Japanese Encephalitis Vaccine Generates Cross-Reactive Memory T Cell Responses to Zika Virus in Humans

Objective Zika virus (ZIKV) and Japanese encephalitis virus (JEV) are mosquito-borne flaviviruses with sequence homology. ZIKV circulates in some regions where JEV also circulates, or where JE vaccination is used. Cross-immunity between flaviviruses exists, but the precise mechanisms remain unclear. We previously demonstrated that T cell immunity induced by the live-attenuated Japanese encephalitis (JE) SA14-14-2 vaccine conferred protective immunity against ZIKV infection in mice, which could even bypass antibody-dependent enhancement. However, the role of T cell immune, especially memory T cell subsets, in cross-reactive immune responses between JE vaccine and ZIKV in humans has not been reported. Methods We examined central and effector memory CD4+ and CD8+ T cell (TCM and TEM) responses (including degranulation, cytokines, and chemokines) in the presence of JEV and ZIKV, respectively, by using qualified peripheral blood mononuclear cell samples from 18 children who had recently received a two-dose course of JE vaccine SA14-14-2 as well as seven children without JE vaccination. Results Cross-reactive CD8+ TCM in response to ZIKV was characterized by secretion of IFN-γ, whereas CD8+ TEM did not show significant upregulation of functional factors. In the presence of ZIKV, IFN-γ and TNF-α expression was upregulated by CD4+ TEM, and the expression signature of CD4+ TCM is more cytotoxic potential. Conclusions We profiled the cross-reactive memory T cell responses to ZIKV in JE vaccine recipients. These data will provide evidence for the mechanism of cross-reactive memory T cell immune responses between JEV and ZIKV and a more refined view of bivalent vaccine design strategy.

some biological characteristics with ZIKV, which mostly causes asymptomatic infection or mild disease. However, JEV infection can also progress to acute Japanese encephalitis (JE), with a fatality rate of 20-30%, and 30-50% of recovered patients have sequelae [6]. Compared with the homology between DENV and ZIKV, JEV is more closely related to ZIKV [7,8]. On average, JEV shares a 56.1% protein sequence identity with ZIKV. JEV is widely distributed in the Asia-Pacifc region, in many countries in East, South, and Southeast Asia [9]. Among them, China was once the most afected country by JEV and one of the earliest countries to initiate JE-vaccination program [10].
Cross-reactive T cell responses induced by prior favivirus exposure or vaccination to heterogeneous faviviruses remains are widely reported between faviviruses [11]. Unlike antibodies that can both prevent and enhance the subsequent infection with heterogeneous faviviruses, the efect of cross-reactive T cell responses may be more inclined to protect against secondary infection [12][13][14][15]. With the ZIKV pandemic, the role of immunodominant protection of cross-reactive CD4 + and CD8 + T cells induced by primary DENV infection in secondary ZIKV infection has been described [13,16,17]. However, the degree of crossreactivity and protective potential is infuenced by factors such as the degree of homology, the sequence of infection, and the interval between primary and secondary infection. JEV has a much wider geographic range than DENV, previously for disease and now for vaccination, in China [18]. Since the implementation of the national Expanded Program on Immunization (EPI) in China in 2008, almost all Chinese people have been immunized against JE [19,20], which prompted researchers to focus on the cross-reactive immune response between JEV and ZIKV. We and other groups have previously characterized the cross-reactive immune response between the two viruses. Tese studies suggest that the cross-protection is mainly conferred by the JEV-induced T cell response [15,21]. Tese conclusions are mostly drawn from experiments in mice. ZIKV-specifc cytotoxic CD8 + T cells can efectively suppress ZIKV infection [22]. In a model of CD8 + T cell adoptive transfer in mice, JE SA14-14-2-vaccination-induced CD8 + T cells can bypass or resist the ADE-mediated by cross-reactive antibodies, biasing the pathogenesis protection balance in global ZIKV infection in favor of protection [21]. Te CD4 + T cells evoked by the JE-vaccination do not serve as the most dominant protective components but trigger T1/T2 cytokine generation through recognizing conserved epitopes [23]. Amplifed cross-reactive clones boosted subsequent ZIKV vaccine responses, resulting in a higher degree of virus clearance [23]. When mapping the DENV-ZIKV crossreactive CD4 + T cell response, it was observed that T1type cytokines played a more prominent role in inhibiting ZIKV replication [13].
Te live-attenuated vaccine SA14-14-2 and the inactivated vaccine (JE-VC/IXIARO) and chimeric vaccine (ChimeriVax-JE) derived from SA14-14-2 strain are widely used and studied worldwide [24,25]. Given the potential for T cells to mediate a cross-protective response, coupled with the fact that ZIKV is still circulating or potentially at risk of spreading [26], understanding the cross-reactive T cell response in children vaccinated with SA14-14-2 against ZIKV will be critical in assessing its potential to protect against ZIKV infection. Such data will also provide important clues for the development of a bivalent vaccine against JEV/ZIKV aimed at inducing a robust T cell response, with good safety profle (avoiding ADE responses) and an efective T cell response. Terefore, in this study, we used peripheral blood mononuclear cell (PBMC) samples collected from children who were vaccinated with two doses of JE vaccine SA14-14-2 to detect cross-reactive central memory (T CM ) and efector memory (T EM ) to ZIKV among CD4 + and CD8 + T cell memory T lymphocyte responses including cytokine secretion and degranulation upon ZIKV antigen stimulation, respectively [27][28][29].

