionary rate of 1.95 x 10(-8) (95% confidence interval 1.32 x 10(-8) to 2.47 x 10(-8)) nucleotide substitutions per site per year for the viral coding genome.

Conclusion: Our work provides evidence for long-term virus-host co-speciation of feline PVs, indicating that viral diversity in slowly evolving viruses can be used to investigate host species evolution. These findings, however, should not be extrapolated to other viral lineages without prior confirmation of virus-host co-divergence.

Figure 1

URR alignment of the λ-PVs. Nucleotide sequence alignment of the upstream regulatory region (URR) or noncoding region (NCR1), from the stop codon of L1 to the start codon of E6, including the feline papillomaviruses (PVs) Felis domesticus PV type 1 (FdPV1), Lynx rufus PV type 1 (LrPV1), Puma concolor PV type 1 (PcPV1), Panthera leo persica PV type 1 (PlpPV1), and Uncia uncia PV type 1 (UuPV1), and the nonfeline carnivore PVs canine oral PV (COPV) and Procyon lotor PV type 1 (PlPV1), all of which belong to the genus λ-PV. Nucleotides are shaded according to the degree of conservation (black = 100% conserved; dark gray = 80% to 99% conserved; light gray = 60% to 79% conserved; no shade = <60% conserved). The conserved and putative E2 binding sites (E2BSs) and the E1 binding site (E1BS) are indicated. In LrPV-1, an E1BS is located at nucleotides 8,151-8,167, flanked by two conserved E2BSs at nucleotides 8,068-8,079 and 8,215-8,226. Two additional modified putative E2BSs were identified at nucleotides 8,117-8,128 (ACC-N6-GTT) and 8,195-8,206 (GCC-N6-GGT). The E1BS in PcPV-1 is located at nucleotides 8,226-8,242, surrounded by conserved E2BSs at nucleotides 8,143-8,154 and 8,289-8,300, and putative modified E2BSs at nucleotides 8,193-8,204 (ACC-N6-GTT) and 8,269-8,280 (CAC-N6-GGT). In PlpPV-1 we identified an E1BS at nucleotide 8,091-4, two conserved E2BSs at nucleotides 8,057-8,068 and 52-63, and modified putative E2BS motifs at nucleotides 8,007-8,018 (ACC-N6-GTT; a position where all other sequences in the alignment exhibited a conserved E2BS) and at nucleotides 32-43 (GCC-N6-GGT). The E1BS in UuPV-1 is located at nucleotides 7,983-7,999, with three conserved E2BSs at nucleotides 7,899-7,910, 7,949-7,960 and 8,047-8,058, and a putative modified E2BS (GCC-N6-GGT) at nucleotides 8,027-8,038. The TATA box of the E6 promotor in LrPV-1, PcPV-1, PlpPV-1, FdPV-1, COPV, and PlPV-1, and a possible degenerate TATA box (TTTAA) in UuPV-1 is indicated.

Figure 2

Phylogenetic relations among felid species and their papillomaviruses. (a) Co-evolutionary relationships revealed by the feline multigene (left) and papillomavirus (PV) complete coding genome (right) phylogenetic trees. Identical tree topologies were reconstructed by using both maximum likelihood (ML) and Bayesian methods for all data sequences from Johnson and coworkers [12] and for the amino acid translation of the concatenated PV genes. Bayesian posterior support values (bold) and ML bipartition values for relationships are shown on the branches. The scale bars indicate the genetic distance (nucleotide substitutions per site). Scientific names and branches in the host phylogenetic tree are color coded to depict recent and historic zoogeographic regions shown on the map (Oriental, Palearctic, Ethiopian, Neotropical, and Nearctic), as inferred by Johnson and coworkers [12] from current distributions, fossil records, and phylogenetic analysis. Branches in black reflect either less certain historical interpretations or geographic distributions beyond one. For the viral phylogenetic tree, the same color code is used for the abbreviated virus name, depicting the geographical region from which the respective virus was isolated. Outgroups are Prionodon linsang (Pli) and Cryptoprocta ferox (Cfe) in the host tree, and Procyon lotor PV type 1 (PlPV1) and canine oral PV (COPV) in the virus tree. (b) Branch length correlation (nucleotide substitutions per site) for the feline and PV phylogenetic trees. Outgroup lineages were excluded from this analysis. FdPV1, Felis domesticus PV type 1; LrPV1, Lynx rufus PV type 1; PcPV1, Puma concolor PV type 1; PlpPV1, Panthera leo persica PV type 1; UuPV1, Uncia uncia PV type 1.

Figure 3

Feline papillomavirus evolutionary rates. Schematic representation of the circular feline papillomavirus (PV) genome with mean evolutionary rates (in nucleotide substitutions per site per year) and confidence intervals for the separate coding genes and the upstream regulatory region (URR). The URR and the second noncoding region (NCR2), which is unique for the PVs of the genus λ-PV, are shown in blue.