Intra‐strain biological and epidemiological characterization of plum pox virus

Abstract Plum pox virus (PPV) is one of the most important plant viruses causing serious economic losses. Thus far, strain typing based on the definition of 10 monophyletic strains with partially differentiable biological properties has been the sole approach used for epidemiological characterization of PPV. However, elucidating the genetic determinants underlying intra‐strain biological variation among populations or isolates remains a relevant but unexamined aspect of the epidemiology of the virus. In this study, based on complete nucleotide sequence information of 210 Japanese and 47 non‐Japanese isolates of the PPV‐Dideron (D) strain, we identified five positively selected sites in the PPV‐D genome. Among them, molecular studies showed that amino acid substitutions at position 2,635 in viral replicase correlate with viral titre and competitiveness at the systemic level, suggesting that amino acid position 2,635 is involved in aphid transmission efficiency and symptom severity. Estimation of ancestral genome sequences indicated that substitutions at amino acid position 2,635 were reversible and peculiar to one of two genetically distinct PPV‐D populations in Japan. The reversible amino acid evolution probably contributes to the dissemination of the virus population. This study provides the first genomic insight into the evolutionary epidemiology of PPV based on intra‐strain biological variation ascribed to positive selection.

Plum pox virus (PPV), an aphid-borne RNA virus belonging to the genus Potyvirus, is the causal agent of a serious viral disease of stone fruit trees (Prunus species), known as sharka (García et al., 2014). PPV is one of the most important and well-studied plant viruses because of its economic impact (Scholthof et al., 2011). Despite stringent quarantine regulations, it had spread throughout a large part of Europe by the 1980s, and has been detected (with a restricted distribution) in South and North America, Africa, and Asia within the past three decades (Rimbaud et al., 2015b). Through extensive research efforts to understand the virus diversity, at least 10 monophyletic strains of PPV (D, M, EA, C, Rec, W, T, CR, An, and CV) have been identified to date (García et al., 2014;Chirkov et al., 2018). Because some strains possess distinct biological properties, such as host preference, aphid transmissibility, disease symptomatology, and geographic distribution (James et al., 2013;Rimbaud et al., 2015b;Sihelská et al., 2017), strain typing has been the sole approach to the characterization of PPV isolates or populations of interest (EPPO, 2006;IPPC, 2018). Meanwhile, biological variation among isolates or populations in a single strain has been recognized as another important but unsolved aspect of PPV epidemiology (Candresse and Cambra, 2006;James et al., 2013;García et al., 2014;Rimbaud et al., 2015b). Several studies have revealed intra-strain variation in biological properties, such as host preference Dallot et al., 1998), aphid transmissibility (Deborre et al., 1995;Glasa et al., 2004;Schneider et al., 2011), and competitiveness (Glasa et al., 2010), among PPV isolates. Although these intra-strain biological variations are believed to be due to viral genetic diversity (James et al., 2013), existing studies have not elucidated the correlations among them based on genome sequences of PPV isolates, with the exception of certain non-aphid-transmissible isolates with mutations in the Pro-Thr-Leu (PTK) and Asp-Ala-Gly (DAG) motifs, which are responsible for aphid transmission of potyvirus (Atreya et al., 1991;Peng et al., 1998) in the helper component proteinase (HC-Pro) (Glasa et al., 2004) and coat protein (CP) Kamenova et al., 2002), respectively. Therefore, it has been suggested that biological information for a range of isolates from each PPV strain is unlikely to be available in the immediate future (Candresse and Cambra, 2006).

PPV-Dideron (D), the most widely distributed strain, is widespread
throughout Europe and has also been detected in South and North America, Africa, and Asia (EPPO, 2006;Chirkov et al., 2016;Oh et al., 2017), including Japan Maejima et al., 2011). In this study, to investigate intra-strain biological variation in PPV-D, we performed molecular epidemiological analysis based mainly on virus samples collected in Japan. Based on complete nucleotide sequence information, we detected five codons under positive selection. One codon was suggested to be responsible for intra-strain biological properties that could be involved in the evolutionary epidemiology of PPV.

