The Zika virus (ZIKV) is a flavivirus transmitted by Aedes
species mosquitoes. It is a single positive-stranded RNA virus closely related to the yellow-fever virus, dengue virus (DENV), and West Nile virus [1
]. Initially isolated in the Zika forest in Uganda in 1947 [2
], it caused an explosive outbreak for the first time in Yap Island, Federated States of Micronesia in 2007 [3
]. Subsequent outbreaks with higher number of cases occurred in 2013–2014 in French Polynesia and other South Pacific Islands, and, more recently, in the Americas [4
]. Although initially believed to only cause mild, self-limiting disease, a causal relationship between ZIKV and neurological complications, such as Guillain-Barré syndrome or congenital malformations, was established during the 2013 and 2015 outbreaks in French Polynesia and Brazil [10
While mutations in ZIKV genome might have contributed to its increased pathogenicity or explosive spread [9
], one of the most important concerns today is related to the high level of DENV seroprevalence in areas where ZIKV is circulating [16
]. Indeed, recent studies have shown that anti-DENV antibodies may enhance ZIKV infection and increase disease severity [17
]. Given these constraints, and the lack of appropriate treatment for ZIKV infection, there is an urgent need to develop a vaccine against this infectious disease.
While antibodies against the E protein of DENV or ZIKV were shown to be highly cross-reactive, T cells can be cross-reactive or not, depending on the targeted peptides. A low degree of CD4 T-cell cross-reactivity between DENV and ZIKV was indeed observed in human donors immune to one of these viruses [18
], whereas DENV/ZIKV cross-reactive T cells were identified in humans and in DENV-immune mice after challenge with ZIKV [22
]. Considering the sequence identity between DENV and ZIKV for the structural proteins capsid and envelope, and the nonstructural proteins NS3 and NS5, which represent the main targets of DENV-specific CD4 and CD8 T cells, respectively, and the protective role of DENV-specific T cells [25
], efforts are currently directed towards the mapping of T-cell epitopes to design new and more effective vaccines against ZIKV [27
]. Predictions of T-cell antigens have been conducted by modeling potential epitopes from the ZIKV proteome that could bind to different HLA class I or class II alleles [23
], or by analyzing ex vivo T-cell responses in transgenic mice expressing human HLA-B*07:02 and HLA-A*01:01 molecules [23
]. More recently, ZIKV epitopes targeted by CD4 and CD8 T cells have also been identified from human donors living in ZIKV- and DENV-endemic regions [22
]. Quite unexpectedly, while the majority of T-cell responses observed upon infection with DENV were directed against the nonstructural proteins NS3, NS4B, and NS5, ZIKV-specific T cells preferentially recognize structural proteins E, prM, and C, with conserved epitopes between DENV and ZIKV representing the main targets for cross-reactive T cells [22
]. Furthermore, in the light of the recent identification of DENV/ZIKV cross-reactive T cells in human and in different animal models [22
], the precise identification of ZIKV T-cell epitopes in the human that activate these cross-reactive T cells is essential to assess the role of these T cells in ZIKV infection and disease. In the present study, we have identified these epitopes from blood donors with a history of ZIKV-only infection or both DENV and ZIKV infection.
Using PBMCs from Colombian blood donors with previous ZIKV infection, we have first established a detailed map of the distribution of ZIKV T-cell epitopes, by quantifying ex vivo IFN-γ responses against peptides covering the whole ZIKV proteomic sequence by enzyme-linked immunosorbent spot (ELISPOT) assay. Measurement of the magnitude of T-cell responses (mediated by CD4 and/or CD8 T cells) against these peptides allowed us to identify immunodominant epitopes that induce strong responses in donors carrying specific HLA alleles. More specifically, we show that the nonstructural proteins NS1, NS3, and NS5 contain most of the immunodominant epitopes that induce a strong T-cell response. In donors with a history of DENV infection, the strongest T-cell responses were directed against peptides of the NS5 protein with a high level of amino acid identity with the four serotypes of DENV, and some matched previously described DENV CD8+ T-cell epitopes, suggesting the activation of cross-reactive T cells.
