Efficacy of Multivalent Dengue Vaccine Candidates Predicted In Silico

TL;DR

Researchers predicted and tested dengue vaccine candidates in silico and found some peptides that effectively stimulate immune responses in mice.

Contribution

The study introduces a computational and experimental approach to identify multivalent dengue vaccine candidates based on immunogenic epitopes.

Findings

Candidate peptides with a score >0.1 induced strong T-cell proliferation and cytokine regulation in mice.

Some peptides stimulated cellular immunity confirmed by plaque reduction neutralization tests.

Selected epitopes from envelope domain III showed high immunogenic potential in both in vitro and in vivo models.

Abstract

Background: Dengue virus (DENV) is becoming a global public health problem, but the immunogenicity of DENV structural proteins is not fully understood. Methods: We predicted the epitope-based immunogenicity of DENV proteins from four serotypes in silico and evaluated their efficacy in vitro (T-cell proliferation assays) and in vivo (ELISpot, qRT-PCR, and plaque reduction neutralization tests using murine splenocytes). We focused on the envelope protein, which contains envelope domain III. Immunogenic B-cell epitopes were predicted using BepiPred-2.0, and regions that induce T cell-mediated immune responses were analyzed using the immune epitope database (IEDB), which validates peptides presented on HLA class I. Results: Nine-amino-acid peptide candidates were selected based on a score of >0.1. The best peptide candidates were tested in T-cell proliferation assays to confirm the in silico data. Subsequently, BALB/c mice were vaccinated with candidate peptides showing immunity in the proliferation assay, and their splenocytes were analyzed. ELISpot and qRT-PCR data showed that some candidate peptides highly regulated cytokines, including interferon-γ, tumor necrosis factor-α, and interleukin-4. Murine sera were collected after peptide boosting 2 weeks apart. Stimulation of cellular immunity was confirmed for some candidates in plaque reduction neutralization tests.

Figures

Full text

1. Introduction

Dengue fever is a viral disease mainly transmitted by mosquitoes infected with the dengue virus (DENV); however, vertical transmission from mother to fetus during pregnancy is primarily associated with maternal viremia and transplacental transmission [1]. The main disease vector, Aedes aegypti, can be found in tropical and subtropical regions, but Aedes albopictus in Korea can also spread the disease [2]. Following an incubation period of 3–14d, the disease symptoms can range from mild fever to dengue hemorrhagic fever and dengue shock syndrome. The various clinical symptoms can be caused by all DENV variants, and symptom severity varies depending on the four DENV serotypes [3]. The primary infection confers lifelong immunity against the virus. However, the cross-reactive antibodies, which do not appear to have a neutralizing response on the virion, predominantly recognize the envelope protein (E) on mature DENV, and precursor membrane (prM) on immature DENV could increase viral entry into host cells [4,5].

DENV consists of several structural proteins, including capsid (C), prM, and E, and nonstructural proteins (NS). Both the E protein, which plays a key role in the entry of DENV, and NS1, which is exposed on the surface of host cells, are currently major vaccine candidates as they induce the production of effective antibodies and suppress DENV infection [6]. Some vaccines against DENV have been developed, but they offer limited protection against all four DENV serotypes. For example, Dengvaxia shows relatively low protection against DENV serotypes 1 (DENV-1) and 2 (DENV-2) [7].

Therefore, we investigated epitopes that appear to be immunogenic for the structural domain of the four DENV serotypes through in silico analysis. The prM and E regions containing the envelope domains (EDs) I, II, and III in DENV-1 to DENV-4 were analyzed, and linear B-cell epitopes and nine amino acid (9-mer) T-cell epitopes, which could effectively recognize HLA class I, were ranked in descending order.

Next, 25 candidate peptides from each serotype were selected based on this rank. To investigate peptide immunity, T-cell proliferation was assayed using the carboxyfluorescein succinimidyl ester (CFSE) method [8]. Candidate peptides showing lower fluorescence in stained human peripheral blood monocytes (hPBMCs) were prepared for animal immunization. BALB/c mice were immunized three times with the final ten candidate peptides, and the spleens of these immunized mice were prepared for cytokine analysis to investigate humoral and cellular immunogenicity.

Collectively, these results may contribute to the development of a new vaccine platform in addition to the development of a dengue vaccine.

2. Materials and Methods

2.1. Cell Lines, Viruses, and Mice

Vero E6 cells (ATCC CRL-1587), isolated from the kidneys of an African green monkey, were grown in Dulbecco’s modified Eagle’s medium (DMEM, Gibco, Grand Island, NY, USA), supplemented with 10% (v/v) fetal bovine serum (FBS, Gibco), 500 units/mL of penicillin, and 500 μg/mL of streptomycin, in a humidified incubator at 37 °C and 5% CO_2_. Master stocks for serotype 1 (DENV-1; NCCP41503, KP406803.1) isolated from the blood of a Korean traveler were received from the National Culture Collection for Pathogens (NCCP); those for serotype 2 (DENV-2; KBPV-VR-29, KP406804), serotype 3 (DENV-3; KBPV-VR-30, KP406805), and serotype 4 (DENV-4; KBPV-VR-31, KP406806) isolated from blood samples of Korean travelers were obtained from the Korea Bank for Pathogenic Viruses (KBPV, Seoul, Republic of Korea). DENV-1, -2, -3, and -4 were amplified at a multiplicity of infection of less than 0.1 in Vero E6 cells, seeded into T-175 flasks at a density of 5 × 10^7^ cells, and cultured in DMEM supplemented with 2% FBS. Viruses were collected and filtered through 0.45-µm membranes. The filtered culture medium was then concentrated using an InnovaPrep system.