| Sequence properties of Japanese PPV isolates
In Japan, PPV-D was first detected from Japanese apricot trees (Prunus mume) in Tokyo in the spring of 2009 .
A preliminary survey identified PPV-infected trees in several areas in and around Tokyo in the summer of 2009, and these PPV isolates were found to be monophyletic (Maejima et al., 2011). Subsequently, annual nationwide surveys involving the inspection of 2.7 million stone fruit trees identified more than 29,000 PPV-infected trees.
Large PPV outbreaks were detected not only in eastern Japan (near Tokyo), but also in western Japan (near Osaka), several hundred kilometres from Tokyo ( Figure 1) (MAFF, 2019).
To elucidate the diversity and phylogenetic relationships of PPV in Japan, we determined the genome sequences of 173 Japanese PPV isolates collected from most of the cities/towns where PPV was detected ( Figure 1 and Table S1) between 2009 and 2014. All isolates shared high sequence identity (>98%) with the 37 Japanese PPV-D isolates reported previously (Maejima et al., 2011), indicating that they also belong to the PPV-D strain. Most of the 173 viral genomes had the same length, with four exceptions: a single nucleotide deletion/insertion was observed in the 5′-or 3′-untranslated region (UTR) of four isolates. Amino acid motifs associated with aphid transmission (Lys-Ile-Thr-Cys [KITC] and PTK motifs in the HC-Pro region, and the DAG motif in the CP region) (Ng and Falk, 2006) were conserved among all of the Japanese isolates, with the exception of the Wa7 isolate, which contained asparagine (N) instead of aspartic acid (D) in the conserved DAG motif. Interestingly, while our previous study showed that the five nucleotide positions (nt 657, 1,598, 2,928, 5,246, and 7,145 of the reference genome sequence of the Ou1 isolate [AB545926]) were conserved only among Japanese isolates (Maejima et al., 2011), they were not necessarily conserved among the 173 newly sequenced Japanese isolates, casting doubt on our previous hypothesis that PPV-D found in Japan originates from a single introduction of infected plant material.

| Phylogenetic relationships among global populations of PPV-D
To elucidate the origin and epidemiological dynamics of PPV-D in Japan, we next conducted a phylogenetic analysis of all Japanese isolates along with non-Japanese PPV-D isolates. In total, 257 full genome sequences of the PPV-D isolates (210 Japanese and 47 non-Japanese) were used in this analysis (Tables S1 and S2). Because the presence of mosaic sequences due to genetic recombination would cause phylogenetic methods to produce misleading results (McGuire et al., 1997), we first confirmed that significant recombination events were not detected among the 257 PPV-D isolates using RDP4 software (Martin et al., 2015). AL11pl, a PPV-Ancestor Marcus (An) isolate, was employed as an outgroup instead of PPV-M for the same reason, that is, because the PPV-Marcus (M) strain was suggested as a recombinant strain between PPV-D and PPV-An (García et al., 2014). Although a recent report has suggested that recombination events occurred in the PPV-An isolate, it remains reasonable to use the isolate as an outgroup in our analysis because PPV-D was not involved in the recombination (Hajizadeh et al., 2019).
Because all phylogenetic trees constructed using the maximum-likelihood (ML), neighbour-joining (NJ), and minimum-evolution (ME) methods showed nearly identical topologies, only the ML tree is shown in Figure 2, with bootstrap values obtained from all three methods. In these trees, a paraphyletic group consisting of isolates from Eastern Europe, Kazakhstan, and Korea was observed. The isolates from Japan, North America, and Western Europe formed a monophyletic group with an isolate from Moldova (Eastern Europe) supported by high bootstrap values (100%) in all tree-building methods ( Figure 2). In the monophyletic group, while Western European isolates branched separately, the isolates from the United States and Canada formed their own monophyletic subgroups, as described in our previous report (Maejima et al., 2011). Intriguingly, Japanese isolates were not monophyletic, and formed two distinct subgroups supported by high bootstrap values in all of the trees ( Figure 2); this suggested that they have distinct origins. One subgroup, which included all 37 of the previously reported Japanese isolates (Maejima et al., 2011), consisted of 134 isolates mainly from the eastern region of Japan (in and around Tokyo). Conversely, the other subgroup consisted of 76 isolates mainly from western Japan (in and around Osaka), with the exception of three isolates from Adachi, Tokyo.
Therefore, we referred to the former subgroup as the "East-Japan" population, and the latter as the "West-Japan" population. Figure S1 shows the fully expanded trees and genetic divergence of the two Japanese populations, and detailed descriptions thereof.

