Introduction

Japanese Encephalitis Virus (JEV) is the most important cause of viral encephalitis worldwide, mostly prevalent in eastern and southern Asia1. JEV is a member of the Flaviviridae family and belongs to the same genus (orthoflavivirus) as dengue, yellow fever and West Nile viruses. It is a single-stranded, positive-sense RNA genome of approximately 11 kb in length. The virion of JEV contains three structural proteins – nucleocapsid or core protein (C), non-glycosylated premembrane protein (M), and glycosylated envelope protein (E), as well as seven non-structural (NS) proteins: NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5 2. JEV is transmitted to humans via mosquito bites from infected Culex species, mainly Culex tritaeniorhynchus. Humans, once infected, do not develop enough viremia to infect the feeding mosquitoes and thus become the dead-end host. The virus spreads via mosquitoes (vector), pigs (amplifying host), and/or water birds (reservoirs)3. Rural and peri-urban areas have the highest prevalence of the disease, where humans live in close proximity to these vertebrate hosts. Japanese encephalitis (JE) primarily affects children while most adults in endemic countries tend to develop immunity after childhood infection, however, individuals can contract infection at any age. Approximately 25% of the patients die and half of the survivors suffer/develop severe neuropsychiatric aftereffects from the disease4.

The most recent common ancestor (TMRCA) of JEV was estimated to have emerged 3255 years ago5 although the first case of Japanese encephalitis viral disease was documented in 1871 in Japan and the virus was first characterized in 1935 from the human brain6. JEV strains are classified into five geographical and epidemiological genotypes including GV, GIV, GIII, GII and GI based on the nucleotide sequence of envelope gene. The GIII genotype was previously dominant, but surveillance data from the last two decades, in the form of nucleotide sequences of collected isolates, have revealed a gradual replacement of GIII by GI in such a manner that GI is either dominant or co-circulating with GIII in many JEV regions of endemicity and epidemicity7. This genotype change is a cause of concern as it may affect disease control measures including vaccination. According to a recent report, the current licensed JE vaccines, which are mostly adapted from a live-attenuated virus (SA14-14-2), belongs to GIII8, fail to offer complete protection against JEV GI infection9. However, the genotype change and spread of the disease in different hosts were not studied country-wise. It is imperative to understand the pattern of genotype change and virus dissemination with respect to both host and geographical region to formulate effective preventive and control measures for JE.

In the present study, we have analyzed complete E protein nucleotide sequences of 731 isolates from different hosts and geographical regions to understand the genotype pattern and the spread of virus across countries in different hosts over the span of 84 years. Further, the dominance of one genotype over the other was assessed along with the divergence time dating within a country.

Materials and methods

Sequence retrieval and creation of data sets

The complete E protein nucleotide sequences (1500 bp in length) of 731 isolates of JEV were retrieved from the GenBank database (https://www.ncbi.nlm.nih.gov/genbank/) for the study. Search parameters included: ((“Japanese encephalitis virus“[Organism] OR Japanese encephalitis virus[All Fields]) AND envelope[All Fields] AND protein[All Fields]) AND (“1499“[SLEN] : “1500“[SLEN]) with a date range of 1935–2019. The final set of nucleotide sequences were of isolates belonging to various hosts including mosquitoes, midges, bats, pigs, horses, and humans with isolation dates ranging from 1935 to the recent past and as well from geographical regions covering China, Japan, Korea, Thailand, Vietnam, India, Indonesia, Taiwan, and some other Asian countries. The sequences with associated background information regarding genotype, year of isolation, host from which it was isolated and country of isolation were retained for the analysis. The isolates were sorted as per the features mentioned in NCBI to create three datasets based on country of isolation, host from which it was isolated and the genotype (GI, GII, GIII, GIV and GV). The countries and hosts with very few isolates were excluded from the country-specific and host-specific datasets. However, all these isolates were considered for generating the genotype-specific dataset. Further analysis is done for the three datasets (Country- wise, Host – wise and Genotype – wise) separately. Also, whole genome sequence data of JEV were downloaded from NCBI Virus database (NCBI Virus). Search parameters included: ((“Japanese encephalitis virus” AND Sequence Length [Min:10000], Nucleotide Completeness [Complete], AND Poxvirus/Integrated [Exclude], AND Lab Passaged [Exclude], AND Vaccine Strain [Exclude])). Sequences with host and country of origin information were selected, resulting in 142 sequences from diverse hosts and geographical regions, with isolation dates ranging from 1952 to the present. Details of the isolates analysed in the study listed in Supplementary File 1.