Female BALB/c mice (4–8 weeks old) were purchased from ORIENT Bio Inc (Seongnam-si, Gyeonggi-do, Republic of Korea). and raised for a week for adaptation. Each group consisted of five mice per cage, of which three mice were prepared for flow cytometry and enzyme-linked immunospot (ELISpot) analysis and two mice for quantitative real-time PCR (qRT-PCR).

2.2. Analysis In Silico

For the computational prediction of epitope discovery, the program BepiPred-2.0 was used to analyze the structural domains of the four DENV serotypes based on the Immune Epitope Database and Analysis Resource (IEDB). Most DENV serotype-specific neutralizing monoclonal antibodies bind to the lateral ridge of domain III of the E protein (EDIII) [9]. Therefore, the entire structural domain, including EDIII, was analyzed. The immunogenic score was displayed in an interactive plot, illustrating the specificity and sensitivity of various epitope thresholds (Figure S1). Points in the plot predicted the score at a given epitope threshold. The epitope was subsequently evaluated as 9-mer peptides for MHC class I-restricted T-cell immunogenicity prediction. The average score for epitopes was higher than 0.1 compared to non-epitopes [10]. All peptides used in this study were synthesized as linear peptides without multimerization or carrier conjugation. The peptides were designed as 9-mer sequences to evaluate MHC class I-restricted T-cell immunogenicity rather than to function as standalone B-cell antigens. A schematic diagram of the bioinformatic workflow used for epitope prediction and candidate selection is provided in Figure S3.

2.3. Immunization Schedule

Female HLA-A02:01 transgenic mice on a BALB/c background (4–8 weeks old, The Jackson laboratory, Bar Harbor, ME, USA) were used for immunization experiments. These mice express the human HLA-A2 molecule, allowing evaluation of HLA-A2-restricted T-cell responses in vivo. Transgenic mice were vaccinated with a mixture of 10 µg peptides per 10 µL of X and 90 µL of the adjuvant Alum-hydroxide (Alhydrogel^®^ 2%, InvivoGen, San Diego, CA, USA). Aluminum hydroxide adjuvant was diluted 5X (1:5 dilution) in PBS and mixed with peptide antigen to yield a final dose of approximately 100 μg of Al^3+^ per mouse. BALB/c mice were randomly assigned to 13 groups (n = 5 per cage) and acclimatized for one week prior to the experiment. Group 1 (G1) consists of a negative control inoculated with phosphate saline only. From G2 to G6, mice were immunized with peptides derived from group E of DENV-1 (1-21 to 1-25; 1-e), respectively. From G7 to G11, mice were immunized with peptides derived from group C of DENV-4 (4-11 to 4-15; 4-c), respectively. For G12 and G13, mice were immunized with pooled peptides with 1-e and 4-c, respectively. After adaptation for a week, mice were primed with 100 µL of vaccine via intramuscular injection. The mice received two booster vaccinations of 10 µg/100 µL two weeks apart. Submandibular blood collection was carried out every two weeks, starting one week after the initial immunization. Mice were euthanized at week 5 after the first immunization, corresponding to one week after the final booster dose, and blood was collected via cardiac puncture followed by isolation of the spleen. For the Plaque Reduction Neutralization Test (PRNT), immunization was performed Three time and animal, antigens, inoculum dose, and immunization cycles were the same as described above. Mice sera were collected at a week and 3 weeks post immunization via facial vein bleeding.

2.4. CFSE Assay

The hPBMCs were obtained from a commercial supplier (Lonza, Basel, Switzerland) and were derived from a healthy adult male donor (CC-2703, 21TL207843, 57 years old). The cells were isolated from donated tissue with informed consent or legal authorization for research use, as provided by the manufacturer. A thawed hPBMC vial was resuspended in 1 mL of serum-free RPMI 1640 and washed twice. A total of 1 × 10^8^ cells were stained with a final concentration of 5 µM CSFE (C3455, Thermo Fisher Scientific, Waltham, MA, USA) for 20 min. Stained cells were seeded into 96-well plates at a density of 2 × 10^6^ cells/well. As a positive control, 5–20 µg/mL of concanavalin A was used. Five candidate peptides each were pooled and diluted with complete RPMI 1640 to a final concentration of 2–10 µg/mL. The seeded hPBMCs were treated with concanavalin A or pooled peptides. The cells were incubated for 72 h at 37 °C in a humidified CO_2_ incubator and then analyzed using flow cytometry (CytoFLEX, Beckman Coulter, Brea, CA, USA).

