Designing potent and immunogenic epitope based peptide vaccine against all serotypes of DENV via structural, physico-chemical and immunoinformatics-based approaches Authors: Aparna Chaudhuri, Bidyut Bandyopadhyay, Buddhadev Mondal, Aniket Sarkar, Sabyasachi Ghosh, Arnim Banerjee, Srijita Ari, Anindya Sundar Panja Published: 2026 ABSTRACT AbstractThe globally severe dengue infection caused by all four dengue virus serotypes continues to be a major public health burden. Thus, the present study aimed to design an epitope based peptide vaccine against the E protein using an immunoinformatic approach. Although there is a licensed vaccine against dengue, its efficacy varies across serotypes. We designed an epitope based vaccine against the E protein of dengue that is effective for all serotypes via rigorous computational screening. Immunoinformatic approaches have been used for identifying active domains and functional pockets of the DENV E protein, followed by epitope prediction, model generation, physico-chemical characterization, antigenicity, allergenicity, toxicity analysis. After screening, epitopes within the 45–65 amino acid range were considered to design the vaccine construct. Our final vaccine construct is highly antigenic (0.7991), non-allergenic, and non-toxic. Molecular docking and molecular dynamics simulations further confirmed the stability and high binding energy of the vaccine construct with TLR2 (ΔG = -13.9 kcal/mol), TLR3 (ΔG = -12.4 kcal/mol), human IgG heavy chain (ΔG = -11.7 kcal/mol, and light chain (ΔG = -11.3 kcal/mol). In silico immune simulations showed enhanced B-cell and T-cell responses. In silico cloning and codon optimization confirmed efficient translation of the vaccine construct and its successful expression in the Escherichia coli host system (E. coli K12 strain). However, in vitro and in vivo validations are required to confirm the above-mentioned computational results for the promotion of the vaccine efficacy process.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-57056-6.Keywords: DENV serotypes, E protein, Epitope based vaccine, Molecular docking, Molecular dynamics simulationSubject terms: Biotechnology, Computational biology and bioinformatics, Immunology, Microbiology Introduction Dengue, a viral ailment, caused by mosquito bites, represents a serious health issue for people across tropical and subtropical regions worldwide. Biological vectors such as Aedes aegypti and Aedes albopictus mosquitoes carry the dengue virus (DENV) to humans. DENV is a +ssRNA virus under the Flaviviridae family1,2. Another name for dengue is “breakbone fever” or “7-day fever” due to severe muscle and joint pain it causes3. DENV has four serotypes: DENV-1, DENV-2, DENV-3, and DENV-4. In 2007, DENV-5 (5th DENV serotype) was identified in Malaysia4,5. The DENV genome is approximately 10.6 kb long and encodes three structural proteins: Capsid (C), Envelope (E), and Membrane (M) proteins as well as seven non-structural proteins, including NS1, NS2A, NS2B, NS3, NS4A, NS4B as well as NS5 2. The DENV E protein, a surface protein, is present as 90 head-tail homodimers6 and plays an essential role in associating with host cell surface receptors, which mediate endocytosis along with the fusion mechanism. The E protein acts as a key target to neutralize antibody formation7. The E protein has three structural domains, including domain I (DI), domain II (DII), and domain III (DIII). The central domain is referred to as DI. The fusion loop - containing DII contributes to dimer formation. The immunoglobulin-like domain, DIII, serves as the receptor-binding domain8,9. DI comprises amino acids 1 to 52, 132 to 192 and 272 to 299. DII contains amino acids 53 to 131 and 193 to 271. DIII contains amino acid residues from 300 to 394 10. High mutation rates in the E protein drive genetic variation and transmission, impacting disease dynamics11. Mutations in the DI-DII hinge affect stability and neutralization sensitivity, while changes in DIII influence infectivity and immune evasion12,13. Dengue-related clinical symptoms include high fever (DF), skin rash, eye pain, headache, stomach pain, diarrhea, nausea, muscle pain, and joint pain14,15. Increased vascular permeability associated with DHF (Dengue Hemorrhagic Fever) may also alter in DSS (Dengue Shock Syndrome), which is a dangerous stage16. Between 2018 and 2024, there was a remarkable hike in dengue cases worldwide, with major outbreaks not only in endemic regions but also in previously unaffected areas. Brazil reported Over 2.9 million cases and about 1,000 deaths17. In 2023, roughly 6.5 million dengue cases were documented globally, affecting the countries across South America and Asia18. European countries like France and Italy reported severe outbreaks19,20. As per data from the National Vector Borne Disease Control Programme, there were 1,062,700 reported dengue cases and 1,526 deaths in India from 2021 to 2025 (https://ncvbdc.mohfw.gov.in/). At present, there are no reliable anti-DENV vaccines or medications. The two currently available dengue vaccines, Dengvaxia (Sanofi, Paris, France) as well as QDENGA® (Takeda, Japan), have limitations. Dengvaxia is suggested for individuals above the age of 9 years21 who have prior dengue exposure, as it may increase the risk of severe disease in seronegatives. So, WHO, in April 2018, advised antibody testing before its use22,23. Qdenga® protects children and adolescents aged 4–16 years in endemic areas but are less effective against DENV3 and DENV4 and offer no protection to dengue-naive individuals24,25. Due to these limitations and incomplete serotype coverage, there is a need to design novel epitope based peptide vaccines using in silico studies. These vaccines may inhibit all DENV serotypes and be cost-friendly, safe, easy to handle and time-efficient. Peptide structures can be modified into multi-epitope formats to enhance efficacy26,27. In this research work, retrieval of DENV E protein sequences, multiple sequence alignment to study active domains and functional pockets, structure prediction, identification of structural similarity, epitope prediction, various analyses, molecular docking along with simulation were done using in silico methods to select a novel epitope based peptide vaccine candidate effective against all DENV serotypes. But validation using both in vivo and in vitro approaches should be performed. Methods In this recent study, we aimed to design a novel epitope based peptide vaccine candidate effective against all DENV