A large acute hemorrhagic conjunctivitis (AHC) outbreak occurred in 2011 in Okinawa Prefecture in Japan. Ten strains of coxsackievirus group A type 24 variant (CA24v) were isolated from patients with AHC and full sequence analysis of the VP3, VP1, 3Cpro and 3Dpol coding regions performed. To assess time-scale evolution, phylogenetic analysis was performed using the Bayesian Markov chain Monte Carlo method. In addition, similarity plots were constructed and pairwise distance (p-distance) and positive pressure analyses performed. A phylogenetic tree based on the VP1 coding region showed that the present strains belong to genotype 4 (G4). In addition, the present strains could have divided in about 2010 from the same lineages detected in other countries such as China, India and Australia. The mean rates of molecular evolution of four coding regions were estimated at about 6.15 to 7.86 × 10−3 substitutions/site/year. Similarity plot analyses suggested that nucleotide similarities between the present strains and a prototype strain (EH24/70 strain) were 0.77–0.94. The p-distance of the present strains was relatively short (<0.01). Only one positive selected site (L25H) was identified in the VP1 protein. These findings suggest that the present CA24v strains causing AHC are genetically related to other AHC strains with rapid evolution and emerged in around 2010.
acute hemorrhagic conjunctivitis
coxsackievirus group A type 24 variant
enterovirus type 70
Hasegawa, Kishino, and Yano
highest posterior density
Markov Chain Monte Carlo
Coxsackievirus group A type 24 variant of the genus Enterovirus and family Picornaviridae causes AHC, as do EV70 and some adenoviruses . The first outbreak of AHC caused by CA24v was reported in Ghana in 1969  and the first isolation of CA24v causing AHC was reported in Singapore in 1970 . Outbreaks of CA24v-associated AHC have occurred sporadically in many areas. Accumulating evidence suggests that AHC outbreaks attributable to CA24v have occurred in many countries, most of which lie between latitude 40° north and 15° south . Hence, since 2000, CA24v-associated AHC had been sporadically reported in Brazil [5, 6], Spain , African countries [8-11] and Asian countries such as Taiwan and China [4, 12]. In Japan, the first AHC outbreak of CA24v occurred in 1985 to 1986 in Okinawa Prefecture; however, its prevalence in other areas was not established .
Coxsackievirus group A type 24 variant has been genetically classified into four genotypes (genotype 1–4) by phylogenetic analyses of the VP1 or 3Cpro coding region [4, 12]. In the phylogenetic tree of the VP1 coding region, genotype 1 (G1) was detected in Singapore in 1970, genotype 2 (G2) in Brazil and Jamaica in 1987 and genotype 3 (G3) in the Dominican Republic in 1993 and in the USA in 1998, whereas genotype 4 (G4) has been detected since 2002 . In the phylogenetic tree of the 3Cpro coding region, G1 was detected in Singapore and Hong Kong from 1970 to 1971, G2 in Singapore and Thailand in 1975 and G3 in France, Asian and African countries from 1985 to 1994, whereas G4 has been detected since 2000 [4, 12]. Previous CA24v associated with AHC outbreaks in Okinawa in 1985 to 1986 was G3 in the 3Cpro tree  and the detected virus was closely genetically related to Taiwanese strains detected in the same years .
VP1 protein of enteroviruses including CA24v contains a number of important neutralization sites [15-17] and the 3Cpro coding region encodes the protease . It has been found that both the VP1 and 3Cpro trees can correctly determine various genotypes of CA24v; however, the VP1 tree is superior to 3Cpro chronologically . Furthermore, recent studies have shown that antigenic changes may occur through positively selected amino acid substitutions caused by selection pressure in the host for respiratory syncytial virus and measles virus [19, 20]. VP1 protein is a major antigen in most enteroviruses; positive pressure in the host may act on this protein, leading to positive selection sites . However, to the best of our knowledge, positive pressure analysis has not been performed in the VP1 coding region of CA24v. Thus, it is important to analyze the VP1 coding region of CA24v causing AHC outbreaks, which have been reported in many countries. In addition, it is important to analyze the VP3 and 3Dpol coding regions because phylogenetic and positive pressure analyses have not been performed in either VP3 protein containing a number of epitopes or VP1 protein [15-17] and the 3Dpol coding region encoding RNA-dependent RNA polymerase .
A large CA24v-associated AHC outbreak occurred in Okinawa in 2011. In the present study, we performed genetic analysis of CA24v isolates from AHC patients in this outbreak by analyzing the VP3, VP1, 3Cpro and 3Dpol coding regions in detail.