Phylogenetic analysis

The multiple sequence alignment was performed for the three datasets of Envelope gene separately using Clustal Omega10. The aligned sequences were then analysed in MEGA X11 software for the construction of the phylogenetic tree. The trees for the three datasets were constructed using the neighbor-joining (NJ) method. For whole genome sequence data multiple sequence alignment was performed using MAFFT (Multiple Alignment using Fast Fourier Transform) and phylogenetic analysis was done using the Maximum Likelihood method in IQ-TREE12 using the best fit model SYM + G4. Bootstrap resampling with 1,000 replicates was used to test the reliability of the phylograms. The output phylogenetic trees generated were then exported to iTOL to view the phylogenetic trees13.

Haplotype analysis and network construction

Number of haplotypes was determined from the Clustal Omega MSA files using DnaSP v6 14. The phylogenetic networking was developed with Network 10.2.0.0 for each category using Median-Joining Network (MJN) algorithm15 with default parameters.

Evolutionary analysis and TMRCA Estimation

Further, using these aligned files, BEAST analysis was done to find the time to the most recent common ancestor of isolates in each country. Evolutionary analysis was conducted by BEAST 2.7.6 16 to generate and run a Bayesian inference of phylogeny with the Markov chain Monte Carlo (MCMC)17 algorithm and a chain length of 100 million iterations. MCMC chain convergence was assessed by evaluating the estimated sampling size by using the Tracer v1.7.2 18. To determine the most suitable substitution and clock models, a rigorous approach was employed. The bModelTest2 was utilized to identify the optimal substitution model, while nested sampling was employed to select the best-fitting clock model. Nested sampling calculates the marginal likelihood for each clock model under consideration. The Bayes Factor (BF) is then computed, comparing the marginal likelihoods of two models: BF = Ml1 / Ml2; where: BF = Bayes Factor, Ml1 = Marginal likelihood of the first clock model and Ml2 = Marginal likelihood of the second clock model. If the BF is greater than 1, the first clock model (Ml1) is preferred; otherwise, the second clock model (Ml2) is selected.” Country wise substitution and clock model parameters are mentioned in Supplementary File 1. The clock rate and the TMRCA estimates were extracted by using Tracer v1.7.2 and the final Maximum Clade Credibility tree (MCC) using a posterior was identified by using TreeAnnotator v2.7.6 with a burnin of 10% discarding the first 10% of the trees. The output generated was visualized in Figtree (https://tree.bio.ed.ac.uk/software/figtree/) to visualize and modify the tree. Temporal signals within each phylogenetic tree were assessed using the Phylostems tool19, which applies root-to-tip regression analysis at every eligible internal node. A rooted tree with branch lengths scaled as genetic distances and tip labels annotated with sampling dates was used as input.

Results

Genotype change pattern across the years in different countries

The full length (1500 bp) envelope gene sequences from 731 isolates were analyzed to identify the genotype change. In most of the endemic countries, it was observed that there was a genotype change from GIII to GI (Fig. 1a). GI, III, and V were found in China (Fig. 1b), with GI dominating the others. GI isolates from China were primarily mosquito-borne, while GIII isolates were from different hosts (Fig. S1a). In Indonesia (Fig. 1c), GIII and IV were found with a spike in GII infection in 1979 and a spike in GIV infection in 1981. It was observed that GIV predominated in Indonesia in the recent past as compared to GIII. The majority of the isolates were of mosquito origin, with the exception of a few GIV isolates from pigs (Fig. S1b).

Fig. 1
figure 1

Genotype change pattern of JEV isolates in different countries: Overall genotype change over the years (a). Genotype change over the years for the isolates within China (b), Indonesia (c), Japan (d), Korea (e), Taiwan (f), Thailand (g), Vietnam (h) and India (i).

Full size image

In Japan (Fig. 1d), GI and GIII, were found with a spike in GI infection in 2008. Most of the GI and GIII isolates were from mosquito, human and swine origin while a few isolates of GIII were of equid origin (Fig. S1c). Since 1990 the GI is predominant in Japan. In Korea (Fig. 1e), GI, II, III and V, were found with a spike in GIII infection in 1983 and between 1987 and 1988. A genotype change from GIII to GI occurred in 1989 and GI became dominant and has remained dominant ever since, except in 2012, when GV dominated the occurrence of GIII. The GI isolates were observed to be from mosquito and swine origin, the GII from human origin, GIII from human and mosquito origin and the GV solely from mosquito origin (Fig. S1d). In Taiwan, GIII infection was dominant till 2007, with spikes in 1965 and 2005 (Fig. 1f). However, after the initial occurrence of GI in 2008, GI dominated GIII. Some of the GIII isolates were of swine origin, while most of the GI and GIII isolates were of human and mosquito origin (Fig. S1e).

