Virus-like particle secretion and genotype-dependent immunogenicity of dengue virus serotype 2 DNA vaccine

Abstract

Dengue virus (DENV), composed of four distinct serotypes, is the most important and rapidly emerging arthropod-borne pathogen and imposes substantial economic and public health burdens. We constructed candidate vaccines containing the DNA of five of the genotypes of dengue virus serotype 2 (DENV-2) and evaluated the immunogenicity, the neutralizing (Nt) activity of the elicited antibodies, and the protective efficacy elicited in mice immunized with the vaccine candidates. We observed a significant correlation between the level of in vitro virus-like particle secretion, the elicited antibody response, and the protective efficacy of the vaccines containing the DNA of the different DENV genotypes in immunized mice. However, higher total IgG antibody levels did not always translate into higher Nt antibodies against homologous and heterologous viruses. We also found that, in contrast to previous reports, more than 50% of total IgG targeted ectodomain III (EDIII) of the E protein, and a substantial fraction of this population was interdomain highly neutralizing flavivirus subgroup-cross-reactive antibodies, such as monoclonal antibody 1B7-5. In addition, the lack of a critical epitope(s) in the Sylvatic genotype virus recognized by interdomain antibodies could be the major cause of the poor protection of mice vaccinated with the Asian 1 genotype vaccine (pVD2-Asian 1) from lethal challenge with virus of the Sylvatic genotype. In conclusion, although the pVD2-Asian 1 vaccine was immunogenic, elicited sufficient titers of Nt antibodies against all DENV-2 genotypes, and provided 100% protection against challenge with virus of the homologous Asian 1 genotype and virus of the heterologous Cosmopolitan genotype, it is critical to monitor the potential emergence of Sylvatic genotype viruses, since vaccine candidates under development may not protect vaccinated humans from these viruses.

Importance: Five genotype-specific dengue virus serotype 2 (DENV-2) DNA vaccine candidates were evaluated for their immunogenicity, homologous and heterologous neutralizing (Nt) antibody titers, and cross-genotype protection in a murine model. The immunity elicited by our prototype vaccine candidate (Asian 1 genotype strain 16681) in mice was protective against viruses of other genotypes but not against virus of the Sylvatic genotype, whose emergence and potential risk after introduction into the human population have previously been demonstrated. The underlying mechanism of a lack of protection elicited by the prototype vaccine may at least be contributed by the absence of a flavivirus subgroup-cross-reactive, highly neutralizing monoclonal antibody 1B7-5-like epitope in DENV-2 of the Sylvatic genotype. The DENV DNA vaccine directs the synthesis and assembly of virus-like particles (VLPs) and induces immune responses similar to those elicited by live-attenuated vaccines, and its flexibility permits the fast deployment of vaccine to combat emerging viruses, such as Sylvatic genotype viruses. The enhanced VLP secretion obtained by replacement of ectodomain I-II (EDI-II) of the Cosmopolitan genotype vaccine construct (VD2-Cosmopolitan) with the Asian 1 EDI-II elicited significantly higher total IgG and Nt antibody titers and suggests a novel approach to enhance the immunogenicity of the DNA vaccine. A DENV vaccine capable of eliciting protective immunity against viruses of existing and emerging genotypes should be the focus of future DENV vaccine development.

FIG 1

Informative sites in the envelope protein of DENV-2 genotypes. (A) Alignment showing the variable amino acid positions and consensus sequence in the E-protein EDs. Gray shading, conservative amino acid substitutions; light blue shading, nonconservative substitutions; brown shading, nonconservative substitutions unique to certain genotypes. (B) Structural locations of surface-exposed informative sites. (Top) Ribbon diagram of the monomeric E protein with several of the known antigenic sites marked in circles; (bottom) top view of the dimeric E-protein crystal model of the mature DENV-2 particle (PDB accession number 3J27) depicted as a surface representation. (C) Side view surface representation of the model of the trimeric E-protein crystal structure of the immature DENV-2 particle (PDB accession number 3C6D). In panels A to C, EDI, EDII, EDIII, and the fusion loop are colored red, yellow, blue, and violet, respectively. Surface-exposed residues (B and C) are highlighted in green. The prM protein (C) is white.

