Inferring Protective CD8+ T-Cell Epitopes for NS5 Protein of Four Serotypes of Dengue Virus Chinese Isolates Based on HLA-A, -B and -C Allelic Distribution: Implications for Epitope-Based Universal Vaccine Design

Dengue is one of the most globally serious vector-borne infectious diseases in tropical and subtropical areas for which there are currently no effective vaccines. The most highly conserved flavivirus protein, NS5, is an indispensable target of CD8+ T-cells, making it an ideal vaccine design target. Using the Immune Epitope Database (IEDB), CD8+ T-cell epitopes of the dengue virus (DENV) NS5 protein were predicted by genotypic frequency of the HLA-A,-B, and-C alleles in Chinese population. Antigenicity scores of all predicted epitopes were analyzed using VaxiJen v2.0. The IEDB analysis revealed that 116 antigenic epitopes for HLA-A (21),-B (53), and-C (42) had high affinity for HLA molecules. Of them, 14 had 90.97–99.35% conversancy among the four serotypes. Moreover, five candidate epitopes, including 200NS5210 (94.84%, A*11:01), 515NS5525 (98.71%, A*24:02), 225NS5232 (99.35%, A*33:03), 516NS5523 (98.71%, A*33:03), and 284NS5291 (98.06%, A*33:03), were presented by HLA-A. Four candidate epitopes, including 234NS5241 (96.77%, B*13:01), 92NS599 (98.06%, B*15:01, B*15:02, and B*46:01), 262NS5269 (92.90%, B*38:02), and 538NS5547 (90.97%, B*51:01), were presented by HLA-B. Another 9 candidate epitopes, including 514NS5522 (98.71%, C*01:02), 514NS5524 (98.71%, C*01:02 and C*14:02), 92NS599 (98.06%, C*03:02 and C*15:02), 362NS5369 (44.84%, C*03:04 and C*08:01), 225NS5232 (99.35%, C*04:01), 234NS5241(96.77%, C*04:01), 361NS5369 (94.84%, C*04:01), 515NS5522 (98.71%, C*14:02), 515NS5524 (98.71%, C*14:02), were presented by HLA-C. Further data showed that the four-epitope combination of 92NS599 (B*15:01, B*15:02, B*46:01, C*03:02 and C*15:02), 200NS5210 (A*11:01), 362NS5369 (C*03:04, C*08:01), and 514NS5524 (C*01:02, C*14:02) could vaccinate >90% of individuals in China. Further in vivo study of our inferred novel epitopes will be needed for a T-cell epitope-based universal vaccine development that may prevent all four China-endemic DENV serotypes.

Introduction ignored, and it raises the possibility that T-cell responses against all DENV serotypes might be beneficial or even required for vaccine protective efficacy. The advent of a T-cell epitope-based vaccine may offer an alternative that avoids ADE. Considering the important role of serotypespecific CD8 + T-cells in controlling DENV infection, a novel strategy for developing prophylactic and therapeutic CD8 + T-cell epitope-based vaccines is needed. Therefore, a T-cell epitope-based universal vaccine that induces a broad dengue-specific, multifunctional, and crossreactive CD8 + T-cell responses among all four DENV serotypes may be a more promising strategy against DENV infections.
The DENV genome consists of a single-stranded RNA of 10.7 kb in length. The open reading frame codes three structural proteins [capsid (C) protein, preM protein, and envelope (E) protein] and seven nonstructural proteins (NS1, NS2a, NS2b, NS3, NS4a, NS4b and NS5) [24,25]. It has been shown that CD8 + T cells preferentially target the NS3 and NS5 proteins, while CD4 + T cells preferentially target the E, C, and NS1 proteins [26]. Notably, NS5 is the largest and the most highly conserved protein encoded by the DENV genome, with approximately 67-82% amino acid sequence identity among the four DENV serotypes [27]. Thus, NS5 proteins could be used as a promising target in the design of a T-cell epitope-based vaccine to induce DENV-specific protective T-cell responses.
A necessary condition for a peptide to be a CD8 + T-cell epitope is that it binds to human leukocyte antigen (HLA) molecules. However, HLA molecules are extremely polymorphic with several thousand variants and can bind distinct sets of peptides [28]. Each HLA variant is expressed at vastly variable frequencies in different ethnic groups and geographic regions [29]. This means that it appears that an extremely large and impractical number of peptides would have to be selected to enable the development of a broadly protective multi-epitope vaccine. A large number of studies focused on predicting epitopes from the E, prM, NS1, NS3, or NS5 proteins allowed the identification of T-cell epitopes in DENV [30][31][32][33][34][35][36]. However, the CD8 + T-cell epitopes of the NS5 protein of DENV Chinese isolates linked with the class I HLA allele in Chinese population have been poorly revealed. Therefore, the identification of CD8 + T-cell epitopes that can induce protective DENV-specific T-cell responses by a feasible immunoinformatics approach is critical and urgent for the development of a T-cell epitope-based vaccine.
In this study, based on the distribution characteristics of HLA class I alleles in Chinese population, we identified putative CD8 + T-cell epitopes of NS5 protein of Chinese DENV isolates using various immunoinformatics approaches. Our results provide putative protective CD8 + T-cell epitope candidates or their combination for the development of a T-cell epitope-based universal vaccine to effectively prevent all four DENV serotypes that are endemic in China.