serotypes. The entire procedure of vaccine design is illustrated below in a flowchart (Fig. 1). A schematic workflow of the epitope based peptide vaccine design against all serotypes of DENV. Envelope (E) protein sequences retrieval FASTA-formatted sequences of DENV E proteins of all serotypes were retrieved from the UniProt (https://www.uniprot.org/) database. These sequences, approximately 400 ± 100 amino acids in length, were categorized by country, continent, and year28. Multiple sequence alignment (MSA) of envelope (E) protein of DENV The retrieved protein sequences underwent MSA utilizing a program named ClustalW available at https://www.ebi.ac.uk/Tools/msa/clustalo for identifying active domains and functional pockets, which utilizes Hidden Markov Models (HMM) profile-profile techniques29,30. HMM profiles detect sequence diversity, remote homolog, as well as domains of proteins31. The analysis was performed using default parameters. Structure prediction along with validation and followed by structural similarity identification 3D peptide structures of DENV were developed via homology modeling platform SWISS-MODEL (https://swissmodel.expasy.org/) 32. Suitable templates were identified through the Basic Local Alignment Search Tool (BLAST) along with the HMM-HMM-based lightning-fast iterative sequence search (HHblits) tools33,34. Protein sequences were compared by BLAST35. HHblits identifies remote homolog by comparing HMM profiles36,34. Structural validation was done with the help of SWISS-MODEL Structure Assessment server (https://swissmodel.expasy.org/assess) in light of the Ramachandran plot, sequence identity, MolProbity score (combination of clash score, Ramachandran outliers along with side-chain rotamer outliers), and QMEAN Z-score. A good model should have > 40% sequence identity, a low MolProbity score, ≥ 98% Ramachandran favored score, as well as >-4.0 Qualitative Model Energy Analysis (QMEAN Z score)37. RMSD (Root Mean Square Deviation) values were determined through PyMOL 0.99 software (http://www.pymol.org/pymol)38 to assess structural similarity by measuring the average distance between atoms in 3D peptide structures39,40. Detecting and evaluating physico-chemical properties, antigenicity, and topological parameters Physico-chemical properties of DENV E proteins were obtained through a server named Expasy ProtParam at https://web.expasy.org/protparam/ to calculate physical as well as chemical properties like molecular weight, amino acid number, predicted protein half-life, theoretical isoelectric point (pI), instability and aliphatic indices, along with the GRAVY (grand average of hydropathicity) value41. The N-end rule was applied to predict proteins’ half-life. This rule states that proteins’ stability in vivo is mainly considered by the nature of its N-terminal amino acid residue42. An instability index below 40 indicates that the protein is stable, but a value above 40 means that the protein may be unstable43. The aliphatic index exhibits the relative volume contributed by aliphatic side chains (particularly alanine, valine, isoleucine, and leucine) which is an indicator of enhanced thermostability of globular proteins44. At http://www.ddg-pharmfac.net/vaxijen/, a server named VaxiJen detected antigenicity by the alignment-free auto cross covariance (ACC) technique. Threshold value greater than 0.4 was considered for identify epitopes45. Transmembrane helices were further identified with the help of another server named TMHMM-2.0 at https://services.healthtech.dtu.dk/services/TMHMM-2.0/ for selecting potent epitopes46. CTL and HTL epitope prediction To identify epitopes of DENV E protein sequences, CTL (Cytotoxic T Lymphocytes) epitopes were detected via IEDB’s MHC-I tool at http://tools.iedb.org/mhci/, and HTL (Helper T Lymphocytes) epitopes were detected via IEDB’s MHC-II tool at http://tools.iedb.org/mhcii/, both utilizing the HLA (Human Leukocyte Antigen) allele reference set. The Length of T-cell epitopes was 9–10 for MHC-I and 15 for MHC-II. The NetMHCpan EL 4.1 prediction method, using artificial neural networks (ANNs) for MHC-I, as well as NetMHCIIpan EL 4.1 for MHC-II, also using artificial neural networks (ANNs) were utilized, respectively, which contain BA (for binding) and EL (for elution) predictors. The EL dataset contains human-derived 177 MHC-I molecules, mainly HLA-A, HLA-B, HLA-C, and HLA-E classes, as well as also contains three human MHC class II isotypes, including HLA-DR, HLA-DQ, and HLA-DP. According to the IEDB server, epitopes with a percentile rank ≤ 1 in the case of MHC-I, as well as ≤ 10 in the case of MHC-II, were selected as strong vaccine candidates47–49. B cell epitope prediction By means of the IEDB server (https://tools.iedb.org/bcell/) including the Bepipred Linear Epitope Prediction 2.0 mode that employs a Random Forest Algorithm, B cell epitopes were identified50. According to the IEDB server, epitopes with a threshold score above 0.5 were selected as potent vaccine candidates47,51,52. Antigenicity, allergenicity, toxicity along with immunogenicity analysis of epitopes Antigenic epitopes were detected by the VaxiJen (http://www.ddg-pharmfac.net/vaxijen/) server which applies alignment-independent auto cross covariance (ACC) method which means conversion of protein sequences into uniform vectors of identical length with threshold value > 0.4 for selection45. Allergenicity was determined using AllerTOP (https://www.ddg-pharmfac.net/AllerTOP/) by ACC method53. Epitope toxicity was assessed via another server named ToxinPred at http://crdd.osdd.net/raghava/toxinpred/54, whereas IEDB (http://tools.iedb.org/immunogenicity/) was used to assess immunogenicity47. BLASTp analysis to prevent potential autoimmune response Comparison of predicted epitopes with the host proteome was employed by BLASTp (https://blast.ncbi.nlm.nih.gov/) to know resemblance for avoiding autoimmune disorders. To avoid potential autoimmune reactions and cross-reactivity, each selected epitope was screened against the UniProt database for homology with the human proteome using BLASTp. Mainly two parameters were considered such as query coverage and percentage identity. Query coverage denotes the percentage of the provided sequence that aligns with the target sequence. On the other hand, percentage identity indicates the extent of similarity over the aligned regions of the sequences55,56. Analysis of cytokine - inducing ability of epitopes Analysis of the cytokine-inducing ability of HTL or MHC class II epitopes is a key step to