Evolutionary reconstructions of the transferrin receptor of Caniforms supports canine parvovirus being a re-emerged and not a novel pathogen in dogs

Abstract

Parvoviruses exploit transferrin receptor type-1 (TfR) for cellular entry in carnivores, and specific interactions are key to control of host range. We show that several key mutations acquired by TfR during the evolution of Caniforms (dogs and related species) modified the interactions with parvovirus capsids by reducing the level of binding. These data, along with signatures of positive selection in the TFRC gene, are consistent with an evolutionary arms race between the TfR of the Caniform clade and parvoviruses. As well as the modifications of amino acid sequence which modify binding, we found that a glycosylation site mutation in the TfR of dogs which provided resistance to the carnivore parvoviruses which were in circulation prior to about 1975 predates the speciation of coyotes and dogs. Because the closely-related black-backed jackal has a TfR similar to their common ancestor and lacks the glycosylation site, reconstructing this mutation into the jackal TfR shows the potency of that site in blocking binding and infection and explains the resistance of dogs until recent times. This alters our understanding of this well-known example of viral emergence by indicating that canine parvovirus emergence likely resulted from the re-adaptation of a parvovirus to the resistant receptor of a former host.

Figure 1. Analysis of dN/dS along each branch of the carnivore TFRC phylogeny.

A) dN/dS was calculated along each branch of the carnivore phylogeny. In this analysis, only the apical domain was analyzed. The number of estimated non-synonymous and synonymous DNA mutations that have occurred along each branch are shown in parentheses (N∶S) after the dN/dS value. B) A secondary analysis was performed with additional canid sequences, and the relevant clade is shown. The lineage leading to dog is highlighted with red branches. Below the phylogeny, key mutations predicted to have occurred along the branches leading to dog (branches 1–3) are shown.

Figure 2. The positions of the positively selected residues in the TfR structure.

A) A listing of the TfR residues in the apical domain under positive selection as revealed by PAML analysis, showing the positions in the different TfRs, and the alternative residues found. B) Residues found to be under positive selection mapped in red onto the crystal structure of the human TfR ectodomain, and those in the apical domain were labeled with the corresponding feline TfR coordinates . The 3 domains of the ectodomain are shown in green (apical domain), blue (protease-like domain), and yellow (helical domain). Strands of the apical domain β-sheet which influence virus binding are labeled. Some residues under positive selection are close to the host-range determinant (feline 383/canine 384) (shown in black) and the leucine at residue 221 (shown in orange) which can be mutated to block capsid attachment, and to discriminate between FPV and CPV in vitro , .

Figure 3. Determining the effects of varying residues in the feline TfR apical domain on parvovirus binding.

CPV (A) or FPV (B) and Tf were incubated with TRVb cells expressing receptors with different combinations of the residues 378, 379, and 380 (feline TfR numbering). The name of one host species which contains the combination of residues shown is also given. Ligands were incubated with the cells at 37°C (white) or 4°C (black). Fluorescence of the labeled capsid was divided by fluorescence of the bound Tf to account for differential receptor expression. The mean of this ratio among all receptor-expressing cells was evaluated for each of three trials. The mean and standard deviation of the three trials is shown. Brackets connect groups of receptors that were statistically different in pairwise comparisons by Tukey's HSD at α = 0.05; i.e. samples not covered by a bracket did not differ at the α = 0.05 level in any comparisons. C) Effects of variant residues in the feline TfR on FPV infection. Cells expressing exogenous TfR with different three-amino acid combinations at residues 378–380 were inoculated and the ratio of infected cells (expressing NS1) and those expressing TfR is shown. The expressed receptors were compared for infection percentage by fitting a generalized linear mixed model to the binomial data and considering replication as a random effect, and those differed at the p = 0.034 level. Only one pairwise comparison was statistically significant after controlling for multiple testing by the Tukey-Kramer method: 34% of cells expressing TfR containing QNR (as seen in the mink TfR) were infected by FPV, while 26% of cells containing RNS (as Pallas' cat) were infected.

Figure 4. Effect of Lys or Asn at position 384 in the black-backed jackal TfR on FPV and CPV binding and infection, compared to feline or canine TfRs under the same conditions.

A) Fluorescently labeled CPV or FPV and Tf were bound to cells expressing empty vector, feline TfR, canine TfR, wild-type jackal TfR, or Lys384Asn mutant jackal TfR at 37°C. The binding of FPV to the 384Asn black-backed jackal TfR exhibits the profile of CPV binding for canine TfR on multiple occasions , . B) Fluorescently labeled FPV or CPV capsids were incubated with cells expressing feline TfR, canine TfR, wild type or mutant black-backed jackal TfR at 37°C. The binding was compared to that of fluorescently labeled Tf. C) Cell expressing these receptors were inoculated with FPV or CPV, and then infection measured by staining for the parvoviral NS1 expression, and the expression of TfR determined by staining for the cytoplasmic tail of the receptor. Error bars = mean ±1SD of three replicates. The wild-type and mutant black-backed jackal receptors were compared by fitting a generalized linear mixed model to the binomial data and considering replication as a random effect; * indicates statistically significant difference in frequency of binding or infection.