Introduction

Currently, dengue is the most important arboviral disease in terms of morbidity and mortality. An estimated 100 million dengue infections are caused yearly by the dengue viruses (D1 to D4), resulting in widespread dengue fever (DF) and at least 250,000 cases of dengue hemorrhagic fever/dengue shock syndrome (DHF/DSS) [1]. In spite of decades of effort, there is no dengue vaccine currently licensed, but numerous vaccines candidates are in development and several are in phase 1/2 of clinical trials [2].

A major obstacle to dengue vaccine development is the epidemiological observation that a dengue virus infection only produces protective immunity against the infecting serotype, and that a secondary infection by a heterologous serotype increases the risk of severe dengue (previously known as DHF/DSS). Antibody-dependent enhancement (ADE) of infection [3] has been postulated to explain disease severity during a secondary infection by a different dengue serotype. In addition, an inadequate pre-existing heterologous cellular immune response has also been implicated in dengue immunopathology [4]. Considering the ADE phenomenon, a dengue vaccine must elicit a long-lasting, balanced and protective immune response to each of the four serotypes to avoid severe dengue in the course of a natural infection. Vaccine evaluations are complex and unreliable because the basis of protective immunity against dengue virus is not well understood, and also because there is no animal model of dengue disease [5].

Domain III of the envelope (E) protein of flaviviruses is suggested to be the receptor-recognition and binding domain of the virus and is one of the targets for dengue vaccine development. The functionality of a recombinant fusion protein, P64k-domain III (DIII-P64k) of the E protein, in terms of induction of serotype-specific neutralizing antibodies and protection in mice and monkeys, has been reported [6–11].

Recently, a novel chimeric protein, domain III-capsid of D2 (DIII-C-2), was shown to induce a functional humoral immune response as well as CMI (cell-mediated immunity) against infecting D2, and it conferred significant protection in immunized mice [12]. Here, we evaluate the ability of a tetravalent formulation based on recombinant proteins DIII-P64k (serotypes 1, 3 and 4) and DIII-C-2 to induce a humoral and cellular immune response against the four dengue viruses in immunized mice. Its protective potential against lethal viral challenge was also determined.

Materials and methods

Tetravalent vaccine formulation

Table 1 shows the components of the tetravalent formulation (TF) employed in this study. The methodology used for genetic construction and expression and purification of the recombinant proteins tested here have been published previously [9, 13, 14].

Table 1 Recombinant proteins of dengue virus employed in the tetravalent vaccine formulation
Full size table

Virus strains

The standard strains, D1 Hawaii, D2 New Guinea C (NGC), D3 H87, and D4 H241 (kindly provided by Dr. Robert Shope, University of Texas Medical Branch, Galveston, TX) were grown in suckling mice. Dengue antigens were extracted from infected mouse brains by the sucrose-acetone method. These antigens were employed for evaluating the humoral immune response by enzyme-linked immunosorbent assay (ELISA). A similar preparation from uninfected mouse brains was used as a negative control (mock) [15]. For animal immunization and virus challenge, the same strains of dengue virus were employed. Neutralization assay was performed using cell culture supernatant from baby hamster kidney (BHK) cells infected with D1 Angola (kindly provided by Dr. Robert Shope, University of Texas Medical Branch, Galveston, TX), D2 A15/1981, D3 116/00 and D4 23/00.

A concentrated preparation of virus was used for the in vitro stimulation of mouse splenocytes. Supernatant (100 ml) from Vero cells infected with 106 pfu/ml of D1 Angola, D2 SB8553 and D3 CS81.1 (kindly provided by Dr. M.J. Cardosa, University of Sarawak, Malaysia) and D4 Dominica (kindly provided by Dr. Robert Shope, University of Texas Medical Branch, Galveston, TX), was concentrated by centrifugation at 80 000 x g for 4 h at 4 °C. The pellet containing the virus was resuspended in 1 ml of PBS (Gibco, Paisley, UK). A mock preparation was similarly prepared from the supernatant of uninfected Vero cells.