design effective vaccines. The cytokine IL-4 inducing ability of epitopes were analyzed with the help of the IL4Pred (http://crdd.osdd.net/raghava/il4pred/) server using the machine learning technique, support vector machine (SVM)57. On the basis of pairing and composition of amino acids, SVM scores were calculated to do prediction. Positive SVM scores mainly denote higher specificity in prediction, and also larger scores specify better confidence that the peptide will stimulate IL-4 production. Peptides capable of triggering the anti-inflammatory cytokine IL-4 are considered to possess anti-inflammatory properties51. Design of epitope based vaccine candidate To construct the epitope based vaccine candidate, mainly CTL, HTL along with linear B-cell epitopes were considered. These CTL, HTL and B-cell representing unique epitopes were screened to develop an epitope based vaccine candidate designed through in silico approaches. After all the previously mentioned analyses, two epitopes were screened within the amino acid ranges of 45–65 (LTKTTAKEVALLRTYCIEASI) and 346–390 (TQNGRLVTTNPIVTDKERPV), respectively. Adjuvant, human β-defensin 2 (HBD-2) (PDB ID: 1FD3), was added at the N-terminal end of the epitopes by the linker EAAAK51. Antigenicity, allergenicity, toxicity analysis of constructed vaccine candidate Antigenicity of constructed vaccine candidate was detected by the VaxiJen server, with a threshold value > 0.4 for selection45. Allergenicity was determined using the AllerTOP server53. Toxicity of vaccine candidate was assessed via the ToxinPred server54. Population coverage Analysis of the selected T cell epitopes Population coverage analysis of the predicted CTL and HTL epitopes was performed by the IEDB Population Coverage Tool (https://tools.iedb.org/population/). The analysis was performed using default parameters. HLA allele frequencies (both MHC-I and MHC-II) were considered to assess global population coverage51. 3D modeling of vaccine candidate The 3D structure of selected vaccine candidates was modeled using the SWISS-MODEL server (https://swissmodel.expasy.org/) 32. Structure refinement along with structural assessment Refinement of the 3D structure of the selected vaccine candidate was performed by Galaxy Refine (https://galaxy.seoklab.org/cgi-bin/submit.cgi?type=REFINE) to improve the vaccine’s structural quality. During refinement, the protein side chains were rebuilt as well as repacked through the CASP10 refinement, and structural relaxation occurred during molecular dynamics simulation58. After that, the structural quality of both the initial and refined structures was assessed by generating Ramachandran plot utilizing the online server PROCHECK (https://www.ebi.ac.uk/thornton-srv/software/PROCHECK/) 52. This analysis evaluates amino acid residues’ dihedral angles for ensuring that they fall within the Ramachandran plot’s favored regions. Besides, ProSA-web (https://prosa.services.came.sbg.ac.at/prosa.php) was also used to assess the predicted models’ structural quality via Z-scores52. Discontinuous B - cell epitope prediction After refinement and validation of the 3D structure of the selected vaccine candidate, conformational (discontinuous) B-cell epitopes were predicted by the ElliPro server (http://tools.iedb.org/ellipro) on the basis of geometrical features of the structure, solvent accessibility, and conformational flexibility of the protein. Here, two parameters, such as minimum score and maximum distance (Angstrom) were set to 0.5 and 6, respectively, for prediction59. Disulfide engineering of the vaccine candidate Disulfide engineering of the vaccine candidate was performed through the Disulfide by Design 2 (DbD2) server (http://cptweb.cpt.wayne.edu/DbD2/index.php) to increase protein stability by incorporating disulfide bonds via the substitution of cysteine residues at selected positions. The analysis was performed using default parameters. The default value of the χ3 angle is + 97° ±30° and − 87° ±30°, and the bond energy is < 2.2 kcal/mol60,61. Molecular docking with binding energy estimation The 3D structures of two Toll-like Receptors, including TLR2 (PDB ID: 2Z7X) and TLR3 (PDB ID: 2A0Z), as well as human IgG (PDB IDs: 6DFJ for DENV-1, 5LCV for DENV-2, 7A3T for DENV-3, and 4AM0 for DENV-4), were retrieved from the RCSB Protein Data Bank (https://www.rcsb.org/. After that, all water molecules as well as ligands connected with these 3D structures were removed. Then, the server named HDOCK (http://hdock.phys.hust.edu.cn/) was employed for molecular docking to predict the interactions of DENV E protein epitopes with TLR2, TLR3, and human IgG62,63. Here, the docking scores were calculated with the help of the server’s knowledge-based iterative scoring functions, such as ITScorePP or ITScorePR. The PRODIGY server (https://wenmr.science.uu.nl/prodigy/) was used to estimate the binding energy values either as Gibbs free energy (ΔG, unit: kcal mol− 1) or dissociation constant (Kd, unit: M) of the docked complexes64. Here, 25 °C was set as the default temperature for calculating the dissociation constant (Kd). The interatomic contacts (ICs) at the protein–protein interface, utilizing a threshold value of 5.5 Å distance, were calculated by the PRODIGY server. According to Vangone and Bonvin (2015), experimentally calculated binding affinity is related to the contact numbers present at the protein–protein interface. PRODIGY follows this rule65. At last, the PDBSum Generate server (https://www.ebi.ac.uk/thornton-srv/databases/pdbsum/Generate.html) was used to visualize the interactions of amino acid residues between the vaccine construct and target receptors utilizing Prot-Prot analysis. The intermolecular interactions of the docked complexes were further observed by Discovery Studio molecular viewer software51,66,67. Molecular dynamics (MD) simulation and calculation of Molecular Mechanics/Poisson-Boltzmann Surface Area (MM/PBSA) Molecular dynamics (MD) simulations were carried out on the finally selected receptor–ligand complexes, such as the vaccine construct containing epitopes within the 45–65 amino acid range, complexed with TLR2, TLR3, and the specified human IgG (4AM0), using the GROMACS simulation package. These docked complexes were fixed in a TIP3P water model under periodic boundary conditions with a 10 Å buffer distance. These were placed at the center of a cubic simulation box (90.0 × 90.0 × 90.0 nm³). The CHARMM27 force field was applied for system parameterization. The system’s explicit solvation was carried out