2.5. ELISpot Assays

The spleens of immunized mice were homogenized through a 40 nm cell strainer using a 3 mL syringe plunger with 5 mL of RPMI 1640 medium supplemented with 10% FBS. The cell suspensions were centrifuged for 5 min at 800 g, and the supernatants were discarded. The collected cells were incubated on ice with 5 mL of red blood cell (RBC) lysis buffer for 3–5 min. After RBC lysis, splenocytes were washed twice with RPMI 1640 containing 10% FBS, resuspended at a density of 3 × 10^6^ cells/100 µL, and then stimulated with 10 μg/100 μL of peptide. A volume of 100 μL was plated into each well of a 96-well polyvinylidene difluoride microplate coated with primary antibodies specific for mouse interferon (IFN)-γ (EL485, 890894, R&D Systems, Minneapolis, MN, USA). The cells were incubated for 18 h at 37 °C in a CO_2_ incubator. Then, the plates were washed and incubated with biotinylated secondary antibody (EL485, 890895, R&D Systems) for 1 h. Then, the cells were treated with streptavidin-alkaline phosphatase (EL485, 895358; R&D Systems) for 30 min. For colorimetric development, Nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate were mixed at a 1:1 ratio, and the cells were treated with the solution until stained spots appeared. The number of spots was quantified using a SpectraMax i3 (Molecular Devices, San Jose, CA, USA) microplate reader.

2.6. Cytokine Analysis in Splenocytes Measured by Flow Cytometry

Splenocytes were seeded at a density of 2.0 × 10^6^ cells/500 µL into 24-well plates, and 500 µL of complete RPMI 1640 containing 200 ng DENV E protein, 1000× diluted monesin (BD GolgiStop, 554724), and brefeldin A (BD GolgiPlug, 555029) was added. As a positive control group, complete RPMI 1640 containing eBioscience cell stimulation cocktail (00-4970-93, Invitrogen, Waltham, MA, USA), monesin, and brefeldin A was used. For the negative control group, only monesin and brefeldin A were added. After 5–6 h incubation at 37 °C in a CO_2_ incubator, fluorescence-activated cell sorting (FACS) buffer containing Dulbecco’s phosphate-buffered saline, 1% bovine serum albumin, 0.1% (w/v) sodium azide was added. The collected splenocytes were stained with fluorochrome-conjugated anti-mouse IFN-γ, anti–TNF-α, and anti–IL-4 antibodies (100× dilution), followed by incubation with an Fc-blocking rat anti-mouse CD16/CD32 monoclonal antibody (1:200, 10 min at 25 °C). The splenocytes were incubated with a fluorochrome-conjugated secondary antibody (1:200 dilution, 30 min at 4 °C; rat anti-mouse IFN-γ: APC, rat anti-mouse TNF-α: BV421, and rat anti-mouse IL-4: PE). All antibodies were purchased from BD Biosciences (Ashland, OR, USA). Cells were fixed with 4% (w/v) paraformaldehyde prior to analysis by flow cytometry using a CytoFLEX LX system, and a minimum of 5000 events for each sample were recorded using FlowJo software (Version 10; BD Biosciences).

2.7. Analysis of Cellular Immunogenicity Measured by qRT-PCR

To investigate the cellular immunogenicity of the vaccine candidates, spleen tissues (approximately 50–100 mg) from two mice of each group were homogenized. Total RNA was extracted using the easy-spin Total RNA Extraction Kit (iNtRON, #17221), following the manufacturer’s instructions. For reverse transcription, 3 μg of total RNA was used in a 50 μL reaction mixture containing 10 μL of 5× Reaction Buffer, 2.5 μL of RiboLock RNase inhibitor, 2 μL of oligo (dT), 2 μL of dNTP mix, 0.5 μL of RevertAid M-MuLV Reverse Transcriptase (#K1622, Thermo Fisher Scientific), and nuclease-free water to the final volume. The reaction was performed for 60 min at 42 °C, followed by enzyme inactivation at 94 °C for 5 min. The synthesized cDNAs were amplified using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad Laboratories, Hercules, CA, USA) with 40 cycles at 95 °C for 15 s and 60 °C for 20 s using the following primers: TNF-αforward:5′-ATG AGC ACA GAA AGC ATG AT-3′ reverse:5′-TAC AGG CTT GTC ACT CGA AT-3′IFN-γforward:5′-AGC AAG GCG AAA AAG GAT GC-3′ reverse:5′-TCA TTG AAT GCT TGG CGC TG-3′IL-4forward:5′-AGC CCT ACA GAC GAG CTC ACT C-3′ reverse:5′-ACG AGG TCA CAG GAG AAG GGA-3′GAPDH (control)forward:5′-TGA TGG GTG TGA ACC ACG AG-3′ reverse:5′-GAT GGC ATG GAC TGT GGT CA-3′

The relative gene expression levels of the target cytokines were quantified. The relative gene expression levels of the target cytokines were quantified using the comparative Ct (2^−ΔΔCt^) method, with normalization to GAPDH as an internal control.