Retrieving the protein sequences
The sequences of the NS5 protein from 155 Chinese isolates belong to all four serotypes of dengue virus (DENV-1, DENV-2, DENV-3 and DENV-4) were retrieved from the National Center for Biotechnology Information (NCBI) protein database (http://www.ncbi.nlm.nih.gov/ protein/). The sequence of DENV-2 NS5 protein (accession number: KC131142.1) was used as an input for various bioinformatics tools for epitope prediction, antigenicity analysis and conservation analysis.

HLA genotypic frequency retrieval
To improve population coverage of the CD8 + T-cell epitopes, it is important to screen epitopes restricted by highly prevalent HLA alleles. Genotypic frequencies of the HLA class I alleles that include HLA-A,-B and-C loci in Chinese population were retrieval from the major histocompatibility complex database (dbMHC) (http://www.ncbi.nlm.nih.gov/projects/gv/ mhc/main.fcgi?cmd=init). For a broad coverage, HLA class I alleles with genotypic frequency >3% in Chinese population were selected for CD8 + T-cell epitope prediction. This parameter setting covers the highly prevalent HLA-A,-B, and-C alleles found in Chinese population.

Epitope prediction
CD8 + T-cell epitope is the minimal amino acid sequence required for CD8 + T-cells activation and recognition by immune system receptors. Since the affinity of an epitope binds to the HLA molecule plays a vital role in determining its immunogenicity. Hence, high affinity between epitopes and HLA molecules tends to be associated with higher immune responsiveness. The prediction of CD8 + T-cell epitopes that interact with different HLA class I alleles were performed using the IEDB analysis resource (http://tools.immuneepitope.org/mhci/) Consensus tool [37], which combines predictions from ANN, aka NetMHC 3.4 [38][39], SMM [40] and Comblib [41], if any corresponding predictor is available for the molecule. Otherwise, NetMHCpan is used. This choice was motivated by the expected predictive performance of the methods in decreasing order: Consensus > ANN > SMM > NetMHCpan > CombLib. For the IEDB-recommended method, a low percentile indicated strong binding affinity to HLA molecules. The threshold of percentile rank was set at 1 in this study [42]. Based on the representative length of peptides that bind to HLA molecules, peptide lengths of 8-11 amino acids were selected for the prediction of epitope-based peptides in this study.

Antigenicity analysis
Antigenicity is a key characteristic of the epitope that is recognized by immune system cells and/or antibodies. Thus, the antigenicity of the predicted epitopes is one of the most important criteria for epitope-based vaccine assessment. Since some of the predicted epitopes may lose antigenicity when analyzed, to ensure that the predicted epitopes could serve as a good CD8 + T-cell epitope, all of the epitopes were screened to assess their antigenicity. VaxiJen v2.0 (http://www.ddg-pharmfac.net/vaxijen/VaxiJen), an online web server used to predict the effective antigens and subunit vaccines, was used to identify and reevaluate T-cell epitope antigenicity. The predicted epitopes were uploaded in a plain sequence format and the virus was chosen as the target organism. The threshold level of an antigen was set at 0.5. The Vaxijen server performed well with 87% accuracy at a threshold of 0.5 antigenic score for viruses [43]. VaxiJen v2.0 allows antigen classification based on the physicochemical properties of proteins without recourse to sequence alignment. Finally, the epitopes with an antigenic score > 0.5 were selected as antigenic for the conservancy analysis.