In Thailand (Fig. 1g), only GI was observed with host origin of mosquito and swine (Figure S1f). In Vietnam (Fig 1h), GI and III were found with a spike in GI infection in 2002 and 2005 and a spike in GIII infection in 2004. The GI has a wider host range, including humans, mosquitoes, and swine, whereas GIII has a human origin only (Fig. S1g). In India (Fig 1i), only GIII was in circulation till 2009 while GI appeared for the first time in 2010. These two genotypes have been co-circulating with GI dominance since 2015. The GIII has a wider host origin, including human, mosquito, pig, and horse while GI has human and pig origin (Fig. S1h).

Genotype and host distribution across the countries

On comparing the genotypes across the countries, GI was found predominantly in China and Japan (Fig. 2a) and in hosts, GII was found in Australia and Indonesia (Fig. 2c), GIII in China, Japan and India (Fig. 2b), GIV in Indonesia and Australia (Fig. 2d) and GV in Korea, Malaysia, Singapore and China (Fig. 2e). On comparing the host origin across the countries among the isolates, mosquito and human isolates were predominantly from China and India respectively ( Fig. S2a and Fig. S2b) and swine isolates from Japan (Fig. S2c).

Fig. 2
figure 2

Phylogenetic tree of JEV isolates of GI (a), III (b), II (c), IV (d) and V (e).

Full size image

Network analysis of JEV isolates

To assess the evolutionary relationship of the isolates across hosts within the same country, network analysis was done. The analysis on isolates in China revealed 175 haplotypes from 254 isolates (Fig. 3a). Within GIII, H_166, H_175 and H_151 had isolates from human and mosquito origin; H_124 had isolates from pig and mosquito origin and H_140 had isolates from pig and human origin. Within the GI, H_30 had isolates from human and mosquito origin. GV has only one isolate H_1 isolated from mosquito. Most of the pig and human isolates were found directly connected to mosquito isolates in both the GI and III. The GI isolates clustered away from the GIII isolates with a 73 nucleotide differences. On a closer look some of the isolates between these hosts had one (H_89), two (H_108, H_115, H_122, H_123, H_58) or slightly greater nucleotide differences. Similar changes were seen among the isolates of GI across hosts. A change in single nucleotide was observed between a pig isolate - H_88 to mosquito isolates - H_89 indicating the evolving nature of the virus as it jumps from one host to the other. Further, this H_88 isolate had one and two nucleotide differences, respectively, with H_86 and H_87 pig isolates. In India (Fig. 3b), 42 haplotypes from 66 isolates were observed. Within GIII, H_21 and H_5 had isolates from human and mosquito origin. Some of the pig isolates in both the genotypes in India were connected directly to the human isolates. One equid isolate H_28 in GIII was connected to H_29 of human origin with 6 nucleotides differences. Within GI, the pig and human isolates had connectivity with 82 nucleotide differences. The GI isolates clearly clustered away from the GIII isolates with 124 nucleotide differences.

Fig. 3
figure 3

Network analysis of JEV isolates across different countries: Network analysis of JEV isolates in China (a), India (b), Japan (c), Indonesia (d), Korea (e), Taiwan (f), Thailand (g) and Vietnam (h).

Full size image

In Japan (Fig. 3c), 114 haplotypes from 213 isolates were observed. The GI isolates clearly clustered away from the GIII isolates with ca. 156 nucleotide differences. In Indonesia (Fig. 3d), 23 haplotypes from 27 isolates were observed. The genotypes IV, II and III clearly clustered from one another. All the isolates from Indonesia are of mosquito origin except H_1 of GIV, which had a swine origin. In Korea (Fig. 3e), 44 haplotypes from 57 isolates were observed. The genotypes clustered distinctly from one another. In Taiwan (Fig. 3f), 50 haplotypes from 58 isolates were observed with GI and III separated by 161 nucleotide differences. In Thailand (Fig. 3g), 14 haplotypes from 14 isolates were observed and all the isolates were GI. In Vietnam (Fig. 3h), 21 haplotypes from 22 isolates were observed and the GI clearly clustered away from GIII with 164 nucleotide differences.