2.8. Plaque Reduction Neutralization Test

Mouse serum samples were obtained by centrifuging the collected blood at 3000 rpm and 4 °C for 7 min and subsequently stored at −80 °C. Vero E6 cells were seeded at a density of 2.5 × 10^5^ cells/well in 12-well plates and incubated for 24 h at 37 °C. The heat-inactivated sera (56 °C for 30 min) were two-fold serially diluted with 216 µL of DMEM (1:10 to 1:10,240), and 120 µL containing 50 plaque-forming units of DENV was added. After 1 h of incubation, the virus–serum mixture was inoculated to infect Vero E6 monolayers for 1 h at 37 °C in a CO_2_ incubator to allow viral adsorption. The virus–serum solutions were then changed to agarose overlay medium containing 50% (v/v) 2× Minimal Essential Medium (Gibco) with 4% FBS, 1% penicillin/streptomycin, and 50% 1% (v/v) agarose. Then, the cells were incubated for 5–6 days. After removal of the overlay medium, plaques were visualized by staining with 5% ethanol–crystal violet solution for 3–4 h.

2.9. Statistical Analysis

Quantitative data were analyzed using appropriate descriptive and inferential statistical methods. Data are presented as mean ± standard deviation (SD) or mean ± standard error (SE). For experiments performed in duplicate, results are shown as descriptive statistics without inferential statistical testing. For experiments conducted in triplicate or more, statistical significance was assessed using Student’s t-test. A p-value < 0.05 was considered statistically significant and is indicated as * (p < 0.05) and ** (p < 0.01).

3. Results

3.1. Structural Analysis and 3D Modeling of Proteins Based on Big Data

Protein structure was analyzed using the TMHMM-2.0 algorithm [11], which was designed for transmembrane helix prediction. The proteins analyzed in this study comprised ten structural and nonstructural proteins (C, prM, E, NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5) derived from four clinically isolated DENV strains identified in Korea. Using in silico analysis, the sequences were examined to predict the structural features and localization of antigenic proteins. Furthermore, the three-dimensional structure of the antigenic protein was modeled using AlphaFold2, a deep learning-based prediction system [12].

3.2. Prediction of B-Cell Epitopes

B-cell epitopes were primarily predicted using BepiPred-2.0. To enhance prediction accuracy, various complementary analytical methods were employed, including Bepipred Linear Epitope Prediction, Chou & Fasman Beta-Turn Prediction, Emini Surface Accessibility Prediction, Karplus & Schulz Flexibility Prediction, Kolaskar & Tongaonkar Antigenicity, and Parker Hydrophilicity Prediction [13,14]. The analysis was performed by either pasting up to 50 protein sequences in the FASTA format into a designated text box or uploading them as a single file. The lengths of protein sequences were between 10 and 6000 amino acids. The amino acid sequences of the queries (DENV serotypes 1–4: prME, KP406803.1, KP406804, KP406805, and KP406806, respectively) were input into FASTA format for this analysis.

3.3. Prediction of T-Cell Immunogenicity

The T-cell immunogenicity prediction model was constructed based on the enrichment of amino acids in immunogenic versus non-immunogenic peptides, as well as the positional importance scores of peptides presented by MHC class I molecules. The impact of the binding affinity on each peptide position was determined based on the specific MHC-I molecule involved. Six positions that contributed minimally to binding affinity were defined as non-anchor positions. Because anchor positions primarily reflect binding affinity rather than differences in immunogenicity, only non-anchor positions were used to evaluate immunogenicity. For this study, positions 1, 2, and 9, commonly recognized as anchor residues across most MHC molecules, were designated as default anchor positions. Peptide sequences were entered into the text field and separated by line breaks. The prediction model was exclusively validated for 9-mer peptides presented by HLA class I molecules. If longer peptides were predicted, additional amino acids were treated as being inserted after position 5 and assumed to possess the same weighting and masking characteristics as the amino acid at position 5. Consequently, peptides were normalized to a uniform 9-mer length in this study, and all candidate peptides were constrained to this length during the analysis.

3.4. Selection and Analysis of Antigen Candidates

The prM and E proteins are known to initially assemble into trimeric spikes protruding from the viral envelope, resulting in the irregular surface morphology of immature viral particles [4]. In the trans-Golgi network, host cell proteases cleave the prM protein to produce the mature M protein and rearrange the E protein trimers into dimers that lie flat on the viral envelope, forming the smooth surface observed in mature, infectious virions.

According to Wahala et al. [9], most serotype-specific neutralizing antibodies bind to the lateral ridge of EDIII. Therefore, antibodies targeting regions other than EDIII may have potential for vaccine development. To investigate this, epitope analysis was conducted sequentially on the structural proteins of DENV, followed by the E protein, specifically its domain III (EDIII).

3.5. Selection of Immunogenic Proteins Inducing B- and T-Lymphocyte Responses

The BepiPred-2.0 server predicted B-cell epitopes using a Random Forest algorithm trained on experimentally verified epitope and non-epitope amino acids derived from protein crystal structures.

Amino acid residues with prediction scores above the default threshold of 0.5 were classified as potential epitopes (Table 1). Based on this criterion, the full antigen sequence was analyzed to extract a list of epitopes that exceeded the threshold score. From this list, epitopes predicted to be located on the outer membrane and longer than nine amino acids were selected for further analysis.