using the SPC water model, followed by the inclusion of counter ions (Na⁺ and Cl⁻) to neutralize the whole charge by replacing proper solvent molecules68–71. The system then underwent energy minimization to eradicate steric clashes, followed by a 100 ns NPT simulation at 300 K along with 1 atm pressure, using Nosé–Hoover temperature coupling and the Martyna–Tobias–Klein barostat. PME electrostatics were applied using a grid spacing of 0.8 Å. Trajectory coordinates along with energy outputs were recorded every 1000 ps using the gmx rms module72. The structural stability and dynamic properties of the complex were analyzed through root mean square deviation (RMSD), root mean square fluctuation (RMSF), radius of gyration (Rg), solvent accessible surface area (SASA), secondary structures, and hydrogen bond formation73. Binding free energy was then assessed from the final 100 ns simulation trajectories using the GMX_MMPBSA tool74,75. The total binding free energy (ΔG_binding) was determined using the following equation: Here, ΔE_MM (ΔE_MM = ΔE_vdW + ΔE_EE) represents the molecular mechanics energy in vacuum, including van der Waals (ΔE_vdW) and electrostatic (ΔE_EE) interactions. ΔG_ps and ΔG_nps stand for the polar and nonpolar solvation free energies, respectively, while TΔS represents the entropic contribution to binding. Codon optimization along with in silico cloning Codon optimization in the Escherichia coli host system (E. coli K12 strain) was performed by the Java Codon Adaptation Tool (JCat) (https://www.jcat.de/). Then the optimized cDNA sequence was evaluated to avoid transcription terminators, ribosome-binding sites, and unwanted restriction sites. GC content and Codon Adaptation Index (CAI) values were calculated to estimate expression efficiency. According to this tool, optimal gene expression and stability require a GC content of 30–70% and Codon Adaptation Index (CAI) value above 0.8 or closer to 1. Additionally, many restriction enzyme sites were added to enable in silico cloning in the pET-28a(+) vector with the help of SnapGene 6.2 software51. In silico immune simulation Immunological simulation of the vaccine candidate was performed by the C-ImmSim server (https://kraken.iac.rm.cnr.it/C-IMMSIM/index.php) to evaluate its immunogenic characteristics along with the potential immunological response within the immune system of human. In this server, position-specific scoring matrices (PSSM) along with machine learning approaches were utilized to evaluate the cytokines, interferons, and antibodies production. It also assesses T helper 1 (Th1) and T helper 2 (Th2) cell responses. The analysis was performed using default parameters, including random seed, simulation steps, and simulation volume. At four-week intervals, corresponding to time steps of 1, 84, and 168, the vaccine was administered in three injections in this study51,76,77. Results Retrieval of DENV envelope (E) protein sequences FASTA-formatted sequences of DENV E proteins of all serotypes were retrieved from the UniProt database. These sequences, approximately 400 ± 100 amino acids in length, were categorized by country, continent, and year. There were twenty DENV-1, twelve DENV-2, five DENV-3, and eight DENV-4 strains. The DENV-1 strains were obtained from Asia, North America, South America, and Australia. DENV-2 strains were from Asia, North America, and South America. DENV-3 strains were from Asia. DENV-4 strains were from Asia and North America (Supplementary Tables 1–4). Multiple sequence alignment (MSA) of DENV E protein sequences MSA of the retrieved DENV E protein sequences were analysed via ClustalW program to specify active domains and functional pockets. On the basis of various literature review, the specified active domains and functional pockets were positioned at amino acid ranges 45–65, 110–140, 190–225, 255–290, 300–345, and also 346–390 78, 79, 2. Fifteen DENV-1, seven DENV-2, three DENV-3 along with five DENV-4 strains within the above-mentioned amino acid ranges were considered after final MSA analysis (Supplementary Fig. 1A-4 C). Structure prediction along with validation Peptides’ model formation was done by the SWISS-MODEL homology modeling server. Validation of structures was done by the SWISS- MODEL Structure Assessment Tool based on QMEAN Z score, sequence identity, Ramachandran Plot, and MolProbity score. According to the SWISS-MODEL server, a good model should have > 40% sequence identity, low MolProbity score, ≥ 98% Ramachandran favored score as well as >-4.0 QMEAN Z score. There were six models for the previously specified 45–65 ranges, ten for 110–140 ranges, six for 190–225, six for 255–290, twelve for 300–345, and thirteen for 346–390 ranges (Supplementary Table 5). After that the RMSD of the predicted structures was calculated. Identification of structural similarity RMSD (Root Mean Square Deviation) calculation helps to analyze structural similarity along with to measure the average distance between the predicted 3D models. On the basis of RMSD calculations, one representative strain (for each specified amino acid range) was considered (Supplementary Tables 6–11). The accession numbers of the finally selected DENV strains’ were A0A0D4D8J0 (45–65 range), A0A0N7H514 (110–140 range), A0A0D4D8J0 (190–225 range), A0A0D4D9E3 (255–290 range), A0A0D4D8J0 (300–345 range), and J9WMM0 (346–390 range) (Supplementary Tables 6–11). Finally, these six strains, corresponding to the six above-mentioned ranges were selected on the basis of > 40% sequence identity, low MolProbity score, >-4.0 QMEAN Z score, ≥ 98% Ramachandran favored score (Table 1) as well as RMSD calculation (Supplementary Tables 6–11). Table 1Details of Sequence identity, MolProbity score, QMEAN Z score, and Ramachandran favored score for the six finally selected DENV strains.Strain with amino acid rangesSequence IdentityMolProbity scoreQMEAN Z scoreRamachandran favored scoreA0A0D4D8J0(45–65 range)62.86%0.50-2.77100%A0A0N7H514(110–140 range)96.77%0.50-2.87100%A0A0D4D8J0(190–225 range)74.29%1.09-0.5296.97%A0A0D4D9E3(255–290 range)91.67%0.50-1.77100%A0A0D4D8J0(300–345 range)47.73%0.580.6997.62%J9WMM0(346–390 range)93.99%0.50-0.04100% Details of Sequence identity, MolProbity score, QMEAN Z score, and Ramachandran favored score for the six finally selected DENV strains. A0A0D4D8J0 (45–65 range) A0A0N7H514 (110–140 range) A0A0D4D8J0 (190–225 range) A0A0D4D9E3 (255–290 range) A0A0D4D8J0 (300–345 range) J9WMM0 (346–390 range) 3D structural details of the six finally selected DENV strains of different