As the majority of isolates are of GIII, a network illustrating the differences across countries was constructed, revealing a detailed picture of genetic relationships and host-specific dynamics (Fig. 4). On analyzing the genotypes vis-a-vis the countries of origin, the isolates seem to be connected across countries. The Chinese cluster was prominent, with a large number of interconnected isolates, indicating significant genetic diversity within GIII in this region. In contrast, isolates from India and Vietnam formed smaller yet distinct clusters, reflecting localized genetic variations and potentially unique transmission dynamics. Indian isolates limited connections to other countries in the network suggest that JEV in India might have regionally restricted transmission patterns, likely influenced by local ecological factors and host interactions. The network connections between Vietnam and China underline potential cross-border transmission events. Japanese isolates exhibited significant diversity, dominated by mosquito and swine hosts, alongside occasional human cases, reflecting endemic transmission. Korea showed smaller clusters, mainly driven by mosquito populations, with limited human and swine involvement. In Taiwan, isolates were uniquely clustered, highlighting region-specific dynamics dominated by mosquitoes and swine. Indonesia, Sri Lanka, and the Philippines exhibited sparse connectivity, with swine and mosquitoes as key hosts, suggesting localized transmission and limited circulation of GIII in these regions.

Fig. 4
figure 4

Network analysis of JEV isolates of GIII across different countries.

Full size image

Additionally, network analysis of JEV isolates was performed across four different host categories, including humans, swine, mosquitoes, and unknown hosts, as depicted in Supplementary Fig. S3. The networks were structured to visualize the relationships between isolates based on their genetic lineages (GI, GII, GIII, GIV, and GV), with nodes representing individual isolates and edges indicating genetic similarity or transmission pathways. The spatial distribution of lineages and their connectivity highlighted lineage-specific and host-specific clustering. In the human network (Fig. S3a), the majority of isolates were grouped into lineages GIII and GI, with GIII dominating the network. Geographic clustering was observed, with isolates from China, Vietnam, and other countries forming distinct clusters. Similarly, the swine network (Fig. S3b) revealed a predominance of GIII and GI lineages, with clear interconnectivity between isolates from various countries, particularly in East and Southeast Asia. These results underscored the role of swine as a major amplifying host for JEV and its close association with human outbreaks. The mosquito network (Fig. S3c) demonstrated greater diversity, with multiple lineages (GI, GIII, GIV, and GV) represented. This diversity highlighted the critical role of mosquitoes in the maintenance and transmission of JEV across different hosts and geographic regions. Additionally, the unknown host network (D) featured isolates primarily belonging to the GIII and GI lineages, suggesting possible spillover events or unidentified reservoirs contributing to the viral circulation. Details of the mutations underlying these phylogenetic relations and the list of isolates belonging to specific haplotypes are mentioned in Supplementary File 2.

BEAST analysis for the JEV isolates within the country

BEAST applies Bayesian analysis on molecular sequences using Markov chain Monte Carlo (MCMC) to average over tree space, with each tree weighted proportional to its posterior probability. The BEAST analysis of isolates from Japan clearly predicted the Most Recent Common Ancestor (TMRCA) back to the year 1808. The GI isolates clearly clustered away from GII isolates. TMRCA of Chinese isolates was predicted to exist back in 1815. Similarly, the existing time of the TMRCA was predicted for India to be 1868, Indonesia to be 1547, Thailand to be 1966, Korea to be 1778, Vietnam to be 1923 and Taiwan to be 1860. Phylogenetic trees for each country from BEAST analysis are provided in Fig. S4.

The temporal signal was evaluated using Phylostrata. The detection of strong and significant temporal signal at multiple nodes (including deep and shallow clades) indicated that the sequences were derived from measurably evolving populations (MEPs)19. The consistent observation of high adjusted R² values (1.0), regression slopes of 1.0, and low p-values (often < 1e-50) further validated the robustness of the temporal signal throughout the tree (Supplementary File 3).

Phylogeny using whole genome sequence data analysis

Whole genome sequences (142) with host, genotype, and country information were analysed to determine whether the observed trends align with those of the envelope gene-based phylogeny. Clustering based on genotypes for isolates from China, Thailand, South Korea, Japan, and Taiwan showed similar patterns when analysed using either the E protein or the whole genome sequences, with some unassigned genotypes clustering between established ones (Fig. S5). However, a distinct genotype-based clustering pattern was not observed for isolates from India when whole genome sequences were used (Fig. S5f). This discrepancy may be attributed to the fact that genotype classification is traditionally based on the E protein, while whole genome-based phylogenies are designed to infer evolutionary history20,21.