Mouse immunization

Groups of 64 female BALB/c mice, 7 weeks old, were purchased from the National Center for the Production of Laboratory Animals (CENPALAB, Havana, Cuba) and housed in appropriate animal-care facilities during the experimental period. Three groups of mice were used. One group of mice was immunized by the intraperitoneal route on days 0, 15 and 30 with 80 μg of the TF adjuvated in Al(OH)3. The second group, the negative control group, received 1.44 mg/ml of Al(OH)3. The third group, the positive control group, consisted of four sets of mice, each one receiving one dose (0.5 mL) of infectious virus (D1-D4) without adjuvant (28 animals per group). Mice were bled 15 days after the last dose, and sera were collected for further immunological analysis.

Antiviral antibodies tested by ELISA

An amplified sandwich ELISA system was used to detect anti-dengue-virus antibodies. Polystyrene plates with 96 wells (MICROLON, Greiner Bio-One, Germany) were coated for 2 h at 37 °C with 100 µL of a mixture of anti-dengue human immunoglobulins in coating buffer (0.16 % Na2CO3, 0.29 % NaHCO3, pH 9.5) per well and then blocked with 2 % BSA (bovine serum albumin) in coating buffer for 1 h at 37 °C. Three washes with PBS (0.8 % NaCl, 0.02 % KCl, 0.014 % KH2PO4, and 0.009 % Na2HPO4, pH 7.4) containing 0.05 % Tween 20 (Merck, Germany) (PBS-T) were performed after each reaction step. The plates were incubated overnight at 4 °C with a saturating concentration of the viral antigen (from dengue-infected suckling mouse brains) and the negative control antigen (from uninfected suckling mouse brains) diluted in PBS-T. After a washing step, immune sera (serially diluted in PBS-T) were added and incubated for 1 h at 37 °C. Later, plates were incubated for 1 h at 37 °C with anti-mouse IgG-peroxidase conjugate (Amersham-Pharmacia). After washing, the substrate solution was added. Plates were kept for 30 min at 25 °C and the reaction was stopped with 12.5 % H2SO4. Absorbance was read at 492 nm in a microplate reader (Bio-Rad, USA). The positive cutoff value was set as twice the mean absorbance value of the negative control sera.

Plaque-reduction neutralization technique

The neutralizing antibody titers to the four dengue viruses were determined by plaque-reduction neutralization technique (PRNT) in a baby hamster kidney cell line (BHK-21) as described by Morens et al. [16] with some modifications of Alvarez et al. [17]. The serum dilution that resulted in a 50 % reduction of plaque count, as determined by probit analysis, was considered the endpoint antibody titer.

Cell culture and viral stimulation

Spleen cells from immunized mice and controls were obtained under aseptic conditions. Erythrocytes were lysed by adding 0.83 % NH4Cl solution. Cells were washed twice with PBS containing 2 % fetal bovine serum (FBS) (PAA Laboratories, Ontario, Canada) and resuspended at 2 × 106 cells/mL in RPMI-1640 medium (Sigma Aldrich) supplemented with 100 U of penicillin and 100 μg of streptomycin per mL (Gibco, UK), 2 mM glutamine (Gibco, UK), 5 × 10−5 M 2-mercaptoethanol (Sigma St. Louis, MO) and 5 % FBS. Finally, 2 × 105 cells/well were cultured in 96-well round bottom plates with the viral antigens at an m.o.i of 1 or with a mock preparation. Concanavalin A (ConA) (Sigma St. Louis, MO) was used as a positive control. In all the experiments, three wells were plated for each antigen. After 4 days of culture, culture supernatants were collected and stored at −20 °C.

IFN-γ detection

The concentration of IFN-γ was determined in the supernatants of splenocytes that had been stimulated with each dengue virus antigen. An ELISA using pairs of MAbs (Mabtech Nacía, Sweden) was employed for IFN-γ determination in duplicate. The protocol recommended by the manufacturers was followed with minor modifications.

