The CD8+ and CD4+ T Cell Immunogen Atlas of Zika Virus Reveals E, NS1 and NS4 Proteins as the Vaccine Targets

Zika virus (ZIKV)-specific T cells are activated by different peptides derived from virus structural and nonstructural proteins, and contributed to the viral clearance or protective immunity. Herein, we have depicted the profile of CD8+ and CD4+ T cell immunogenicity of ZIKV proteins in C57BL/6 (H-2b) and BALB/c (H-2d) mice, and found that featured cellular immunity antigens were variant among different murine alleles. In H-2b mice, the proteins E, NS2, NS3 and NS5 are recognized as immunodominant antigens by CD8+ T cells, while NS4 is dominantly recognized by CD4+ T cells. In contrast, in H-2d mice, NS1 and NS4 are the dominant CD8+ T cell antigen and NS4 as the dominant CD4+ T cell antigen, respectively. Among the synthesized 364 overlapping polypeptides spanning the whole proteome of ZIKV, we mapped 91 and 39 polypeptides which can induce ZIKV-specific T cell responses in H-2b and H-2d mice, respectively. Through the identification of CD8+ T cell epitopes, we found that immunodominant regions E294-302 and NS42351-2360 are hotspots epitopes with a distinct immunodominance hierarchy present in H-2b and H-2d mice, respectively. Our data characterized an overall landscape of the immunogenic spectrum of the ZIKV polyprotein, and provide useful insight into the vaccine development.


Introduction
As a mosquito-borne virus belonging to the flavivirus genus of the Flaviviridae family, Zika virus (ZIKV) was firstly isolated in 1947 from rhesus macaque (Macaca mulatta) in the Zika forest, Uganda [1,2]. Although human infection was reported as early as 1964, the first major ZIKV outbreak did not occur until 2007 in Yap Island, where over 70% of the population within the island became infected [3]. Infection with ZIKV in humans is often asymptomatic or mild, consisting of skin rashes, conjunctivitis, fever and headaches [4]. However, the outbreak of Zika virus, emerging since 2016 in French Polynesia and in South America and spreading immediately globally, was linked to Guillain-Barre syndrome in adults as well as an increase in fetal abnormalities, including placental insufficiency, microcephaly, making ZIKV infection a global health crisis by the World Health Organization [5][6][7][8][9][10]. Additionally, ZIKV can be transmitted by sexual, blood-borne and maternal-fetal routes [11][12][13], and male infertility has been reported in mouse and human studies [14][15][16].
Studies from mouse models and exposed humans have demonstrated a strong adoptive virus-specific T cells response in clearance of ZIKV [17][18][19][20]. CD4+ T cells proliferate rapidly and have been shown to have an essential role in protection against primary ZIKV infection through assisting B cells to generate neutralizing antibodies and producing polyfunctional cytokines in a murine model [17,[21][22][23]. Concomitantly, CD8+ T cells eliminate ZIKV infection by recognizing conserved viral proteins presented by major histocompatibility complex (MHC) class I glycoproteins [24,25], becoming activated and expressing antiviral cytokines, suggesting a protective cytotoxic T-cell response [26][27][28]. Moreover, the depletion of CD4+ and CD8+ T cells or deficiency of T cells in Rag1 −/− mice resulted in higher viral loads after infection of ZIKV, but adoptive transfer of CD8+ T cells from ZIKVinfected mice reversed this effect [18,27,28], thus, indicating a pivotal role of T cells in the anti-ZIKV immunity. However, the immunodominant hierarchy of the ZIKV polyprotein is still largely unknown.
The ZIKV genome contains a single open reading frame encoding a polyprotein consisting of 3410 amino acids, which would be post-translationally processed into structural (C, prM/M, and E) and non-structural (NS1, NS2a, NS2b, NS3, NS4a, NS4b, and NS5) proteins by cellular and viral proteases [29]. The antigenic characteristics of the different ZIKV proteins are not well determined. CD4+ and CD8+ T-cell responses to capsid, envelope proteins and non-structural protein 1 (NS1) have been observed in ZIKV-infected monkeys and humans [30,31]. In mice, several CD8+ T-cell epitopes restricted to H-2 b have been identified, with a significant portion derived from envelope proteins, including E 294-302 [27,28]. Moreover, Wen et al. identified HLA-B*0702 and HLA-A*0101-restricted epitopes in Ifnar1 −/− HLA transgenic mice after ZIKV infection [32]. However, the profile of antigenic peptides spanning the whole ZIKV proteome has not been defined.
In this study, we characterized the immunogenic hierarchy of ZIKV based on the peptides synthesized spanning all the structural and nonstructural proteins. The profile of immunodominant antigens and epitopes was mapped among the H-2 b and H-2 d mice. The CD8+ and CD4+ T cell recognition features of the epitope spectrum were characterized. These findings suggest a clear cell-mediated antigenic profile with epitope hotspots among the whole proteome of ZIKV and have important implications for designing vaccines and evaluating T-cell assays.