MATERIALS AND METHODS
Okinawa Prefecture, a subtropical archipelago composed of over 100 islands, is the southwestern-most prefecture of Japan. Okinawa Island, the most populous island, has an area of 1208 km2 and is located about 2000 km southwest of Tokyo. Ishigaki Island, a small island with an area of 223 km2, is located about 400 km southwest of Okinawa Island and about 280 km east of Taiwan. From May to December 2011, about 4000 cases of AHC were reported in these islands. To investigate the AHC agent, 26 conjunctival swab samples were collected from AHC patients on Okinawa Island in June and 10 conjunctival swab samples from AHC patients on Ishigaki Island in October.
Virus isolation and detection
To isolate the virus agent of AHC, samples were subjected to cell culture using four cell lines (HEp-2, RD18S, Vero and Vero E6 cells), as previously described . The cells were checked daily for cytopathic effects and culture supernatant fluids harvested when clear cytopathic effects had been observed. To identify the pathogen, viral nucleic acid was extracted from 140 μL of 10 culture supernatant fluids harvested and 26 clinical samples negative for virus isolation using an available kit, the QIAamp Viral RNA Mini Kit (Qiagen, Tokyo, Japan), and suspended in DNase or RNase-free water, respectively. To detect enteroviruses, cDNA was synthesized using a PrimeScript RT reagent kit (Takara, Shiga, Japan) after RNA extraction, and amplified using primers as previously described [23, 24]. Additionally, adenoviruses were detected by a PCR method, as previously described . Amplicons were electrophoresed on 2.0% (w/v) agarose gel stained with ethidium bromide, purified using a QIAquick PCR Purification Kit (Qiagen) and nucleotide sequences determined by a direct sequencing method.
Determination of the full genome sequences of the VP3, VP1, 3Cpro and 3Dpol coding regions in CA24v isolates
To determine the full sequences of the VP3 (720 nt), VP1 (915 nt), 3Cpro (549 nt) and 3Dpol (1383 nt) coding regions of CA24v isolates, viral RNA of strains isolated in this study was extracted and cDNA synthesized as described above. Nucleotide sequences were amplified using primers as previously described  and determined as described above.
Phylogenetic analysis and estimation of evolutionary rate of the VP3, VP1, 3Cpro and 3Dpol coding regions in CA24v isolates by the MCMC method
To estimate the rate of molecular evolution and evolutionary relationships, phylogenetic analysis of the nucleotide sequence of the VP3, VP1, 3Cpro and 3Dpol coding regions in CA24v was performed using the sequences obtained in this study and 13, 25, 33 and 14 reference strains, respectively. A phylogenetic tree was constructed using the Bayesian MCMC method, as previously described [26, 27]. The dataset of the VP3 coding region was analyzed under a strict clock using the Hasegawa, Kishino, and Yano (HKY) models with the proportion of invariant sites model. The dataset of the VP1 coding region was analyzed under a strict clock using the HKY model with the gamma distributed rates across sites model. The dataset of the 3Cpro and 3Dpol coding regions was analyzed under a lognormal relaxed uncorrelated clock using the general time-reversible model with the gamma distributed rates across sites model. Each model was selected by the KAKUSAN4 program (version 4.0). The MCMC chain of the VP3, VP1, and 3Dpol coding regions was run for 10,000,000 steps and sampled every 1000 steps. The MCMC chain of the 3Cpro coding region was run for 2,000,000,000 steps and sampled every 200,000 steps. The phylogenetic tree was viewed in FigTree (version 1.3.1; available at: http://beast.bio.ed.ac.uk).
Calculation of similarity plots and p-distance of the VP3, VP1, 3Cpro and 3Dpol coding regions in CA24v isolates
Sequence similarities of the VP3, VP1, 3Cpro and 3Dpol coding regions of representative genotype strains were calculated based on the VP1 coding region (genotype G2 [BRA87-10629 strain], G3 [DOR93-10630 strain] and G4 [CA24v/Okinawa/5/2011 and Hangzhou13-02 strains]) against a prototype strain (EH24/70 strain). They were visualized by calculating pairwise sequence identity in a sliding window of 200 nucleotides and progressively advancing the window in 20 nucleotide steps across the length of the alignment using SimPlot software (version 3.5.1) . The Kimura substitution model was used to perform similarity plot calculations. In addition, to assess the p-distance distribution for the VP3, VP1, 3Cpro and 3Dpol coding regions between the present strains and other strains (as reference strains), their values were calculated as previously described . The picornavirus database was comprehensively searched to calculate the p-distance, as previously described . As a result, the full sequences of the VP3, VP1, 3Cpro and 3Dpol coding regions of CA24v with 13, 25, 33 and 14 reference strains, respectively were obtained.
