Molecular characterization of echovirus 9 strains isolated from hand-foot-and-mouth disease in Kunming, Yunnan Province, China

Echovirus 9 (E9) belongs to the species Enterovirus B. So far, 12 whole genome sequences of E9 are available in GenBank. In this study, we determined the whole genomic sequences of five E9 strains isolated from the stools of patients with hand-foot-and-mouth disease in Kunming, Yunnan Province, China, in 2019. Their nucleotide and amino acid sequences shared 80.8–80.9% and 96.4–96.8% identity with the prototype Hill strain, respectively, and shared 99.3–99.9% and 99.1–99.8% mutual identity, respectively. Recombination analyses revealed that intertype recombination had occurred in the 2C and 3D regions of the five Yunnan E9 strains with coxsackieviruses B5 and B4, respectively. This study augmented the whole genome sequences of E9 in the GenBank database and extended the molecular characterization of this virus in China.

) was recovered from Vero cells, whereas none were from human rhabdomyosarcoma (RD) cells. Of these isolates, 115K3 and 115V3 were isolated from the same sample but from KMB17 and Vero cells rather than human rhabdomyosarcoma (RD) cells. They were isolated from three boys and one girl, ranging in age from 0.9 to 4.5 years. The whole VP1 sequences (918 nucleotides) of the five Yunnan strains showed the greatest identity (94.4%-94.8%) with E9 strain Echo9/FJPT176/CHN/2016 (MG922545), isolated from a patient with HFMD in China. The five isolates shared 78.3%-78.5% nucleotide identity and 85.7%-86.4% amino acid identity with the whole VP1 sequence of the E9 prototype Hill strain, which was isolated from a healthy child in Cincinnati in 1953 21 , and 80.8%-94.8% nucleotide and 88.1%-99.6% amino acid identity with other E9 strains. The whole-VP1 nucleotide and amino acid identities among the five Yunnan isolates were 99.6%-99.9% and 99.3%-100%, respectively.
The 53 whole VP1 sequences available in GenBank were included in an analysis of the five isolates collected in this study (Fig. 1). According to the approximate mean 15% cutoff divergence value used to genotype enterovirus A71 (EV-A71) 22 , the E9 strains were divided into eight clusters (A-H). The main epidemic strains belonged to the D, F, and H clusters. Of these, cluster D contained most E9 strains ( Whole-genome sequence analysis. The whole genome sequences of the five strains (115K3, 115V3, 123K3, 133K3, and 121K3) isolated in Yunnan Province in 2019 were determined. The genome sequences were 7445-7450 nucleotides in length, containing an ORF of 6612 nucleotides, which encoded a polyprotein of 2203 amino acids. The ORF sequence was flanked by a noncoding 5′-UTR of 733-738 nucleotides and a noncoding 3′-UTR of 103-106 nucleotides. The whole-genome nucleotide and amino acid identities of the five isolates were 99.3-99.9% and 99.1-99.8%, respectively. The total base compositions were 28.2-28.4% A, 24.6-24.7% G, 23.1-23.2% C, and 23.9-24.1% U. Because the mutual identities of the whole-genome nucleotide and deduced amino acid sequences of the five strains were > 99.1%, strain 115V3 was selected as the representative strain for further analysis.
Pairwise comparisons of the nucleotide and amino acid sequences of strain 115V3 and the E9 prototype Hill strain and other E9 strains are shown in Table 2. Strain 115V3 shares 79.0% and 80.5-88.7% nucleotide identities with the whole genomes of the E9 prototype Hill strain and the other E9 strains, respectively, and derived amino acid sequence identities of 94.4% and 93.5-97.6%, respectively. Phylogenetic analysis of P1, P2, and P3 regions. Phylogenetic trees were constructed for P1, P2, and P3 regions of all E9 strains and EV-B prototype strains available in GenBank, the five Yunnan isolates, and three EV-B strains (CV-B5/P727/2013/China, CVB4-B4M063015, and E11-1000/ISR/1999) (Fig. 2). In the P1 region, the Yunnan isolates grouped together with all E9 strains, including the E9 prototype Hill strain, thus further confirming the initial typing results. However, in the P2 and P3 coding regions, the E9 strains clustered with different EV-B strains and formed different clusters (Fig. 2). In the P2 region, the Yunnan isolates clustered with two E9 isolates (MSH/KM812/2010 and E-9/PMKA1322/THA/2011) and one CV-B5 strain, CV-B5/P727/2013/China (KP289438). In the P3 region, the Yunnan isolates clustered with one CVB4 strain, B4M063015 (MG845888). The prototype Hill strain formed a lineage with the E18 prototype strain Metcalf (AF317694), which is a recombinant with the Hill strain 23 . The results indicated that several potential recombination events have occurred between the E9 strains, including the Yunnan isolates, and other EV-B strains in the non-capsid coding regions, and that these E9 isolates may be the different recombinant strains.