Conservancy and population coverage analysis
To obtain the universal T-cell epitopes of four DENV serotype variants, the conservancy of candidate epitopes should be considered prior to other criteria, even population coverage rate. As a universal T-cell epitope, it should be highly conserved in all viral variants. Hence, to determine the conservation level of the predicted epitopes among the NS5 protein sequences of the different DENV strains, the predicted epitopes were analyzed for their conservancy using the IEDB epitope conservancy tool (http://tools.immuneepitope.org/tools/conservancy/iedb_ input) with a sequence identity threshold of 100%. The conservancy level of each potential epitope was calculated by seeking identities in all NS5 protein sequences of the four DENV serotype variants retrieved from the NCBI protein database. The epitopes that were 100% conserved in >90% of the sequences analyzed in four serotypes were selected as candidate epitopes. These highly conserved epitopes were selected and used to determine the population coverage by the IEDB population coverage calculation tool (http://tools.immuneepitope.org/ tools/population/iedb_input). Finally, all of the selected epitopes were analyzed for similarity with human proteome using the BLAST program (http://www.ncbi.nlm.nih.gov/BLAST/) to verify that they would not trigger autoimmunity.

Results and Discussion
Retrial of NS5 protein sequences of four DENV serotypes The NS5 proteins sequence of all four DENV serotypes circulating in China were retrieved in FASTA format from the NCBI protein database. A total of 155 sequences of the NS5 protein of Chinese isolates of the four DENV serotypes was obtained (S1 File, S2 File, S3 File and S4 File) and used for the further epitope analysis.

HLA class I alleles analysis
Since specific HLA alleles are expressed at variable frequencies in different ethnic groups and different geographic regions. Therefore, HLA allele frequencies prevalent in dengue hyperendemic areas must be considered in vaccine design. Here we focused on DENV-specific T-cell epitopes that are associated with the highly prevalent HLA alleles in a Chinese population. To this end, the genotypic frequency of the highly prevalent HLA class I alleles found in this Chinese population (>3%) that include HLA-A,-B, and-C loci were obtained from the dbMHC database.  (Table 1).  Prediction and antigenic analysis of CD8 + T-cell epitopes CD8 + T-cell responses play a substantial role in eliminating DENV infected cells that cannot be managed by antibody responses. An effective vaccine that can provide protection against dengue virus infection requires robust, broad, and multi-functional CD8 + T-cell responses. Therefore, the CD8 + T-cell epitopes that bind to different HLA class I alleles with varying affinities must first be identified. Here a large number of the antigenic epitopes with a high binding affinity score of <1 percentile and an antigenicity score > 0.5 were obtained from NS5 proteins of DENV Chinese isolates against HLA-A,-B, and-C alleles. Most epitopes bind with high affinity to single HLA-A,-B, or-C molecules. As a consequence, a total of 21 antigenic epitopes were obtained against the seven alleles of the HLA-A loci ( Table 2). A total of 53 antigenic epitopes were obtained against the eight alleles of the HLA-B loci (Table 3), while 42 antigenic epitopes were obtained against the eight alleles of the HLA-C loci (Table 4). Surprisingly, none of the epitopes bind to HLA-A Ã 02:07. Notably, some epitopes, like 92  LMYFHRRDLRL 524 (C Ã 01:02 and C Ã 14:02), and 362 FTNMEAQL 369 (C Ã 03:04 and C Ã 08:01) can be presented by multiple HLA molecules, suggesting that they can cover a broader population and may be better epitope vaccine candidates. Further, the potential of the epitope and HLA binding is essential in the assessment of the immunogenic potential of epitopes. Hence, as a ligand of the HLA-A molecule, epitopes 225 WYMWLGAR 233 and 224 IWYMWLGARF 233 presented by HLA-A Ã 24:02 were the best binders based on their 0.2 percentile. As ligands of HLA-B molecules, epitopes 89 TQMAMTDTTPF 99 and 530 CSAVPSHW 537 , which were presented by HLA-B Ã 15:01 and HLA-B Ã 58:01, respectively, were the best binders based on their 0.1 percentile. Likewise, as a ligand of HLA-C molecule, epitope 91 MAMTDTTPF 99 presented by HLA-C Ã 03:02 was the best binder based on its 0.1 percentile. Additionally, HLA-A Ã 24:02 has the highest number of binding epitopes (9/21), followed by A