Selective pressures analysis of the VP3, VP1, 3Cpro and 3Dpol coding regions in CA24v isolates
To evaluate selective pressure on the VP3, VP1, 3Cpro, and 3Dpol coding regions of the present strains and 13, 25, 33, and 14 reference strains, the dS and dN rates at every codon in the alignment were estimated by the following three different methods: single likelihood ancestor counting, fixed effects likelihood, and internal fixed effects likelihood methods using DATAMONKEY (http://www/datamonkey.org), as previously described [27, 31]. To examine the dN and dS rates, these methods were performed incorporating the HKY85 model of nucleotide substitution for the VP3 and VP1 coding regions and the general reversible Markov process model of nucleotide substitution for the 3Cpro and 3Dpol coding regions, and the phylogenetic tree deduced using the neighbor-joining method. Positive (dN>dS) and negative (dN<dS) selection was determined by a P-value of <0.1 (single likelihood ancestor counting, fixed effects likelihood and internal fixed effects likelihood methods).
Nucleotide sequence accession numbers
The sequence data obtained in this study have been registered with GenBank (accession no. AB769150–AB769165).
Virus isolation and detection
Twenty-six and 10 conjunctival swab samples were collected from AHC patients on Okinawa and Ishigaki Islands, respectively, in 2011 and seven and three strains of CA24v, respectively, were isolated in the RD18S cell line. In addition, amplicons of CA24v were obtained from six clinical samples collected on Okinawa Island by RT-PCR. Other enteroviruses and adenoviruses were not detected.
Phylogenetic analysis and estimation of evolutionary rate by Bayesian MCMC method of the VP3, VP1, 3Cpro and 3Dpol coding regions in CA24v isolates
The phylogenetic tree constructed using the Bayesian MCMC method based on the full sequences of the VP3 (720 nt), VP1 (915 nt), 3Cpro (549 nt) and 3Dpol (1383 nt) coding regions in CA24v isolates is shown in Figures 1-4. Overall, the present phylogenetic tree constructed by MCMC based on the VP3, VP1, 3Cpro and 3Dpol coding regions forms four genotypes (G1–4). The locations and genotypes of these strains are compatible with a previously reported phylogenetic tree constructed by the neighbor-joining method (Figs. 1-4) [4, 12]. Phylogenetic trees based on VP3 and VP1 sequences showed that the present strains are genotype G4 (Figs., 2 1). In addition, phylogenetic trees based on 3Cpro and 3Dpol sequences showed that the present strains can be classified as the same genotype (Figs., 4 3). Furthermore, a phylogenetic tree based on the VP1 sequences suggested that the present strains are derived from the same lineages as have been detected in other countries, such as China, India and Australia (Fig. 2).
Next, the year of the first major division, the subdivision of the cluster belonging to G4, the subdivision of the present strains, and each rate of molecular evolution of the VP3, VP1, 3Cpro, and 3Dpol coding regions are shown in Table 1. The year of the first major division of the VP3, VP1, 3Cpro and 3Dpol coding regions was estimated as 1969, 1966, 1964 and 1964, respectively. The year of the G4 strain division of the VP3, VP1, and 3Cpro coding regions was 1997, 1996 and 1995, respectively, whereas that of the 3Dpol coding regions was 1976. The year of the present strain division of their coding regions was 2009, 2010, 2009 and 2008, respectively. These findings suggest that the present strains have derived from a common ancestor Chinese strain (China/GD46/2010 and China/GD01/2010) in about 2008 to 2010 and formed various lineages. Furthermore, the rate of molecular evolution of their coding regions was estimated at about 6.15 to 7.86 × 10−3 substitutions/site/year.