Viral Strains and Mice
ZIKV strain ZIKA-SMGC-1 (GenBank accession number: KX266255) [15] was amplified in C6/36 mosquito cells and harvested from cell supernatants 7-10 days after infection. Virus was titrated using baby hamster kidney (BHK)-21 cell-based focus-forming units (FFUs). Specific-pathogen-free wild-type mice C57BL/6 (H-2 b ) and BALB/c (H-2 d ) were purchased (Vital River Co., Ltd. Beijing, China), and bred at Laboratory Animal Center, Chinese Center for Disease Control and Prevention. All of the animals were housed in groups of three to five animals in Eurotype II long clear-transparent plastic cages with autoclaved dust-free sawdust bedding. They were fed a pelleted and extruded mouse diet ad libitum and had unrestricted access to drinking water. The light/dark cycle in the room consisted of 12/12 h with artificial light. All experiments were performed following institutional Animal Care and Use Committee-approved animal protocols. C57BL/6 and BALB/c female mice between 6 and 8 weeks of age were intraperitoneally inoculated (i.p.) with 10 4 focus forming units (FFUs) of ZIKV in a 200 µL volume of 10% FBS/PBS buffer.

Splenocyte Isolation
Isolation of splenocytes was performed as described previously [33]. ZIKV-infected mice were killed 14 days after infection. The spleens were perfused with PBS immediately and disrupted and passed through a 40 µm sieve mechanically. Red blood cells were lysed with RBC lysis solution (Solarbio, Beijing, China) before cryopreservation. Splenocytes were isolated and used for ELISPOT assays and intracellular cytokine staining (ICS) assays.

Peptide Prediction Approaches and Peptide Synthesis
ZIKV polyprotein sequences of Asian lineages (Brazil 2015 strain, GenBank: KU365777) were obtained from the NCBI protein database. Peptides (20-mer) that overlapped by 10 amino acids were designed using online software (www.hiv.lanl.gov, accessed on 12 December 2016) [34]. A total of 364 overlapping polypeptides were designed and synthesized. The 8-to 12-mer epitopes that bound H-2 b and H-2 d were predicted within the 20-mer peptides using the NetMHC 4.0 Server (http://www.cbs.dtu.dk, accessed on 4 July 2017), as previously described [35]. For each mouse allele, the lists of peptides obtained above were sorted by predicted affinity and restricted to the top 1~3. Overlapping 20-mer peptides and 8-to 12-mer epitope candidates were synthesized by Scilight Biotechnology Co., Ltd. (Beijing, China). The purity of the synthesized peptides was 95%, as determined by high-performance liquid chromatography. Peptides were dissolved in DMSO at 20 mg/mL and stored at −20 • C.

ELISPOT Assays
Positive overlapping peptides of the ZIKV polyprotein were detected by 2-D matrix pool analysis and further verified with individual peptides. The 364 overlapping peptides were coded and mixed in 80 matrix peptide pools (X-axis:1-1 to 4-12, Y-axis:1-A to 4-G) Table S1 (Supplementary Materials) and detected using an IFN-γ ELISPOT assay (BD Pharmingen, San Diego, CA, USA) [34,36]; the positive peptides were detected and verified additionally. Briefly, a total of 5 × 10 5 mouse splenocytes was stimulated with matrix peptide pools (with 2 µM of each peptide) or 10 µM of individual peptide in 96-well flat-bottom plates that were coated with anti-IFN-γ mAb. Phorbol-12-myristate-13-acetate (PMA) and ionomycin were used as a positive control, whereas DMSO with the mean concentration in peptide/splenocytes co-incubation well was added into the control well (splenocytes alone). After incubation for 20 h, biotinylated IFN-γ mAb was added, followed by streptavidin-HRP. Then 3-amino-9-ethylcarbazole substrate solution was added to the wells and incubated for 5 to 20 min in the dark at room temperature. Finally, IFN-γ spotforming cells (SFCs) were counted using an ELISPOT reader. Responses are expressed as number of SFCs per 1 × 10 6 splenocytes and were considered positive if the magnitude of the response was SFCs > 40, the magnitude of the positive well should have 2-folds than the control well.