3.6. Prediction of T-Cell Immunogenicity Based on Derived B-Cell Epitopes

The identified B-cell epitopes were fragmented into overlapping 9-mer peptide sequences, and T-cell immunogenicity scores were calculated for each sequence. Higher scores indicated a greater likelihood of eliciting a T cell-mediated immune response. Based on the findings of Calis et al. [10], it was confirmed that epitopes generally exhibited significantly higher immunogenicity scores than non-epitopes, with the average score for epitopes being approximately 0.1. Thus, 38–60 structural epitopes each were identified in DENV-1 to DENV-4 (Figure 1A–D). Some epitopes in EDI/II were identified in all DENV serotypes; however, no epitopes were predicted in the EDIII region of DENV-3 and DENV-4. Taken together, these results suggest that the EDIII domain is not essential for establishing a dengue-specific multivalent vaccine. To assess the real effects of vaccine candidate epitopes in comparison to in silico prediction, 25 epitopes per serotype were selected according to their scores (Figure 2). Based on these criteria, the final peptide sequences were determined and are listed (Table S1).

3.7. Establishment of an In Vitro Immunogenicity Assay to Evaluate Dengue Vaccine Candidates

To identify potential DENV vaccine candidates, a T-cell proliferation assay was conducted using CFSE labeling. This method utilizes the principle that CFSE-labeled T cells exhibit reduced fluorescence intensity upon antigen-driven proliferation, as previously described [8].

For each DENV serotype 1–4, 25 peptides were pooled into sets of five, resulting in 20 experimental groups. To prioritize peptide candidates for further immunological characterization, a predefined selection criterion was applied based on the magnitude of T-cell proliferation measured by CFSE dilution. Peptides inducing a ≥2.0-fold increase in T-cell differentiation relative to the negative control were considered to exhibit robust immunogenicity and were therefore selected for downstream analyses. In contrast, peptides showing only marginal increases (<2.0-fold) were excluded, as these responses were close to background levels observed in the assay. Following CFSE staining, immunogenicity was assessed for each group (Figure 3). In the T-cell proliferation assay, the levels of T-cell differentiation in the 1-e and 4-c groups were increased by approximately 2.2-fold and 2.1-fold, respectively, compared with the negative control group. In contrast, the 2-a to 2-e and 3-a to 3-e groups exhibited only modest increases in T-cell differentiation (1.1- to 1.5-fold), indicating relatively low immunogenicity. The gating parameters for IFN-γ, TNF-α, and IL-4 are shown in Figure S2.

To identify specific epitopes within these peptide pools that may contribute to the observed proliferative response, individual peptides from groups 1-e and 4-c were further evaluated in hPBMCs in a dose-dependent manner (Figure 4).

CFSE-stained hPBMCs were treated with increasing doses (2, 5, and 10 μg) of each peptide. Among the tested candidates, peptide 1-23 (EKEVAETQH) derived from the EDIII domain of DENV-1 and peptide 4-11 (GLETRAETW) derived from the matrix (M) protein of DENV-4 consistently exhibited increased CFSE dilution compared to the negative control. Based on duplicate experiments, these peptides showed approximately 1.6-fold and 1.3-fold increases in T-cell proliferation, respectively. Although these results are based on descriptive analysis, they suggest that discrete epitopes from distinct DENV serotypes may contribute to enhanced T-cell activation.

Taken together, these findings indicate that individual epitopes derived from viral structural proteins have the potential to elicit immunogenic responses and warrant further investigation.

3.8. Evaluation of Cellular Immunogenicity of Candidate Peptides in Balb/c Mice

To investigate the correlation between in vitro and in vivo immunogenicity, both the cellular and humoral immune responses to the selected peptides were evaluated in control mice.

No mortality or abnormal clinical signs were observed in any group during the experimental period. Furthermore, the body weight was not significantly different between the treatment groups (G2–G13) and the negative control group (G1; Figure 5B). The weight change in the mice according to the date is indicated in Appendix S1. However, in the IFN-γ ELISpot assay, several peptide-treated groups, specifically G6, G7, G8, G9, and G10, showed increased cytokine expression compared to the negative control group (Figure 5C). The raw data performed in triplicate using ELISpot assay is shown in Appendix S2. These findings suggest that epitopes derived from envelope domain III (EDIII) of DENV-1 and from non-EDIII structural regions of DENV-4, even when administered individually, enhance IFN-γ expression, indicating potential cellular immunogenicity.