The neutralizing or enhancing activity of ZIKV-induced antibodies from the same donors were also analyzed using a flow cytometry-based assay. Results show that ZIKV infection in DENV-immune individuals resulted in increased levels of neutralizing antibodies against ZIKV, in comparison with DENV-naïve individuals, and in reduced potential for antibody enhancement of ZIKV infection, in comparison with DENV-immune ZIKV-naïve individuals. Altogether, these data strongly suggest that a sequential DENV and ZIKV infection may improve the immune protection against ZIKV infection but not against DENV infection.
2. Materials and Methods
2.1. Ethics Statement
Human blood samples were taken after obtaining the informed consent from healthy adult donors from the Fundación Hematológica Colombia (Bogotá D.C., Colombia) in accordance with the tenets of the Declaration of Helsinki. All protocols described in this study were approved by the institutional review board (IRB) of the EL Bosque University (Colombia). Donors were of both sexes and between 20 and 60 years of age. A total of 82 samples were obtained from different ZIKV-endemic areas near Bogotá D.C. (mainly from Villavicencio, Meta) between September and November 2016.
2.2. Human Blood Samples
PBMCs were purified by density gradient centrifugation (Lymphoprep™, Stemcell Technologies, Vancouver, BC, Canada), resuspended in fetal bovine serum (FBS) (Gibco Invitrogen, Carlsbad, CA, USA) containing 10% dimethyl sulfoxide and cryopreserved in liquid nitrogen. Eleven of the 82 blood samples obtained had to be excluded from the study due to poor viability of cells.
2.3. Viruses and Cell Lines
The in vitro assays were conducted using the DENV1 KDH0026A strain (provided by Dr L. Lambrechts, Institut Pasteur, Paris), DENV2 R0712259 strain (provided by Dr. A.-B. Failloux, Institut Pasteur, Paris), DENV3 KDH0010A strain (provided by Dr. L. Lambrechts, Institut Pasteur, Paris), DENV4 VIMFH4 (from the Institut Pasteur Collection), and ZIKV FG15 strain (provided by Dr. D. Rousset, Institut Pasteur, Cayenne). All viruses were grown using the Aedes Albopictus mosquito cell line C6/36 cultured in Leibovitz’s L-15 medium supplemented with 10% fetal bovine serum containing 0.1 mM nonessential amino acids and 1 × tryptose phosphate broth. Vero-E6 and DC-SIGN-expressing U937 cells were kindly provided by Dr M. Flamand and Dr B. Jacquelin (Institut Pasteur, Paris), respectively.
2.4. HLA Typing
Genomic DNA isolated from PBMCs of the study subjects by standard techniques (QIAmp, Qiagen, Hilden, Germany) was used for HLA typing. High-resolution Luminex-based typing for HLA class I and class II molecules (alleles A, B, C, and DRB1, respectively) was used according to the manufacturer’s protocol (Sequence-Specific Oligonucleotides (SSO) typing; Immucor, Inc., Peachtree City, GA, USA).
