Antigenic characteristics and genomic analysis of novel EV-A90 enteroviruses isolated in Xinjiang, China

Enterovirus A90 (EV-A90) is a novel serotype of enterovirus A species that is rarely reported. Here, we isolated five enteroviruses from patients with acute flaccid paralysis in Hotan and Kashgar cities in Xinjiang, China that were identified as EV-A90 by molecular typing. The VP1 sequences of these Xinjiang EV-A90 strains showed 88.4–89% nucleotide sequence identity to the prototype EV-A90 strain; however, genome analysis indicated complex recombination events in P2 and P3 regions. Next, the seroprevalence of EV-A90 was examined in 49 serum specimens collected in Hotan and Kashgar, and 37.5% were EV-A90 antibody positive (>1:8), with a geometric mean titre (GMT) of 1:10.47. The low positive rate and GMT suggest a low-level EV-A90 epidemic in Xinjiang. Two of the five Xinjiang EV-A90 strains were temperature sensitive, and three were temperature resistant, and a comparative genomics analysis suggested that an amino acid substitution (H1799Y) in the 3Dpol region was related to temperature sensitivity. Although the epidemic strength is low, some EV-A90 strains were temperature resistant, which is suggestive of strong virulence and transmission capacity. This study expanded the number of EV-A90 in GenBank and provided basic data that may be useful for studying the molecular epidemiology of EV-A90.

Enterovirus type EV-A90 is a novel enterovirus that has been associated with acute flaccid paralysis (AFP). It was firstly identified and classified in 2005 with a molecular typing method, and the first EV-A90 strain, BAN99-10399, was assigned as the prototype strain 6 . Currently, there are six full-length EV-A90 genome sequences and several entire VP1 sequences in GenBank [6][7][8][9][10] . In mainland of China, EV-A90 has been rarely reported, and the only previous reports were in Shandong province, where EV-A90 was isolated from three children in 2001 and 2003 11 .
In this study, we determined the full-length genome sequences of five EV-A90 strains (

Results
Virus harvest and viral titre determination. Stool specimens were inoculated into rhabdomyosarcoma (RD) cells. When a complete cytopathic effect (CPE) was observed during a five-day observation period, viruses were harvested 12 . After a fixed gradient dilution and re-inoculation into a 96-well plate, the viral titre was calculated according to the Behrens-Kärber formula. All of the samples were determined to have a cell culture infective dose of 10 5 -10 6 in 0.05 mL of cell culture (CCID 50 /0.05 mL). Molecular typing and sequence analysis. Using molecular typing methods, we identified all five isolates in this study as EV-A90. The VP1 regions of these isolates showed 88.4-89.0% nucleotide sequence similarity and 98.2-98.9% amino acid sequence similarity to the prototype strain. The genomes of these five EV-A90 strains were 7,423-7,425 nucleotides in length, with a 5′-NTR of 747-749 nucleotides, a 94-nucleotide 3′-NTR preceding the poly (A) structure, and a single ORF of 6,582 nucleotides encoding 2,194 amino acids.
In the comparative analysis, 81 enterovirus prototype strains were used as the reference sequences, including 18 EV-A strains, 55 EV-B strains, 7 EV-C strains, and 1 EV-D strain. As shown in Table 1, the Xinjiang EV-A90 isolates were highly similar to each other and to some of the other EV-A in some regions. For example, they showed 82.6% nucleotide (100% amino acid) identity to EV-A91 (AY697461) in the VP4 region, 88.2-89.2% nucleotide (97.9-99.3% amino acid) identity in the 2B and 2 C regions, and 87.2-92.1% nucleotide (96.4-100% amino acid) identity in the P3 region. They also showed >85% nucleotide (90% amino acid) identity to EV-A121 (KU355877), EV-A76 (AY607458), EV-A89 (AY697459), and EV-A92 (EF667344) in the 2B region or other non-structural regions. This suggested that complex recombination events probably occurred in the P2 and P3 regions of the Xinjiang EV-A90 strains.