serotypes across various amino acid ranges, obtained from the SWISS-MODEL server, and after RMSD analysis, are depicted in supplementary figure (Supplementary Fig. 5–10). Assessment of physico-chemical properties, antigenicity and topology By the Expasy ProtParam tool, the physico-chemical characteristics of DENV E proteins were analyzed to calculate molecular weight (MW), amino acid number, predicted protein half-life, theoretical isoelectric point (pI), aliphatic indices, instability indices (values below 40 means stable), and GRAVY (Grand average of hydropathicity). Antigenicity was evaluated through the Vaxijen server while the TMHMM v2.0 server was employed for identifying transmembrane helices (TM Helix) for selecting potential epitopes. The previously selected six strains (the accession number details and serotype names were provided in supplementary Tables 1–4), obtained after RMSD calculation, showed more favorable physico-chemical properties than other strains. Details of physico-chemical properties were depicted in supplementary table (Supplementary Table 12). The finally selected six strains showed favorable pI, high stability (as their instability indices were below 40) along with prolonged half-life exceeding 30 min under in vitro conditions in mammalian cells. Their high aliphatic indices indicate good thermal stability throughout an extensive temperature range. The finally selected six DENV strains were A0A0D4D8J0 (45–65 range) of DENV-1 and DENV-4, A0A0N7H514 (110–140 range) of DENV-3, A0A0D4D8J0 (190–225 range) of DENV-1 and DENV-4, A0A0D4D9E3 (255–290 range) of DENV-4, A0A0D4D8J0 (300–345 range) of DENV-1 and DENV-4, and J9WMM0 (346–390 range) of DENV-1. Antigenicity scores and number of TM helices of the selected dengue peptides are represented below (Table 2). Table 2Antigenicity score and TM helix number of dengue peptides of different serotypes within specified amino acid ranges.Strain with amino acid rangeAntigenicity(Threshold value 0.4)TM Helix numberA0A0D4D8J0(45–65)0.66380A0A0N7H514(110–140)0.71620A0A0D4D8J0(190–225)0.69990A0A0D4D9E3(255–290)0.51820A0A0D4D8J0(300–345)0.60770J9WMM0(346–390)0.87460 Antigenicity score and TM helix number of dengue peptides of different serotypes within specified amino acid ranges. The six finally selected DENV strains were highly antigenic, with scores surpassing the threshold value of 0.4. The predicted number of TM helices was also zero. This is good because proteins containing more than two TM helices are not considered for vaccine design. A key trait of an ideal vaccine candidate is maximum surface exposure. Screening of CTL, HTL along with B cell epitopes Effective vaccine development requires mimicking natural immunity by stimulating a persistent adaptive immunological response, primarily involving the activity of T cells and B cells. The humoral immune response is activated by B cell epitopes through the production of antibodies, whereas cellular immunity is triggered by Cytotoxic T Lymphocytes (CTL) epitopes, which provide long-term protection. Helper T lymphocytes (HTL) epitopes also help to activate both T and B cells. To identify epitopes within DENV E protein sequences, CTL epitopes were predicted using the IEDB’s MHC-I tool, and HTL epitopes were predicted using the IEDB’s MHC-II tool, both utilizing the HLA allele reference set. Two methods named NetMHCpan EL 4.1 and NetMHCIIpan EL 4.1 were used for MHC-I and MHC-II prediction, respectively, which contains both BA (for binding) and EL (for elution) predictors. According to the IEDB server, epitopes with percentile rank ≤ 1 in the case of MHC-I and ≤ 10 in the case of MHC-II were selected as strong vaccine candidates. The best epitopes were screened utilizing MHC-I alleles (HLA-B*08:01; HLA-B*15:01; HLA-B*35:01; HLA-B*40:01; HLA-B*44:03; HLA-B*51:01; HLA-B*58:01; HLA-A*11:01; HLA-A*26:01; HLA-A*30:01; HLA-A*31:01; HLA-A*33:01; HLA-A*68:01) and MHC-II alleles (HLA-DRB1*09:01; HLA-DRB1*12:01; HLA-DRB1*13:02; HLA-DRB1*15:01; HLA-DRB3*02:02). With the help of the IEDB server, including the Bepipred Linear Epitope Prediction 2.0 method, that comprises Random Forest Algorithm, B cell epitopes were acknowledged. According to the server named IEDB, epitopes with a threshold score above 0.5 were selected as potent vaccine candidates. A total of six HTL epitopes (MHC-II), six CTL epitopes (MHC-I) along with six B-cell epitopes were obtained within the specified amino acid ranges. On the basis of literature review, the selected amino acid ranges were 45–65, 110–140, 190–225, 255–290, 300–345, and also 346–390 78, 79, 2. The predicted T cell epitopes (MHC-I : Supplementary Table 13; MHC-II : Supplementary Table 14) along with B-cell epitopes (Supplementary Table 15) from the selected six strains within six specified amino acid ranges are represented in tabular form. Epitopes in the given sequences were highlighted by yellow colour. Antigenicity, allergenicity, toxicity along with immunogenicity analysis of epitopes On the basis of antigenicity, allergenicity, toxicity along with immunogenicity, six epitopes from the selected DENV strains of different serotypes within specified amino acid ranges for epitope based peptide vaccine design were considered. Antigenicity analysis was done by the VaxiJen v2.0 server which employs an alignment-free method which means conversion of protein sequences into uniform vectors of identical length using auto cross covariance (ACC). Threshold values greater than 0.4 was considered for identify antigenic epitopes. This action was crucial for ensuring that the selected epitopes could elicit immune responses without harming host tissues. AllerTOP was utilized to determine allergenicity of epitopes based on the ACC method. Epitope toxicity was assessed via the ToxinPred server. Immunogenicity of epitopes was studied through the IEDB server. The selected epitopes showed highest binding affinity with both MHC I and MHC II molecules as well as recognition by B cells. Highly immunogenic, highly antigenic, non-allergenic, and non-toxic epitopes of DENV strains A0A0D4D8J0 (45–65 range), A0A0N7H514 (110–140 range), A0A0D4D8J0 (190–225 range), and J9WMM0 (346–390 range) were considered (Table 3). Table 3Details of antigenicity, allergenicity, toxicity, and immunogenicity of epitopes of finally selected six DENV strains of different serotypes within amino acid ranges.Strain within amino acid rangesEpitopesAntigenicityAllergenicityToxicityImmunogenicityA0A0D4D8J0(45–65) LTKTTAKEVALLRTYCIEASI 