Conservancy and population coverage of CD8 + T-cell epitopes
As an effective vaccine formulation, the epitope-based universal vaccine must include highly conserved CD8 + T-cell epitopes among all DENV serotypes to induce cross-reactive T-cell responses based on the fact that the conserved epitope candidates are more likely to confer cross-protection between pathogen variants. Here, conservancy analysis revealed a total of 14 highly conserved epitopes with !90% protein sequence matching in a total of 155 NS5 protein sequences from four DENV serotypes of Chinese isolates. For the HLA-A allele, five of the 21 epitopes were conserved with !90% conservancy (   Since the HLA allele frequencies vary among populations due to different genetic backgrounds, to design an effective vaccine, we should consider the candidate epitopes that specifically bind with the prevalent HLA molecules in the target population where the vaccine will be employed. Therefore, here we examined the population coverage of the proposed epitope vaccine candidate in a Chinese population. The results showed that the epitope 92 (Table 5). It is worth noting that the combination of the epitopes 92 AMTDTTPF 99 (B Ã 15:01, B Ã 15:02, B Ã 46:01, C Ã 03:02 and C Ã 15:02), 200 CVYNMMGKREK 210 (A Ã 11:01), 362 FTNMEAQL 369 (C Ã 03:04 and C Ã 08:01), and 514 LMYFHRRDLRL 524 (C Ã 01:02 and C Ã 14:02) could vaccinate >90% of the Chinese population, suggesting that the four epitopes are better candidates for a multiple T-cell epitope-based vaccine. These highly conserved HLA restricted epitopes with acceptable population coverage could be putative epitope vaccine candidates in their combinations to elicit DENV-specific T-cell responses. Finally, to avoid the autoimmune response, all of the predicted class I HLAbinding antigenic epitopes were analyzed for their homology with human proteome, but no epitope was homologous with human proteome. Based on these results, we proposed that the combination of these highly conserved epitopes could be as universal CD8 + T-cell epitope vaccine candidates to induce DENV-specific T-cell responses against four DENV serotypes that are endemic in China.

Conclusion
HLA-restricted epitopes for prophylactic or therapeutic vaccines against infectious diseases to induce a T-cell response that eliminates infected cells is a promising vaccine strategy. In this study, we identified 14 universal CD8 + T-cell epitope candidates using immunoinformatic approach, and they are highly conserved among all four DENV serotypes that are endemic in China. The combination of four epitopes, including 92 AMTDTTPF 99 (B Ã 15:01, B Ã 15:02, B Ã 46:01, C Ã 03:02 and C Ã 15:02), 200 CVYNMMGKREK 210 (A Ã 11:01), 362 FTNMEAQL 369 (C Ã 03:04 and C Ã 08:01), and 514 LMYFHRRDLRL 524 (C Ã 01:02 and C Ã 14:02), could vaccinate >90% of individuals in China. These epitopes are valuable T-cell epitope-based vaccine candidates for the development of a universal dengue vaccine that is capable of eliciting specific and robust protective T-cell responses against four DENV serotype variants. In conclusion, our study highlights that it is possible to design an epitope-based universal vaccine against all four DENV serotypes based on protective CD8 + T-cell-mediated cellular immune responses. Bold and italic-indicates the percentage of epitope that is 100% conserved in more than 90% of the sequences analysed in four serotypes. doi:10.1371/journal.pone.0138729.t003 Inferring CD8 + T-Cell Epitopes for DENV NS5 Protein