Animal protection assay

One month after the last immunization, mice were inoculated intracerebrally (i.c.) with 20 μl of a preparation of infectious dengue virus containing 50 median lethal doses. Mice were observed daily for 21 days, and mortality and morbidity were recorded.

The maintenance and care of experimental animals used in this research complied with the Cuban Institute of Health Guidelines for the humane use of laboratory animals.

Statistical analysis

Data were processed using the Graph Pad Prism program (version 5, 2007) using the Kruskal-Wallis non-parametric test with Dunn’s corrections or ANOVA parametric test followed by Newman-Keuls for multiple comparisons in relation to the analysis of normality and variance homogeneity. Data from the protection assay were analyzed by the log-rank test.

Results

Immunogenicity in BALB/c mice

In order to determine the immunogenicity of the tetravalent formulation (TF), 4- to 6-week-old BALB/c mice were immunized. Serum samples from each animal obtained 15 days after the last dose were tested by ELISA to determine the anti-dengue antibody titer. As shown in Fig. 1, sera from mice immunized with TF displayed similar high levels of antiviral antibodies against the four serotypes (1:60,000 to D1, 1:20,000 to D2, 1:140,000 to D3 and 1:70,000 to D4). Nevertheless, the IgG antibody response against D2 was significantly lower than the response against D3 (p<0.05). The positive control groups showed similar levels of antibodies against the four dengue serotypes, although recognition was significantly higher for the homologous serotype for D1 (p<0.001), D2 (p<0.001) and D4 (p<0.05).

Fig. 1
figure 1

Anti-dengue antibody response in mice immunized with the tetravalent formulation (TF) as determined using an ELISA system. The analysis of data was performed using a Kruskal-Wallis non-parametric test with Dunn’s multiple comparison test. Data represent the geometric mean with 95 % confidence intervals. Different letters (a and b) indicate significant differences

Full size image

Mouse sera collected 15 and 30 days after the last TF dose were tested by PRNT (Fig. 2). Sera collected 30 days after the last dose from animals immunized with D1 and D2 showed significantly higher (p<0.05 and p<0.01, respectively) geometric mean neutralizing antibody titers (GMT) than sera collected 15 days after the last dose. Fig. 3 shows the GMT of neutralizing antibodies to the four serotypes in sera collected both from immunized animals and positive controls. At 30 days after the last TF dose, sera showed antibodies that neutralized D1, D2, and D3 virus (1:50, 1:181 and 1:19, respectively), with the highest titers to D2 (p<0.001). A similar analysis in the positive control groups revealed a higher neutralizing antibody response against D2, following by D1. Neutralizing antibodies against D3 and D4 were practically undetectable.

Fig. 2
figure 2

Kinetics of neutralizing antibodies, as determined by PRNT, in sera collected 15 and 30 days after the last dose in mice immunized with the tetravalent formulation (TF). The analysis of data was performed using the Mann-Whitney test. Data represent the geometric mean with 95 % confidence intervals. Different letters (a and b) indicate significant differences

Full size image
Fig. 3
figure 3

Neutralizing antibody response, as determined by PRNT, in sera collected 30 days after the last dose in mice immunized with the tetravalent formulation (TF) and in positive control groups. The analysis of data was performed using a Newman-Keuls multiple comparison test. Data represent the mean with 95 % confidence intervals. Different letters (a, b and c) indicate significant differences

Full size image

Cellular immune response

Culture supernatants of treated splenocytes from immunized animals were tested by ELISA to measure the concentration of IFN-γ (Fig. 4). Compared to the negative control, higher levels of IFN-γ secretion were observed in splenocytes from animals immunized with the TF and stimulated with each dengue serotype (7509.53 ± 3846.37 for D1, 8574.54 ± 2036.06 for D2, 3281.99 ± 1134.49 for D3, and 4885.99 ± 1302.20 for D4). Higher concentrations of IFN-γ were observed in supernatants of treated splenocytes from the positive control groups, with higher figures for D2 and D4.