FIG 2

Vaccine design, analysis of recombinant antigen expression and secretion, and immunization schedule. (A) Schematic representation of plasmid vectors containing DNA of the different DENV-2 genotypes: pVD2-Asian 1, pVD2-American, pVD2-American/Asian, pVD2-Cosmopolitan, and pVD2-Sylvatic. These plasmid constructs include the human cytomegalovirus early gene promoter (^PCMV), the JEV signal sequence, prM, 80% of the E-gene region (amino acids 1 to 396) of each DENV-2 genotype, 20% of the homologous E-gene region of JEV (amino acids 397 to 495), and a bovine growth hormone poly(A) signal (BGH pA). (B) Immunofluorescence analysis of recombinant protein expression in COS-1 cells transfected with plasmids containing DNA of the different DENV-2 genotypes. After fixation and permeabilization, recombinant proteins were detected with anti-DENV-2 MHIAF, followed by incubation with goat anti-mouse IgG-fluorescein isothiocyanate and DAPI to counterstain the nucleus. The characteristic green fluorescence in the cells indicates positivity for the intracellular expression of recombinant proteins. (C) Detection and quantification by an Ag-capture ELISA of secreted VLPs in culture supernatants harvested from transiently transformed COS-1 cells. Data are presented as means ± SEMs of two independent repeat transfection experiments. Means were analyzed by one-way ANOVA (P < 0.05), and Holm-Sidak's posttest significance is indicated with asterisks (***, P = 0.0002; ****, P < 0.0001). (D) Schedule of immunization with different plasmid vectors containing DNA of the different DENV-2 genotypes. Groups of 4-week-old BALB/c mice were immunized with three doses of 100 μg DNA intramuscularly. At 4 weeks following the last immunization, blood samples from each group of mice were taken and pooled for further serum antibody analyses. Black arrows, immunization times; red arrows, blood sampling times.

FIG 3

Naturally occurring amino acid variation affects the immunogenicity of DENV-2 DNA vaccines. (A) Total IgG antibody responses in mice injected with different DENV-2 DNA vaccines were determined by a standard Ag-capture ELISA 4 weeks following the last immunization. **, P = 0.0057; ***, P = 0.0007, ****, P < 0.0001. (B) The total IgG ELISA titer is positively and significantly correlated with the VLP secretion titer, as determined by a regression correlation analysis (r = 0.9511; P = 0.0129). (C) Percent blocking of an HRP-labeled anti-prM MAb (155-49) by sera from mice vaccinated with different DENV-2 genotypes containing antibodies to VLPs of the different DENV-2 genotypes determined by an epitope-blocking ELISA. *, P < 0.05. (D) The ratio of uncleaved prM and E among the VLPs of the different DENV-2 genotypes was measured by ELISA and is expressed as a percentage relative to that for immature pVD2-Asian 1 VLPs. *, P = 0.0366. (E) Cross-neutralization activity of antisera elicited by vaccines containing the different DENV-2 genotypes against a set of viruses of different DENV-2 genotypes determined by FRμNT. Nt₅₀ titers are presented as a heatmap generated using the web tool of the HIV sequence database. Lower Nt₅₀ values are red, and the colors progress to orange and then yellow to bright yellow at higher values. GMT, geometric mean titer. *, P < 0.05; ****, P < 0.0001. (F) The total IgG ELISA titer is negatively correlated with the Nt₅₀ titer upon regression correlation analysis (r = 0.5673; P = 0.3186). (G) Ratio of Nt₅₀ titers to DENV-2-specific ELISA titers as a measure of the functional neutralizing activity of the elicited immune sera. ELISA data are means ± SEMs of two independent repeat experiments. **, P = 0.0038; ***, P = 0.003; ****, P < 0.0001. Nt₅₀ titers are expressed as the mean ± 95% CI. Means were analyzed by one-way ANOVA (P < 0.05), and Holm-Sidak's posttest significance is indicated with asterisks.

FIG 4

Characterization of type-specific and cross-reactive activity and domain-specific responses of antibodies from sera from mice vaccinated with the different genotypes. (A) ELISA reactivities of MAbs against E-protein EDIII mutant DENV-1 and DENV-2 VLPs; (B) domain-specific IgG responses to the EDIII and non-EDIII epitopes of sera from mice vaccinated with the different genotypes; (C) blocking of an HRP-labeled DENV-2-specific MAb (3H5-1) and cross-reactive MAbs (6B6C-1 and 1B7-5) by sera from mice vaccinated with the different DENV-2 genotypes containing antibodies to VLPs of the different DENV-2 genotypes determined by an epitope-blocking ELISA. ELISA data are the means ± SEMs of two independent repeat experiments. Means were analyzed by one-way ANOVA (P < 0.05), and Holm-Sidak's posttest significance is indicated with asterisks (*, P < 0.05; **, P < 0.01).

