Cross-Protection Against Zika Virus Infection Conferred by a Live Attenuated Japanese Encephalitis SA14-14-2 Vaccine

Zika virus (ZIKV) and Japanese encephalitis virus (JEV) are closely related mosquito-borne flaviviruses. Japanese encephalitis (JE) vaccine SA14-14-2 has been in the Chinese national Expanded Program on Immunization since 2007. The recent recognition of severe disease syndromes associated with ZIKV, and the identification of ZIKV from mosquitoes in China, prompts an urgent need to investigate the potential interaction between the two. In this study, we showed that SA14-14-2 is protective against ZIKV infection in mice. JE vaccine SA14-14-2 triggered both Th1 and Th2 cross-reactive immune responses to ZIKV; however, it was cellular immunity that predominantly mediated cross-protection against ZIKV infection. Passive transfer of immune sera did not result in significant cross-protection, but did mediate antibody dependent enhancement in vitro, though this did not have an adverse impact on survival. This study suggests that SA14-14-2 vaccine can protect against ZIKV through a cross-reactive T cell response. This is vital information in terms of ZIKV prevention or precaution in those ZIKV-affected regions where JEV circulates or SA14-14-2 is in widespread use, and opens a promising avenue into developing a novel bivalent vaccine against both ZIKV and JEV. Importance Japanese encephalitis is a controllable disease in many countries in Asia, especially in China, where many people have Japanese encephalitis virus (JEV) immunity due to extensive JEV vaccination campaigns or natural exposure. Live-attenuated SA14-14-2 strain is a safe and effective vaccine recommended by the World Health Organization and has been vaccinated more than 600 million doses since 1989. As the prevalence of Zika virus (ZIKV) and rising risk in above regions, the cross-reactive immune response between these two antigenically closely related flaviviruses, JEV and ZIKV, should also be fully recognized, which is presumed to be based on those ambiguous cross-reactive immunity between dengue virus and ZIKV. In this study, we found that JEV SA14-14-2 vaccine conferred cross-protection against ZIKV challenge in mice, which is mainly due to cellular immunity rather than neutralizing antibody response. However, specific protective components or cooperation between components warrant to be explored in subsequent experiments. In conclusion, this study can provide important evidence for those who live in JEV-endemic areas and are at risk for ZIKV infection.


116
Adult female mice were used at six weeks of age, and neonatal Ifnar -/mice were used 117 between 24 h and 36 h after birth.

124
Cross-reactive protection against ZIKV in SA14-14-2-immunized Ifnar -/mice 125 Three weeks after the final vaccination, at 15 weeks of age, the Ifnar -/mice were 126 challenged i.p. with a lethal dose of JEV or ZIKV (10 3 PFU for both). Body weight and 127 mortality were monitored daily for 14 consecutive days.

128
Plaque reduction neutralization test (PRNT) 129 Sera were collected from the C57BL/6 mice three weeks after the final vaccination.

130
Neutralizing antibody (nAb) titers were detected by measuring plaque reduction 6 neutralization titer (PRNT) as previously reported (18,19). Serum samples were heated at 132 56 °C for 30 minutes to inactivate complement and then two-fold serially diluted from 1:10 133 to 1:1,280. Diluted sera were mixed 1:1 with virus suspension containing 50 134 plaque-forming units (PFU), and incubated at 37 °C for 1 h. The mixture was transferred 135 to a confluent monolayer of Vero cells in a 24-well plate, and incubated at 37 °C for 136 another 1 h. After washing, the infected Vero cells were overlaid with MEM containing 137 1.2% methylcellulose followed by incubation at 37 °C for five-eight days. Plaques were 138 visualized by crystal violet counter staining and counted. The reciprocal highest serum 139 dilution yielding a 50% reduction of the average number of plaques as compared with the 140 virus infection wells was calculated as the 50% neutralization titer (PRNT 50

150
To detect IgG antibodies and their subclasses, ELISA was performed according to the 151 method as previously described (18). Sera were collected from C57BL/6 mice three 152 weeks after the final vaccination. Each well of 96-well plates was coated with purified viral 153 particles (10 5 PFU) from JEV or ZIKV and then blocked with 3% bovine serum albumin.

