Japanese encephalitis remains a serious public health threat in Asia, where more than two billion people are at risk. The complex JEV zoonotic cycle prevents any direct action to reduce viral transmission; therefore, human infection can only be prevented by vaccination campaigns with children . Apart from the live-attenuated JEV vaccine SA14-14-2 developed for childhood application , no affordable JEV vaccine is recognized for mass vaccination. Several vaccine candidates inducing neutralizing antibodies are under clinical trial and they may meet the immunogenicity and protective capacity requirements after a single dose regimen . However, the use of NS1 protein as an immunogen to reinforce the immune response to JEV structural proteins has been discussed, although the induction mechanisms of protective JEV immunity using a NS1 hexameric antigen form have not been investigated. Drosophila S2 cells have been used for JEV NS1 protein expression and purification from the cell supernatant harvested after serum-free cell culture . Previous recovery of DENV E protein crystals in S2 cells indicated that this system expressed native-like and well-folded glycoproteins . More than 50 mg L–1 of the hexameric form of NS1 glycosylated protein were expressed in the cell culture supernatant , which was shown in this study to contain carbohydrates similar to the native protein produced in infected cells  and was suitable for recombinant vaccine subunit preparation. In this study, two doses of 1 μg of glycosylated hexameric JEV NS1 were used to immunize mice. High antibody titers were detected in the sera of two groups of mice immunized with aqueous or emulsified NS1 with ISA-51-VG formulations. The adjuvant ISA-51-VG was a mix of mineral oil and a surfactant, which has been previously tested for human vaccination . This adjuvant was tested for its capacity to boost the T-cell response against NS1 in immunized mice. The titers of IgG1 subclass were 4–9 times higher than those of IgG2a in both groups of mice, while titers of IgG1 in mice immunized with adjuvanted NS1 were 2–4 times higher when compared with those of mice immunized with aqueous NS1, whereas IgG2a titers remained low. Similarly, 100 times higher IgG1 titers compared with IgG2a were observed in mice that were immunized with E. coli-expressed JEV NS1 protein mixed with Freund’s adjuvant . This difference was presumably due to the use of Freund’s adjuvant, the high antigen dose and the number of antigen injections (three instead of two). In vitro splenocyte stimulation by NS1 elicited T-cell proliferation and IFN-γ secretion, although higher IFN-γ secretion was observed in mice immunized with aqueous NS1. These results suggested that the NS1 proteins are engulfed by antigen presenting cells (APCs) in vivo. The peptides derived from NS1 digestion are subsequently presented by MHC class II molecules to T helper cells. The IgG1 subclass antibody response corresponds to Th2 cell activation, whereas the IgG2a response reflects Th1 cell activation. The IFN-γ produced by Th1 cells inhibits Th2 cell proliferation and IgG1 production . These results indicate that NS1 immunization elicited both Th1 and Th2 cell responses and that ISA-51-VG improved IgG1 production, and reduced Th1 activation and INF-γ expression. Mice inoculated with TBE NS1, which was expressed by a recombinant adenovirus, stimulated in vivo IFN-γ production . A previous study showed that soluble DENV-1 NS1 is a lipoprotein  and was internalized by mouse hepatocytes in vivo and by cultured cell lines in vitro . We observed that purified soluble JEV NS1 attached to many types of cell membranes in vitro and was subsequently internalized (Li and Deubel, unpublished data). This feature may at least partially explain why aqueous NS1 rapidly stimulated the Th1 cell response via APCs, whereas NS1 mixed with ISA-51-VG was slowly released from the emulsion and preferentially activated a Th2 response [31, 34]. The cell-mediated immune response induced by SA14-14-2 immunization and NS1 stimulation was also studied. Compared with the NS1 group, the SA14-14-2 group showed higher cell proliferation SI but lower IFN-γ production. These differences may be due to different epitope presentation from native protein or replicative virus cytokine profiles in response to NS1 and SA14-14-2 immunization, respectively. IL-2 was detected in SA14-14-2, but not in NS1-immunized mice (data not shown), confirming a Th1 stimulation pathway induced by replicative viruses. Production of cytotoxic T lymphocytes (CTLs) is another protective immune response induced by NS1 immunogens issued from replicative viruses . Mice primed with JEV-infected cells  or JEV NS1 expression recombinant viruses  stimulated mice to generate CTLs against JEV. Further study showed that CTLs recognize peptides derived from NS1 and NS3 . One study showed that NS1-expressing DNA immunogen could stimulate CTLs against JEV in mice, but induced low protection . However, we did not expect any CTL response since NS1 was not presented in a replicative system that could stimulate class I antigen presentation.
The protective immune responses elicited against the purified JEV NS1 hexameric protein was investigated by challenging mice with JEV SA14 injected i.n. Immunization with recombinant subunits required several injections, but the mice were not highly susceptible to JEV at the end of the immunization schedule. However, the i.n. method adopted in this study killed more consistently mice aged over three months than i.p. inoculation (Additional file 2: Figure S2A and B). Another approach would have been to infect mice by an intracerebral or an i.p. route , but we felt these techniques did not simulate a natural transmission, as they mechanically break the blood brain barrier. In our study, two doses of 1 μg of hexameric NS1 emulsified in adjuvant provided significant protection (83%) against viral challenge when compared with soluble NS1 (72%), as it reduced morbidity and mortality and increased the survival period after infection. Anti-NS1 immune responses were induced by YFV, DENV, TBEV and WNV immunization with NS1 protein, an NS1-encoding DNA gene, or viral vectors expressing NS1 [13, 14, 18]. Mouse immunization with JEV NS1 produced in Spodoptera frugiperda Sf9 insect cells induced little or no protection , whereas NS1 carried by viral vectors induced an antibody response [26, 37], but low protection [26, 37]. DNA vaccination of mice with the NS1 gene  induced more than 80% protection, whereas DNA vaccination with the NS1-NS2A gene protected only 20% of mice against a JEV challenge . Interestingly, 100 μg of non-folded and non-glycosylated JEV NS1 was produced in E. coli and injected three times to induce protective immunity in 87.5% of vaccinated mice . Recently, a study using protein E fragments or NS1 produced in E. coli and injected at 50 μg seven (i.n.) or five (i.p.) times showed better protection when the proteins were injected i.n. rather than i.p. . Only two much lower doses of the hexameric NS1 form provided protective effects, which may be due to the elicitation of antibodies that bind to conformational epitopes.
