The role of evolutionary intermediates in the host adaptation of canine parvovirus

Abstract

The adaptation of viruses to new hosts is a poorly understood process likely involving a variety of viral structures and functions that allow efficient replication and spread. Canine parvovirus (CPV) emerged in the late 1970s as a host-range variant of a virus related to feline panleukopenia virus (FPV). Within a few years of its emergence in dogs, there was a worldwide replacement of the initial virus strain (CPV type 2) by a variant (CPV type 2a) characterized by four amino acid differences in the capsid protein. However, the evolutionary processes that underlie the acquisition of these four mutations, as well as their effects on viral fitness, both singly and in combination, are still uncertain. Using a comprehensive experimental analysis of multiple intermediate mutational combinations, we show that these four capsid mutations act in concert to alter antigenicity, cell receptor binding, and relative in vitro growth in feline cells. Hence, host adaptation involved complex interactions among both surface-exposed and buried capsid mutations that together altered cell infection and immune escape properties of the viruses. Notably, most intermediate viral genotypes containing different combinations of the four key amino acids possessed markedly lower fitness than the wild-type viruses.

Fig 1

Ligand-binding properties and relative fitness of wild-type (WT) and intermediate viruses. The left-most columns indicate the virus background and combination of four VP2 residue changes used for each intermediate virus, with white squares indicating CPV-2 residues (87Met, 101Ile, 300Ala, 305Asp) and black squares indicating CPV-2b residues (87Leu, 101Thr, 300Gly, 305Tyr). (A) Virus binding profiles for 8 MAbs as determined by HI assay. Strong (black), intermediate (gray), and weak (white) binding for each MAb was defined by its wild-type CPV-2 and CPV-2b binding titers, as wild-type virus specificity for these MAbs is well characterized (22). (B) Virus binding and uptake in feline and canine cells measured by flow cytometry. The median fluorescent intensity (MFI) relative to CPV-2 binding from three independent experiments was averaged, and the standard error of the mean is shown. (C) Relative fitness of intermediate viruses relative to wild-type CPV-2 (top half) or CPV-2b (bottom half) following 7.5 days of replication in feline cells. Average fold changes in intermediate relative to wild-type PHRs from three or more competition assays are shown. Asterisks indicate that the change in PHR is significantly different from zero or that there is no change over time, with P being <0.05 (*) or <0.005 (**).

Fig 2

(A) Sequence trace data for the variable nucleotide in the VP2 codon 87 showing an increase in CPV-2 (green peak for adenine) and a decrease in CPV-2b (red peak for thymine) over time at each of the three input ratios of virus (10:1, 1:1, and 1:10). Traces are displayed using Geneious software (5). (B) Graphical representation of the change in PHR of CPV-2b relative to CPV-2 over time at each of the input ratios shown in panel A. The decrease in PHR over time shows that the second virus (CPV-2) replicates to higher levels than the first (CPV-2b). pi, postinfection.

Fig 3

(A) Asymmetric unit of the CPV-2 capsid structure, showing the spatial relationships among the surface-exposed residues 87, 300, and 305 in red. Each VP2 peptide has a different color, and three key surface loops are outlined in black. (B) A closer, three-dimensional representation of the CPV-2 structure, showing the four residues of interest in red and indicating with yellow arrows all potential hydrogen bonds that may be altered when these residues are changed to the CPV-2b sequence.

Fig 4

Graphical representations of the relative fitness of CPV-2 intermediate viruses. (A) Adaptive landscape of the CPV-2 mutational intermediates tested here, showing the fitness effects of the four key VP2 residues that define the difference between the newer CPV-2a clade and the older CPV-2 clade. Each individual blue bar and corresponding black error bar represent PHR values following 7.5 days of infection in cultured feline cells (top half of Fig. 1C). Wild-type CPV-2 (left) is assigned a PHR value of zero; other PHRs are normalized to this value. On the right is the quadruple mutant in the CPV-2 background, representing a potential ancestral state of the CPV-2a clade. The x axis represents the number of mutations from wild-type CPV-2, and the various combinations of single, double, and triple mutations are distributed along the y axis in an arbitrary fashion. The numbers below each bar indicate which residues are changed from the CPV-2 sequence. An overlay colored to match PHRs provides a graphical representation of the landscape, showing fitness valleys and peaks. (B) Bubble plot showing the evolutionary pathways available to CPV-2 during its adaptation to dogs. The x and y axes are as described above. Peak height ratios were shifted onto a positive scale and multiplied by the same arbitrary scaling factor to allow the smallest and largest values to be visualized on the same plot as circles of various diameters. The largest circle for each number of mutations is shown in red. Two intermediate viruses in the CPV-2 background had either significantly reduced fitness (changes at VP2 87/300) or were nonviable (changes at VP2 87/300/305, marked by an X), effectively eliminating evolutionary pathways that use these combinations from plausible mutational trajectories. Lines indicate the remaining potential pathways, with the most parsimonious (i.e., shallowest) evolutionary pathway highlighted in red.