166
Antibody-dependent enhancement (ADE) assay 7 Three weeks after the final immunization, sera were collected from C57BL/6 mice. Serially 168 ten-fold-diluted sera were incubated with JEV or ZIKV at a multiplicity of infection of 2 for 169 one hour at 37 °C before adding to THP-1 cells. Samples were incubated for two hours at 170 37 °C and gently shaken every 15 min. Cells were centrifuged and washed three times, 171 followed by resuspension in fresh RPMI-1640 medium and incubated for three days at 172 37 °C. Supernatants were collected and viral titer measured by plaque assay on Vero

219
1B and 1C). Vaccinated mice maintained a normal body weight whereas control mice 220 showed significant weight loss (approximately 16%) and most of them died within 9 days

225
There is a clearly established role for nAb in protection from flaviviruses. Therefore, in 226 order to evaluate cross-reactive neutralization of ZIKV induced by SA14-14-2 vaccination, 227 three weeks after the final immunization, sera were collected from C57BL/6 mice and the 228 nAb titers were assayed by PRNT ( Fig. 2A). As expected, mice administered three doses 229 of JEV SA14-14-2 developed a high level of JEV-specific nAb, with a GMT of 1:494 (Fig.

230
2B). In contrast, JEV SA14-14-2 antisera from the immunized mice failed to neutralize 231 ZIKV, and the PRNT 50 titer of antisera was comparable to that of controls. This result 232 suggests that nAb elicited by JEV SA14-14-2 yielded no neutralizing activity against ZIKV.

233
To further test whether JEV SA14-14-2 antisera would cross-protect mice against ZIKV 234 challenge in vitro, for example by a mechanism distinct from neutralization, immune sera 235 were harvested from SA14-14-2-vaccinated C57BL/6 mice on week three after the final 236 immunization. A serum/virus (JEV or ZIKV) mixture was prepared and inoculated i.c. into 237 naïve neonatal C57BL/6 mice ( Fig. 2A). Control sera from C57BL/6 mice injected with 238 PBS were included in the same experiment. SA14-14-2-immune serum protected 239 neonatal mice (8/8) from JEV challenge, after which the mice exhbited steady growth and 240 development, manifesting a normal and continuous weight increase, reaching 15.4 g at 9 the end of observation ( Fig. 2C and D). In contrast, none of the neonatal mice (0/8) 242 receiving non-immune sera survived JEV challenge, and exhibited severe growth delay, 243 with an endpoint weight of only 2.0 g. ZIKV challenge was less pathogenic in this model 244 than JEV, with control mice (ZIKV infected receiving no sera) developing subnormally but 245 with indistinctive body weight change (Fig. 2C). No effect of SA14-14-2-immune serum 246 could be detected in this experiment, the terminal weights of the mice receiving JEV 247 SA14-14-2-immune serum and control mice were 9.8 g and 9.1 g, respectively. Survival in 248 the two groups was also the same, 10.0% (1/10) in mice incoculated with mixture of 249 SA14-14-2-immune sera and ZIKV and 8.3% in mice injected with control mixture (1/12,

257
Whilst we could detect no protective role of cross-reactive antibody against ZIKV infection 258 from SA14-14-2-immune sera, we sought to determine whether there was any binding to 259 ZIKV, as cross-reactive, binding but non-neutralizing antibodies have been described (23).