0.6638Non-allergenNon-toxin0.18147A0A0N7H514(110–140) FQCLESIEGKVVQHENLK 0.6832Non-allergenNon-toxin0.03925A0A0D4D8J0(190–225) KMKKKTWLVHKQWFLDLPLPWTAGAD 0.5471Non-allergenNon-toxin0.13369A0A0D4D9E3(255–290) SQEGAMHTALTGAT 0.3929Non-allergenNon-toxin0.1183A0A0D4D8J0(300–345) MAETQHGTTVVKVKYEGAGAPCKVPIEIRDVNK 0.6888Non-allergenNon-toxin-0.37524J9WMM0(346–390) TQNGRLVTTNPIVTDKERPV 0.696Non-allergenNon-toxin0.32095 Details of antigenicity, allergenicity, toxicity, and immunogenicity of epitopes of finally selected six DENV strains of different serotypes within amino acid ranges. A0A0D4D8J0 (45–65) A0A0N7H514 (110–140) A0A0D4D8J0 (190–225) A0A0D4D9E3 (255–290) A0A0D4D8J0 (300–345) J9WMM0 (346–390) BLASTp of epitopes BLASTp analysis was done through local alignment to know similarity between the epitopes and the host proteome, to avoid autoimmune disorders. Epitopes of DENV strains A0A0D4D8J0 (45–65 range) and J9WMM0 (346–390 range) were selected. We excluded the epitopes whose number of BLAST scores was comparatively higher (Supplementary Table 16). Cytokine - inducing ability analysis of epitopes The cytokine - inducing ability of HTL (MHC class II) epitopes is a key step to design effective vaccines. The IL-4 inducing ability of epitopes was analyzed with the help of the IL4Pred server. Epitopes of the DENV strain A0A0D4D8J0 (45–65 range) were IL-4 inducer as showed positive Support Vector Machines (SVM) score (1.08) but epitopes of the DENV strains J9WMM0 (346–390 range) were non IL-4-inducer as showed negative SVM score (-0.22). Design of epitope based vaccine candidate To construct the epitope based vaccine candidate through an in silico study, CTL, HTL along with linear B-cell epitopes were considered. These CTL, HTL and B-cell representing unique epitopes were utilized in this study. Adjuvants are added frequently at the N-terminal end of a vaccine candidate for enhancing immune responses, mostly cellular immunity. In epitope-based peptide vaccines, they help to develop long-term immune memory as well as strengthen impaired immune responses in older individuals. Human β-defensin is a potent adjuvant, normally added to the N-terminal end of a peptide or protein antigen for increasing immune responses with the help of the EAAAK linker. The EAAAK linker boosts the formation of a firm α-helical structure, enhances structural stability, and decreases steric hindrance between functional domains80,81,51. Human β-defensin adjuvant (HBD-2) along with the specific linker EAAAK, was incorporated into the construction of the vaccine candidate to enhance protein stability, proper folding along with the separation of functional regions. Human β-defensin 2 (HBD-2) (PDB ID: 1FD3) was added at the N-terminal end of the epitopes by the linker EAAAK51. The design of our epitope based peptide vaccine construct is depicted below: The total sequence of this epitope represents CTL, HTL, and B cell epitopes. Within the 45–65 amino acid range, the sequence of the first vaccine construct (HBD-2 + EAAAK linker + Epitope) is: “GIGDPVTCLKSGAICHPVFCPRRYKQIGTCGLPGTKCCKKPEAAAKLTKTTAKEVALLRTYCIEASI”. and within the 346–390 amino acid range, the second vaccine construct (HBD-2 + EAAAK linker + Epitope) is: “GIGDPVTCLKSGAICHPVFCPRRYKQIGTCGLPGTKCCKKPEAAAKTQNGRLVTTNPIVTDKERPV”. We assigned two short names to the first and second vaccine constructs, such as DSVAX 1 and DSVAX 2, respectively. Antigenicity, allergenicity, toxicity of constructed vaccine candidate Antigenicity was detected by the VaxiJen server which applies an alignment-independent auto cross covariance (ACC) method with a threshold value > 0.4 for selection. Allergenicity was determined with the AllerTOP server by the ACC method. Epitope toxicity was assessed via another server named ToxinPred. DSVAX 1 was highly antigenic (score 0.6690), non-allergenic, and non-toxic. DSVAX 2 was antigenic (score 0.6480), non-allergenic, and non-toxic. Population coverage Analysis of the selected T cell epitopes Population coverage analysis of the predicted CTL and HTL epitopes was performed by the IEDB Population Coverage Tool. HLA allele frequencies (both MHC-I and MHC-II) were considered to assess global population coverage, which helps to know the capability of the vaccine epitopes to provide protection across various geographical regions. The result exhibited that the total epitope based peptide vaccine candidate, incorporating all selected CTL and HTL epitopes, achieved population coverage of 95.41%. This finding indicates the strong potential of the designed vaccine candidate for broad-spectrum immunization. A graphical representation of world population coverage is provided in the supplementary file (Supplementary Fig. 11). Structure prediction of vaccine candidate The 3D structure of the vaccine candidates DSVAX 1 and DSVAX 2 were predicted directly with the help of the SWISS-MODEL server (Supplementary Fig. 12). Structure refinement along with structural assessment Refinement of the 3D structures of the vaccine candidates DSVAX 1 and DSVAX 2 was performed by Galaxy Refine. During refinement, the protein side chains were rebuilt as well as repacked through CASP10 refinement, and structural relaxation occurred during molecular dynamics simulation. After that, the structural quality of initial and refined models of both DSVAX 1 and DSVAX 2 was assessed by generating Ramachandran plots utilizing the PROCHECK server. This analysis evaluates amino acid residues’ dihedral angles for ensuring that they fall within the Ramachandran plot’s favored regions. According to the Ramachandran plot analysis, DSVAX 1 exhibited higher structural quality than DSVAX 2. Initial model of DSVAX 1 showed 92.3% of residues in the most favoured regions along with 7.7% in allowed regions whereas refined model of DSVAX 1 showed 96.2% of residues in the most favoured regions along with 3.8% in allowed regions (Supplementary Fig. 13 a-b). In comparison, initial model of DSVAX 2 showed 90% of residues in the most favoured regions, 6.7% in allowed regions along with 3.3% in disallowed regions whereas refined model of DSVAX 2 showed 96.7% of residues in the most favoured regions along with 3.3% in disallowed regions (Supplementary Fig. 14 a-b). As per the analysis of Ramachandran plot, the total residues in the most favoured regions were within the optimal range. More than 90% of residues in the most favoured regions indicate any model’s reliability. These findings conclude that DSVAX 1 represented a maximum proportion of residues in the favoured regions which indicated superior