Waterfowl parvovirus infections of geese and ducks have been reported in China, France, Hungary, Germany, Israel, and the USA. GPV causes Derzsy’s disease (gosling plague), which is characterized by high mortality, watery diarrhea, lethargy, anorexia, prostration, and weight loss in young Muscovy ducklings and goslings. MDPV only affects young Muscovy ducklings (2–4 weeks old) and cause 3-W disease, which is characterized with ascites, enteritis, myocarditis, hepatitis, high mortality, and morbidity (Glávits et al., 2005).

Normally, Cherry Valley ducks and mule ducks are resistant to classical goose parvovirus infection (Chen et al., 2016). However, in recent years, a distinct GPV-related parvovirus was successively isolated in mule duck and Cherry Valley duck around the world. The causative agent could proliferate in duck embryos and duck embryo fibroblast cells, but cannot cause a cytopathic effect in primary goose embryo fibroblast cell. The infected ducks show typical symptoms of short beak and dwarfism, with high morbidity and low mortality rates; therefore, the disease was named as beak atrophy and dwarfism syndrome (BADS) (Palya et al., 2009; Li et al., 2016). Since its occurrence in France in 1971, BADS has spread to many European and Asian breeding areas in mule duck, Muscovy duck, and Cherry Valley ducks (Palya et al., 2009). Our surveys show that the recent outbreak of BADS in China has caused severe infection in mule duck and Cherry Valley duck flocks, resulting in immense economic losses (Chen et al., 2015, 2016). Although the GPV-related parvovirus has been isolated, the evolutionary processes of its initial infection and subsequent spread in the new host remain poorly understood.

Several studies have proven that entering and replicating in new host cells are two key factors for a pathogen to infect a novel host (Li et al., 2014; Longdon et al., 2014). Host shift of a pathogen could be monitored by surveillance of potential donor species, observation of specific mutations of a pathogen, and the pathogen’s basic reproductive number (R₀) (Longdon et al., 2015). Moreover, analyses of evolution rate and selection pressure are also necessary for monitoring host shift of a pathogen (Streicker et al., 2010; Mollentze et al., 2014).

According to the International Committee on Taxonomy of Viruses (ICTV) rules, waterfowl parvovirus is classified as a member of Parvoviridae family, Parvovirinae subfamily, and Dependoparvovirus genus. GPV and MDPV all belong to the Anseriform dependoparvovirus species. Parvoviruses, characterized with single-stranded DNA genomes, are widely spread from invertebrates to mammal and show a remarkable evolutionary capacity that is different from other DNA viruses (Lópezbueno et al., 2006). The waterfowl parvovirus genome contains two open reading frames (ORFs). The non-structural proteins (NS1 and NS2) are encoded by the left ORF, whereas the right ORF encodes three capsid proteins (VP1, VP2, and VP3). VP2 and VP3 are contained in the carboxyl-terminal portion of VP1 (Istvan et al., 2014). Capsid proteins are important determinants for parvovirus tropism, pathogenicity, and host range (Salganik et al., 2014; Tu et al., 2015). For example, residues on the threefold spike in the capsid proteins determine the host range differences between canine parvovirus (CPV) and feline parvovirus (FPV) (Hueffer and Parrish, 2003). During endosomal trafficking, the minute virus of mice (MVM) virions will expose the N-terminal of VP1 (Mani et al., 2006). Meanwhile, a recombinant capsid VP1 N-terminal with a unique region (VP1u) of human parvovirus B19 became readily recognizable by a specific monoclonal antibody (MAb), whereas the natural capsid was not recognized (Quattrocchi et al., 2012). Therefore, research on VP gene evolutionary processes provided us an opportunity to study the waterfowl parvovirus host range and pathogenicity.

Human, animals, and plants are under constant threat from the emerging viral infectious diseases. The key processes of emerging new viruses are infection and transmission in new hosts (Delaney et al., 2012). To understand these processes, we need to consider the biological processes occurring around the emerging disease. For example, the new host must be exposed to the pathogen, and then the pathogen must be able to infect the new host; finally, the pathogen will keep spreading in the new host population (Longdon et al., 2011).

Parvovirus infections need successful entries into the host cells and many kinds of glycolipids, glycans and glycoproteins might function as receptors. The choice of receptor would determine animal specificity and tissue tropism; and it might influence the endosomal trafficking in host cells (Harbison et al., 2008). Therefore, phylogenetically conserved cell receptors is most important for the virus to enter the cell (Woolhouse, 2002). Previous studies showed that the VP protein of GPV and MDPV has high sequence similarity (70.2 to 70.3%) with AAV-2, in which five amino acids in the VP protein (Arg-484, Arg-487, Lys-532, Arg-585, and Arg588) are directly involved for its receptor binding (Opie et al., 2003; Wu et al., 2006). Shien et al. (2008) suggested that changes in sequence near the receptor binding sites (Arg-487, Lys-571, and Lys/Arg-575) of VP1 might be responsible for attenuation of GPV by using AAV-2 as the model system. As parvoviruses are non-enveloped viruses, the indentations or projections in the capsid surface are responsible for receptors recognition (Smith and Helenius, 2004). In our research, the VP protein of NDPV had two important amino acid changes (Asn-489 and Asn-650) when compared with GPV. The changed amino acid at position 489 is located in the NDPV putative receptor binding site of VP protein. Although the amino acid at position 650 is not near the receptor binding site, the exposed amino acid may form projections or indentations that bind the receptors of duck cells. These changes might lead to the host jump from goose to duck. Similarly, CPV evolved from a virus that is closely related to FPV. The virus got the capability to infect dogs through the acquisition of mutations in capsid. The mutations changed the interactions between virus capsids and the transferrin receptor type 1 (TfR) on canine cells surface (Stucker et al., 2011). However, further investigations should be conducted to verify this hypothesis, such as directed mutagenesis and study of virus–host-cell interactions.