Phylogenetic analysis of global EV-A90 strains and evolutionary distance estimation. We collected all the available VP1 sequences and full-length genome sequences of EV-A90 strains from GenBank. Then, we constructed phylogenetic trees based on the sequences of the VP1 region ( Fig. 1) after sequence screening, multiple alignment, and shape optimization. Bootstrapping values greater than 80% were considered statistically significant 13 . According to the phylogenetic tree based on the VP1 region, the EV-A90 strains are divided into 3 clusters (A-C), and the strains in cluster C are further divided into 2 sub-clusters (C1 and C2) based on nucleotide divergence. Strains belong to C genotype become major prevalent cluster, which is different from the early genotypes that have only a few sequences published. We inferred that viruses in Cluster C are possibly more actively varying, especially those in the C2 sub-cluster. As shown in Table 2 p-distance within Cluster C was 0.1138 ± 0.0088 (95% CI), and the maximum difference between sub-clusters C1 and C2 was 12.0%. According to the strategies for the classification of EV-A71 genotypes 14 , clusters A, B, and C were three different genotypes (Genotype A, B and C), and genotype C was further divided into two sub-genotypes (C1 and C2).
Recombination analysis of the Xinjiang EV-A90 strains. We used Simplot (V3.5.1) to analyse the five Xinjiang EV-A90 strains for potential recombination, and the results were quite similar, so one EV-A90 sequence (HT-HTH08F) was selected as a reference strain. The similarity plot and genome bootscanning analysis of HT-HTH08F were shown in Fig. 2 indicated that the P1 regions of the five strains were homologous to that of the prototype EV-A90 strain, whereas the P2 and P3 regions, specifically from the second half of the 2B region to the 3′-NTR, were more similar to EV-A121 and EV-A91. In addition, the sequence from 3D to the 3′-NTR was also fairly similarity to EV-A89 and EV-A76. This suggested that recombination events occurred in the Xinjiang EV-A90 strains, likely with some other EV-A types, between 2B and 3D. By verifying of maximum likelihood (ML) tree (Fig. 3), we found an obvious topological exchange between EV-A90 and EV-A121 from P3 region (or 3CD region). It possibly proved a recombination occurred. We hardly found the same phenomenon in P2 region (or after 2B region) even though the EV-A90 strains also partially lost their original genetic characteristics in P2 region.
Neutralization assay and seroprevalence evaluation. In our study, 49 serum samples were collected from healthy children in Xinjiang who were ≤5 years old. Of these samples, 25 were collected from Hotan city, and 24 were collected from Kashgar city. In the neutralization assay in 96-well plates, only 18 samples (>1: 8) were positive for EV-A90, for a total positive rate of 36.73%. The geometric mean titre (GMT) was 1:10.47 for all positive samples. The EV-A90 neutralization antibody titres are shown in Table 3. The results of the neutralization   Table 2. The nucleotide (deduced amino acid) similarity and the P-distance between groups (sub-groups) were calculated. The similarity data was shown in the first two parts. Another is the P-distance between groups (subgroups) in the 95% confidence interval. And the standard error was shown below each data. assay showed an obvious difference between the two cities (Z = 3.2344, P < 0.01). In Hotan, the GMT was 1:9.19, while in Kashgar, it was 1:20.16. However, in general, the seroprevalence of EV-A90 in both cities was lower than that of other enterovirus serotypes, such as EV-A71 15 .
Xinjiang EV-A90 strains showed different temperature-related phenotypes. Temperature sensitivity is an important characteristic of enteroviruses. It usually serves as an in vitro marker for the attenuation of poliovirus vaccine strains and EV-A71, although the link between temperature sensitivity and attenuation may not be straight forward, it could serve as an indicator of virulence in Enterovirus 16,17 . The Poliovirus Sabin strain is a good example for understanding the importance of temperature sensitivity. Based on our observations, three strains (HT-MYH12F, HT-HTH08F, and KS-KSH13F) were temperature resistant and two others (HT-PSH40F and KS-SCH05F) were temperature sensitive (Fig. 4). After a 48 h incubation, the maximum viral titre of KS-SCH05F and HT-PSH40F at 36 °C was >10 5 TCID 50 , whereas their titres at 39.5 °C were ≤10 3 TCID 50 .
In contrast, the titre of the other three strains did not differ significantly when they were incubated at these two temperatures. Generally speaking, we considered a strain to be temperature sensitive when the logarithm of the viral titre of the strain at 36 °C was twice the viral titre of the strain at 39.5 °C. The results showed that HT-MYH12F, HT-HTH08F, and KS-KSH13F replicated to a greater degree at the higher temperature (39.5 °C).
To investigate the possible mechanism underlying the difference in temperature sensitivity among these strains, they were screened for nucleotide and amino acid substitutions. A total of 26 nucleotide substitutions in the NTRs and 29 amino acid substitutions in the ORF were detected and are summarized in Table 4. The substitutions were randomly distributed throughout the genome; however, one amino acid substitution, a histidine to tyrosine in 3Dpol was interesting. In the two temperature-sensitive strains, the amino acid at this locus is a histidine, whereas in the three temperature-resistant strains, this amino acid is a tyrosine. We also compared the base composition and codon bias between temperature-sensitive and temperature-resistant strains; however, little information was obtained. The strains in both groups showed 44.6% G + C content, and the amino acid total compositions of the five strains were identical.

