Sculpting humoral immunity through dengue vaccination to enhance protective immunity

Abstract

Dengue viruses (DENV) are the most important mosquito transmitted viral pathogens infecting humans. DENV infection produces a spectrum of disease, most commonly causing a self-limiting flu-like illness known as dengue fever; yet with increased frequency, manifesting as life-threatening dengue hemorrhagic fever (DHF). Waning cross-protective immunity from any of the four dengue serotypes may enhance subsequent infection with another heterologous serotype to increase the probability of DHF. Decades of effort to develop dengue vaccines are reaching the finishing line with multiple candidates in clinical trials. Nevertheless, concerns remain that imbalanced immunity, due to the prolonged prime-boost schedules currently used in clinical trials, could leave some vaccinees temporarily unprotected or with increased susceptibility to enhanced disease. Here we develop a DENV serotype 1 (DENV-1) DNA vaccine with the immunodominant cross-reactive B cell epitopes associated with immune enhancement removed. We compare wild-type (WT) with this cross-reactivity reduced (CRR) vaccine and demonstrate that both vaccines are equally protective against lethal homologous DENV-1 challenge. Under conditions mimicking natural exposure prior to acquiring protective immunity, WT vaccinated mice enhanced a normally sub-lethal heterologous DENV-2 infection resulting in DHF-like disease and 95% mortality in AG129 mice. However, CRR vaccinated mice exhibited redirected serotype-specific and protective immunity, and significantly reduced morbidity and mortality not differing from naїve mice. Thus, we demonstrate in an in vivo DENV disease model, that non-protective vaccine-induced immunity can prime vaccinees for enhanced DHF-like disease and that CRR DNA immunization significantly reduces this potential vaccine safety concern. The sculpting of immune memory by the modified vaccine and resulting redirection of humoral immunity provide insight into DENV vaccine-induced immune responses.

Keywords: DNA vaccines; antibody-dependent enhancement; dengue; dengue hemorrhagic fever; dengue vaccines; immune refocusing; original antigenic sin; vaccine safety.

FIGURE 1

Structural locations of the pVD1-CRR envelope (E) protein cross-reactive epitope knock out substitutions. The locations of the DENV-1 E substitutions introduced to construct pVD1-CRR mapped on the crystal structure of the homologous DENV-2 E protein dimer (Modis et al., 2003). (A) Locations of pVD1-CRR substitutions (spheres) on a ribbon diagram of the mature E dimer as it appears looking straight down toward the virion surface. The three structural domains are labeled and colored red for E domain I (EDI), yellow for EDII, and blue for EDIII. The EDII fusion peptide (EDII_FP) is also labeled and colored green. The glycans in EDI (N67) and EDII (N153) are depicted as stick representations and colored brown. (B) Side view of the E protein dimer with all colorations the same as in panel A. (C) Close up of the EDII_FP from one monomer and EDIII region of the other at the dimer interface, as it appears in panel B. The side chains of two residues in the EDII_FP and three residues in EDIII where CRR substitutions were introduced are depicted as ball and stick representations and labeled with the introduced substitutions. Structural domains are colored as in panel A and the two antigenic regions EDII_FP and EDIII_CR are noted with circles roughly representing the size of an IgG footprint binding to these regions.

FIGURE 2

Protective Efficacy: Both pVD1-WT and pVD1-CRR vaccines protect AG129 mice against lethal DENV-1 challenge. (A–D) 12 week post vaccination immunogenicity of AG129 mice immunized (i.m.) with 100 μg of pVD1-WT or pVD1-CRR vaccines at 0, 4, and 8 weeks. Geometric means and 95% confidence intervals are depicted unless otherwise noted. All endpoint titers were log₁₀ transformed and statistical significance was determined using the Mann–Whitney U test to account for non-normality of some transformed data. Panels B and C both used a one-tailed test since there was an a priori expectation that pVD1-CRR immunized mice would exhibit reduced EDII_FP recognizing IgG. (A) DENV-1 total IgG endpoint titers 12 weeks post immunization. (B) Percent of total DENV-1 IgG recognizing immunodominant EDII_FP epitopes, arithmetic means and 95% CI depicted. (C) Calculated DENV-1 EDII_FP IgG endpoint titers. (D) DENV-1 50% antigen focus-reduction micro neutralization titers (FRμNT₅₀). A t test was utilized here as both data sets were normally distributed. (E) pVD1-WT and pVD1-CRR immunized AG129 mice were challenged with 1.1x10⁵ FFU of DENV-1 (Mochizuki strain, i.p.) at 12 weeks. Kaplan–Meier survival curves are shown. Both pVD1-WT (n = 18) and pVD1-CRR (n = 18) vaccinated animals had the same survival (94%), which was highly significant in comparison to naїve mice (n = 4, p = 0.0003). (F) Four mice from each vaccine treatment (not included in the survival curves) were a priori scheduled to be euthanized 3 and 8 DPC. DENV-1 titers (FFU/mL) were log₁₀ transformed and analyzed using two-way ANOVA; error bars represent standard error of the mean (SEM). Vaccine treatment was highly significant (p < 0.0001). Bonferroni post hoc tests indicated that viremia of each vaccinated group was significantly lower than for naїve mice (n = 2) 3 DPC (p < 0.001), only pVD1-CRR vaccinated mouse viremia was significantly lower than naїve mice (n = 2) 8 DPC (p < 0.01) and there was no difference between WT or CRR vaccinated mouse viremia 3 or 8 DPC.