The Distinct Immunogenic Hierarchy of Structural and Nonstructural Proteins of ZIKV in H-2 b and H-2 d Mice
To identify the specific peptides and epitopes of ZIKV in C57BL/6 (H-2 b ) and BALB/c (H-2 d ) mice, we designed 364 overlapping peptides from the full-length sequence (3423 amino acids) of ZIKV. Peptides (20-mer) that overlapped by 10 amino acids were synthesized to ensure that shorter peptides (e.g., 8 to 11-mers) were represented in at least one peptide ( Figure 1A,B). Next, we tested T-cell responses to the ZIKV protein libraries mixed with peptides from proteins using IFN-γ-ELISPOT assays in H-2 b and H-2 d mice infected with ZIKV for 14 days ( Figure 1C). Robust T-cell reactions can be observed in H-2 b mice against E, NS2, NS3, NS4 and NS5 protein libraries, while NS1, NS3, and NS4 protein libraries can induce strong T-cell reactions in H-2 d mice To further validate the profile of the immune reaction to these ZIKV-derived protein libraries, intracellular cytokine staining (ICS) was performed. Splenocytes were stimulated with all eight protein libraries and the frequency of IFN-γ/TNF-α/IL-2-producing CD8+ and CD4+ T cells was determined. E, NS2, NS3 and NS5 protein libraries induced a high frequency of IFN-γ-expressing CD8+ T cells, while, E and NS4 induced a high frequency of IFN-γ-CD4+ T cells in H-2 b mice ( Figure 2A). In H-2 d mice, NS1, NS4 protein libraries induced the highest expression of three cytokines (IFN-γ, IL-2 and TNF-α) in CD8+ T cells, which was similar to the ELISPOT assay results, while NS4 induced the highest IFN-γ-expressing CD4 + T cells (Figure 2A,B). Thus, generally, ZIKV E protein in H-2 b mice, and NS1 and NS4 in H-2 d mice were the dominant antigens for inducing a high frequency of IFN-γ-expressing CD8+ T cells, while NS4 for both mouse alleles dominate the IFN-γexpressing CD4+ T cell responses. These results demonstrated distinct dominance features of ZIKV protein libraries to induce virus-specific CD8+/CD4+ T cells among different mouse alleles ( Figure 1D). To further validate the profile of the immune reaction to these ZIKV-derived protein libraries, intracellular cytokine staining (ICS) was performed. Splenocytes were stimulated with all eight protein libraries and the frequency of IFN-γ/TNF-α/IL-2-producing CD8+ and CD4+ T cells was determined. E, NS2, NS3 and NS5 protein libraries induced a high frequency of IFN-γ-expressing CD8+ T cells, while, E and NS4 induced a high frequency of IFN-γ-CD4+ T cells in H-2 b mice ( Figure 2A). In H-2 d mice, NS1, NS4 protein libraries induced the highest expression of three cytokines (IFN-γ, IL-2 and TNF-α) in CD8+ T cells, which was similar to the ELISPOT assay results, while NS4 induced the highest IFN-γ-expressing CD4 + T cells (Figure 2A

The Profile Mapping of Antigenic Peptides across the Whole ZIKV Polyprotein in Mice
To verify the map of the T-cell response to ZIKV, all 364 peptides spanning the ZIKV proteome were tested by IFN-γ-ELISPOT assays using matrix peptide pools in ZIKV-in-

The CD8+ and CD4+ T Cell Recognition Features of the ZIKV Antigens
To further validate the immune reaction and cytokines induced by these above-positive peptides, splenocytes were stimulated with a positive peptide individually and the IFN-γ, TNF-α, and IL-2 secreting of the antigen-specific CD8+ and CD4+ T cells was detected. For H-2 b mice, 3 peptides presented positive for three cytokines of IFN-γ/TNFα/IL-2 in CD8+ T cells and 13 peptides in CD4+ T cells (Figure 4). Peptides such as E640-659 and NS52955-2973 in CD8+ T cells performed strongly, producing three cytokines. For H-2 d mice, six peptides presented positive for three cytokines in CD8+ T cells and three peptides in CD4+ T cells ( Figure S1). Peptides such as NS11054-1071 and NS42349-2367 performed strongly, with three cytokines producing in CD8+ T cells. Other peptides performed immune activation with production of two or individual cytokines. Taken together, these results demonstrate a distinct CD8+ and CD4+ T cell recognition of the epitope spectrum of ZIKV.