Discussion
The novel enterovirus serotype is a classification that first appeared after the traditional serotyping method was replaced by molecular typing methods 18 . Isolates with >75% nucleotide similarity (or >85% amino acid similarity) in the VP1 region are considered to be of the same serotype 5,19 . If this condition is not met, then the strains may belong to different serotypes. Based on molecular typing, many novel enterovirus serotypes have been defined and re-classified.
In mainland of China, many novel enterovirus serotypes were identified through the AFP case surveillance system. We reviewed the novel enterovirus serotypes identified in China in recent years, including EV-A76 20 , EV-A89 21 , EV-B81 13 , EV-B111 22 , and EV-C96 23 , and found that a sensitive surveillance system plays an very important role in finding novel enteroviruses, even when the surveillance system is not specific for enteroviruses.
The gold standard to distinguish enteroviruses serotypes is molecular typing method that based on the difference of VP1 sequences similarity. Although there are no global criteria for the classification of EV genotypes, rather, the ranges of percentage similarities between (intra-typic) clusters and subclusters often overlap, suggesting that percentage similarities cannot always be used for unambiguous genotype/genogroup designations, 15%  figure (a-d). Isolates in this research were marked with and the prototype of EV-A90 (BAN00-10399) was marked with . The phylogenetic tree was reconstructed using maximum likelihood method, in which GTR + I + G model was used in P1 fragment and TN93 + I + G model was used in the rest fragment. We also put forward a hypothesis between the major parental type (yellow) and the minor parental type (blue) because of recombination.  difference in nucleotide sequences for identifying enterovirus genotypes plays an important role in molecular epidemiology research. Brown and Oberste published an article 14 that has strong impact on classification of enterovirus genotype and subgenotype. EV-A71, an enterovirus serotype, use this standard for genotyping. Followed that, E-11 24 , CV-A16 25 as well as CV-A6 26 has been classified to subgenotype used the same standard. Besides, the genotyping and subgenotyping standards for EV-A71 and CV-A16 have been accepted by RIVM, who developed enteroviruses molecular typing tool 27 . As to the global distribution of EV-A90, previous studies showed that it was divided into two genotypes, and genotype B was the major type, which is spreading in Asia and Europe 11 . However, a Latvian EV-A90 strain 7 (accession number EF392677) and a Russian EV-A90 strain 10 (accession number KC879486) clearly form a separate branch and should be considered carefully. Whether these strains are genotype A or B is difficult to determine as strains in this branch have at least 16% nucleotide difference among them. According to the criteria for the differentiation of enterovirus genotypes, EV-A90 strains should to be divided into three genotypes (A, B, and C), with sub-genotypes C1 and C2 in genotype C (previously classified as genotype B). And the C genotype of EV-A90 is presently the major type in circulation. However, the definition and number of genotypes are affected by the numbers of the reported cases or the published sequences, thus, these results may change as viruses evolve and as more sequences are reported. It is generally accepted that mismatch mutations and recombination are the principal mechanisms driving the evolution of enteroviruses 28,29 . Recombination, especially intertypic recombination, is often observed in the non-structural genomic regions encoding enzymes and small active molecules, which may lead to changes in their environmental adaptations and disease severity 29 . An early study showed high nucleotide sequence similarity among EV-A90, EV-A76, EV-A89, EV-A91, and EV-A121 in the 3D region 30 . Not only can these studies help to identify recombination events in enteroviruses, but it is also important for screening of possible parental donors. In our study, the similarity and genome bootscanning analysis suggested that all of the five Xinjiang EV-A90 strains were recombinants, especially in P3 (or 3 CD) region, we proved them a recombination with EV-A121. In addition, we found that the sequences of the Xinjiang strains in part of the P2 region was highly similar to to EV-A76, EV-A89, and EV-A91. However, we rarely found the slightly different topological characteristics between P1 and P2 regions. Certainly, we can't found the major parental type in P2 region, which is possibly affected by some unidentified EV-A types. Therefore, the possibility that Xinjiang EV-A90 recombined with an unidentified EV-A serotype enterovirus cannot be excluded. Detailed surveillance data for novel enteroviruses in different areas over different time periods are needed for further study.