| Positively selected sites in the PPV-D genome
The phylogenetic analysis based on large-scale genome sequencing clearly showed that there are two PPV-D populations, East-Japan and West-Japan, with closely related but independent origins in Japan ( Figure 2). To determine if there were distinct differences at the molecular level between these two populations, as well as between Japanese and non-Japanese PPV-D isolates, we evaluated the selective pressure on each codon position in the large open reading frame (ORF) and pretty interesting potyviridae ORF (PIPO) of the PPV-D genome using the HyPhy package  in MEGA (Tamura et al., 2013). While no significant selection bias was detected in the PIPO, five amino acid residues in the large ORF appeared to be subjected to significant positive selection (Table 1). This result was supported by positive selection analyses using fixed effects likelihood (FEL) , single-likelihood ancestor counting (SLAC) , and fast unbiased Bayesian approximation (FUBAR) (Murrell et al., 2013) available on the Datamonkey server (http://www.datam onkey.org/) (Weaver et al., 2018) (Table 1). Two of these positively selected residues (amino acid positions 2,303 and 2,635) were located in nuclear inclusion protein b (NIb), a potyvirus RNA-dependent RNA polymerase. The other three residues (amino acid positions 2,855, 2,872 and 2,873)

F I G U R E 1
The distribution map of PPV in Japan shown based on the 2014 nationwide survey. In total, 210 PPV-infected Prunus spp. were collected from 40 of 43 municipalities (plotted on the map) in 11 prefectures where PPV was detected between 2008 and 2014. They included 37 samples reported in our previous study (Maejima et al., 2011). Numbers in parentheses indicate the numbers of sequenced samples. Detailed information for each sample is shown in Table S1 were located in the N-terminal variable region of CP. The leucine at 2,303, cysteine at 2,635, and phenylalanine at 2,855 (aa2303L, aa2635C, and aa2855F, respectively) were conserved in the most recent common ancestor (MRCA) of the PPV-D isolates estimated using MEGA, as well as other known PPV strains, except for the PPV-CV strain with aa2635G (Table 1). Moreover, aa2303L was also conserved in other potyviruses, while aa2635C and aa2855F were not ( Figure S2). Moreover, in the West-Japan population, there were two steps involved in the amino acid change at position 2,635: first, a single nonsynonymous mutation (UGC to CGC transition) at codon 2,635, resulting in an amino acid substitution from cysteine to arginine at F I G U R E 2 Maximum-likelihood (ML) phylogenetic tree based on complete genome sequences of the PPV-D strain. A PPV-An isolate, AL11pl, was used to root the tree. For the sake of clarity, interior branches representing distinct clusters are collapsed into filled triangles. The detailed topology of the Japanese clusters is shown in Figure S1. Numbers at the nodes, or in the triangles, represent the percentage of bootstrap values obtained for 1,000 replicates in ML/neighbour-joining/minimum-evolution methods (values above 70% for ML are shown). Amino acid changes at the three positively selected residues in the N-terminal variable region of CP were estimated to be six L2855F, four F2855L, and one L2855I ( Figure S3b); five S2872P, three S2872L, and one S2872A ( Figure S3c); and four Q2873R, one P2873Q, and one Q2873K ( Figure S3d).

| Biological characterization of the positive selection sites in NIb from the Japanese PPV-D isolates
Because the positive selections at amino acid positions 2,303 and 2,635 were specific to the Japanese isolates among the five positively selected sites, we hypothesized that they are responsible for intrastrain biological variations among these isolates through effects on the function of the corresponding protein NIb. Given that NIb is a potyvirus RNA-dependent RNA polymerase (Hong and Hunt, 1996), At 3 wpi, 75% (9/12) of the plants inoculated with PPV-aa2635R and PPV-aa2635H retained mixed infection, while 17% (2/12) and 8% (1/12) were dominantly infected by PPV-aa2635R and PPV-aa2635H, respectively, in leaves infected systemically (Table 2) (Table S4).