With a similarity plot and a bootscanning analysis, several recombination events were confirmed in the genomic sequence of strain 115V3 (Fig. 3). In the P1 region, strain 115V3 showed greatest identity (> 90%) with the E9 strain MSH/KM812/2010. However, in the 2C-3A regions and the 3B-3D region, strain 115V3 shared greatest homologies (> 92%) with CVB5 strain P727/2013 and CVB4 strain B4M063015, respectively. Furthermore, RDP4 analysis revealed that the 115V3 isolate underwent a recombination event involving the strain CVB4/B4M063015/MG845888. The recombination event started at approximately 5300 nt and ended at approximately 7350 nt, in the P3 region of the isolates' genomes ( Fig. 4).

Discussion
Since 2008, HFMD caused by EVs has become a serious infectious disease in mainland China, mainly occurring in children under 5 years of age. Many EVs, including echoviruses, cocirculated in both sporadic and epidemic cases of HFMD [14][15][16][17][18] . Since the introduction of an inactivated EV-A71 vaccine, the prevalent etiological agents of HFMD have changed, and CV-A6 has become the main pathogen in mainland China 24,25 . EVs, and especially echoviruses, are the major agents of aseptic meningitis 3-5 , and EV-B, including echoviruses, can also cause  www.nature.com/scientificreports/ HFMD [18][19][20] . The serotypes of the EV pathogens causing HFMD are the same as those causing aseptic meningitis. Therefore, the persistent surveillance of the pathogens responsible for HFMD and aseptic meningitis is very important, and may allow the main serotypes of EVs associated with outbreaks to be predicted. Because of their error-prone replication, EVs form highly polymorphic populations within their hosts. Since the prototype Hill strain was isolated in 1953 21 and the Barty strain was isolated from a child with aseptic meningitis in 1957 26 , E9 isolates have evolved into eight clusters, thus indicating that E9 is genetically diverse. On the basis of analysis of the entire VP1 gene, the Chinese E9 strains were divided into D1 (2008-2010) and D3 (2010-2019) clusters, and the five Yunnan isolates in the study formed a single cluster. Furthermore, the average VP1 nucleotide and amino acid sequence divergence was 8.3% (5.3-11.2%) and 2.75% (0.3-5.2%) between the five Yunnan isolates and all whole VP1 gene sequences of other Chinese strains available in GenBank, respectively. In particular, the only whole genome of the Chinese E9 strain available in GenBank, MSH/KM812/2010, which was isolated from a patient with HFMD in 2010 in Yunnan Province, had the nucleotide and amino acid sequence divergence of 12.65% (12.5-12.8%) and 3.1% (2.4-3.8%), respectively. This finding indicated that the Chinese strains have evolved. Therefore, we speculate that the five Yunnan strains have adapted to a changing environment, in response to selection pressures.
The VP1 capsid protein of the EVs is located together with many immunodominant-serotype-specific epitopes in the exposed B-C loop. Although VP1 is the most variable of the capsid proteins, the N-terminus of VP1, which contains the B-C loop, is highly conserved within individual enteroviral serotypes 27 . The B-C loop is located inside the viral particle and is negligibly influenced by the immune pressure exerted by the host. The sequence of the E9 B-C loop, GDPESTDRFDA (amino acids 83-93 of the VP1 protein) 28  This indicates that the B-C loop is highly conserved in the Chinese strains. Moreover, amino acid A81, which is exposed at the surface near the B-C loop in the VP1 protein, is shared by most E9 strains, including the five Yunnan strains. A previous study reported that an E9 strain carrying the T81A substitution in VP1 had lytic potential toward pancreatic cells 28 . However, the non-lytic E9 strains Hill and Barty, including the five Yunnan strains in the study (dada not shown), contain alanine at this site. Thus, additional genetic substitutions in the viral genome may be very important in the pathogenicity toward pancreatic islets, and further research is required 29 .