ZIKV seropositivity was determined using a recombinant antigen-based (ZEDIII antigen) indirect ELISA, as previously described [35
]. Briefly, 96-well plates (Nunc, Life Technologies, Rochester, NY) were coated overnight at 4 °C with 50 ng of antigen in PBS. After washing, 200 μL PBS containing 3% skimmed milk and 0.1% Tween-20 were added for 1 h at 37°. The blocking solution was replaced by 100 μL of plasma diluted 1:500 in PBS containing 1.5% BSA and 0.1% Tween-20, and plates were incubated at 37 °C for 60 min. After 3 washes, bound antibodies were detected with a horseradish peroxidase-conjugated goat antihuman IgG immunoglobulin (Rockland Immunochemicals Inc. Limerick, PA, USA). Following incubation at 37 °C for 1 h and 3 washes, 100 μL of a substrate solution containing TMB (KPL SeraCare, Milford, MA, USA) was added. After 15 min incubation, the optical density (OD) was determined at 650 nm with an automated plate reader (Tecan infinite 200 pro). Each plasma sample was tested in duplicate. Plasma samples, obtained from individuals with positive DENV IgG serology collected before the ZIKV outbreak, were used as negative controls. The cut-off was calculated from the negative controls and was 0.196. DENV seropositivity was determined by indirect ELISA for IgGs (Panbio; Alere Inc., IL, USA), and by capture ELISA for IgM (Tecnosuma, Havana, Cuba) following the manufacturer’s instructions. The quantification of neutralizing and enhancing activities of antibodies against DENV and ZIKV infections was determined using a flow cytometry-based assay, as described previously [36
]. Briefly, 10-fold serial dilutions of plasma samples were incubated at 37 °C for 1 h with either a dilution of virus inducing 7–15% of infection (for neutralization assay) or a dilution of virus inducing between 0.5% and 2% infection (for Antibody-Dependent Enhancement (ADE) Assay). Virus–antibody mixture was then added to 5 × 104
cells (U937-DC-SIGN cells for neutralization of DENV1-4 infection, Vero cells for neutralization of ZIKV infection, or K562 cells for the ADE assay), for 2 h at 37 °C, after which cells were washed 2 times with fresh medium and then incubated for 24 h. The cells were then fixed with 4% paraformaldehyde, stained with 4G2 antibody conjugated to Alexa Fluor™ 488, and the percentage of infected cells was measured by flow cytometry. The neutralization titer of antibodies was expressed as the reciprocal dilution of plasma at which 50% of the virus was inhibited. Plasma samples from donors collected before ZIKV outbreak or from negative samples provided from the Kits to detect anti-DENV antibodies did not reveal any neutralization activity against ZIKV or DENV infection, respectively. For the ADE assay, the peak titer was expressed as the logarithm of reciprocal dilution of plasma at which the percentage of infection was maximal. Following the ELISA and neutralization assays, from the 71 plasma samples selected for this study, a total of 9 samples from ZIKV-immune and DENV-naïve individuals and 11 samples from DENV- and ZIKV-immune (DENV/ZIKV-immune) individuals were further selected for ELISPOT analysis.
2.6. RT-PCR Assays for Detection of DENV and ZIKV
RNA was extracted from plasma using the QIAamp Viral RNA Mini kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. Samples were tested for DENV and ZIKV using a specific nested-PCR assay, as previously described [38
]. Detection of ZIKV was confirmed in 3 out of the 9 plasma samples from ZIKV-immune DENV-naïve donors and in 6 out of the 11 plasma samples from DENV/ZIKV-immune donors, whereas DENV was detected in 2 out of the 11 plasma samples from DENV/ZIKV-immune donors (Table 1
2.7. Viral Sequences
The identical amino acid sequence of 2 Zika viruses from Colombia (KX087102 and KU820897) was used as a reference for the set of overlapping 15-mer peptides. A total of 50 full-length protein-coding DENV sequences from Colombia (serotype 1: 14 sequences; serotype 2: 16 sequences; serotype 3: 13 sequences; serotype 4: 7 sequences) were retrieved from GenBank and used for pairwise sequence identity comparisons.
All peptides were synthesized by Mimotopes (Victoria, Australia). A total of 853 15-mer peptides overlapping by 11 amino acids and 197 9-mer peptides overlapping by 8 amino acids were tested by ELISPOT assay. For the identification of T-cell epitopes, 15-mer peptides were combined into pools of 12 peptides, and individual peptides from the positive pools were tested in a second ELISPOT assay. Following the identification of the positive 15-mer peptides, and according to their HLA class I or class II restriction potential (predicted or shared between at least two donors), 9-mer peptides were synthesized and tested individually.