FIGURE 3

DENV-1 CRR vaccine stimulates reduced levels of immunodominant cross-reactive EDII_FP IgG and reduces enhanced DENV-2 mortality. (A–C) 12 week post vaccination immunogenicity of AG129 mice immunized (i.m.) with 100 μg of pVD1-WT or pVD1-CRR vaccines. Geometric means and 95% confidence intervals are depicted. Endpoint titers were log₁₀ transformed and statistical significance was determined using the two-tailed Mann–Whitney U test. Panels B and C both used a one-tailed test since there was an a priori expectation that pVD1-CRR immunized mice would exhibit reduced EDII_FP recognizing IgG. (A) DENV-1 total IgG endpoint titers 12 weeks post immunization. (B) Percent of total DENV-1 IgG recognizing immunodominant EDII_FP epitopes, arithmetic means and 95% CI depicted. (C) Calculated DENV-1 EDII_FP IgG endpoint titers. (D) Survival of pVD1-WT and pVD1-CRR immunized AG129 mice following sub-lethal heterologous DENV-2 challenge. 12 weeks following vaccination, immunized mice (n = 22) or age matched naїve controls (n = 8) were challenged (i.p.) with 4.2 × 10⁵ FFU of DENV-2 S221. Kaplan-Meier survival curves and p values are shown. All naїve mice survived virus challenge with minimal signs of morbidity, pVD1-WT vaccinated mice suffered 95% mortality from enhanced DENV-2 disease (p < 0.0001, compared to naive). pVD1-CRR immunized mice exhibited 68% survival which did not differ from naїve mouse survival (100%, p = 0.0769) yet was significantly greater than pVD1-WT immunized mouse survival (4.5%, p < 0.0001). The Bonferroni multiple comparison adjusted α = 0.017. (E) pVD1-CRR vaccinated mice exhibited a rapid, large magnitude increase in DENV-2 neutralizing antibody titers following DENV-2 challenge. FRμNT₅₀ titers determined with DENV-2 16681 virus on Vero cells, log₁₀ transformed and the transformed data analyzed by two-way ANOVA (p < 0.0001). Bonferroni post hoc test significance is depicted with asterisks, error bars represent SEM. 0, 3, 4, and 5 DPC n = 4, 4, 5, and 9 for WT and 4, 4, 1, and 3 for CRR immunized mice respectively. (F) Viremia of pVD1-WT vaccinated mice increased rapidly 3–5 days after DENV-2 S221 challenge, whereas pVD1-CRR vaccinated mice had 30-fold, 40-fold, and at least 1800-fold lower viremia 3, 4, and 5 DPC (all CRR mice <10 FFU/mL) than did WT immunized mice. Virus titers were log₁₀ transformed and the transformed data analyzed by two-way ANOVA (p < 0.0001). Bonferroni post hoc test significance is depicted with asterisks, error bars represent SEM. 3, 4, and 5 DPC n = 4, 5, and 6 for WT and 4, 1, and 3 for CRR immunized mice respectively.

FIGURE 4

Pathophysiology of enhanced DENV disease in vaccinated AG129 mice. (A) Enhanced DENV disease associated vascular leak pathology of pVD1-WT and -CRR vaccinated mice following sub-lethal heterologous DENV-2 infection. No pathology was visible in naїve or vaccinated mice 3 DPC, yet by 4 and 5 DPC mice succumbing to enhanced disease exhibited severe vascular leakage associated pathology. The same mouse is shown in the bottom two photos to highlight the severe intestinal hemorrhage observed in pVD1-WT but not in pVD1-CRR vaccinated mice (arrows). (B) Histology of uninfected, pVD1-WT, and pVD1-CRR vaccinated mouse livers 3 DPC. The top row is hematoxylin and eosin stained liver sections and the bottom row is immunohistochemistry for DENV-2 NS1 protein of the same individual animals. NS1+ mononuclear inflammatory cells and sinusoidal endothelial cells stain red in their cytoplasm and are visible (arrows) in liver tissue from vaccinated mice but not uninfected mice. Multiple sections from multiple animals were examined and single representatives are shown. All photos were taken at 400x magnification. (C) In vitro DENV-2 enhancement by pVD1-WT and pVD1-CRR vaccinated serum 12 weeks following vaccinations and one day prior to DENV-2 challenge. Consistent with the In vivo results, pVD1-WT vaccinated serum significantly enhanced DENV-2 infection whereas pVD1-CRR vaccinated serum did not (p = 0.0012). The data are representative of two independent experiments of 4 pools of 6–7 individual serum specimens for each vaccine treatment and were analyzed with two-way ANOVA and Bonferroni post hoc test significance at individual dilutions is depicted with asterisks.