Ethical Approval.
Written informed consent in Chinese was obtained from all guardians of vaccinated children before enrollment, and the ethical approval was given by Beijing Children's Hospital, Capital Medical University (approval number: 2020-k-85). All procedures performed were in accordance with the Declaration of Helsinki. Te study was explained in detail, and a section of the consent granted the investigators' permission for possible future use of the serum and PBMC samples.

Study Cohort.
In total, we took peripheral venous blood samples and separated sera and PBMCs from 18 apparently healthy children (2 years old) who had previously received a prime and boost vaccination with live-attenuated JE SA14-14-2 vaccine for less than half a year from Jan through Feb 2022. Seven unvaccinated children's (6 months old) PBMCs were used as system controls. None of the subjects had visited an area where ZIKV was endemic and had no history of seeking medical attention for symptomatic ZIKV infection. Te sex and age of study individuals are shown in Table S1. Before we analyzed the induction of JEV-specifc and ZIKV cross-reactive CD4 + or CD8 + memory T cells among JE vaccinated individuals, hemogram parameters were analyzed in an automatic analyzer (Lifotronic Technology Co., Ltd.) within one hour after the blood samples were taken. Tere were no individuals with elevated Creactive protein above the threshold. Seroconversion was confrmed by both enzyme-linkedimmune-sorbent assay (ELISA) and plaque reduction neutralization test (PRNT).

ELISA.
Te presence of JEV-specifc IgG antibodies was measured by using an indirect ELISA kit (Shanghai B&C Biological Technology, China) according to the manufacturers' instructions, and it was previously described [20,30]. Briefy, serum samples were diluted at 1 : 41 dilution with bufer that goes with the kit. Te diluted test sera and control samples (100 μL/well) were added to each well and incubated at 37°C for half an hour, followed by fve washes. Ten, 100 μL of horseradish peroxidase-conjugated mouse antihuman IgG monoclonal antibody was added to each well at 37°C for half an hour. Te initiation of the peroxidase reaction occurred after incubation at 37°C for 15 min in the dark. Te reaction was then halted by the addition of 50 μL of 2 M sulfuric acid per well. Te result was represented as optical density, which was read at 450 nm using an ELISA plate reader (Termo, USA). Te cut-of value was calculated based on the manufacturers' instructions. Te optical density of recipients' sera greater than the cut-of value was considered positive.

Cell and Lines
Viruses. C6/36 cells and Vero cells were used for virus propagation and PRNT, respectively. Vero cells were cultured in the MEM medium containing 5% fetal bovine serum; C6/36 cells were cultured in the RPMI-1640 medium containing 10% fetal bovine serum. JEV (Beijing-1 strain) and ZIKV (SMGC_1 strain) were propagated in C6/ 36 cells and stored in a −80°C freezer. Te virus was inactivated by UV irradiation for 1 h. Inactivated viral particles were harvested from the culture supernatant of C6/36 cells that had been infected by JEV or ZIKV, concentrated by 8% polyethylene glycol precipitation and then purifed from clarifed extracts by ultracentrifugation at 100, 000 × g for 3 h at 4°C.