2.9. Ex Vivo IFN-γ ELISPOT Assay
PBMCs (2 × 105) were incubated in 96-well flat-bottom plates (MSIPS 4510, Millipore, Millipore, Burlington, MA, USA) coated with anti-IFN-γ mAb (clone 1-D1K, Mabtech, Stockholm, Sweden) with 0.2 mL of complete RPMI containing 10% human AB serum with pools of 12 peptides (2 g/mL, final concentration) or individual peptides (1 g/mL final concentration) for 20 h. Following a 20 h incubation at 37 °C, the wells were washed with PBS/0.05% Tween 20 and then incubated with biotinylated anti-IFN-γ mAb (clone 7-B61-, Mabtech, Stockholm, Sweden) for 1 h 30 min. The spots were developed using streptavidin-alkaline phosphatase (Mabtech, Stockholm, Sweden) and BCIP/NBT substrate (Promega, Madison, WI, USA) and counted using an automated ELISPOT reader (Immunospot, Cellular Technology Limited, Cleveland, OH, USA). The number of IFN-γ-producing cells was expressed as spot-forming cells (SFC) relative to 1 × 106 PBMCs. Values were calculated by subtracting the number of spots detected in the nonstimulated control wells. Values were considered positive if they were equal to or greater than 20 spots and at least three times above the means of the unstimulated control wells. As a positive control, cells were stimulated with CEF peptide pool (Mabtech, Stockholm, Sweden).
2.10. Immunogenicity and HLA Restrictions Prediction
The evaluation of binding possibilities of peptides to MHC class I and class II alleles was analyzed using the NetMHCpan3.0 and NetMHCIIpan3.1 servers, respectively [39
All data were analyzed with Prism software version 7.0 (GraphPad Software Inc., La Jolla, CA, USA). Statistical significance was determined using the nonparametric two-tailed Mann–Whitney test to compare two independent groups. Differences were considered significant at p < 0.05.
In this study, using PBMCs from ZIKV-infected human blood donors, we have identified numerous T-cell epitopes specific to ZIKV or shared between DENV and ZIKV. While the DENV-specific T-cell responses are predominantly directed against NS3, NS4B, and NS5, the response against ZIKV mainly targets epitopes in the NS1, NS3, and NS5 proteins. The stronger and broader IFN-γ response against peptides from the NS5 protein observed in donors previously infected with DENV led us to postulate that this region contains more peptides recognized by cross-reactive T cells, whereas the NS1 protein is preferentially targeted by ZIKV-specific T cells, which is consistent with the higher percentage of identity observed between ZIKV and DENV sequences in the NS5 protein, in comparison with the NS1 protein.
Among the epitopes activating cross-reactive T cells, several peptides in the NS5 protein matched the T-cell epitopes recently identified from ZIKV-positive donors, such as the NS52983–06
, and NS54614–75
peptides recognized by CD8 and CD4 T cells, in the context of HLA-B*35:01 and HLA-DRB1*07:01, respectively [22
]. For several epitopes, the 15-mer or 9-mer peptides matched epitopes recently identified in transgenic mice expressing human HLA molecules, or in humans exposed to ZIKV, thus confirming the class I allele restriction for these peptides. This is the case for 15-mer peptide VARVSPFGGLKRLPA inducing a response in a donor expressing the HLA-B*07:02 allele, which contains the C253–5
peptide SPFGGLKRLPA shown to elicit a significant response in HLA-B*07:02 transgenic mice infected with ZIKV [23
]. The same correlations have been established with NS3 (FPDSNSPIM), NS4B (RGSYLAGASLIYTVT), and NS5 (NQMSALEFYSY) peptides that induced a strong response in human donors expressing the HLA-B*07:02 and HLA-A*01:01 alleles, respectively (Table S1
), and in transgenic mice expressing these alleles [23
]. In other cases, the epitopes identified in HLA-B*07:02 and HLA-A*01:01 transgenic mice were also identified in responding donors that nevertheless do not express these alleles, such as the NS32192–33
peptide (Table 2
), and the NS1193–3
or the NS5132–7
peptides (Table 3
and Table S1
), which elicit a response in donors that express neither of the two alleles, HLA-B*07:02 or HLA-A*01:01. For these donors, one possibility could be that the epitope identified in transgenic mice has a higher affinity for a human HLA allele different from the allele expressed by the transgenic mice, or that the 15-mer peptide contains another epitope that binds to a different allele. Binding studies with 9-mer epitopes and HLA class I stabilization assays using TAP-deficient cells should discriminate between these possibilities.