Titers
Based on the seroepidemiological results of this study, EV-A90 enteroviruses are present at low levels, and the low average seroprevalence of EV-A90 antibodies suggests that there has not been a large-scale epidemic in Xinjiang. Because the levels of neutralization antibody levels are dynamic, the level within a population is an objective indicator of prevalence and transmission. Even though there are significant differences between Kashgar and Hotan, none of the strains were likely acquired in a large-scale epidemic. Because the strain (KS-SCH05F) used in the neutralization assay was isolated from Kashgar, it was expected that the overall level in Kashgar would be slightly higher than that in Hotan.
It is very interesting that, among the five tested Xinjiang EV-A90 strains, three (HT-MYH12F, HT-HTH08F, and KS-KSH13F) were temperature resistant and two (strain HT-PSH40F and KS-SCH05F) were temperature sensitive. Among the nucleotide and amino acid differences in the five Xinjiang EV-A90 strains, it seemed that an amino acid substitution (histidine to tyrosine) in the 3D pol region may be related to the temperature-sensitive phenotype, because the amino acid in the two temperature-sensitive strains is histidine while that in the three temperature-resistant strains is tyrosine. Interestingly, this amino acid substitution was also observed in a type I attenuated poliovirus vaccine strain (type I Sabin strain) 31 . In previous studies, substitutions in 3C and 3D were often related to changes in biological characteristics, including virulence, temperature-sensitivity, etc. 32 . However, additional evidence is needed to support this hypothesis, and our team is currently using reverse genetics to verify the role of this site.
In conclusion, we reported the full-length genome sequences of five EV-A90 strains isolated during AFP surveillance in the Xinjiang Uygur Autonomous Region of China. This is the first report of EV-A90 in Xinjiang, and at present, the number of EV-A90 strains worldwide is still very limited. Sequence analysis suggested that these five EV-A90 strains have high genetic diversity compared with the prototype strain, suggesting intertypic recombination within the non-structural protein-encoding region of all five strains, and extensive genetic exchange with other EV-A serotypes, such as EV-A89, EV-A76, EV-A91, and EV-A121. According to the seroepidemiology  18  19  32  64  117  119  122  149  171  142  195  196  407  459  470  548  585  669  708   T  -A  C  T  T  -A  A  HT-HTH08F  N  A  I  T  V  T  K  N  H  T   G  -. . 234  260  265  338  538  645  683  691  701  711  855  865  925  967  1044  1141  1166 1283 1452 .
. . survey, EV-A90 remains a low-level epidemic in Xinjiang. However, these viruses may have strong virulence or transmission capacity, since some of the isolated Xinjiang EV-A90 strains were temperature resistant. Hence, EV-A90 has the potential to become a more common strain. This study expands the number of EV-A90 whole genome sequences in GenBank and provides valuable information regarding the molecular epidemiology of EV-A90.

Sample collection and virus isolation. This study involved no human participants or experimentation.
The only materials used in our study were stool samples that were collected during a long-term AFP surveillance program for public health purposes. The AFP surveillance system is run by the national laboratory network in support of the polio eradication strategy. Written informed consent for the use of their clinical samples was obtained from all individuals involved in this study. This study was approved by the Ethics Review Committee of the National Institute for Viral Disease Control and Prevention (NIVDC), Chinese Center for Disease Control and Prevention. All experimental protocols were approved by the NIVDC, and the methods were carried out in accordance with the approved guidelines. The Xinjiang EV-A90 strains HT-MYH12F, HT-HTH08F, and HT-PSH40F were isolated from stool samples of patients with AFP residing in Hotan city, Xinjiang, China. The other two strains KS-KSH13F and KS-SCH05F were similarly isolated from Kashgar city, Xinjiang, China. We used the RD cell line for virus isolation. Each strain was purified by plague assays and cultivation of serial dilution. The cell line was provided by the WHO Global Poliovirus Specialized Laboratory in the USA and was originally purchased from American Type Culture Collection (Manassas, VA, USA). Infected cell cultures were harvested after complete CPE was observed.
In the seroprevalence study of EV-A90 antibodies, 49 healthy children ≤5 years of age were surveyed. Serum samples were collected from these 49 children in 2013 for seroepidemiological analysis of enteroviruses, with informed parental consent, by the Xinjiang Center for Disease Control and Prevention. The samples included 25 from Hotan and 24 from Kashgar, which are the two cities where the enterovirus strains were isolated. The same serum samples were used previously for EV-A89 and EV-B106 seroepidemiology studies 21,33 . None of the children showed any signs of disease at sample collection.