PRNT.
Te PRNT is considered as gold standard for detecting neutralizing antibodies (nAbs) against faviviruses after vaccination or natural infection [20]. Seroconversion of nAbs is an indicator that favivirus vaccine-induced immune protection has been successfully established. Heat inactivated sera were two-fold serially diluted from 1 : 10 to 1 : 160. Te diluted serum was mixed with an equal volume of 100 plaque forming units (PFU) of JEV and incubated for 1 h at 37°C. Te mixture was incubated with Vero cells for 1 h. Cells with removal of the inoculum were cultured under the MEM overlay medium and visualized by crystal violet staining. PRNT 50 was defned as the reciprocal of the highest serum dilution that produced a 50% reduction in mean plaque number serum compared to control wells containing virus alone. With reference to the guideline, PRNT 50 titers ≥1 : 10 are considered positive [31].

Statistical Analysis.
Statistical analysis was performed with SPSS Statistics version 17.0 (SPSS Software Inc., USA), and the fgures were made with GraphPad Prism version 6 (GraphPad Software Inc., USA). Te Mann-Whitney U test or Kruskal-Wallis test (for multiple comparisons) were used to compare variables between two groups. Te chi-square test was used to assess diferences in the composition ratio of pluripotent T PF between stimuli. Statistical signifcance was set at * P < 0.05, * * P < 0.01, and * * * P < 0.001. All of the tests were two tailed.

Complete Blood Count, IgG, and nAb Results.
After preliminary testing of blood and isolated serum samples, complete blood count results of the 16 vaccinated children were all within the reference interval. Both IgG binding and nAb antibody measurement against JEV showed seroconversion, following two doses of SA14-14-2 vaccine among these children (Table 1).

Discussion
Te role of T cell-mediated adaptive immune system in controlling viral infection should be of interest [35]. In addition to nAbs, T cells play an important role in host defense against viruses. As well as helping antibody responses, CD4 + T cells also aid in the initiation of cytotoxic T cells, the generation and maintenance of memory CD8 + T cells, as well as direct killing of target cells. CD8 + T cells can clear viruses from infected tissues by killing infected cells.
For the optimal vaccine design, simultaneous activation of Results are expressed as mean ± SD. Diferences between unmatched groups were compared using an unpaired t-test, the Mann-Whitney U test, or the Kruskal-Wallis rank-sum test with Dunn's post hoc test for multiple comparisons. * P < 0.05, * * P < 0.01, and * * * P < 0.001.