We also reported the identification of several peptides that share common sequences with DENV and are preferentially targeted by cross-reactive T cells, after DENV and ZIKV infection. Among these peptides, the NS52933–07
peptides contain the amino acid sequence HPYRTWAYH that shares seven amino acids with an epitope identified in DENV1-positive or in DENV-positive and ZIKV-positive donors [22
]. Similarly, the NS53253–39
peptide contains the amino acid sequence KPWDVVTGV, which is 67% identical to the epitope KPWDVIPMV identified in individuals infected with DENV1 [46
]. Finally, the strongest T-cell responses in DENV/ZIKV-immune donors were observed with the NS54814–95
peptide or the NS53453–59
and the NS54654–79
peptides (Table 3
), which contain 9-mer epitopes identified previously in DENV-infected individuals [44
] or, more recently, in ZIKV-positive donors, respectively [22
]. Altogether, these data reveal the activation of DENV/ZIKV cross-reactive T cells that dominate the response following sequential DENV and ZIKV infection. Notably, although these cross-reactive peptides exhibit a high degree of sequence identity with DENV and can stimulate a T-cell response after DENV infection, they do not induce a response after primary infection with ZIKV, suggesting that they are immunodominant in the context of DENV but not in the context of ZIKV infection. This result is expected, as the immunodominance of an epitope or its relative abundance depends on the other epitopes expressed by the protein. This is also in agreement with previous observations showing that epitope production correlates with cleavability of flanking residues expressed in the protein sequence [51
Importantly, for these cross-reactive epitopes, the absence of a T-cell response in ZIKV-infected donors is not simply due to the absence of the presenting HLA allele in this population, as most of the alleles expressed in responding DENV/ZIKV-immune donors were also expressed in ZIKV-immune DENV-naïve donors (Table S1
). This is what we observed for the NS5132–7
, and NS55465–60
epitopes, predicted to be strong binders to the HLA-B*35:01 and HLA-B*40:02 alleles, respectively, that were frequently expressed by our ZIKV-immune DENV-naïve donors (Table S1
). Altogether, these results show that, in the case of initial ZIKV infection, there is a preferential recognition of ZIKV-specific epitopes, whereas there is a more frequent and stronger T-cell response against cross-reactive epitopes after sequential heterologous DENV/ZIKV infection. Interestingly, the strong T-cell response observed in DENV/ZIKV-immune donors against these NS5 epitopes relies primarily on donors that express the HLA-B*3501 allele, an allele associated with high-magnitude responses against DENV, and a stronger protection against DENV infection and disease [25
]. As all blood samples were obtained from donors with asymptomatic ZIKV infection history, we cannot relate the strength of the ZIKV-specific T-cell response obtained in HLA-B*35:01 donors to the protection against the disease. Further studies with more subjects with a higher susceptibility to disease following primary ZIKV infection are required to determine whether, as for DENV, there is an HLA-linked protective role for T cells in ZIKV infection. Likewise, it would also be important to compare disease severity in donors having experienced a previous DENV infection or not, to determine whether cross-reactive T cells induced after DENV infection could mediate a better protection against ZIKV infection and disease, as recently suggested in mice [23
]. As both CD4+ and CD8+ T cells were shown to contribute to protection against DENV infection, a comprehensive analysis of MHC class II-restricted response is needed to determine the role of CD4 in ZIKV infection and disease protection. Finally, further phenotypic analyses of ZIKV-specific T cells in asymptomatic or symptomatic donors will help in defining correlates of protection in natural immunity and vaccination against ZIKV infection and disease. It will be particularly important to determine whether, as for DENV-specific T cells, strong responses against ZIKV-specific peptides are more frequent for specific HLA alleles and are associated with multifunctionality [25
From the same individuals, we have also shown that, while a primary DENV infection induces cross-reactive antibodies with enhancing activity against the heterologous DENV serotypes and against ZIKV, a primary ZIKV infection in naïve donors mainly elicits ZIKV-specific antibodies with no enhancing activity against DENV. These results do not coincide with the enhancing activity described in the plasma of one donor infected with ZIKV only [52