|Coding region||Year of first major division||Year of G4 strain division||Year of present strain division||Rate of molecular evolution (×10−3 substitutions/site/year)|
|VP3||1969 (1967–1970)a||1997 (1996–1998)||2009 (2008–2010)||7.86 (6.03–9.99)|
|VP1||1966 (1962–1970)||1996 (1994–1997)||2010 (2009–2010)||6.37 (5.17–7.60)|
|3Cpro||1964 (1958–1970)||1995 (1993–1997)||2009 (2008–2010)||6.15 (4.38–8.07)|
|3Dpol||1964 (1954–1970)||1976 (1967–1983)||2008 (2007–2010)||7.49 (4.87–10.28)|
Analyses of similarity plots, p-distance, amino acid substitutions and positively selected sites in the VP3, VP1, 3Cpro and 3Dpol coding regions in CA24v isolates
The similarity plots of the full sequences of the VP3, VP1, 3Cpro and 3Dpol coding regions of representative genotype strains based on the VP1 coding region (genotype G2 [BRA87-10629 strain], G3 [DOR93-10630 strain] and G4 [CA24v/Okinawa/5/2011 and Hangzhou13-02 strains]) are shown in Figure 5. The similarities of these strains were compared with a prototype strain (EH24/70 strain). Among these strains, over 0.8 similarity was seen in the VP3 and VP1 coding regions. On the other hand, two partial coding regions of 3Dpol (nucleotide position 1–250 and around 300–600) of G2 and G3 strains showed low similarity. The similarity plots showed that the similarities between the present strain (CA24v/Okinawa/5/2011 strain) and prototype strain (EH24/70 strain) in all coding regions were 0.77 to 0.94 (Fig. 5). In addition, in the present strains, the degrees of identity of nucleotides and amino acids of the VP3, VP1, 3Cpro and 3Dpol coding regions were >99%.
The p-distance values for VP3, VP1, 3Cpro and 3Dpol coding regions obtained in this study are shown in Table 2. The p-distance values of the VP3, VP1, 3Cpro and 3Dpol coding regions of the present strains were <0.01, whereas the p-distance values of all CA24v strains were <0.20. Thus, although the present strain values are short, the p-distances of CA24v strains may be relatively long.
|Coding region||P-distance value (mean ± SD)|
|Present strains||G4 strains||All CA24v strains|
|VP3||0.003 ± 0.002||0.036 ± 0.028||0.062 ± 0.042|
|VP1||0.002 ± 0.003||0.033 ± 0.018||0.057 ± 0.042|
|3Cpro||0.003 ± 0.002||0.041 ± 0.026||0.075 ± 0.039|
|3Dpol||0.001 ± 0.001||0.027 ± 0.019||0.076 ± 0.066|
Amino acid substitutions were compared between the present strains and a prototype strain (EH24/70). Many amino acid substitutions were found in these coding regions. The results were as follows: K78A and D146E in the VP3 coding region; S11T, L25H, A27V, S32L, V51A, I56V, M89I, E100D, K103R, T146A, Y151H, I196M, F250Y, I256T, I297T and N301D in the VP1 coding region; I15V, V54I, K78R, V114I, N139H, I151V and M160I in the 3Cpro coding region; and H32Y, A37V, K139R, T147A, A197T, V262I, V371I, V388I and V436I in the 3Dpol coding region. Among these, only one positively selected site was estimated as L25H in the VP1 coding region. This positively selected site was also found in 11 reference strains detected since 2005, excluding BR-ES/93/2005 strain in the VP1 tree (Fig. 2). The present strains and these 11 strains were classified into one cluster: the year of division of this cluster was around 2002 (Fig. 2).
We performed full nucleotide sequences of the VP3, VP1, 3Cpro and 3Dpol coding regions in CA24v isolates from AHC patients in Okinawa Prefecture in 2011. The present study may show for the first time the evolution of four major genes (VP3, VP1, 3Cpro and 3Dpol coding regions) of CA24v from patients with AHC in Asian countries, including Japan. We can summarize our findings as follows: (i) the present strains were classified into G4 and may be derived from the same lineage strains as have been detected in China, India and Australia (Figs., 2 1); (ii) division of the present strains occurred around 2010; (iii) the rate of molecular evolution of these coding regions was relatively rapid (6.15 to 7.86 × 10−3 substitutions/site/year); (iv) nucleotide similarities between the present strains (CA24v/Okinawa/5/2011 strain) and a prototype strain (EH24/70 strain) were 0.77 to 0.94 and the p-distance values of the present strains were short (<0.01); and (v) only one positively selected site (L25H) was found in the VP1 coding region and all CA24v strains at this site belonged to one cluster that was derived around 2002 (Fig. 2). These findings suggest that the rate of molecular evolution of CA24v obtained in the present phylogenetic tree was relatively rapid and the prevalent CA24v strains in Okinawa Prefecture are genetically related to other CA24v strains detected in China, India and Australia since 2002.
