Discussion

This study shows that CPV-2b has emerged for the first time as the dominant antigenic CPV variant circulating in dogs with parvoviral enteritis in Australia. This finding is consistent with an increasing selective pressure for CPV-2b in Australia that was identified by analysis of time-stamped CPV isolates in a previous study in which samples were collected from all Australian States and territories (Clark et al., 2017). Since our samples were largely collected in three States (NSW, Victoria and Western Australia) we compared strain prevalence in these States between the two studies and identified that CPV-2b predominance was not the result of sampling bias; between 2008 and 2016 CPV-2b comprised 44% of strains from NSW, Victoria and Western Australia (Clark et al., 2017), whereas between 2015 and 2019 CPV-2b comprised 57%.
It is surprising that CPV-2c was not identified among the VP2 sequences in this study. CPV-2c was identified in faecal samples from three dogs hospitalised with acute gastroenteritis in 2015 in Australia (Woolford, Crocker, Bobrowski, Baker, & Hemmatzadeh, 2017). However, testing of faecal samples obtained 1-2 days earlier from the same dogs in the laboratories of two of the authors did not confirm this and the viruses were characterized as CPV-2a or CPV-2b (data not shown). A possible explanation for this discrepancy is that the dogs were co-infected with multiple CPV variants, with a change in strain predominance over time. Alternatively, the PCR and sequencing methodologies used in the three laboratories may have preferentially amplified different variants.
In the Asia Pacific region, the CPV-2c variant is not widespread although it has become the dominant strain in Taiwan (Lin et al., 2017) and Vietnam (Hoang et al., 2019). By contrast, in the most recent molecular survey of 70 CPV isolates collected in New Zealand, one of Australia’s closest neighbouring countries, CPV-2c was not identified (Ohneiser, Hills, Cave, Passmore, & Dunowska, 2015).
Some of the CPV strains we identified had VP2 sequences that were intermediate between CPV and FPV (Figure 3). Although CPV and FPV share >99% nucleotide identity they have specific host ranges, antigenic and haemagglutination properties, which are controlled by the capsid VP2 gene (Chang, Sgro, & Parrish, 1992; Shackelton et al., 2005; Truyen, Agbandje, & Parrish, 1994). Subsequent to adaptations of an FPV-like ancestral virus, CPV emerged as a new virus in dogs in the late 1970s, although a specific evolutionary pathway has not been identified (Allison et al., 2016). On phylogenetic analysis, most of the viruses in Clade 2 in this study contained two FPV-defining residues, 564-Asn and 568-Ala (Figure 1 and Table 2). CPV differs from FPV at 7 VP2 residues and substitutions at three of these (564 (Asn to Ser), 568 (Ala to Gly) and 80 (Lys to Arg)) are associated with loss of replication ability of CPV-2 in cats, as determined using viruses recombinant between FPV and CPV-2 (U. Truyen et al., 1994). The 300 VP2 loop region (residues 299 to 301), structurally proximal to residues 80, 564 and 568, is a critical determinant of host range due to its interaction with the transferrin type I receptor (TfR) to mediate infection (Allison et al., 2016). Mutations in residues in close proximity to the 300 VP2 loop region can alter the efficiency of TfR binding and infection. One unique CPV-2b strain in our study, in Clade 2, contained another FPV-defining residue, 323-Asp (Table 2). VP2 aa residues 93 and 323 are critical in controlling CPV host range and a CPV-specific antigenic site on the capsid (Chang et al., 1992; Hueffer et al., 2003).
We also identified CPV-2a-like viruses that contained substitutions typical of CPV-2 in Clade 2, including four with 305-Asp and two with 87-Met (Figure 3). CPV-2a-like viruses isolated from dogs differ from CPV-2, at VP2 residues 87, 101, 300, and 305. The CPV-2a-specific residues 87-Leu and 101-Thr were likely acquired during evolution of the virus in raccoons, while the substitutions at 300 (Gly) and 305 (Tyr) were acquired when the virus transferred back to the canine host (Allison et al., 2012). Importantly, residues 87, 300, and 305 all lie within the binding footprint of the TfR, while residue 101 lies close to residue 87, just below the capsid surface (Hafenstein et al., 2007). The evolutionary trajectory between CPV-2 and CPV-2a-like viruses was likely facilitated by passage through an alternative host (Stucker et al., 2012). Another interesting finding of this study is that several Australian CPV strains displayed signatures in the VP2 protein that are typical of Asian CPVs, including 5-Gly, 267-Tyr, 324-Ile (Mira et al., 2018; Vannamahaxay et al., 2017; J. Wang et al., 2016), suggesting the introduction of CPV strains from Asia or, at least, for CPV circulation between Asia and Australia.
Our finding of intermediates between FPV and CPV-2 is unusual since intermediates have rarely been identified in molecular surveillance studies in the field. One exception was the detection of 568-Ala in a CPV-2a strain from Italy (Battilani et al., 2019), which was not accompanied by 564-Asn as seen here. The lack of identification of intermediates in other studies is likely due to the limited number of epidemiological investigations carried out before the 1990s, which restricts information on the possible geographical and temporal variations of CPV worldwide. The reversion of some VP2 aa residues to those of FPV, as identified here, could reflect alternate cycles of replication in dogs and cats. While there are no endemic wildlifeFelidae in Australia, there are large populations of free-roaming feral cats (Felis catus ). Alternatively, these changes could represent residues retained from CPV-2 intermediates during adaption from an FPV-like ancestor to CPV-2 rather than reverse mutations. We consider this explanation less likely since Australia does not harbor endemic wild Felidae or Canidae hosts (apart from a small population of dingoes (Canis lupus dingo )), thought to be required in viral evolution (Allison et al., 2012). Also, the strains with these “reverse” mutations were characterised by 297-Ala and 555-Val, both recent acquisitions among CPV-2a-like viruses (Meers et al., 2007). Finally, since recombination has been shown to play a role in the evolution of CPV (Shackelton et al., 2005), these findings could be the result of multiple evolutionary mechanisms, confounding the actual evolutionary patterns. Site-directed mutagenesis of CPV-2 virus to CPV-2b and vice versa has shown that some mutations in a given background are not well tolerated and markedly decrease virus fitness (Stucker et al., 2012). Accordingly, the discovery of intermediates with a vast repertoire of combinations of key mutations between FPV, CPV-2 and CPV-2a-like variants could inform a better understanding of the trajectory of evolution of CPV.