]. Apart from the fact that the ADE activity detected in vitro could result from the presence of cross-reactive antibodies induced not exclusively after DENV or ZIKV infection, this discrepancy could be due to the too-low number of ZIKV-immune DENV-naïve donors with low neutralizing-antibody titers (three donors with Neut50
values < 300). Alternatively, as the ADE activity in one donor was revealed from day 48 post-onset of symptoms [52
], and since all donors from our cohort were asymptomatic, it is possible that the samples from ZIKV-immune donors were collected at an early time after infection, prior to the appearance of enhancing activity against ZIKV infection. In agreement with previous studies, we have also observed a higher level of cross-reactive antibodies between DENV and ZIKV in donors who have been recently infected with DENV [17
], as in the case of donors 55, 63, and 66 from group C donors, donors 26 and 53 from group D donors, and donors 20 and 69 from group E donors (Figure 5
). Our study also clearly reveals a significant increase in the titer of neutralizing antibodies against ZIKV infection in ZIKV-immune donors with previous or current DENV infection (donors from groups C, D, and E), in comparison with ZIKV-immune DENV-naïve donors (Figure 3
B). However, while the increased titer of neutralizing antibodies between donors from B and C is still significant, we did not observe any significant difference in the titer of ZIKV-neutralizing antibodies between donors from groups B (with sequential DENV and ZIKV infections) and D (DENV-immune with acute ZIKV infection), or from groups B and E (acute DENV and ZIKV infections). A higher number of donors from groups D and E should allow us to determine whether there is also an increase in the level of ZIKV-neutralizing antibodies in donors with acute DENV and/or ZIKV infection. Strikingly, while ZIKV infection in DENV-immune donors was shown to increase the level of neutralizing antibodies against ZIKV, in comparison with DENV-naïve donors, it decreases the enhancing activity of antibodies against ZIKV, in comparison with donors infected only with DENV. These results are consistent with the inverse correlation observed between the level of neutralizing antibodies and enhancing activity in DENV-immune donors [54
]. Interestingly, while ZIKV infection in DENV-immune donors does not modulate the enhancing potential of antibodies against DENV1-3 infection, it decreases the ADE activity against DENV4 infection in vitro, a result that could be explained by the higher amino acid sequence identity in the E protein between ZIKV and DENV4, in comparison with DENV1-3 [56
Taken together, these results strongly support the activation and clonal expansion of memory B cells induced after sequential DENV and ZIKV infection, as shown recently in DENV1-infected individuals [57
]. These data also show that, while a sequential DENV and ZIKV infection results in increased neutralization potential and decreased enhancing activity against ZIKV, it does not affect the level of neutralizing or enhancing antibodies against DENV, suggesting that ZIKV infection in DENV-immune individuals should not have a significant impact on dengue disease following subsequent DENV infection. However, as a small fraction of individuals infected with ZIKV were only shown to express antibodies with enhancing activity against DENV4, it will be important in the future to analyze in these individuals the impact of ZIKV infection on the outcome of dengue disease following DENV4 infection. In this sense, it is also worth mentioning that the modulation of ADE activity measured in vitro does not necessarily reflect the in vivo situation, given in particular the conflicting results on the impact of pre-existing immunity to DENV on ZIKV pathogenesis [21
]. Finally, in light of the stronger T-cell response and the higher titer in neutralizing antibodies against ZIKV observed in individuals sequentially infected with DENV and ZIKV, it would be important to determine whether antigen-specific T follicular-helper (Tfh) cells could be detected in these individuals, which could account for the activation of memory B cells, as shown recently in PBMC of DENV-infected patients [61
]. In this sense, our study provides a framework for detailed analyses of cell populations potentially linked to the activation of memory B cells with high neutralization potential against DENV and ZIKV.