Higher antibody titers in the mouse group with a higher survival rate suggested that NS1 had induced an antibody response that might play an important role in preventing JEV infection. In order to verify this hypothesis, different antibody doses induced during mouse immunization were injected into naïve mice. One-month-old mice were used for the adoptive immunity test, so it was possible to apply an i.p. challenge in C3H mice, which caused 100% mortality with a low dose of virus . Passive transfer of 100 μl of anti-NS1 antisera (titrated 1:3000) to one-month-old mice protected 100% from morbidity and death, whereas 30 μl was less protective and 10 μl was not protective, suggesting that an important NS1 function in JEV pathogenesis could be inhibited by IgG-NS1 immunocomplexes. However, anti-NS1 has no neutralizing activity on virus infection in vitro. The antibody-dependent complement-mediated cytolysis of JEV-infected cells associated with anti-NS1 antibodies was demonstrated by in vitro experiments . Another study showed that the depletion of C3 complement components in mice did not affect the anti-NS1 passive protection capacity against YFV . In addition, immunization with a genetic deletion in C5 using an adenovirus expressing NS1 protected mice from a TBEV lethal challenge . These experiments suggested that antibody-dependent complement-mediated cytolysis may play a minor role in antibody-associated immune protection in vivo. However, WNV anti-NS1 MAb through Fc-γ receptor-dependent and C1q-independent pathways could be an alternative protection route involving the scavenging of infected cells by macrophages [23, 24]. In our study, five JEV anti-NS1 MAbs generated in a previous study  that exhibited a high binding affinity against different NS1 epitopes were used in a passive protection test. Two MAbs exhibited no capacity to prevent death, although they significantly extended survival by 2–6 days. MAb 3E10 was directed against a conformational epitope, while MAb 4C4 recognized the NS1 N-terminus and 7C2 recognized the NS1 C-terminus and provided significant delay and prevention against JEV mortality. However, 500 μg of MAbs 7H5 and 8F1 significantly delayed death after JEV infection in a dose-dependent manner and were probably directed against a linear epitope of the C-terminus fraction of NS1 that cross-reacted with DENV NS1 (Tables 4 and 5). Whether those MAbs recognize the same epitope is not known, but they presented different NS1 binding affinities. Interestingly, anti-WNV MAbs that bound to a similar NS1 fragment also provided mice with up to 90% protective immunity from a lethal challenge . Identification of the epitope(s) targeted by the three JEV anti-NS1 MAbs would facilitate further study of the NS1 antibody interaction and its function in anti-flaviviral immunity. Further study using a mixture of those MAbs and their possible role in scavenging NS1-IgG complexes would bring a better understanding of the function of anti-NS1 antibodies in protective immunity.
The mechanism by which NS1 contributes to in vivo flaviviral pathogenesis is largely unknown, which hinders the elucidation of the mechanism whereby anti-NS1 provides protective activity. Anti-NS1 antibodies interact with protein NS1′, an elongated form of NS1 found in flaviviruses of the JEV serogroup , which is involved in neuroinvasion by WNV subtype Kunjin . However, the mechanism involved remains unknown. It is likely that, in addition to interacting with NS1, anti-NS1 antibodies may block the in vivo deleterious activity of NS1′ in WNV and JEV infections. Understanding this interaction would open up new research avenues.
NS1 alone cannot induce a sterilizing immunity, but it contributes to the consolidation of flavivirus neutralizing immunity that is primarily elicited by protein E. Several vaccine candidates are known to provide better protection when NS1 is included in the vaccine preparation [38, 45–50]. A recent study demonstrated the enhancement of anti-E antibody neutralization titer by mouse co-immunization of protein E with NS1, suggesting that NS1 might be usefully added to the viral structural components in a JEV vaccine .
A previous study showed that autoantibodies induced by DENV NS1 recognized coagulation factors including fibrinogen and platelets, as well as integrin/adhesion proteins . Binding of anti-DENV NS1 antibodies to endothelial cells induced inflammation and apoptosis . The possibility that other flaviviral NS1 proteins may elicit autoantibodies when used as components of a vaccine should not be underestimated. Whether JEV NS1 induced autoantibodies needs further study. However, JEV anti-NS1 antibodies did not bind to endothelial cells like DENV anti-NS1 antibodies did  and JEV infection does not cause hemorrhagic manifestations, minimizing the role of NS1-elicited autoantibodies.
In summary, immunization with a hexameric NS1 elicited Th2 and low Th1 cell responses, and induced a partial protective immune response in a mouse model. The ISA-51-VG adjuvant improved the performance of NS1 immunization by contributing to the production of higher antibody titers and increased the mouse survival rate. Passive transfer of sufficient anti-NS1 antibodies provided full protection to mice from lethal JEV challenge. However, anti-NS1 MAbs provided poor protection, although they significantly extended the survival period before death. These results support the hypothesis that anti-NS1 antibodies participate in immune protection against JEV infection.