After crossing the species barrier, the pathogen must maintain itself successfully in the new host for its continuous spread, that is, the pathogen should have a reasonable R₀, which satisfies R₀ > 1 (May et al., 2001). If R₀ < 1, despite that the new host repeatedly acquires the pathogen, the spread of infection in the new host will be limited. Conversely, if R₀ > 1, an epidemic may occur (Woolhouse et al., 2005). Few studies about the growth of waterfowl parvovirus epidemic have been reported. Our demographic analysis showed that GPV clades 1 and 2 grew at a high rate (r = 1.717 year^-1), which might cause an epidemic since the time of their radiation. The BSP analysis predicted the effective population size of GPV which showed a constant value from 1966 to 2000, however, the values has been increasing during the recent 15 years. Furthermore a relationship has been found between the effective population size and epidemics of GPV. Since 2000, many studies about GPV outbreak in Sweden, Britain, and China have been reported (Jansson and Palya, 2007; Irvine et al., 2008; Chen et al., 2016). We estimated the R₀ of GPV clades 1 and 2 of 1.043 (assuming D = 0.3 month) and 2.001 (assuming D = 1 month). Similar R₀ values for the parvovirus B19 (between 1.1 and 1.8) and human bocavirus (between 1.08 and 1.59) were previously estimated by statistical methods (Zehender et al., 2009). These values are lower than the R₀ estimates of pandemic influenza virus but they are similar to the new influenza A virus that has caused outbreaks worldwide (Ferguson et al., 2005; Fraser et al., 2009). These results indicate that NDPV and GPV have high potential for widespread transmission.

B-cell-recognized immune-epitopes of NDPV may be helpful for predicting the antibody binding sites. Previous studies suggested that binding sites in VP domains of parvovirus are related to infections of host cells (Yu et al., 2012). For example, different amino acids in VP2 of CPV compared with amino acids of FPV and mink enteritis virus (MEV) were found (Langeveld et al., 1993). Of these amino acids, aa93–aa103 were found to be critical in immune-epitope recognition for a CPV specific and neutralizing MAb (Parrish et al., 1988). Residues Ala-300 and Thr-301 are critical amino acids for neutralizing MAb specific for both CPV and FPLV (Parrish and Carmichael, 1986). In the present study, many immune-epitopes were predicted in the VP proteins of waterfowl parvovirus. Among them, an immune-epitope of NDPV (aa515–aa528) was estimated as the binding site of VP3 domain, which may lead to the generation of a neutralizing antibody. This prediction provides important information for the development of BADS vaccine.

Most cross-species viral transfers lead to “dead-end” infections in their new hosts without subsequent transmission (Shackelton et al., 2005). The lack of subsequent transmission implies that the transfer of a virus variant that has the capability to replicate in the new host is most important in disease emergence. The association of species jump and subsequent transmission of NDPV with preexisting genetic variation or post-transmission adaptation remains unclear. Phylogenetic analysis showed that NDPV cases and recent wild attenuated GPV strains had close genetic relationship and NDPVs were diverged 20–15 years ago, indicating that NDPV may have been present in some localities for years before it was associated with BADS in China in 2015. Furthermore, our results indicated that the recent wild attenuated GPV was diverged from NDPV, which suggested that NDPV can infect goose with inconspicuous symptoms. This finding might be the reason for the wide malfunction of vaccine in goose industry in Southeastern China (Shao et al., 2015). Although the study demonstrated that most sites are subject to strong purifying selection for NDPV VP gene along with the fast evolutionary process, nine sites had dN/dS values of >1. Thus, these specific amino acids were under positively selected pressure. Especially, Tyr365Leu and Ala366Arg occurred in GPV→MDPV branches resulting in dN/dS value of infinity, indicative of positive selection in MDPV. In general, a strong antigenicity viral protein may undergo strong selection pressure, leading to many positively selected sites present in the antigenic viral protein (Nielsen, 2005). In our study, positively selected sites (Tyr365Leu, Ala366Arg) are located within the predicted immune-epitopes (aa360–aa374) in MDPV. This result suggested that the immune system of host defense mechanisms may act as a selective pressure to MDPV. Nevertheless, the association of species jump of NDPV to selection pressure requires further study.

SS and LY designed the study. WF and DX contributed analytic tools. WF, ZS, and TS analyzed data. SS and WF wrote the paper. JZ and KH revised the manuscript. All authors reviewed the results and approved the final version of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.