260
Thus, to determine the presence of JEV-specific and ZIKV-cross-reactive IgG antibody 261 and its subclasses, including IgG1, IgG2a, IgG2b, and IgG3, induced by 262 SA14-14-2-vaccination, sera were collected three weeks after the last immunization from 263 C57BL/6 mice and analyzed by ELISA (Fig. 3A). Levels of both JEV-specific and 264 ZIKV-cross-reactive IgG antibodies were higher in the sera of immunized mice than those    The effect is common among flaviviruses (25). Therefore, in order to determine whether 279 SA14-14-2-immune sera could promote ADE of ZIKV infectivity in vitro, we exposed 280 FcγRI/II-bearing cell line THP-1 to ZIKV in the presence or absence of 281 SA14-14-2-immune sera. We observed dose dependent enhancement of infection from a 282 dilution of 1:100, which peaked at 1:10,000 dilution, with ZIKV infection enhancement up 283 to 79.7-fold (Fig. 3D). In contrast, sera from control mice did not significantly enhance the 284 infectivity of ZIKV, although modest enhancement at a dilution of 1:10,000, likely to due to 285 non-specific effect. Meanwhile, when THP-1 cells were infected in the presence of the 286 SA14-14-2-immune sera, we found that they also yielded a 17.0-fold greater infection of 287 JEV at a dilution of 1:10,000 than those in the absence of sera (Fig. 3D). This homotypical 288 enhancement is most likely the result of sub-or non-neutralizing titer of serum dilution. As

293
Having provided indirect evidence that both Th1 and Th2 responses were made following 294 SA14-14-2 vaccination, and that these responses could cross-react with ZIKV, albeit not 295 with protective nAb, we sought to determine the nature of the T helper response after 296 SA14-14-2-vaccination. Three weeks after the third and final immunization, splenocytes 297 were collected from the C57BL/6 mice. The levels of splenocyte-derived IL-2, IL-4, and 298 IFN-γ were determined by ELISPOT assay (Fig. 4). Notably, when pulsed with ZIKV 299 antigen, splenocytes from SA-14-14-2-immunized mice responded by making all three 300 cytokines, although the levels were lower when compared with responses of 301 SA-14-14-2-immune splenocytes upon stimulation with JEV antigen. IL-2 and IFN-γ are 302 predominant markers of the Th1 response, IL-4 expression is defined as a marker of the 303 Th2 response. The results indicated that either the Th1-or the Th2-type cross-reactive 304 immune responses against ZIKV was evoked by administration of the JEV SA-14-14-2 305 vaccine.

307
Although multiple cytokines can be produced under ZIKV antigen stimulation, it was 308 unclear whether cellular immunity was indispensible or essential for the in vivo 309 cross-reactive protection. Therefore, adoptive transfer of immune splenocytes from 310 SA14-14-2-immunized C57BL/6 mice into naïve Ifnar -/recipient mice was performed 311 three weeks after the final immunization, followed by viral challenge with a lethal dose of 312 JEV or ZIKV (Fig. 5A). After JEV challenge, mice receiving splenocytes derived from 313 control group showed marked and up to 26.7% body weight loss, all mice died (0/6) within 314 nine days, whereas 100% (6/6) of mice receiving SA14-14-2-immune splenic lymphocytes 315 11 survived without obvious weight change ( Fig. 5B and C), suggesting that a protective 316 prototype of splenocytes activated by SA14-14-2 was successfully established.

Discussion 324
It is becoming increasingly apparent that the pre-existing immunity triggered by primary

338
Previously, we found that live attenuated JE vaccine SA-14-14-2 conferred 339 cross-reactive nAbs which contributed to the cross-protection against DENV challenge 340 (16). In contrast to this result, in this study, mice immunized with SA14-14-2 showed a 341 cross-reactive IgG antibody response to ZIKV without the presence of neutralizing activity 342 (Figs. 2B and 3B). Although the ADE in SA14-14-2-immune sera was detected in vitro ( Fig.   343 3D), we found no evidence that this resulted in a harmful effect; indeed, in subsequent 344 ZIKV challenge, SA14-14-2 vaccination slightly lengthened the median survival (Fig. 2D).