The CD8+ and CD4+ T Cell Recognition Features of the ZIKV Antigens
To further validate the immune reaction and cytokines induced by these above-positive peptides, splenocytes were stimulated with a positive peptide individually and the IFN-γ, TNF-α, and IL-2 secreting of the antigen-specific CD8+ and CD4+ T cells was detected. For H-2 b mice, 3 peptides presented positive for three cytokines of IFN-γ/TNF-α/IL-2 in CD8+ T cells and 13 peptides in CD4+ T cells (Figure 4). Peptides such as E 640-659 and NS5 2955-2973 in CD8+ T cells performed strongly, producing three cytokines. For H-2 d mice, six peptides presented positive for three cytokines in CD8+ T cells and three peptides in CD4+ T cells ( Figure S1). Peptides such as NS1 1054-1071 and NS4 2349-2367 performed strongly, with three cytokines producing in CD8+ T cells. Other peptides performed immune activation with production of two or individual cytokines. Taken together, these results demonstrate a distinct CD8+ and CD4+ T cell recognition of the epitope spectrum of ZIKV.

The Immunodominant Hotspots of ZIKV Recognized by CD8+ T Cells in Mice
To further identify the exact short epitopes (8-11 amino acids) recognized by CD8+ T cells within the overlapping 20-mer peptides that tested positive in the screening, we predicted the potential short CD8+ T cell epitopes through the binding motif of H-2 class I molecules (D b , K b , D d and K d ). A total of 102 short epitope candidates were predicted, 45 were specific for H-2D b , 33 for H-2K b , 6 for H-2D d and 18 for H-2K d . Through the IFN-γ-ELISPOT using the splenocytes from mice infected with ZIKV, a total of 20 H-2D b ,15 H-2K b , 2 H-2D d , 12 H-2K d and 2 H2-I restricted epitopes were identified (Tables 1 and 2). For H-2 b mice, the positive epitopes distributed among prM (3), E (9), NS1 (1), NS2 (4), NS3 (5), NS4 (7) and NS5 (6) ( Figure 5A); for H-2 d mice, the positive epitopes distributed Wild-type C57BL/6 (H-2 b ) mice were infected with 10 4 FFU of ZIKV, splenocytes were harvested at 14 d.p.i. and stimulated with above-positive peptides to assess cytokines production by ICS. The percentages and heat map analysis of IFN-γ, TNF-α and IL-2 produced in CD8+/CD4+T cells in H-2 b and H-2 d mice (n = 3 per peptide). Dashed lines between red and yellow are weakly positive, beyond red are strongly positive.

Discussion
C57BL/6 and BALB/c mice models are widely used for the pathogenesis study of ZIKV infection and vaccine development [27,28,38]. Yu, et al. compared the neurological manifestation for Zika virus infection in C57BL/6, Kunming, and the BALB/c mouse model, and found C57BL/6 owned the highest susceptibility and pathogenicity to the nervous system, while BALB/c associated with similar ocular findings to clinical cases [36]. Additionally, the strain of two mice had a different immune responses preponderance, Th1 immune response and IFN-γ production are dominant for C57BL/6, while Balb/C triggers more of the Th2 immune response and humoral response [37]. The difference in the T-cell response could be due to the fact that the MHC I locus of Balb/c mice is H-2 d , while C57BL/6 is H-2 b [38].