It has been reported that the VP1 tree is superior to 3Cpro chronologically . However, we did not find significant superiority of the VP1 classification in the present phylogenetic tree. The year of G4 strain division of the VP3, VP1 and 3Cpro coding regions was between 1995 and 1997; however, that of the 3Dpol coding region was 1976. These results might be affected by the smaller number of isolate sequences available compared with the VP1 and 3Cpro coding regions. On the other hand, similarity analysis showed that the similarities of partial sequences (nucleotide position 1–250 and around 300–600) in the 3Dpol coding regions of G2 and G3 strains are relatively low. Thus, genetic divergence of CA24v might be seen in this region. In addition, recombinations are well-known in enteroviruses [21, 33]. These recombinations may occur in nonstructural rather than structural or VP proteins [21, 33]. Although genetic divergence of the 3Dpol coding region was seen only in some sites, the findings of the present study show that recombinations in CA24v may have occurred between the 3Cpro and 3Dpol coding regions.
We used the MCMC method to assess the rate of molecular evolution. The rates of evolution of the VP3, VP1, 3Cpro and 3Dpol coding regions in the present CA24v strains were 6.15 to 7.86 × 10−3 substitutions/site/year, which are similar to that of the 3Cpro coding region of CA24v reported previously (3.0 to 3.7 × 10−3 substitutions/site/year) [34, 35]. Previous reports have estimated the VP1 coding region of enterovirus type 68, EV70 and echoviruses to be 3.84 to 7.05 × 10−3 substitutions/site/year [36-39] and the 3Dpol coding region of enterovirus type 71 to be 5.53 × 10−3 substitutions/site/year . Thus, the evolution rate of the VP1 and 3Dpol coding regions in CA24v may be the same as that in other enteroviruses.
The major antigenic epitopes (virus neutralizing) of enteroviruses including CA24v are contained within VP1 protein and also sometimes within VP3 protein [15-17]; however, the antigenicity may be different in each virus. The 3Cpro and 3Dpol coding regions encode protein protease and RNA polymerase, respectively ; these nonstructural proteins play important roles in viral genome replication and viral proliferation. In the present study, we found many amino acid substitutions in their proteins; however, we found only one positively selected site (L25H) in the deduced VP1 protein. The number of amino acid differences in their proteins may have occurred by chance or be attributable to greater tolerance to changes, rather than to immune responses. In the present strains, a deduced amino acid (aa 25) substitution in VP1 protein was found as the positively selected site. Previous reports have suggested that aa 1–30 of amino acid sequences in VP1 protein in poliovirus may be associated with cell entry and inducing neutralizing antibodies in the host [16, 41]. In addition, in these strains, other amino acid substitutions found in the deduced VP3 (K78A) and VP1 (S32L, T146A, Y151H, I256T, I297T and N301D) proteins may also be associated with induction of neutralizing antibodies or changes in the structures of VP proteins. In addition, the amino acid substitutions found in the deduced 3Cpro (N139H) and 3Dpol (H32Y, T147A, A197T) proteins in these strains may also be associated with changes in the structures of nonstructural proteins. Thus, the possible effects of VP3–1 (structural proteins) and 3Cpro/3Dpol (non-structural proteins) substitutions might be more clearly differentiated. Taken together, further studies are needed to elucidate how these amino acid substitutions affect induction of neutralizing antibodies and changes in nonstructural protein structures.
In the present study, we detected CA24v strains in 16/36 samples; no other enteroviruses or adenoviruses were detected. Thus, we suggest that CA24v was the viral agent that caused the AHC outbreak in Okinawa Prefecture in 2011. However, although about 4000 cases of AHC were reported, only a few samples were collected. Further large-scale studies may be needed to identify the viral agent(s) causing AHC outbreaks. In addition, it is important to comprehensively analyze the CA24v genome, because this data may clarify the detailed molecular epidemiology of CA24v. However, a relatively large amount of genetic information is needed, very little of which is currently available for CA24v. Thus, we were not able to perform adetailed molecular epidemiologic study of CA24v, which is a limitation of the present study. Accumulation of more genetic data is needed to enable a comprehensive genetic study of CA24v.
In conclusion, the present CA24v strains emerged in around 2010 and evolved as various lineages, causing AHC in Okinawa Prefecture in 2011. CA24v may evolve rapidly, as do other enteroviruses such as EV70, which is associated with AHC. To better understand the properties of AHC agents, further and larger studies regarding the genomics of both agents (EV70 and CA24v) are needed.
We thank the eye clinics collaborating with the local health authority of Okinawa Prefecture for reporting the weekly number of AHC patients to Okinawa Prefectural Infectious Disease Surveillance Center and providing us with conjunctival swab samples from AHC patients. This work was partly supported by a Grant-in-Aid (H25-Shinko-Ippan-015) from the Japan Society for the Promotion of Science and for Research on Emerging and Re-emerging Infectious Diseases from the Ministry of Health, Labour and Welfare, Japan. There were no external funding sources for this study.
No authors have any financial relationships or interests to disclose.