| D ISCUSS I ON
A number of studies on the molecular epidemiology of PPV have aimed to identify viral genomic signatures explaining biological features of the virus (Bousalem et al., 1994;Dallot et al, 2001;Glasa et al., 2002Glasa et al., , 2004Glasa et al., , 2013Maejima et al., 2011;Schneider et al., 2011;Chirkov et al., 2018). However, identification of such genomic determinants at the single amino acid level has been unsuccessful owing to the high sequence divergence among strains (Salvador et al., 2008), even among populations or isolates within a single strain (except for certain obvious point mutations associated with aphid transmissibility) Kamenova et al., 2002;Glasa et al., 2004). In the current study,

| Origin and dispersal routes of PPV-D in Japan
This study revealed that all 210 Japanese isolates, which were collected throughout most of the areas where PPV was found from 2008 to 2014, belong to the PPV-D strain, but are divided into two genetically distinct populations (East-Japan and West-Japan) (Figure 2). Both   Figure 2). This suggests that PPV-D in Korea (Oh et al., 2017) has a different origin from the PPV-D populations in Japan, even though these two Asian countries are geographically close.
In the East-Japan population, the Tokyo isolates sampled in and around Ome city were distributed in all six subclades with the highest sequence diversity ( Figure S1a, c), suggesting that this area was the source of the plum pox disease outbreak in the East-Japan region.
Supporting this theory, there are trading records of scions or nursery stocks from Ome city to remote cities such as Odawara, Mito, Nagahama, Suita, and Nara, where isolates belonging to the East-Japan population were identified. In the West-Japan population, only isolates from Osaka prefecture were distributed in all three clades and the ungrouped branch (Figure S1b); these isolates had the greatest sequence diversity ( Figure S1c). However, the origins and dispersal routes of the West-Japan population remain unknown. Further sample collection and sequence analysis may answer this question.

| Evidence of intra-strain positive selection in PPV
Our study provides strong evidence of intra-strain positive selec- Therefore, it would be valuable to determine whether amino acid substitutions at these sites affect the antigenicity of the virus and compromise diagnosis and strain typing by the monoclonal antibodies.
The amino acid substitution from leucine to isoleucine at position 2,303 (UUA to AUA transversion) in NIb occurred independently several times in both the East-Japan and West-Japan populations (Table 1 and Figure S3a). This substitution was not detected for non-Japanese PPV-D isolates, nor in any other PPV strain (Table 1). This indicates that the positive selection at amino acid position 2,303 has important implications for the evolution of PPV in Japan. However, PPV-aa2303I did not outperform PPV-aa2303L in terms of systemic accumulation (Figure 4a), or in the competitive assay (Table S4) using the experimental host N. benthamiana.
Because the only discernible difference between the Japanese and non-Japanese PPV-D isolates is whether or not the major host plant was Japanese apricot (Tables S1 and S2), the L2303I amino acid substitution may be involved in the adaptation of PPV to Japanese apricot. Although the systemic accumulation of PPV with aa2303I was not greater than that of PPV with aa2303L in Japanese apricot, it should be noted that this result was based on rather limited numerical data (only five isolates with aa2303I compared to 86 isolates with aa2303L) (Figure 4c). Inoculation tests using a larger sample size may clarify the significance of the positive selection at amino acid position 2,303 for the evolution of the PPV-D strain.

| Biological and epidemiological implications of amino acid position 2,635
The positive selection at amino acid position 2,635 in NIb was peculiar to the West-Japan population (Table 1), where the amino acid TA B L E 2 Competitive assay of mixinoculated PPV aa2635 variants substitution events were divided into two stages. The first amino acid substitution (C2635R) occurred in the MRCA of the West-Japan population, and the second substitution (R2635C/H/S) occurred frequently in its progeny ( Figure 3). Interestingly, the viral accumulation level for the isolates with aa2635C was significantly higher than for aa2635Rtype isolates in Japanese apricot, the natural host (Figure 4d). This is consistent with the results of assays using the experimental host N. benthamiana (Figure 4b). Because amino acid position 2,635 is in the vicinity of a conserved Gly-Asp-Asp (GDD) motif that is essential for the activity of NIb (Li and Carrington, 1995), amino acid position 2,635 may contribute to the efficiency of virus replication.
It has been demonstrated that there is a significant positive correlation between viral accumulation level and the efficiency of insect acquisition/transmission (Simons, 1958;Pirone and Megahed, 1966;Banik and Zitter, 1990;Escriu et al., 2000;Matsukura et al., 2013;Li et al., 2014) including potyviruses (De Bokx et al., 1978;Romanow et al., 1986;Wosula et al., 2012). This has led us to postulate that aa2635C-type PPV is more readily transmitted by aphids than aa2635R-type PPV. Furthermore, in the competitive assay of the amino acid position 2,635 variants, PPV-aa2635C was more abundant in systemically infected leaves than PPV-aa2635R ( Table 2), suggesting that aa2635C-type PPV is more likely to be acquired and transmitted by aphid vectors than aa2635R-type PPV, even in coinfection conditions. aa2635C-type PPV is therefore considered to be more infectious than aa2635R-type PPV, given that its high titre and competitiveness in plants is probably associated with high aphid transmissibility. Another variant, PPV-aa2635S, was also significantly superior to PPV-aa2635R in terms of viral titre ( Figure 4b) and competitiveness (Table 2) in N. benthamiana, implying that the aa2635Stype PPV also has higher epidemic potential than aa2635R-type PPV.
Although there was no significant difference in accumulation levels between PPV-aa2635H and PPV-aa2635R in N. benthamiana, the isolates with aa2635H tended to accumulate at higher levels than those with aa2635R in the natural host, Japanese apricot ( Figure 4d).