Discussion
C57BL/6 and BALB/c mice models are widely used for the pathogenesis study of ZIKV infection and vaccine development [27,28,38]. Yu, et al. compared the neurological manifestation for Zika virus infection in C57BL/6, Kunming, and the BALB/c mouse model, and found C57BL/6 owned the highest susceptibility and pathogenicity to the nervous system, while BALB/c associated with similar ocular findings to clinical cases [36]. Additionally, the strain of two mice had a different immune responses preponderance, Th1 immune response and IFN-γ production are dominant for C57BL/6, while Balb/C triggers more of the Th2 immune response and humoral response [37]. The difference in the T-cell response could be due to the fact that the MHC I locus of Balb/c mice is H-2 d , while C57BL/6 is H-2 b [38].
Here, we developed a whole genome peptide library of ZIKV to investigate the overall antigen-specific T-cell-mediated immunity in wild-type model mice (C57BL/6 and BALB/c). Previous studies indicated that DENV (Dengue virus) dominant epitopes were within NS3, NS4B, and NS5 [39,40], whereas the major T-cell antigens of HCV (Hepatitis C virus) were located in NS3, NS4A and NS5 [41][42][43]. However, only a few studies have demonstrated T-cell epitopes of ZIKV from envelope proteins [28,38]. Our data shows that T-cell response-targeted ZIKV protein profiles in H-2 b and H-2 d mice were obviously different. Both structural and non-structural proteins appeared to be targets of the anti-ZIKV T-cell response in H-2 b mice, with E protein the primary target. However, non-structural proteins (NS1, NS3, NS4) showed a strong T-cell reaction in H-2 d mice.
The difference in the T-cell response to immunodominant proteins (E protein) between ZIKV and other flaviviruses is very interesting. This is mainly possible due to the difference of species or alleles that we mentioned above. Additionally, there were 11/47 peptides from E protein inducing a high frequency of IFN-γ of CD8+ T cells in H-2 b mice, which means shorter immunodominant epitopes of E protein recognized by H-2 b than non-structural protein after ZIKV infection.
We provide a broad map of the T-cell response to ZIKV with identification of 91 and 39 peptides that target all viral proteins in H-2 b and H-2 d mice, respectively. The difference of MHC I locus may affect the recognition of peptides for T cells. The E, NS2, NS3 and NS5 protein induced a high frequency of IFN-γ-expressing CD8+ T cells, while E and NS4 responded to CD4+ T cell. Here we have a systematic analysis of the different activation characteristics of ZIKV proteins in CD8+ and CD4+ T cells with cytokines secreting, the NS4 protein libraries had more immunodominant peptides responding to CD4 subsets, which corresponds to the immune-thermogram analysis. These results demonstrated distinct dominance features of protein libraries to induce virus-specific CD8+/CD4+ T cells.
Moreover, multiple immunodominant epitopes such as E 294-302 recognized by CD8+ T cells in H-2 b mice were highly conserved to other flaviviruses. Previous studies have found that T-cell immunity to ZIKV and DENV induced responses that are cross-reactive with other flaviviruses in both humans and HLA transgenic mice [44]. Peptides and epitopes of ZIKV we identified in C57BL/6 and BALB/c mice were important for understanding the characterization of ZIKV cross-protective immunity.
Among the positive peptides in H-2 b and H-2 d mice, respectively, the dominant epitopes of E 283-302, NS1 796-815 , NS4 2130-2149 , NS5 2519-2536 and NS4 2387-2406 were located at the junction of proteins. ZIKV, in the same way as like other flaviviruses, encodes a single polyprotein that is cleaved co-and post-translationally by cellular and viral proteases [45]. Identification of CD8+ T cell epitopes through proteasome cleavage site predictions reveals peptides that can bind to major histocompatibility complex (MHC I) molecules; the C-terminus of peptides presented by MHC I molecules result from proteasome cleavage [46,47]. It is possible that the cleavage sites of adjacent proteins are more susceptible to the protease; therefore, the processed epitopes are abundant for presentation by H-2 molecules and recognized by T cells on the surface of the flavivirus-infected cells.

Conclusions
In summary, our current study characterizes the mouse allele-dependent immune hierarchy against the whole ZIKV proteome, broaden the whole map, and draw the hotspots of the CD8+ T cell and CD4+ T cell epitope recognition profile of the virus. Our results serve to understand the T-cell immunogenic feature of ZIKV and may shed light on vaccine development.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/v14112332/s1. Figure S1: Peptides immune-thermogram analysis of CD8+ /CD4+ T cell in H-2b and H-2d mice.; Table S1: 2-D matrix pool.  Data Availability Statement: All data required to interpret the data are provided in the main document or the Supplement Materials. Further data are available from the corresponding author upon reasonable request.