Genetic diversity and recombination of enterovirus G strains in Japanese pigs: High prevalence of strains carrying a papain-like cysteine protease sequence in the enterovirus G population

To study the genetic diversity of enterovirus G (EV-G) among Japanese pigs, metagenomics sequencing was performed on fecal samples from pigs with or without diarrhea, collected between 2014 and 2016. Fifty-nine EV-G sequences, which were >5,000 nucleotides long, were obtained. By complete VP1 sequence analysis, Japanese EV-G isolates were classified into G1 (17 strains), G2 (four strains), G3 (22 strains), G4 (two strains), G6 (two strains), G9 (six strains), G10 (five strains), and a new genotype (one strain). Remarkably, 16 G1 and one G2 strain identified in diarrheic (23.5%; four strains) or normal (76.5%; 13 strains) fecal samples possessed a papain-like cysteine protease (PL-CP) sequence, which was recently found in the USA and Belgium in the EV-G genome, at the 2C–3A junction site. This paper presents the first report of the high prevalence of viruses carrying PL-CP in the EV-G population. Furthermore, possible inter- and intragenotype recombination events were found among EV-G strains, including G1-PL-CP strains. Our findings may advance the understanding of the molecular epidemiology and genetic evolution of EV-Gs.


Identification and genomic characterization of EV-Gs with a PL-CP sequence
Complete or nearly complete aa sequences of the coding-sequence region (CDS) of 59 Japanese EV-G strains were aligned and compared. We found that 17 of the 59 strains contain extra 633 to 651 nt (211 to 217 aa) within the 2C-3A coding region. According to BLAST analysis, these sequences have sequence homology to the PL-CP sequence variants that were recently identified in EV-G strains from the USA and Belgium [24][25][26]. Each inserted sequence is located between the coding regions 2C and 3A as in the USA and Belgium strains. The insertion sequences were aligned and compared with those of the PL-CP of EV-G strains from the USA and Belgium and with the PL-VP sequences in the genome of nidoviruses including porcine and bovine toroviruses by phylogenetic analysis and pairwise sequence comparison (Fig 2 and S2 Table). The PL-CP sequence of Japanese EV-G1 and that of Japanese EV-G2 revealed !74.0% nt and !74.6% aa sequence identities to each other and to the USA and Belgium EV-Gs and clustered in one group but are distantly related to those of porcine and bovine toroviruses, showing lower sequence identities (57.0% to 64.6% in the nt sequence and 49.6% to 58.7% in the aa sequence). EV-G HgYa2-1 and porcine torovirus HgYa2-2 were identified on the same farm at the same time; however, the nt and aa sequence identity between the PL-CP sequences of those strains was 62.3% and 54.3%, respectively (S2 Table). Japanese EV-G strains carrying PL-CP were subdivided into G1-PL-CP lineage 1, G1-PL-CP lineage 2, and G2 in the VP1 phylogenetic tree (Fig 1); however, these groups were not clearly detectable in the PL-CP tree (Fig 2)

Phylogenetic analysis and similarity plot evaluation for the nearly full genome of EV-Gs
To further investigate the genomic relations among EV-G strains, phylogenetic trees based on nt sequences of three regions (VP4-VP3, VP1, P2, and P3) were constructed. The tree for VP4-VP3 was similar to that of VP1, but the P2 and P3 trees showed topologies different from each other and no clear-cut EV-G types could be defined (Fig 3A). EV-G1-PL-CP lineage 3 was found to be related to G1-PL-CP lineage 2 and 3 in the trees for VP4-VP3 and VP1, whereas G1-PL-CP lineage 3 was closely related to G3 and G9 strains in the P2 tree and to G3 and G1-PL-CP strains in the P3 tree. The G2-PL-CP strain HgYa2-1 showed high homology   Genetic diversity and recombination of enterovirus G strains in Japanese pigs  to G2 strains in the tree for VP4-VP3 and VP1, whereas HgYa2-1 showed high similarity with G1-PL-CP lineage 1 strains in regions P2 and P3 (Fig 3A). By SimPlot analysis, the crossover site was mapped to the 2A region. The G2-PL-CP strain HgYa2-1 revealed that the downstream region of the crossover site has high similarity to the G1-PL-CP strain MoI2-2-1 ( Fig  3C). To find a possible recombination breakpoint, a bootstrap scanning analysis was conducted. A possible recombination breakpoint was identified in the middle of the 2A region (Fig 3D).
Ishi-Ka2 branched independently in the trees for VP4-VP3 and VP1, whereas Ishi-Ka2 clustered with G3-lineage 1 and formed a cluster with G3 strains identified on the same farm (S1A Fig). SimPlot analysis suggested that Ishi-Ka2 has extremely high similarity to G3-lineage 1 strain Ishi-Ka7 in regions 2C and P3 (S1C Fig). G3-lineage 2 strains showed a topology similar to that of the VP1 tree in VP4-VP3; however, the G3-lineage 2 strains branched separately from the G3-lineage-1 strains in the P2 and P3 trees (S1A Fig) and were found to be closely related to each other throughout the genome (S1D Fig).