Journal of Tropical Medicine 5
CD4 + and CD8 + T cells is an ideal strategy for vaccineinduced cellular immunity. An earlier study reported that DENV-specifc CD8 + and CD4 + T cells could produce IFN-c upon favivirus stimulation and lyse infected target cells [36]. Indeed, evidence accumulated from our group and other group's studies in mouse models suggests that T cells are actually protective against the favivirus infection in both infection and vaccination settings, both in specifc and cross-reactive responses [15,32,37,38]. Te immunization of immunodominant CD8 + T cell epitopes of DENV can improve viral clearance and protection during primary DENV infection [39]. CD8 + T cells can even confer protection from ADE-mediated infection with DENV and ZIKV in mice [21,40,41]. Te protective efect of CD4 + T cells against faviviruses has been clearly demonstrated in a mouse model [42]. Protective and long-lived immunity is closely related to the production of CD4 + T cells [43], which includes cytokine production, recruitment and activation of innate immune cells, enhancement of CD8 + T cell responses, promotion of immune memory, and direct cytotoxicity to infected cells [44]. Although some studies indicate that CD4 + T cells are not required for the control of primary DENV infection, their induction by epitope immunization nevertheless contributes to virus clearance and reduces tissue viral burden [38].
In humans, the exact role of JEV-induced T cells in preventing ZIKV infection and pathogenesis is unclear. We found that JEV-ZIKV cross-reactive T cells were detected in PBMC samples from children vaccinated with JEV, similar to our results in mice, and reported that these cells responded after restimulation in vitro. JEV-ZIKV crossreactive CD8 + T CM is only IFN-c-producing upon ZIKV stimulation, but this cytokine appears to be critical in crossprotection in the mouse model. ZIKV cross-reactive CD8 + T EM did not have detectable potential for cytotoxic and chemotactic activity. Here, in terms of CD4 memory T cells, we showed that among those who received two doses of SA14-14-2 vaccine, peripheral CD4 + T CM and T EM cells were characterized by the expression of three markers following ZIKV stimulation. Normally, T CM are highly sensitive to antigenic stimulation, while the dependence on costimulatory signals is reduced. After homing to the T cell area of secondary lymphoid organs, T CM cells present reactive memory and proliferate rapidly. Tey have almost no effector function but can proliferate stably and diferentiate into efector T cells in the presence of antigen [45]. T CM cells mainly produce IL-2, and a small amount of IFN-c and perforin through T cell receptor signaling [46]. In this study, we observed a similar polyfunctional feature in crossreactive CD4 + T CM to that in the JEV group itself, with a high level of CD107a, IFN-c, and TNF-α, indicating an important role for CD4 + T CM in cross-reactive T cell responses. Cross-reactive CD4 + T EM mainly expresses the markers IFN-c, TNF-α, and IL-2. Studies have shown that immunity generated by faviviruses sharing the CD4 + T cell epitope promotes protection during subsequent heterologous infection [12], which is speculated to be mediated by the NS3 protein [47,48]. T cells express two or more of the above fve markers, namely, polyfunctional T cells (T PF ). We measured the coexpression of more than two markers; however, we found that the frequency of induction of ZIKV cross-reactive T PF by JEV-vaccination was low. We detected two or more cytokine repertoires only in CD4 + T CM and T EM but not in CD8 + T CM and T EM . In the CD4 + T CM of JEV-vaccinated individuals, the frequency of the IFN + TNF + population was 0.02%, 0.01%, and 0.01% in the JEV-specifc, ZIKV crossreactive, and unstimulated groups, respectively, without diferences across groups; in CD4 + T EM , the frequency of IFN + TNF + population was 0.04%, 0.01%, and 0% in these three groups, respectively.
It should be noted that existing anti-JEV antibody tests cannot completely rule out isolated ZIKV infection, as the available kits do have partial cross-reactivity to ZIKV in specifcity, albeit at a very low level. In addition, the cases of latent infection with ZIKV have been detected in the population of Guangxi Province, a border province in southern China [26], bringing some uncertainty to the immune background of the study individuals in this study. However, we took into account the following three points: (1) the existing reported local cases of ZIKV infection were in border provinces in southern China but not yet prevalent in northern China, and these subjects did not travel outside of China in those ZIKV endemic areas; (2) they did not travel to the Chinese provinces (Yunnan and Guizhou) where ZIKV was detected in wild mosquitoes but no domestic ZIKV cases were reported; and (3) theoretically, it is unlikely that the level of T cell immune response caused by primary infection with ZIKV is lower than that of the cross-reaction elicited by JEV vaccination. Moreover, the COVID-19 pandemic poses considerable difculties in the availability of larger sample sizes; thus, larger sample sizes would be benefcial to the frmness of the aforementioned conclusions.
Two issues to be considered in future investigations are explained. (1) To ensure that immune responses restricted by diferent HLA alleles and diferent species of JEV vaccines are adequately represented in JEV-ZIKV cross-reactive T cell response studies, the subjects of this study were all individuals vaccinated with the live-attenuated SA14-14-2 vaccine, but did not include individuals vaccinated with inactivated vaccines or recombinant chimeric vaccines, whose immune characteristics were diferent [49]. Results are expressed as mean ± SD. Diferences between unmatched groups were compared using an unpaired t-test, the Mann-Whitney U test, or the Kruskal-Wallis rank-sum test with Dunn's post hoc test for multiple comparisons. * P < 0.05, * * P < 0.01, and * * * P < 0.001.
needs to be investigated, which has implications for the vaccine design, cross-reactivity, and immune escape by cross-reactive immune response. T cell epitopes of faviviruses are generally conserved [43], and there are very few instances of T cell epitopes causing acute infections in viral escape (such as those caused by faviviruses). In contrast, viruses that drive the progression of chronic viral infection evade T cell epitope recognition, which is due to a fundamental diference in selection pressure [50]. Given the importance of T cells in cross-reactive immune responses, boosting T cell responses to improve vaccine efcacy is desirable. Tis can be achieved by generating broad favivirus cross-reactive T cell responses by sequential immunization against faviviruses that share T cell epitopes.

Conclusions
In conclusion, we enrich our current understanding of how T cells induced in JEV-vaccinated children cross-react with ZIKV through experiments, and we put forward that the role of JEV-specifc and ZIKV cross-reactive T cells in the infection control may be the strategy for the development of bivalent vaccines that induce dual protection with safety and efcacy. Further expansion of these fndings will signifcantly improve our understanding of T cell function and highlight the potential clinical beneft of incorporating JEV-ZIKV cross-reactive T cell epitopes into experimental vaccine formulations to improve cellular immune responses.

Data Availability
Te data used to support the fndings are included within the article.

Conflicts of Interest
Te authors declare that they have no conficts of interest.