3.9. Cytokine Analysis to Evaluate the Cellular Immunogenicity of Peptide Candidates

To validate the reproducibility of the cellular immunogenicity of the selected candidate peptides, cytokine production was analyzed ex vivo using splenocytes obtained from three mice per group. Intracellular staining of IFN-γ, TNF-α, and IL-4 was analyzed using flow cytometry. Among the experimental groups, mice immunized with peptides G3, G7, G8, G10, G11, or the DENV-4 combination (G7/G8/G9/G10/G11; G13) exhibited significantly elevated IFN-γ production compared to the negative control group (G1; p < 0.05; Figure 6B). Additionally, increased TNF-α expression was observed in groups immunized with peptides G3, G6, G7, the DENV-1 combination (G2/G3/G4/G5/G6; G12), or G13, relative to the negative control (Figure 6A). However, a downward trend in IL-4 expression was noted in the G4, G5, G6, G7, and G8 groups, although the differences were not significant (Figure 6C). Collectively, cytokine profiling of splenic lymphocytes revealed a trend toward enhanced IFN-γ and TNF-α production in the G3, G6, G7, G8, G10, G11, G12, and G13 groups. Notably, the G3, G7, and G13 groups consistently showed elevation of both IFN-γ and TNF-α, suggesting that these peptide formulations may effectively induce pro-inflammatory and antiviral immune responses. These findings suggest that domains derived from DENV-1 and DENV-4 EDIII and M regions may serve as potent immunogens by upregulating cytokines associated with antiviral defense and Th1-type cellular immunity.

3.10. EDIII of DENV-1 Regulates IFN-γ at the Transcriptional Level

To investigate whether the ED and M regions of DENV modulate host cytokine expression at the transcriptional level, we performed qRT-PCRs using probes specific for IFN-γ, TNF-α, and IL-4 using splenocytes obtained from two mice per group.

As shown in Figure 7A–C, a slight increase in IFN-γ mRNA expression was observed in the spleens of mice from the peptide G6, G7, G9, and G12 groups, compared to the negative control. Notably, TNF-α expression was significantly elevated in the peptide G3, G6, G12, and G13 groups. Increased TNF-α transcript levels were also observed in the peptide G9, G10, and G11 groups, although these differences did not reach statistical significance. A mild increase in IL-4 mRNA expression was detected in the peptide G6, G7, G9, G10, G11, and G12 groups, compared to the negative control group. However, none of these increases were significant. Collectively, the transcriptional analysis revealed that the G3, G6, G7, G9, G12, and G13 groups exhibited elevated expression of IFN-γ and TNF-α transcripts. Among these, groups G6, G9, and G12 demonstrated the most consistent and reproducible upregulation of both cytokines across experiments.

Taken together, these findings suggest that the tested candidate peptides, particularly those derived from the DENV ED and M regions, may promote the transcriptional activation of antiviral and pro-inflammatory cytokines such as IFN-γ and TNF-α, supporting their potential as effective immunogens in eliciting cellular immune responses.

3.11. Evaluation of Neutralizing Antibody Responses and Cross-Reactivity Between DENV-1 and DENV-4

To evaluate the correlation between humoral and cellular immune responses elicited by the selected antigen candidates in immunized naïve mice and to assess the potential induction of cross-neutralizing activity against DENV-1 and DENV-4, a plaque reduction neutralization test was performed. Following the primary immunization with the selected antigens, the titers of neutralizing antibodies against DENV-1 and DENV-4 were measured. No significant difference in neutralizing activity against DENV-1 and DENV-4 was observed in the sera collected after the first immunization (Figure 8A,B). However, after the third immunization, sera from the experimental groups G2 to G5, which received the DENV-1-derived antigen, showed a 1.3- to 2.8-fold increase in neutralizing activity against DENV-1 compared to pre-boost levels (Figure 8C). Theoretically, sera from groups G7 to G11, immunized with DENV-4-derived antigen, were expected to show an increase in neutralizing titers against DENV-4 following the third immunization; however, no significant change was observed at either the primary or tertiary time point (Figure 8D).

These findings suggest that cross-reactive neutralizing responses are not readily elicited between DENV-1 and DENV-4 when a single structural protein antigen is used. Furthermore, EDIII of DENV-1 may contribute to T cell-mediated immunity and induction of B cell-mediated neutralizing antibody responses.

4. Discussion

In this study, we evaluated peptide vaccine candidates for DENV based on computational predictions. BepiPred-2.0, an in silico prediction analysis tool, identified 100 potential 9-mer peptides from structural proteins of DENV (C, prM, M, E) that were expected to promote antigen-specific T-cell differentiation via MHC class I.

For candidate selection, we acknowledge that this selection strategy, which prioritizes peptides inducing strong CD4^+^/CD8^+^ T-cell proliferation in a CFSE-based in vitro assay, may have excluded epitopes that elicit moderate but potentially protective immune responses. In particular, epitopes derived from other serotypes or from distinct structural or non-structural protein regions may not have met the predefined cutoff despite their relevance in vivo.

However, the primary aim of this study was to identify highly immunogenic peptide candidates capable of inducing robust T-cell responses as a first-pass screening. Therefore, a stringent selection criterion was intentionally applied to reduce false-positive candidates and focus subsequent analyses on peptides with the greatest translational potential.

Further studies will incorporate a broader panel of peptides spanning additional serotypes and protein domains, as well as complementary immunological readouts, such as cytokine profiling and in vivo challenge models, to comprehensively assess epitope coverage and protective efficacy.

All peptides used in this study were synthesized as linear peptides without multimerization or carrier conjugation. The peptides were designed as 9-mer sequences to evaluate MHC class I (HLA-A2)-restricted T-cell immunogenicity rather than to function as standalone B-cell antigens. However, the in vivo results in mice may differ from these predictions, suggesting the need for further testing.