345
This is an important result, suggesting that cross-reactive antibodies from SA14-14-2 346 vaccination are not pathogenic in vivo, because many people in Asia have received 347 inactivated JE vaccine which will generate anti-JEV nAb but may not contain many of the 348 cross-reactive T cell epitopes, which lie in the NS proteins (28). Here, we use neonatal 349 C57BL/6 mice instead of neonatal Ifnar -/mice because the former is susceptible to ZIKV 350 and can mimic the signs (18). Although SA14-14-2 vaccine triggered both Th1 and Th2 351 responses (Figs. 4B and 3C), adoptive splenocyte transfer was superior to serum transfer 12 in protection against ZIKV infection, implying a strong correlation between 353 SA14-14-2-induced cellular immunity and the ZIKV-cross-protective capacity. However, 354 the limitation in this study is that we did not elucidate the functional components which are 355 cross-protective by further sorting of splenocytes for adoptive transfer, such as purified T 356 cells or their subpopulations (CD8 + or CD4 + alone), or cross-reactive memory B cells.

357
Furthermore, we did not determine whether the cross-reactive nAb response was present 358 in recipient mice transferred with SA14-14-2-immune splenocytes, although this is unlikely 359 to change our primary conclusions.

360
Traditionally, based on cross-neutralizing activity, flaviviruses have been subdivided 361 into distinct serocomplexes. Cross neutralization between different serocomplexes is 362 usually not observed (4). The flavivirus E protein is the principal antigen against which the 363 nAb response is directed. The extent of cross-neutralization correlates with the amino acid 364 sequence identity of E protein: when the sequence identity in E protein is less than 40%, 365 cross neutralization is lost (29). JEV and ZIKV belong to distinct serocomplexes, although 366 the homology of the E protein amino acid sequence between the JEV SA14-14-2 strain 367 and the ZIKV SMGC-1 strain was 53.4%, the distinct epitopes in E proteins of JEV and 368 ZIKV within different flaviviruses that dominate antibody responses are presumably 369 responsible for the unavailable cross-neutralization (30). For the sequence homology of 370 the E protein, ZIKV is more closely related to the DENV than to the JEV serocomplex (4).

371
However, most B cell epitopes are conformational (31), and therefore, sequence 372 homology may not fully reflect "relatedness" as measured by cross-reactive humoral 373 responses. Interestingly, one structural model of the relatedness of flavivirus surface 374 topology in fact placed ZIKV and JEV closer to each other than either was to DENV (32).

375
Nevertheless, despite this structural similarity, this did not account for protection in the 376 model we describe here.

377
In contrast, NS proteins among flaviviruses are more conserved with up to 68.0% 378 identity than structural proteins. Weiskopf et al. indicated that, following heterologous 379 DENV infection, memory CD8 + T cells expanded that recognized conserved NS proteins 380 (33). In fact, NS3 and NS5 represent the main targets of the CD8 + T cell response to 381 flaviviruses (34, 35). One study reported more cross-reactive T cell responses to full 382 length NS3 helicase, because of higher sequence homology, than that to the protease 383 region alone (36). Also, a homologous analysis based on the NS5 protein would place 384 ZIKV closer to the JEV serocomplex than to DENVs (37). In our study, we hypothesise 385 that it is the cross-reactive response to shared T cell epitopes in the NS proteins that 386 contributes to protection mediated by adoptive transfer of immune splenocytes.

602
to naïve adult Ifnar -/mice, and one day later mice were challenged with a lethal dose of either 603 JEV or ZIKV. As a control, splenic lymphocytes from PBS-treated mice were transferred to 604 naïve Ifnar -/mice prior to challenge. Results were evaluated for (B) body weight change and 605 (C) survival rate of mice 14 consecutive days post challenge (n = 6). Mice exhibiting more than 606 25% loss in weight were humanely euthanized for ethical reasons. Each experiment was 607 independently repeated three times. Results are expressed as mean +/-SD. * P < 0.05; *** P 608 < 0.001.