overall structural quality compared to DSVAX 2. Besides, ProSA-web was also used to assess predicted models’ structural quality via Z-scores. According to ProSA-web server, reliable model’s Z-score must lie within the range of native protein. The ProSA analysis also supported the best quality of DSVAX 1, which showed the Z-score of − 4.01 in initial model and most favourable Z-score of -5.49 in refined model (Supplementary Fig. 13 c-d). In contrast, DSVAX 2 initial model showed a Z-score of -3.50 and DSVAX 2 refined model showed a Z-score of -3.77 (Supplementary Fig. 14 c-d). More negative Z-score value indicates better quality of a model. All these findings indicate that DSVAX 1 is the most suitable candidate. Discontinuous B - cell epitope prediction After refinement and validation, conformational (discontinuous) B-cell epitopes were predicted by the ElliPro server on the basis of geometrical features of the structure, solvent accessibility, and conformational flexibility of the protein. The minimum score and maximum distance (Angstrom) were set to 0.5 and 6, respectively, for prediction. Conformational (discontinuous) B-cell epitopes of DSVAX 1 and DSVAX 2 showed predicted scores above 0.5. But the score of DSVAX 1 was higher than DSVAX 2. It signifies that DSVAX 1 possesses strong antigenic potential as well as suitable surface accessibility. The Graphical presentation of the findings was incorporated in the supplementary file (Supplementary Fig. 15–16), and the details of the scores and regions of epitopes were represented in tabular form (Supplementary Tables 17–18). Disulfide engineering of the vaccine candidates Disulfide engineering of the vaccine candidates (DSVAX 1 and DSVAX 2) was performed through the Disulfide by Design 2 (DbD2) server to increase protein stability by incorporating disulfide bonds via the substitution of cysteine residues at selected positions. As a result, protein stability was enhanced by strengthening its structural integrity, and also resistance was increased to denaturation. The DbD2 server helps to detect residue pairs with proper geometric along with energy properties for cysteine substitution as well as disulfide bond formation in the model of vaccine candidates. The default value of the dihedral angle (χ3) is + 97° ±30° and − 87° ±30°, and the bond energy is < 2.2 kcal/mol. The expected four residue pairs along with suitable parameters in case of DSVAX 1 were as follows: Alanine 13 paired with Lysine 39 at + 86.69° and bond energy was 2.19 kcal/mol; Cysteine 20 paired with Tyrosine 24 at + 120° and bond energy was 2.98 kcal/mol; Glycine 28 paired with Cysteine 37 at + 103.09° with a bond energy of 2.26 kcal/mol; and Cysteine 30 paired with Threonine 35 at -98.32° and bond energy was 3 kcal/mol. The expected three residue pairs along with suitable parameters in case of DSVAX 2 were: Aspartic acid 4 paired with Threonine 7 at + 103.30° and bond energy was 4.27 kcal/mol; Cysteine 38 paired with Tyrosine 24 at + 81.18° and bond energy was 6.77 kcal/mol; and Leucine 32 paired with Threonine 35 at + 101.55° with a bond energy of 2.3 kcal/mol. The designed residue pairs in both DSVAX 1 and DSVAX 2 enhance the stability and rigidity of the modelled proteins by forming disulfide bonds. But capability of DSVAX 1 is higher as per the greater number of residue pairs, more favourable dihedral angles, and lower bond energy values. The images of these analyses are included in the supplementary file (Supplementary Fig. 17–18). Molecular docking The HDOCK (Highly Integrated DOCKing) server was employed for molecular docking to evaluate the interactions of the constructed vaccine candidates DSVAX 1 and DSVAX 2 with Toll-like Receptor 2 (TLR2), Toll-Like Receptor 3 (TLR3), and human Immunoglobulin G (IgG). Intrinsic scoring functions were utilized in the HDOCK server. The binding energy of the docked complexes was calculated by utilizing the PRODIGY (PROtein binDIng enerGY prediction) server. The two vaccine candidates were individually docked with TLR2, TLR3, and human IgG. The 3D structures of TLR2 (PDB ID: 2Z7X), TLR3 (PDB ID: 2A0Z), and human IgG (PDB IDs: 6DFJ for DENV-1, 5LCV for DENV-2, 7A3T for DENV-3 as well as 4AM0 for DENV-4) were retrieved from the RCSB Protein Data Bank. Four human IgG antibodies were used for individual docking. These antibodies’ heavy chains (H) as well as light chains (L) were used for docking with the specified vaccine candidates. Discovery Studio Molecular viewer and the PDBSum Generate server (Prot-Prot analysis) were used to visualize the interactions of amino acid residues between the vaccines constructs and target receptors. All these docking images are represented below (Figs. 2, 3, 4 and 5). After docking, it was found that DSVAX 1 exhibited higher docking score as well as high binding energy when interacting with TLR2, TLR3, and the DENV-4 inhibiting human IgG (PDB ID:4AM0). The docking scores along with their corresponding binding energy values are depicted below in tabular form (Tables 4 and 5). Table 4Details of docking scores along with their corresponding binding energy values of DENV vaccine constructs with TLR2 and TLR3 docked complexes.Docked complexDocking ScoreBinding Energy(kcal/mol)Vaccine Candidate with amino acid rangeTLRDSVAX 1TLR2-199.84-13.9TLR3-218.26-12.4DSVAX 2TLR2-181.2-11.6TLR3-211.12-11.3 Details of docking scores along with their corresponding binding energy values of DENV vaccine constructs with TLR2 and TLR3 docked complexes. Binding Energy (kcal/mol) Table 5Details of docking scores along with their corresponding binding energy values of DENV vaccine constructs -human IgG docked complexes.Docked complexDocking ScoreBinding Energy(kcal/mol)Vaccine Candidate with amino acid rangePDB ID of human IgG with chainDSVAX 14AM0 (H)-245.98-11.74AM0 (L)-198.49-11.3DSVAX 24AM0 (H)-241.06-114AM0 (L)-235.46-10.9 Details of docking scores along with their corresponding binding energy values of DENV vaccine constructs -human IgG docked complexes. Binding Energy (kcal/mol) After docking, the vaccine candidate DSVAX 1 was selected as a novel epitope based peptide vaccine candidate based on its cytokine-inducing ability, high binding affinity, and favorable Ramachandran score. Docking images of the vaccine construct (DSVAX 1) - TLR2 (Fig. 2), vaccine construct (DSVAX 1) - TLR3 (Fig. 3), vaccine construct (DSVAX 1) - IgG heavy chain (Fig. 4), and vaccine construct (DSVAX 1) – IgG light chain complexes (Fig. 5) are represented below. Prot-Prot analysis of PDBSum Generate server calculates 2 salt bridges, 4 hydrogen bonds, and 119 non-bonded contacts between TLR2 and the vaccine construct. Similarly, 1 salt bridge, 7 hydrogen bonds, and 149 non-bonded contacts were identified between TLR3 and the vaccine construct. 