The RGD motif at the C-terminus of the VP1 protein is very important to the pathogenicity of E9 30 . The interaction between the virus and the cell occurs via the contact between the RGD motif and the host cell receptor 29 . The Barty strain, which is highly virulent in newborn mice, and other E9 strains, including three of the Yunnan strains, contain this motif, whereas the Hill strain, which is nonpathogenic to newborn mice, does not. This suggests that these three Yunnan strains would display the same pathogenicity as the Barty strain. www.nature.com/scientificreports/ However, the RGD → GGD substitution is present in strains 115V3 and 133K3, although the five Yunnan strains were all isolated from patients with HFMD, without aseptic meningitis. Therefore, the RGD → GGD substitution requires further study. For EV, VP1 is the most immunogenic protein and is involved in its host cell with receptor-mediated entry 31 . However, one negative selection site (VP1 286 H → R) was found in the study. We speculate that although the amino acids are basic, this substitution may affect viral function, although this possibility requires further study. Recombination is a major mechanism of enteroviral evolution, particularly that of EV-B 32,33 . The E9 prototype Hill strain is a recombinant between the E9 strain Barty and the E18 prototype strain Metcalf 23 . Therefore, we used BLAST online to screen the strains sharing the highest identity with the E9 strain 115V3 in different regions of the E9 genome. We consequently identified several putative recombination events in different coding regions between strain 115V3 and the CVB5 strain CV-B5-P727/2013/China, the CVB4 strain B4M063015, and the E11 strain 1000/ISR/1999. On the basis of the very high (approximately 90%) sequence identities, we inferred that these viruses are recombination partners. Among them, CBV5 strain CV-B5-P727/2013/China was isolated from patients with HFMD 34 , the CVB4 strain B4M063015 was isolated from raw sewage 35 , and the E11 strain 1000/ISR/1999 was isolated from a chronically infected immunodeficient patient. Es play major roles in natural recombination events among the coxsackieviruses B (CVBs) 36 . CVB infections are associated with HFMD, aseptic meningitis, acute myocarditis, and fatal neonatal infections [37][38][39] . CVB5 is one of the five most common EVs in the USA 40 , and it has the highest prevalence in Germany and Spain 3,41 . CVB5 was also involved in an outbreak of neurological HFMD in China 42 . The numbers of cases of HFMD caused by CVB4 and CVB5 have been reported to be increasing in China 36 . These viruses also frequently cocirculate in patients with HFMD www.nature.com/scientificreports/ in China 43 . Therefore, their epidemiological characteristics may offer these viruses sufficient opportunities for mixing and recombination.
In conclusion, the E9 strains were highly genetically diverse, and intertypic recombination events have occurred in the genomic regions encoding nonstructural proteins. This serotype is globally widespread, and frequently cocirculates with other EVBs. Recombination and mutation drive the evolution of E9. Although the five Yunnan E9 strains analyzed in this study were isolated from HFMD patients without aseptic meningitis, other possible pathogenicities, including aseptic meningitis, cannot be ruled out, and the pathogenicity of these strains warrants further study. Systematic epidemiological surveillance is required to assess the links between E9 and its associated diseases. The serotype was identified by comparison of the nucleotide sequence with known sequences using BLAST 45 . Two long-distance PCR amplifications were performed with a PrimeScript™ One Step RT-PCR Kit Ver.2 (TaKaRa, Dalian, China) and the following primer pairs: E201F (TTA AAA CAG CCT GTG GGT TG) and E93R (TCC ACA TCA AAG CGC AAG TA) for 5´ end amplification, and E93F (AGG CAT GTG AAA AAT TAC CA) and E98R (ACC GAA TGC GGA GAA TTT AC) for 3' end amplification. The primers used for sequencing of the whole-length genome were designed by a primer-walking" strategy 46 . All primers used in the study are listed in Table 2. The PCR products were purified with a QIAquick PCR purification kit (Qiagen, Germany), and sequenced in both directions at least twice with an ABI 3130 Genetic Analyzer (Applied Biosystems, USA).

Materials and methods
Selection pressure analysis of the E9 VP1 gene. The selection pressure of E9 VP1 was predicted with the Datamonkey online Application 46 (http:// www. datam onkey. org) and calculated with the following four   Table 3.
Nucleotide sequence accession numbers. The whole genomes of the five E9 strains isolated in this study have been submitted to GenBank under accession numbers MZ488277-MZ488281.