We identified two CPV-2 sequences, original type, from dogs without available vaccination history. These are most likely vaccine strains that were not the cause of enteritis in these dogs, since CPV-2 is no longer circulating in the field but attenuated vaccines containing this strain are used commonly in Australia. Reversion to virulence of the vaccine strains is theoretically possible but in the absence of corroborating evidence considered unlikely. Also, three dogs for which vaccination history was available, were infected with a CPV-2b strain closely related to a vaccine strain (SAH strain, accession no. FJ222822; Figure 1, Clade 4). All three dogs were from different litters from a breeding facility and had been vaccinated with the same CPV-2b SAH vaccine strain. The VP2 sequences of the virus from these dogs also had 100% nucleotide identity with two clinical isolates collected from dogs in the US in 2003 (AY742951) and 2009 (JN867605), for which accompanying vaccination data were not available (Allison et al., 2012; Shackelton et al., 2005). Viruses from these dogs differ from the SAH vaccine strain at one residue, 570-Lys, which is 570-Glu in the vaccine strain. Reversion to virulence of the vaccine strain in these cases is possible, or alternatively the CPV-2b strain in these cases may have been a vaccine mutant, but not the cause of disease. Another possible explanation is that these were wild-type viruses with a similar VP2 sequence to the vaccine virus.
Primary immunization failures (3.8%) among adult dogs that had completed a primary puppy vaccination course were rare, supporting the results of a previous Australian study where the failure rate was 3.3% (Ling, Norris, Kelman, & Ward, 2012). Although immunization failures can occur for many reasons including vaccine factors (e.g. improper storage or expiry) and various animal factors (e.g. immune-suppression due to disease or poor nutrition) (Roth & Spickler, 2010), the most likely cause of immunization failure of young adult dogs that have completed their primary vaccination course and been re-vaccinated is a genetic inability to synthesise antibodies against CPV, resulting in “non-responders” (Day et al., 2016). Some reports suggest that dogs infected with CPV-2c that have been vaccinated with a CPV-2 containing product, may be at higher risk of immunization failure compared to those receiving a CPV-2b vaccine (Decaro et al., 2008). If this is the case, then the rate of CPV immunization failures identified in Australia may not be broadly representative since only a single report exists of CPV-2c in Australia (Woolford et al., 2017). Commercial vaccines containing CPV-2c are not available. However, whether CPV-2 vaccines confer less protection in vivo than those containing CPV-2b, is controversial. It has been shown in vitro that dogs vaccinated with CPV-2 develop significantly lower antibody titres against CPV-2a-like viruses compared to antibody titres against CPV-2 (Cavalli et al., 2008). Whether the duration of immunity to CPV-2c is shorter in animals vaccinated with CPV-2 has not been investigated.
Almost half of the dogs with available vaccination histories were incompletely vaccinated. These dogs were likely permissive to infection due to interference of vaccine virus by maternally derived antibodies (MDA), since disease onset occurred before the completion of the primary vaccination series or in dogs completing the series before the age of 16 weeks. MDA in dogs generally persist until 13 to 15 weeks of age, and interfering titres of MDA have long been recognised as the most common cause of CPV immunisation failure (Decaro, Buonavoglia, & Barrs, 2020; Pollock & Carmichael, 1982).
The large proportion of dogs with parvoviral disease that were incompletely vaccinated in this study reflects the importance of final vaccination of puppies at 16-weeks of age or older to reduce immunization failure rate. However, almost half of the incompletely vaccinated dogs developed disease before completion of the primary vaccination series. Taken together, these findings indicate that strategies to overcome titres of MDA capable of interfering with vaccine protection in puppies are as important as increasing vaccination coverage overall to reduce the incidence of canine parvovirosis. Such strategies include measurement of MDA titres in puppies using haemagglutination assays to determine the optimal time for vaccination. Another tool, which has been shown to be at least partially effective in an experimental setting is the use of intranasal or oral MLV CPV vaccines (Cavalli et al., 2020; Martella et al., 2005).
In conclusion, passive surveillance between 2015 and 2019 shows that CPV-2b has emerged, for the first time, as the dominant antigenic CPV variant circulating in dogs with parvoviral enteritis in Australia. Disease was diagnosed in unvaccinated or incompletely vaccinated dogs in near equal proportions and immunization failures were uncommon. CPV-2c strains were not identified in this study. However, analysis of translated VP2 sequences revealed a vast repertoire of aa mutations. Strains of CPV that were intermediate between CPV and FPV, displayed reverse mutations, or, more likely, residual mutations retained from CPV-2 during adaptation from an FPV-like viral ancestor. Similarly, intermediates between CPV-2a-like viruses and CPV-2 were also identified. These findings could be helpful to inform a better understanding of the evolution of CPV in dogs.