| Possible contribution of reversible evolution to geographical dissemination
The significant positive selection at amino acid position 2,635 in the West-Japan population (Table S3), as well as experimental evidence from reverse genetics (Figure 4b and Table 2) and field samples ( Figure 4d), strongly suggest that aa2635R-type PPV is less adaptive than aa2635C-type PPV. This raises the following question: why did the MRCA of the West-Japan population acquire the reversible amino acid change from cysteine to arginine?
Non-adaptive amino acid changes followed by back mutation (reversion) at positively selected sites are known as "reversible evolution", and have been reported for epitopes of several human viruses (Leslie et al., 2004;Timm et al., 2004;Botosso et al., 2009). In such cases, the less adaptive amino acid change contributes to viral escape from host immune responses by altering viral antigenicity, and reversion to restore the original antigens occurs after infection of new hosts that have not yet developed immunity to them. This is because viruses with immune escape mutations are inherently less adaptive than those with the original antigens in the absence of immune pressure. Genetic drift with the so-called founder effect could also have contributed to the introduction of the less-adaptive MRCA of the West-Japan population into Japan. To the best of our knowledge, this study provides the first molecular evidence of reversible evolution of plant pathogens that may be related to a mechanism for overcoming plant quarantine regulations. A recent study reported that the incubation period (time from infection to symptom expression) and latent period (time from infection to onset of infectiousness) are almost synchronous for PPV (Rimbaud et al., 2015a). It will be interesting to see how the amino acid substitutions at position 2,635 affect these features. using both the immunochromatography assay kit (Nippon Gene) and

| Sample collection
the RT-LAMP assay kit (Nippon Gene). Table S1 shows the data for the Japanese samples (isolate name, geographical origin, host plant, year of sampling, and GenBank accession number). A PPV-D-infected sample from the UK (GBR1) was kindly provided by MAFF PPS. Other PPV-D-infected European samples, from Bulgaria, Serbia, and Slovenia (N1, N9, N28, S13, and SVN1), were described previously (Maejima et al., 2014). These non-Japanese samples were imported with permission from MAFF. All samples were stored at −80 °C until use.

| Sequencing of the viral genome
Genome sequences of PPV isolates were determined as described previously (Maejima et al., 2011). The 24 nt at the 5′ end and 26 nt at the 3′ end corresponding to the primer sequences used for RT-PCR amplification were not determined. Mixed infection was improbable because very few ambiguous sites (average 0.67 sites per isolate), and no discrepancy in the overlapping region of the amplified fragments, were identified among the 173 Japanese isolates and six non-Japanese isolates sequenced in this study.

| Phylogenetic analysis
The 210 genome sequences of the Japanese isolates were aligned with those of 47 non-Japanese PPV-D isolates and a PPV-An isolate, as an outgroup sequence (Table S2), using MEGA 6 (Tamura et al., 2013).
Recombination among sequences was analysed using the RDP (Martin and Rybicki, 2000), GENECONV (Padidam et al., 1999), BOOTSCAN (Salminen et al., 1995), MAXCHI (Smith, 1992), CHIMAERA (Posada and Crandall, 2001), SISCAN (Gibbs et al., 2000), and 3SEQ (Boni et al., 2007) methods in RDP4 (Martin et al., 2015), with the default settings and a p = .05 threshold. Events detected by fewer than two methods were ignored. The best nucleotide substitution model (GTR + G+I) was selected using MEGA 6. Phylogenetic analysis was performed with MEGA 6 using the ML method under the GTR model, with the complete deletion option for gap sites. The robustness of the tree topology was confirmed using NJ and ME algorithms with MEGA 6. The ancestral sequences of PPV-D isolates were estimated using the phylogenetic tree constructed using the ML method. The phylogenetic trees were edited using Adobe Illustrator CS4.