Importantly, the use of human HLA-A2-based in silico prediction in combination with in vivo immunogenicity testing in BALB/c mice represents an inherent limitation of the present study. While HLA-A2 was selected to prioritize peptide candidates with translational relevance to human immune responses, BALB/c mice express murine MHC class I molecules (H-2), which differ in peptide-binding preferences from human HLA alleles. As a result, peptide–MHC interactions and T-cell activation profiles observed in mice may not fully recapitulate those predicted for human HLA-A2-restricted responses.

Nevertheless, this cross-species evaluation was intentionally adopted as an initial screening strategy to identify broadly immunogenic peptides capable of eliciting measurable cellular immune responses in vivo. Similar approaches have been widely used in early-stage vaccine development to balance translational relevance with experimental feasibility. Future studies employing HLA-A2 transgenic or humanized mouse models will be required to more accurately assess HLA-restricted T-cell responses and to further validate the immunogenic potential of the selected peptide candidates.

The apparent overlap between B-cell epitope prediction and CD8^+^ T-cell peptide selection warrants clarification. In the present study, B-cell epitope prediction using BepiPred-2.0 was primarily employed as an initial filtering step to identify antigenic regions that are likely to be surface-exposed and immunologically relevant, particularly in the context of antibody recognition. This analysis was not intended to directly predict CD8^+^ T-cell epitopes.

Subsequently, candidate peptides derived from these antigenic regions were subjected to a separate in silico immunogenicity prediction framework specifically optimized for MHC class I-restricted CD8^+^ T-cell responses. This second step focused exclusively on 9-mer peptides and evaluated immunogenic potential based on amino acid enrichment and positional importance within HLA class I binding motifs. Thus, B-cell epitope prediction and CD8^+^ T-cell peptide selection were conducted as complementary but conceptually distinct steps within a hierarchical screening strategy.

Immunogenicity assessments were performed in both standard human HLA-A2 in hPBMC and splenocytes of transgenic mice. The vaccine candidates that consistently showed immunogenicity in both humans and mice included LEKEVAETQ (EDIII, G3), KEVAETQHG (EDIII, G6), GLETRAETW (M, G7), ICRRDVIDR (EDI/II, G9), and the two combination groups (G12 and G13). Additional immunogenicity evaluations using these candidate peptides are required to further optimize immune responses. Based on these findings, future studies should focus on the application of circular RNA platforms for in vitro and animal model testing to confirm immunogenicity.

Although EDIII, particularly its lateral ridge region, is known to be highly strain-specific and variable among DENV serotypes, EDIII-derived peptides were intentionally included in this study to experimentally evaluate whether such regions could still elicit measurable cellular immune responses across different serotypes. Consistent with previous reports, EDIII-derived peptides from DENV-3 and DENV-4 failed to induce significant immunogenicity in our in vitro proliferation assay (Figure 4), supporting the notion that EDIII-based epitopes may have limited cross-serotype applicability. These findings experimentally validate prior observations and underscore the importance of empirical screening when selecting epitope candidates.

The vaccine candidates G3, G6, G7, G9, and the peptide combinations G12 and G13 significantly increased T-cell proliferation. Furthermore, the cellular immune responses observed in standard animal models showed that these antigens tended to increase the expression of pro-inflammatory cytokines such as IFN-γ and TNF-α.

It is well established that effector cytokine production, such as IFN-γ secretion, can be regulated independently of extensive T-cell proliferation [14] and that individual T cells exhibit substantial heterogeneity in their functional cytokine output [15,16]. Moreover, even a limited population of highly functional effector T cells can disproportionately contribute to aggregate cytokine readouts measured by ELISpot assays [17,18,19]. This provides explanations for the elevated IFN-γ ELISpot response observed for peptide G2, despite its limited induction of T-cell proliferation in the CFSE-based assay.

The increase in IFN-γ expression may indicate that these vaccine candidates activate M1 macrophages, which play a key role in inflammatory responses and antiviral activity [20]. M1 macrophages promote inflammation and enhance antigen presentation. In contrast, M2 macrophages, which are polarized by IL-4, antagonize M1 macrophage activity and induce anti-inflammatory responses [21]. Interestingly, all vaccine candidates showed a tendency to increase IL-4 expression, suggesting that a positive feedback from activated M1 macrophages may temporarily increase IL-4 levels to counterbalance their activation. To further investigate these observations, future studies should use qRT-PCR or flow cytometry to assess IL-4 expression in spleen tissues at various time points following antigen administration.

In general, significant increases in neutralizing antibody levels by antigen challenge were observed in the serum from mice vaccinated with peptides G2–G5 (prM, EDIII, stem). These results align with cellular immunity data for G3. However, no change in neutralizing ability was observed with peptides G7–G11 (EDII + M) derived from DENV-4, suggesting distinct mechanisms for activating plasma B cells or cytotoxic T cells through different structural proteins of DENV. G3 (EDIII), G6 (EDIII), G7 (M), and G9 (EDI/II) showed strong cellular immune responses, whereas G2–G5 (prM, EDIII, stem) showed strong humoral immune responses. Therefore, the combination groups (G12 and G13) showing both strong cellular and humoral immunity are considered the most promising vaccine candidates. However, a single epitope can induce changes in immunogenicity by increasing IFN-γ and TNF-α levels [22].