Fig. 4
figure 4

Concentration of IFN-γ in culture supernatant of dengue-stimulated splenocytes from mice immunized with the tetravalent formulation (TF) and positive control groups as measured by ELISA

Full size image

Mouse protection assay

One month after the last TF dose, mice were inoculated by i.c. injection with live neuroadapted D1, D2, D3, and D4 and observed daily. After 21 days, 40 % of the TF-immunized mice challenged with D1 survived. This level of protection did not differ from the results seen with the negative control group, where there was only one survivor among the mice in this group. In contrast, 100 % of the mice from the D1-immunized positive control group were protected (Fig. 5a). Fifty-six percent of TF-immunized mice were protected after challenge with D2; this protection was statistically significant when compared to the negative control group (p<0.05) but otherwise undistinguishable from the high levels of protection exhibited by the D2 positive control group (Fig. 5b). All mice inoculated with the neuroadapted D3 strain survived, including those in the negative and positive control groups. Since the virus was not neurovirulent enough as to make a significant difference between the positive and negative control groups in terms of mortality, we observed the animals for signs of morbidity. We followed the manifestation of any symptoms of disease, including mild (motor weakness and loss of appetite) and severe ones, among the different groups of animals. The negative control group showed the lowest percentages of completely healthy animals (60 % from day 12 to day 15). In contrast, 100 % of the mice in the TF-immunized group remained completely healthy during the length of the observation period, similar to the positive control group. In this last group, only 1 of 10 mice showed signs of disease after receiving the virus (Fig. 5c). At the end of the observation period, 67 % of animals immunized with TF and challenged with D4 survived. There was no statistically significant difference in mortality between the TF-immunized group and the D4 positive control group Only one mouse survived in the negative control group (p<0.01) (Fig. 5d).

Fig. 5
figure 5

Survival and morbidity curves for mice immunized with the tetravalent formulation (TF) and those challenged with a) D1 lethal virus, b) D2 lethal virus, c) neuroadapted D3 strain, and d) D4 lethal virus. Different letters (a and b) mean significant differences by the log-rank test (number of animals per group = 12)

Full size image

Discussion

A preventive dengue vaccine should provide long-lasting protective immunity to the four dengue serotypes. Some of the challenges of a dengue vaccine are (i) the possible induction of cross-reactive non-neutralizing antibodies and ADE, (ii) stability problems of a tetravalent formulation, (iii) DV-associated reactogenicity, (iv) possible recombination and dissemination by mosquitoes, and (v) that it must be absolutely safe for children and able to be rapidly implemented on a large scale [18].

To address these challenges, we designed a tetravalent formulation combining DIII-P64k recombinant proteins of D1, D3 and D4 with DIII-C-2 designed to induce neutralizing antibodies and CMI against the four dengue virus serotypes. As a proof-of-concept of this strategy, we previously evaluated in mice the recombinant capsid protein of D2 (C2) co-immunized with D4 recombinant protein DIII-P64k (PD24). The animals receiving the PD24–C2 mixture showed the highest levels of antiviral antibodies. Similar results were obtained for IFN-γ secretion levels, and consistently, the percentage of mice surviving after viral challenge was significantly higher for those immunized with the mixture than for those inoculated with PD24 protein alone [19]. In addition, we assessed the immune response elicited by the formulation PD10-C2 (D1 recombinant protein DIII-P64k and recombinant capsid protein of D2) and its protective capacity against D1 and D2. The humoral immune response was mainly directed against D1, while high levels of IFN-γ secretion were detected after stimulation with both D1 and D2. Consistently, animals immunized with the bivalent formulation were significantly protected against challenge with either serotype [20]. These findings demonstrate the adjuvant capacity of the capsid protein and also highlight the potential of these particles for enhancing the immune response, primarily the cellular response, against heterologous recombinant proteins.