| Selection, transition/transversion ratio, and genetic divergence analysis
Natural selection was estimated for each codon of the large ORF (nt 147-9,569) and PIPO (nt 2,915-3,217) of 257 PPV-D genome sequences using the Hyphy package under the General Time Reversible model (Nei and Kumar, 2000) in MEGA 6, and using FEL, SLAC, and FUBAR methods on the Datamonkey server. The transition/transversion ratio (R) was estimated for the aligned PPV-D complete genome sequences using the ML method with MEGA 6. For the analysis of genetic divergence within each geographical group of Japanese populations, the number of base substitutions per site was calculated using a pairwise distance matrix in MEGA 6, and was plotted on a graph using Microsoft Excel.

ACK N OWLED G EM ENTS
We thank the staff of MAFF PPS and of each prefecture for sample collection in Japan. We also thank Hideo Hoshi (Tokyo Metropolitan

DATA AVA I L A B I L I T Y S TAT E M E N T
The data supporting these findings are available in DDBJ/EMBL/ GenBank at https ://www.ncbi.nlm.nih.gov/genba nk/ under the accession numbers LC374954-LC375132.

R E FE R E N C E S S U PP O RTI N G I N FO R M ATI O N
Additional supporting information may be found online in the Supporting Information section.  (Maejima et al., 2011). In this population, the isolates sampled in and around Ome city, where PPV was first found in Japan, belonged to every subclade and included all of the ungrouped isolates. Clade A was further divided into four subclades (A-I, A-II, A-III, and A-IV) with high bootstrap values and four ungrouped isolates (Ak1, Ok1, Ou6, and Ou16). Subclade A-I consisted of 46 isolates from 6 out of 11 prefectures where PPV was found: Kanagawa (Od), Ibaraki (Mi), Nara (Na), Osaka (Su), Shiga (Ng), and Tokyo (Ou). Interestingly, all of the isolates in subclade A-I were found in ornamental Japanese apricot gardens and all of the Tokyo isolates in the subclade were collected from an ornamental Japanese apricot garden (OuX) in Ome city.
As shown in our previous study (Maejima et al., 2011) Tree of the West-Japan population. The West-Japan population consisted of three clades (clades X, Y, and Z) with high bootstrap values, and an ungrouped isolate (Kc1). In this population, the isolates from Osaka exclusively constituted clade X, branched at the roots of clades Y and Z, and included all of the ungrouped isolates.
Clade X consisted entirely of isolates from Osaka (Hg, Os, Su, Tn, and Ya) and was further divided into two subclades (subclades X-I and X-II). Subclade X-I consisted of isolates from two adjacent cities, Osakasayama (Os) and Tondabayashi (Tn), located in the southeastern part of Osaka. Three isolates from Osakasayama city (Os1, Os2, and Os3) formed a monophyletic cluster. Subclade X-II consisted of two isolates (Su81 and Su83) from private gardens in Suita city located near the ornamental garden where Su1, 3, 14, 46, and 49 (subclade A-I) were collected, as well as single isolates from Yao city and Higashiosaka city (Ya1 and Hg1, respectively).
Clade Y was composed of three subclades (subclades Y-I, Y-II, and Y-III) and an ungrouped isolate (Sk3). Subclade Y-I consisted of 10 isolates from Nara (Na and Sa) forming a monophyletic cluster and four isolates from Aichi (Fs and In). The other isolates from Aichi (c), and aa2873 (d) of the PPV-D strain. The topology of phylogenetic tree is the same as Figure 2, and AL11pl was used to root the tree.
The colours of branches and isolate names indicate the amino acid residues of the ancestors and isolates, respectively. † L (UUG); † † S (UCU); † † † P (CCA)

TABLE S1
List of Japanese PPV-D isolates used in this study