A major limitation of the present study is the absence of direct in vivo challenge experiments to evaluate the protective efficacy of the selected peptide vaccine candidates against viral infection. While cellular immune responses, such as IFN-γ and TNF-α production, together with partial humoral immunity assessed by PRNT, provide important indicators of immunogenicity, these parameters do not necessarily correlate with protective immunity. The primary aim of this work was to prioritize highly immunogenic peptide candidates through a stepwise screening strategy, and therefore comprehensive viral challenge experiments were beyond the scope of the current study. Future investigations will focus on directly validating the protective capacity of the selected peptides using in vivo viral challenge models, including assessments of viral burden, disease severity, and survival following infection.

To further interpret the observed discrepancy in neutralizing antibody responses between DENV-1– and DENV-4-derived peptides, several mechanistic considerations should be taken into account.

The differential induction of neutralizing antibody responses between DENV-1– and DENV-4-derived peptides observed in this study may be attributed to intrinsic differences in the antigenic properties and functional roles of the respective protein domains. The DENV-1 peptide that elicited neutralizing activity is derived from envelope domain III (EDIII), a region well known to harbor both linear and conformational epitopes that are critical for receptor binding and virus neutralization [23,24,25]. In contrast, the DENV-4-derived peptide originated from the matrix (M) protein, which primarily contributes to virion assembly and maturation and is less frequently associated with the induction of potent neutralizing antibodies [26,27].

In addition, effective induction of neutralizing antibody responses often depends on the preservation of conformational epitopes, which may not be fully recapitulated by short linear peptide antigens [28,29]. This limitation may partially explain the lack of significant neutralizing activity observed in mice immunized with the DENV-4-derived peptide, despite its ability to elicit detectable cellular immune responses.

Although only one peptide (G5) induced a measurable neutralizing antibody response against DENV-1 following repeated immunization, this observation does not necessarily indicate a limitation of the in silico prediction strategy itself. The computational approach employed in this study was primarily designed to prioritize peptides with high T-cell immunogenic potential, rather than to directly predict neutralizing antibody epitopes, which are often dependent on conformational determinants and higher-order protein structure.

In this context, the identification of G5 as a peptide capable of eliciting neutralizing activity may be viewed as a downstream functional validation step, highlighting the necessity of experimental screening to complement in silico predictions. Thus, the current results underscore the complementary roles of computational epitope prioritization and in vivo immunological validation, rather than suggesting insufficient predictive power of the in silico approach.

Furthermore, the generation of robust neutralizing antibody responses requires adequate CD4^+^ T cell help, particularly through T follicular helper (Tfh) cell-mediated germinal center reactions that support B cell maturation and affinity maturation [30,31]. Although DENV-4-derived peptides induced cellular immune responses, these responses may not have sufficiently supported the differentiation of B cells into antibody-secreting plasma cells capable of producing high-affinity neutralizing antibodies.

Collectively, these findings suggest that cellular immune activation does not necessarily correlate with the induction of neutralizing humoral immunity, especially when single linear peptide antigens derived from distinct structural proteins are used [32]. The observed discrepancy between DENV-1 and DENV-4 highlights the importance of antigen selection and structural context in peptide-based vaccine design and underscores the need for further optimization to elicit balanced cellular and humoral immune responses.

Antibody-dependent enhancement (ADE) represents a critical safety concern in dengue vaccine development, whereby sub-neutralizing or cross-reactive antibodies facilitate viral entry into Fc receptor-bearing cells, potentially exacerbating disease severity upon secondary infection. This phenomenon has been extensively documented in both clinical and experimental settings and has posed a major challenge to the development of safe and effective dengue vaccines. In the present study, ADE was not directly evaluated, as the primary objective was to identify peptide candidates capable of inducing robust cellular immune responses and preliminary neutralizing activity as a first-pass immunogenicity screening. Importantly, the majority of candidate peptides evaluated here were short linear epitopes designed to preferentially stimulate T-cell responses rather than to elicit high titers of broadly cross-reactive antibodies, which are more commonly implicated in ADE. Nevertheless, the induction of neutralizing antibodies by selected peptides, particularly those derived from EDIII, underscores the necessity of carefully assessing both the magnitude and quality of antibody responses. Future studies will incorporate ADE-specific assays, including Fc receptor-mediated infection models and heterologous serotype challenge systems, to comprehensively evaluate the safety profile of these vaccine candidates. Such analyses will be essential to determine whether the immune responses elicited by these peptides confer protection without increasing the risk of ADE.

5. Conclusions

In silico, many immunogenic candidates showed high specificity against human T lymphocytes; however, this does not fully correspond to our in vitro results. Nonetheless, some of the predicted candidates regulated antiviral cytokines in vivo. This study will contribute to the development of vaccines against DENV and is a reference for discovering novel vaccine candidates.