FIG 5

Naturally occurring mutations affect the neutralizing ability and physical binding of antibodies. (A) Cross-FRμNT titers of pVD2-Asian 1 postvaccination serum against a set of strains of different DENV-2 genotypes. Dashed line, geometric mean Nt₅₀ titer; solid lines, 95% confidence interval of the titer of the serum pool. Error bars represent SEMs. (B) Binding ELISA reactivity of pVD2-Asian 1 postvaccination serum to different VLP antigens. The percentage of the reactivity of sera with antibodies to heterologous and mutant VLP antigens remaining relative to that for the wild-type Asian 1 genotype VLPs was calculated. (C) The percentage of the reactivity of MAb 1B7-5 to the chimeric Asian 1 VLP antigens remaining relative to that for the wild-type Asian 1 genotype VLPs was calculated. All data represent the means ± SEMs of two independent experiments.

FIG 6

The in vivo protective efficacy of the DENV-2 DNA vaccine is genotype dependent. (A to C) Groups of 2-day-old suckling mice obtained from pVD2-Asian 1-vaccinated ICR female mice (Nt₅₀ titer, >1:200) were infected intracranially with 2,000 LD₅₀s of DENV-2 strain 16681 (A), 300 LD₅₀s of DENV-2 strain 1222 (B), and 1,000 LD₅₀s of DENV-2 strain DakHD 10674 (C). The survival of the mice was monitored for 21 days after infection. Kaplan-Meier survival curves were analyzed by the log-rank test. P values of <0.05 were considered significant. The statistical significance of the difference between mice from vaccinated mothers and mice from naive mothers is indicated with asterisks. *, P = 0.0232; ****, P < 0.0001.

FIG 7

Enhancement of VLP secretion and immunogenicity by manipulation of ectodomain of E protein. (A) An increase in the VLP secretion titer in pVD2-Cosmopolitan/Asian 1 EDI-II-transfected COS-1 cells after replacement of the EDI-II gene of the WT pVD2-Cosmopolitan plasmid vector with the corresponding gene segment from the pVD2-Asian plasmid was determined by ELISA. ****, P < 0.0001. (B) pVD2-Cosmopolitan/Asian 1 EDI-II-vaccinated mice exhibit a larger total IgG antibody response. **, P = 0.0017. (C) Improvement of FRμNT titers against the different DENV-2 genotypes by sera from pVD2-Cosmopolitan/Asian 1 EDI-II-vaccinated mice. *, P = 0.0195; **, P = 0.0031; ***, P = 0.0004. All data represent the means ± SEMs of two independent experiments. Means were analyzed by an unpaired t test, and significance (P < 0.05) is indicated with asterisks.

FIG 8

Naturally occurring mutations affect the structural stability of the E protein of pVD2-Cosmopolitan VLPs. (A) Stable free energy (ΔΔG) calculations for E-protein substitutions in Cosmopolitan genotype strain BC145/97 based upon the Protein Data Bank structural coordinates for the mature dengue virus particle (PDB accession number 3J27) and soluble E protein (PDB accession number 1OAN) (25, 27). (B to E) Amino acid residues involved in E-protein intra- and interdimer polar contacts. (B) E71 (in a red ball-and-stick representation) and its intradimeric contact with amino acid residues S72, R73, and V114 (in green ball-and-stick representations), connected by yellow dashed lines. (C) The construct with the E71A substitution shows a loss of intradimeric contact with R73. (D) K88 (in a red ball-and-stick representation) and its contact with various amino acid residues (in green ball-and-stick representations), connected by yellow dashed lines. K88 of chain A of the E-protein asymmetric unit has intradimeric contacts within chain A at positions Q86 and K234 and has an interdimeric contact with E85 of chain C. K88 of chain C has intradimeric contacts within chain C at positions Q86, N230, and K234. (E) K88E of chain A shows a loss of interdimeric contact with E85 of chain C and a steric clash with Q86 of chain C. K88E of chain C shows a loss of intradimeric contact with N230 of chain C.