FIGURE 5

DENV-1 CRR vaccination sculpts immunity to redirect secondary immunity to heterologous dengue infection. AG129 mice were immunized (i.m.) with 100 μg of pVD1-WT or pVD1-CRR vaccines and challenged at 12 weeks with heterologous DENV-2 S221 (4.2x10⁵ FFU, i.p.). Arithmetic means and SEM are depicted in A, B, D, and E and analyzed with two-way ANOVA. Bonferroni post hoc test significance is depicted with asterisks. (A) DENV-1 total IgG endpoint titers of pVD1-WT and -CRR immunized mice 3, 4, and 5 DPC with DENV-2. Three, 4, and 5 DPC n = 4, 3, and 8 and 4, 1, and 3 for WT and CRR vaccines respectively. (B) DENV-2 total IgG endpoint titers of pVD1-WT and -CRR vaccinated mice. (C) Percent DENV-1 epitope-specific responses pre- and post-challenge for IgG recognizing immunodominant cross-reactive EDII_FP epitopes. pVD1-WT immunized mice had larger populations of EDII_FP post DENV-2 challenge than did pVD1-CRR immunized mice (ave = 35.5 and 7.65%, respectively, p = 0.0004). pVD1-WT immunized mice also exhibited an increase in EDII_FP IgG from pre- to post-DENV-2 challenge (ave = 17.7 and 35.5%, respectively, p = 0.0057. Geometric means and 95% CI are depicted. Statistical significance was determined with a one-tailed Mann–Whitney U. The Bonferroni α = 0.025. (D,E) DENV-2 epitope specific IgG responses post DENV-2 challenge of pVD1-WT and pVD1-CRR immunized AG129 mice; 3, 4, and 5 DPC n = 4, 5, and 9 and 4, 1, and 2 for WT and CRR immunized mice respectively. (D) Percent DENV-2 IgG recognizing immunodominant cross-reactive EDII_FP epitopes. (E) Percent DENV-2 IgG recognizing E protein epitopes outside of immunodominant EDII_FP and EDIII_CR antigenic regions (Non-EDII_FPEDIII_CR). (F) FRμNT₅₀ titers of pVD1-WT and -CRR immunized mouse sera 3 DPC with DENV-2 against DENV-1, DENV-2, DENV-3, and DENV-4. Bars represent GMT, statistical significance was determined with a one-tailed t-test. All titers were log₁₀ transformed prior to analysis.

FIGURE 6

DENV-1 CRR vaccination redirects immune responses to DENV-2 to increase a diversity of neutralizing antibodies. Percent blocking of labeled monoclonal antibodies (MAbs) by pVD1-WT and pVD1-CRR immunized mice 3 DPC to DENV-2 as determined by blocking ELISA. AG129 mice were immunized (i.m.) with 100 μg of pVD1-WT or pVD1-CRR vaccines and challenged at 12 weeks with a sub-lethal dose of heterologous DENV-2 S221 (4.2 × 10⁵ FFU, i.p.). 3 DPC sera of the four mice from each vaccine treatment were pooled and percent blocking was determined independently four times in duplicate. To facilitate comparison between vaccine treatments across the different MAbs, x-fold increases in blocking by pVD1-CRR vaccinated sera relative to pVD1-WT vaccinated sera are depicted above the x-axis for each MAb. Below the MAbs, serotype specificity is depicted (DENV-2, DENV-2 type-specific; subcomp, DENV sub-complex cross-reactive; complex, DENV complex cross-reactive and complex + supercomplex cross-reactive). Below specificity, p-values from a two-tailed t test comparing percent blocking for each vaccinated sera are depicted. Below that the presence or absence of the epitope recognized by each MAb on the DENV-1 CRR VLP antigen is depicted (all MAbs recognize epitopes found in the DENV-2 challenge virus). For example, 3H5 and 9A3D-8 as DENV-2 specific antibodies are absent on both pVD1-WT and pVD1-CRR VLP antigens; 9D12 and 1A1D-2 reactivity are ablated by the EDIII_CR substitutions in the pVD1-CRR plasmid (Table 2) but present in the WT DENV-1 antigen; and 10-D35A, D3-5C9-1, and 1B7 reactivities were not altered by the substitutions introduced into in the pVD1-CRR plasmid and hence are present on both DENV-1 WT and CRR immunization antigens. FRμNT₅₀ titers for each MAb against DENV-2, -1, -3, and -4 in μg/mL are also shown.