The geographic analysis of Hunan Province and study design
The geographic relationships between Hunan Province and the DENV outbreak areas in China were analysed. The results showed that Hunan, which is surrounded by Yunnan, Guangdong, Guangxi, Hainan, Fujian, Zhejiang and the other dengue outbreak areas, became a central area of the DENV epidemic (Fig. 1) The map in the figure was drawn by the authors of this study. Part of the data in the figure was cited from Zhao et al. [26], and part of the data was provided by the Centers for Disease Control and Prevention of Hunan Province.
During the DENV outbreak in Qiyang County, Hunan, in September 2018, a total of 260 serum samples of fever patients were collected, and 96 cases were confirmed to be NS1-positive by colloidal gold testing. Seven strains were proliferated in C6/36 cells for over 6 days to build a viral seed pool of Hunan DENV. Eighty-nine viral RNA genomes were successfully extracted directly from the 96 NS1-positive serum samples, followed by gene sequencing of the DENV structural protein C/prM/E genes. Phylogenetic analysis, recombination and selection pressure analysis, and potential secondary structure prediction based on structural gene sequences originating from epidemic strains were performed to understand the genetic characterization, potential source, and evolution of the epidemic DENV strain. The study design is shown in Fig. 2.
Phylogenetic analysis
The E protein gene sequences of 123 representative DENV-2 strains and four serotypes of standard strains were selected to construct phylogenetic trees with MEGA software version 7.0. The results showed that all 89 strains in this study were of the cosmopolitan DENV-2 genotype. The closest relative was the Zhejiang epidemic strain (MH010629, 2017), followed by strains isolated from Malaysia (KJ806803, 2013), Bali (KT806318, 2014), Indonesia (KT781561, 2014), and the Philippines (KU517847, 2015) (Fig. 3). Among the neighbouring provinces of Hunan Province, the provinces of Zhejiang, Yunnan and Guangdong each had more than 1000 reported cases of dengue fever in 2017, and all four serotypes were detected in each province [15, 20]. There were also reported cases in Fujian Province in 2017. Only one amino acid mutation (I431V/A) was observed in all 89 epidemic strains compared with the nearest related strain from Zhejiang (MH010629, 2017). These data suggest that the causative agent of the DENV outbreak in Hunan Province in 2018 may have come from the epidemic strains in Zhejiang Province in 2017.
Bases and amino acid mutations
Three structural protein-overlapping fragments from epidemic strains were obtained by PCR amplification. After sequencing, the proteins were effectively spliced, and the length of the coding nucleotide sequences was 2,325 nt, which encoded 775 amino acids. The homology between isolates was 99.7–100%, and the amino acid (AA) sequence of the E protein was highly conserved. By comparison, the homology of nucleotide and amino acid sequences between the 89 epidemic strains and DENV-2SS was 93.5% and 97.8%, respectively. Compared with DENV-2SS, the epidemic strains had two hundred fifteen mutated bases in the structural protein region, among which 195 were synonymous mutations and 20 were non-synonymous mutations, leading to 17 AA substitutions (Fig. 4). Two AA substitutions at the 104th (C104: M→I) and 108th (C108: L→M) amino acids were observed in protein C in the isolated strains; six amino acid mutations, including at the 143rd (M29: D→N), 166th (M52: K→N), 196th (M82: T→A), 241st (M127: I→V), 262nd (M148: H→Y), and 266th (M152: A→V) amino acids, occurred in the structural protein prM/M; and nine amino acid mutations, including at the 332nd (E52: Q→H), 351st (E71: D→A), 406th (E126: K→E), 409th (E129: V→I), 429th (E149: H→N), 444th (164: I→V), 602nd (E322: I→V), 670th (E390: N→S), and 742nd (E462: I→V) amino acids, were observed in structural protein E (Fig. 4).
Potential secondary structure of the structural protein region
The protein secondary structures of the DENV-2 standard strain KM204118 and three randomly selected sequences (HNQY2018014, 021, and 028) from the 89 isolated strains were predicted. Compared with DENV-2SS, the Hunan epidemic strains lacked one nucleotide-binding site (site 6) and one DNA-binding site (site 18) as well as one protein binding region (sites 4 and 5) in the capsid protein (Fig. S1), while one new DNA binding site (site 74) and two new protein binding sites (sites 19 and 29) were observed in the isolated strains. Moreover, variations in the disordered region were found among the Hunan epidemic strains, DENV-2SS and the Zhejiang/2017 strain (Fig. S1). In the prM/M region, which contained 166 amino acids, the protein secondary structure of the epidemic strains was highly consistent with that of the Zhejiang/2017 strain (Fig. S2). However, compared to DENV-2SS, three protein binding regions were missing in the Hunan epidemic strains (sites 122, 133, and 220), and one novel protein binding region had emerged (site 144). Additionally, one helical transmembrane region of the isolates visibly differed from DENV-2SS, and eight significant changes were observed in the buried and exposed regions, while no noticeable variations were found in the strand or helix regions (Fig. S2). Three protein binding sites (sites 584, 596, and 642) were missing at the 495-AA locus of the E protein, one novel protein binding location (site 377) was observed in the Hunan isolates, four considerable alterations were also observed in the exposed and buried regions, and minor changes were found in the helical transmembrane and disordered region (Fig. 5). Moreover, there were 22 changes in strand regions. Of these, 11 were new (120, 166, 192, 309, 334, 347, 446, 455, 512, 582–584, 591), 11 were missing (101, 102, 124, 141, 207, 290, 294, 553, 607, 636, 651, 692–695), and nearly 70% of the changes occurred in E proteins. Nevertheless, compared with the Zhejiang 2017 strain, there was no significant change in the protein binding region or the polynucleotide binding region in the structural proteins (C, prM/M, and E) (Fig. 5, Fig. S1, and Fig. S2).
Possible three-dimensional structure of the structural protein E genes
The possible three-dimensional structures of the structural proteins of the representative epidemic strains (HNQY2018014, 021, and 028) were predicted and compared with those of the DENV2-SS and Zhejiang/2017 strains. Homology modelling revealed that the five strains had the same three-dimensional structure. In addition, binding sites were predicted by the 3DLigandSite ligand binding site prediction server, and four protein binding sites (HIS429, ALA430, THR435, and GLY436) were observed in DENV-2SS (Fig. 6E). The Hunan epidemic strains and the Zhejiang/2017 strain had the same binding sites at ASN429, THR435, and GLY436 (Fig. 6D). HNQY2018028 had two different binding sites (429 and 430) compared to DENV-2SS (Fig. 6) and one different binding site (429) compared to the Zhejiang/2017 strain.
Recombination and selection pressure analysis
RDP4 software was used to analyse potential recombination events among HNQY2018001–HNQY2018089 and other representative DENV-2 virus strains. Preliminary analysis results showed that no recombination event occurred in these DENV-2 strains (p<0.05). The structural proteins of 202 strains were analysed, including 113 representative strains of DENV-2 and the 89 isolated strains. The MEEM method identified the maximum number of actively selected sites (n = 16). However, the FEL, IFEL and FUBAR methods indicated that all 775 sites were under negative pressure (Table 1). Therefore, no significant evidence of positive selection was found with at least three different methods, so positive selection pressure at these sites cannot be determined.