Discussion

In the present study, available full length nucleotide sequences of envelope gene of JEV isolates from endemic countries were compiled and analyzed to study, the genotype change pattern over the period of nine decades, spread of virus across the countries in different hosts and divergence time dating. The previous studies7,21 focused mainly on GI and III of JEV till the year 2009 and 2011, respectively. Here, all 5 genotypes of JEV were analyzed and the genotype change pattern observed across different countries over the years provided valuable insights into the epidemiology of JEV. Analysis of full-length envelope gene sequences from a diverse range of isolates spanning several decades revealed intriguing trends in genotype distribution.

To understand the divergence of the isolates, analysis was done vis–à–vis the country, host, and genotype. Within each country all genotypes diverged into distinct clades. The TMRCA of the Japanese isolates was predicted to be 1808 and this was in concordance with the first outbreak in Japan that occurred in 1871 22. The TMRCA – 1547, of the Indonesian isolates further established the previous evolutionary studies that JEV originated from its ancestral virus around 1500 AD23 in the Indonesia-Malaysia region. The distribution of the genotypes over the years indicated the evolution of genotypes I, II, and III in recent years over genotypes IV and V. The distribution of isolates among different countries showed differential prevalence of genotypes. Thailand had only GI, Taiwan GIII until 2007 and India GIII until 2009. On the other hand, multiple genotypes of JEV have been isolated in Korea (GI, GII, GIII, GV), Indonesia (GII, GIII, and GIV), and China (GI, GIII, and GV). The ancestral genotypes GIV and GV23 were prevalent in the eastern part of Asia (China, Indonesia, Korea). However, the genotypes that are more recently evolved (I, II, and III) were found geographically distributed across regions of Asia than the ancient genotypes (IV and V). It is possible that GV has been geographically restricted to fewer countries like China and Korea due to its narrow host range, since mosquitoes seem to be the only known vector for GV. However, Genotypes I, II, and III had a wider host range (GIII-humans, mosquitoes, pigs, and horses; GI-pigs and humans), attributing to their distribution pattern. geographically broader regions.

Further, a notable change from GIII to GI was observed in many countries. This change was witnessed in 1991 in Japan and Korea, in 2001 in Vietnam24. The emergence and dispersion of GI was also consistently noted in other regions, including China in 1979, Taiwan in the 1980s25, and Australia in 2000 26. According to the current study, GI started to replace GIII during early 1990s in Japan and Korea, in late 1980s in Thailand and in 2001 in Vietnam. In Japan and Korea, it was observed that the cases of GII came to a halt after the incidence of GI. However, in China, India, Taiwan and Vietnam both GI and GIII were found for a longer time. This pattern explains the circulation of both the genotypes of JEV in these regions27,28,29.

The re-emergence of GV in China after a dormancy of 57 years, and thousands of kilometers away from Malaysia and Singapore and its current circulation in Korea warrants appropriate control measures. Also, a change in host specificity of GV from human to mosquito was observed in Korea and China. However, further analysis of human isolates from these countries would strengthen the finding. The clustering of GV away from GIII in our study corroborates the distinct features of GIII and GV in in vitro and in vivo studies23,30,31.

A concern of re-emergence surfaced in our study, in the form of GIV outbreak in Indonesia after a span of 37 years in 2017, which was limited to a narrow host range of only mosquitoes and pigs. Unlike GV, this was not limited to a particular country, but has moved to Australia in 2019 32 marking the first GIV JE case in humans. This continued in the form of indigenous GIV outbreaks in Australia during 2022–2023, marking the largest JEV outbreak in this region33.

Conclusion

By analyzing full-length nucleotide sequences of the envelope gene, this study revealed intriguing trends in genotype distribution and spread of the virus across various hosts and regions. A notable change from GIII to GI is observed in many countries and distinct emergence and dispersion patterns across different regions and time periods are documented. The re-emergence of GV and IV after a dormancy of 57 years and 37 years, respectively, suggests a potential change in host specificity. TMRCA analysis revealed insights into the evolutionary history of JEV validating its origin in the Indonesia-Malaysia region and subsequent diversification into various genotypes. The network analysis revealed significant genetic diversity and lineage-specific clustering of JEV isolates across countries and hosts, highlighting complex transmission dynamics. Since the envelope gene is the key determinant of viral entry and JEV can be classified into five genotypes according to the full-length genome sequence or E gene sequences, the observed trends are reliable. Analyzing whole genome sequences further reinforced these findings. These findings provide valuable insights into the epidemiology and evolutionary dynamics of JEV would aid in shaping future control and prevention strategies.