Taking these results into account, here, we determined the immunogenicity and protective capacity of a TF containing DIII-C-2 and the recombinant proteins DIII-P64k of D1, D3 and D4. An ELISA using dengue virus as the coating antigen revealed that antibodies induced by the tetravalent antigen do in fact recognize and bind to each one of the dengue viruses. Most importantly, we found that the antibodies elicited by the TF also neutralized the infectivity of D1, D2 and D3 as determined by PRNT. The other conclusion that we derive from our data is that the DIII-specific neutralizing antibody response in mice appears to differ somewhat for the different dengue virus serotypes, with D4 DIII uniformly stimulating the lowest titers. The induction of a low neutralizing response against D4 has been reported after the evaluation of different vaccine candidates based on live attenuated viruses, fusion proteins including DIII, or the whole envelope protein [21–23]. Moreover, based on epidemiological data, D4 has been described as a naturally attenuated virus, since there are few reports of primary D4 outbreaks [24]. Another important issue in this sense is that the PRNT has been described as having low sensitivity for measuring neutralizing antibodies against D4 [25]. Nevertheless, the balance among serotype specific neutralizing antibodies could be modified by adjustments in the relative amount of each dengue DIII protein in tetravalent formulations, e.g., D4 neutralization potency could be significantly increased by increasing the D4 DIII dose.

It is generally accepted that neutralizing antibodies are a major contributor to protective immunity and are considered surrogate markers of protective efficacy [26, 27]. For dengue vaccines in clinical trials, neutralizing antibody titers ≥1:10 are taken as evidence of seroconversion [26]. Thus, we may conclude that the immunized animals in this study successfully seroconverted with respect to dengue 1, 2 and 3 serotypes.

The T-cell response was studied by monitoring the production in vitro of IFN-γ in splenocytes obtained from immunized mice in response to virus stimulation. The immunized mice also showed broad cell-mediated immunity specific for the four dengue serotypes, similar to that induced by viral infection. It is known that spleen cells stimulated in vitro with the virus produce high levels of IFN-γ, which may play a role in antiviral activity [28]. Additionally, the secretion of this cytokine came from a subset of CD8+ cells, as reported for other related viruses [29, 30]. Specifically for dengue viruses, it has been reported that there is a relationship between CD8 cytotoxic activity and IFN-γ secretion [31, 32]. In previous reports, a cytotoxic T-cell epitope within domain III from D2 that conferred protection in the absence of a neutralizing humoral response has been described [33]. The results obtained for D4 support this observation. Stimulation with this serotype induced high levels of IFN-γ secretion in the mice previously immunized with the TF. In addition, the protection obtained in the challenge experiments was higher (~70 %) despite the fact that neutralizing antibodies against this serotype were not detected. This result constitutes an example of the important role of the cellular immune response in protecting against dengue virus in the mouse model, confirming the evidence provided by Van der Most et al. and Gil et al. [32, 34]. Recent results in vaccinated individuals showing high levels of neutralizing antibodies to dengue who were not protected suggests the complexity of the protection mechanisms and the importance of the cellular immune response [35].

Nonetheless, a contribution of antiviral antibodies to protection should not be dismissed. Despite the lack of neutralizing antibodies, the antiviral response might be involved in protection through other mechanisms such as complement activation or antibody-dependent cell-mediated cytotoxicity.

Suckling mice are employed to evaluate the protective efficacy of dengue vaccines by observing morbidity and mortality after viral challenge [36]. However, a lack of correlation between the neutralizing antibody titers and the level of protection has been reported [7, 37]. This observation suggests a possible mechanism of survival other than the neutralizing activity, such as CMI. Here, we obtained different protection levels after challenge with the four dengue serotypes. The case of D1 is of interest. The lowest level of protection was observed with this serotype despite the fact that the TF-immunized mice generated adequate levels of neutralizing antibodies and high levels of IFN-γ after D1 stimulation. At present there is not an appropriate animal model for vaccine evaluation or pathogenesis studies. Although mice and monkeys are used for vaccine evaluation, the results do not always predict the usefulness of a vaccine candidate [38].

In light of the results obtained in this and previous reports we can conclude that a future dengue vaccine should contain a mixture of the four dengue recombinant DIII-C proteins to guarantee the induction of serotype-specific antibodies and a CMI response, both of which are required by the adaptive immune system to control dengue infection [8, 39].