Molecular typing and full-length genome sequencing. Viral RNA was extracted from 100 µL of viral isolate with the QIAamp Viral RNA Mini Kit (Qiagen, USA). The primer pair E488 and E486 34 was used to amplify a portion of the VP1 region. The PCR products were purified using the QIAquick PCR purification kit (Qiagen, Germany) and sequenced in both directions from each strand using an ABI 3130 Genetic Analyser (Applied Biosystems, Foster City, CA, USA). The obtained partial VP1 sequences were analysed by comparison to the sequences in GenBank using the BLAST server and were typed using the EV Genotyping Tool.
Amplification of the full-length genome was performed in three steps, in which three overlapping RNA fragments were amplified. Reverse transcription-polymerase chain reaction (RT-PCR) was performed using the SuperScript III One-Step RT-PCR System with Platinum Taq High Fidelity DNA Polymerase (Thermo Fisher Scientific, USA). Then amplicons were purified with the QIAquick PCR purification Kit (Qiagen, Germany) and sequenced using an ABI 3130 Genetic Analyser (Applied Biosystems, Hitachi, Japan) by the dideoxy chain termination method. Each nucleotide was sequenced at least once. Some primer pairs used for PCR and sequencing were designed with Primer 5.0 software, and their parameters were assessed on the OligoCalc website (http:// biotools.nubic.northwestern.edu/OligoCalc.html) 35 . The rest of genome was determined with a primer walking strategy. The primer pairs used for each step are listed in Table 5.
Sequence bioinformatics analysis. The nucleotide (or deduced amino acid) sequences of the Xinjiang EV-A90 strains and other enteroviruses were compared by pairwise alignment using MEGA (version 6.06) 36 . The similarity matrix was processed with BioEdit (version 7.0.9.0) 37 . The neighbour-joining method with the maximum likelihood composite model was used to construct the phylogenetic trees. The bootstrap analysis was performed with 1000 pseudo-replicates. For the potential recombination analysis, we used Simplot 3.5.1 38 . The result is shown in a 200-nt sliding window, in which each point represented similarity between the Xinjiang EV-A90 strain and other EV-A types. Besides, we also constructed ML tree for verifying recombination using suitable substitution model tested by MEGA.
Neutralization assay against EV-A90. In our study, antibody detection and seroprevalence evaluation were performed by micro-neutralization assay 15 . Before use, each serum sample was inactivated by incubation at 56 °C for 30 min. Then, each sample was serially diluted (1:4 to 1:1024), and aliquots (50 µL) of five different concentrations were placed in duplicate into a 96-well plate. Next, we added virus (50 µL) to each well and confirmed that each well was 100 TCID 50 . After 2 h of neutralization at 36 °C in a CO 2 incubator, we added the same volume of RD cells to each well. Two wells in each column were used as serum controls. We also used a new plate for a cell control and virus control that we called a virus back-titration. All of the plates were incubated for 7 days. Only when the virus titre in the back-titration plate was between 32 and 320 TCID 50 and when the cell control was acceptable did we judge the result as effective. Then, the highest dilution of serum where 50% of the cultures were protected from CPE was recorded. The titre was calculated statistically with SAS version 9.4 using Wilcoxon rank sum test, for reasons that we barely collected the lowest serum titers during data collected when the titers of sample are lower than 1:4. The dataset contains ranked data, so rank sum test is helpful to evaluate the difference and data significance. A serum sample was considered positive if the neutralization antibody was observed at a dilution of 1:8, and then the GMT was calculated. Temperature sensitivity assay. The temperature sensitivities of the five Xinjiang EV-A90 strains were assayed on a monolayer of RD cells in 24-well plates. To observe the growth of EV-A90 at different temperatures, each virus was inoculated into 10 tubes; 5 of them were incubated at 36 °C, and the rest were incubated at 39.5 °C. After absorption at 36 °C or 39.5 °C for 1 h, the unabsorbed virus was removed, and maintenance medium (100 μL) was added to each well. The plates were incubated again at 36 °C or 39.5 °C, respectively, and harvested at 5 different time points post-infection (4,8,12,24, and 48 h). The 50% cell culture infectious dose (CCID 50 ) was calculated by the end-point dilution method on an RD cell monolayer in a 96-well plate at 36 °C. Virus isolates showing a greater than two-logarithm difference in titre at the two temperatures were considered to be temperature-sensitive. The series of viral titres were recorded, and the growth curves were drawn in Origin, version 9.