Discussion
Although we did not initially aim to determine EV-G prevalence among pigs in Japan in this study, contigs that were longer than 5,000 nt were found in 22.5% (50/222) of pigs and on 23.4% (18/77) of farms, suggesting that EV-Gs are widespread among Japanese pigs. Fortyfour strains out of 59 (74.6%) were detected in healthy pigs, indicating that EV-Gs seem not to be associated with diarrhea in pigs, in accordance with other reports [13][14]38]. Because the detection limit of the method was not tested, a true prevalence study is needed in the future.
EV-G genotyping is based on >25% divergence between VP1 nucleotide sequences [14,39]. In the present study, according to the criteria, seven genotypes (G1, G2, G4, G6, G9, G10, and G?) were found in the feces samples from Japanese pigs, and the predominant genotypes were G3 (37.3%; 22/59) and G1 (28.8%; 17/59; Table 1, Fig 1). There are few studies on the genotyping of EV-Gs in pigs, and limited information is available from DDBJ/EMBL/GenBank databases. G1 and G6 types are predominant genotypes in Vietnam, whereas G3, G2, and G4 types appear to be common genotypes in Spain (however, that study did not analyze complete VP1 sequence) [40]. To date, G1-G16 genotypes and at least three EV-Gs with an unassigned genotype, including Ishi-Ka2 in this study, have been reported [13-14, 25, 41]. Owing to the limited number of reports on a specific geographic area, probably not all genotypes of EV-Gs are known at present. Further studies are needed for a comprehensive understanding of the genetic diversity of EV-Gs in other geographic areas.
Picornaviruses show significant genetic variability driven by both mutations and recombination events [42][43]. Ishi-Ka2 manifested >25% VP1 nucleotide sequence divergence from other strains; therefore, Ishi-Ka2 can be considered a new serotype of EV-Gs. Nonetheless, Ishi-Ka2 shares high sequence homology with the G3-lineage 1 strain Ishi-Ka7, which was identified in a pig kept on the same farm, except for the P1 region. It is likely that Ishi-Ka2 emerged by possible recombination events; however, the putative recombination points could not be identified, and the origins of these recombination events are unclear because the recombination counterparts of these strains could not be found in the DDBJ/EMBL/GenBank databases or our dataset. G3-lineage 2 strains have sequence homology to G3-lineage 1 in the P1 region, but they are distantly related to G3-lineage 1 on the basis of regions P2 and P3. Because VP1 induces a host immune response, serological properties can be hypothesized based on sequence homology of the VP1 gene. On the other hand, these results suggest that full genome analysis may be needed in addition to the genotyping approaches based solely on the VP1 gene for precise EV-G classification.
RNA recombination events contribute to genetic diversity and may lead to changes in virulence, escape from host immunity, and adaptation to a new host [42][43][44][45][46][47][48][49][50]. EV-G strains carrying PL-CP in pigs with diarrhea have been reported in the USA and Belgium [24][25][26]. In these cases, EV-G-PL-CPs were detected solely or with low abundance of PEDV. Shang et al. constructed an infectious clone of the EV-G-PL-CP strain, 08/NC_USA/2015, and compared it with a PL-CP knockout recombinant virus. They found that the PL-CP knockout virus showed impaired growth and induced higher expression levels of innate-immunity genes, suggesting that EV-G-PL-CP strains acquire pathogenesis via a recombination event [25]. Four out of 17 Japanese EV-G-PL-CP strains were detected in diarrheic cases of pigs; however, 13 EV-G-PL-CP strains were isolated from healthy pigs. In all cases of detection of EV-G-PL-CP in Japan, EV-G-PL-CP strains were identified together with other enteric viruses, such as astrovirus, sapelovirus, posavirus, rotavirus, picobirnavirus, sapovirus, teschovirus, torovirus, PEDV, St-Valerien virus, or kobuvirus (Table 1). Mixed infection with EV-G-PL-CP and other enteric viruses may influence the pathogenicity of EV-G-PL-CP strains.
The sequences of PL-CP of Japanese EV-G PL-CP strains are distantly related to the sequences derived from ORF1 of toroviruses, even though they were simultaneously identified on the same farm, and they have homology to those of USA and Belgium strains (Fig 2), suggesting that a recombination event between an EV-G and torovirus occurred in the past. By recombination analyses, possible recombination events between EV-G-PL-CP strains were uncovered and a recombination breakpoint was identified in the middle of region 2A (Fig 3), in agreement with another report that describes a recombinant event between EV-G8 and EV-G9 [14], suggesting that this point may be a hotspot of recombination events of EV-G. Furthermore, VP1-2A junction is a known recombination hot-spot in human enteroviruses and this was discussed in many papers [51][52][53][54][55]. The present recombination profile in EV-G described here apparently mirrors that in human enteroviruses. EV-Gs that received PL-CP have been evolving independently and gaining genetic diversity via recombination events.

Conclusions
By a metagenomics approach, high genetic diversity of EV-Gs, including new genotypes and high prevalence of EV-Gs carrying PL-CP, was observed among EV-G isolates from the feces of Japanese pigs. EV-Gs comingle and pose a risk of coinfection in the current growing and high-density pig husbandry system of Japan. Coinfection of a single animal with multiple EV-Gs, including EV-G-PL-CP strains, may lead to recombination events, which may in turn promote genetic diversity of EV-Gs and EV-G-PL-CPs. These findings may improve our understanding of the molecular epidemiology and evolution of EV-Gs.