1 salt bridge, 3 hydrogen bonds, and 98 non-bonded contacts were found between IgG heavy chain and the vaccine construct. 1 salt bridge, 6 hydrogen bonds, and 123 non-bonded contacts were observed between IgG light chain and the vaccine construct. The evaluated binding energy along with the above-mentioned intermolecular interactions indicates that the vaccine candidate DSVAX 1 shows high binding affinity toward TLR2, TLR3, IgG heavy chain, and IgG light chain. These findings indicate the potential effectiveness of the vaccine construct. Molecular docking and interaction profiling of the vaccine construct (DSVAX 1)-TLR2 docked complex. (A) Model of the docked complex. The vaccine construct is shown in red and TLR2 in purple, (B) Key interacting residues of the docked complex visualized by Discovery studio molecular viewer, (C and D) Colored lines exhibit different types of intermolecular (protein-protein) interactions within the docked complex, as indicated in the key. Molecular docking and interaction profiling of the vaccine construct (DSVAX 1)-TLR3 docked complex. (A) Model of the docked complex. The vaccine construct is shown in red and TLR3 in purple, (B) Key interacting residues of the docked complex visualized by Discovery studio molecular viewer, (C and D) Colored lines exhibit different types of intermolecular (protein-protein) interactions within the docked complex, as indicated in the key. Molecular docking and interaction profiling of the vaccine construct (DSVAX 1)-IgG heavy chain docked complex. (A) Model of the docked complex. The vaccine construct is shown in red and IgG heavy chain in yellow, (B) Key interacting residues of the docked complex visualized by Discovery studio molecular viewer, (C and D) Colored lines exhibit different types of intermolecular (protein-protein) interactions within the docked complex, as indicated in the key. Molecular docking and interaction profiling of the vaccine construct (DSVAX 1)-IgG light chain docked complex. (A) Model of the docked complex. The vaccine construct is shown in red and IgG light chain in pink, (B) Key interacting residues of the docked complex visualized by Discovery studio molecular viewer, (C and D) Colored lines exhibit different types of intermolecular (protein-protein) interactions within the docked complex, as indicated in the key. Molecular dynamics (MD) simulation and calculation of Molecular Mechanics/Poisson-Boltzmann Surface Area (MM/PBSA) MD simulations are an effective method for investigating biomolecular complexes’ dynamics, providing critical insights beyond those offered by rigid docking approaches. Unlike static docking methods, MD simulations allow the evaluation of receptor-ligand interactions in a solvated, neutralized, and temperature-controlled environment, closely mimicking physiological conditions82,83. This ability is a key for understanding the conformational dynamics and stability of the finally considered receptor-ligand complexes over time. In the present study, a 100 ns MD simulation was conducted by GROMACS 2021.2 to examine the docked complexes. The CHARMM27 force field was employed for parameterization of the system. Prior to the simulation, energy minimization was conducted to resolve any steric clashes. Subsequently, the systems were equilibrated under NVT (constant number of particles, volume, and temperature) and NPT (constant number of particles, pressure, and temperature) conditions. Neutralization of the solvated complex system was achieved by adding counter-ions such as Na+ and Cl−. The trajectory was analyzed throughout the 100 ns simulation period under NPT conditions to assess key structural parameters. Many GROMACS tools, such as gmx rmsd, gmx rmsf, gmx gyrate, gmx SASA, and gmx hbond were used to analyze RMSD, RMSF, Rate of Gyration, Solvent accessible surface area, and hydrogen bond formation, respectively. The RMSD study revealed a consistent and stable binding conformation between the vaccine construct and the antibody throughout the simulation. Both the ligand atoms and the receptor backbone exhibited minimal fluctuations in their RMSD profiles, indicating a stable interaction interface that resisted significant conformational drift82,84. The RMSF study was also conducted for evaluating the receptor’s flexibility upon ligand binding. The results demonstrated low fluctuation values for most residues, particularly at the interaction interface, confirming that ligand binding contributed to the stabilization of the receptor protein structure85. These findings suggest that the docked complexes maintained a highly stable interaction profile over the 100 ns time frame, withstanding thermal and structural perturbations. The Radius of Gyration (Rg) determines the overall shape along with compactness of the protein by correlating with experimentally obtainable hydro-dynamic radius. The Rg results showed maintenance of compactness during the 100 ns simulation which indicates stable folding. Hydrogen bond formation is essential to know binding specificity as well as molecular interactions of the receptor-ligand complex. All the findings of hydrogen bond numbers showed minimal fluctuations but the bond counts were remains comparatively high. This signifies structural integrity, stability of binding of the receptor-ligand, and well-maintained intermolecular interactions. The Surface accessible solvent area (SASA) estimates the portion of the protein surface that is exposed to the solvent during simulation. The steady turn down of SASA findings over time exhibited that the protein became more compact, stable, and less exposed in solvent. Also, the number of secondary structural elements was determined during the simulation over time which showed minimal fluctuations. This indicated negligible conformational rearrangements and overall structural preservation86–88. The MD simulations reveal the structural stability as well as the dynamics of the docked complexes, providing key insights into TLR-vaccine candidate as well as antigen-antibody recognition and binding mechanisms. Furthermore, such simulations can significantly inform the rational design of therapeutic antibodies or peptide-based vaccine candidates by identifying stable interaction hotspots and conformationally rigid regions. In summary, the simulation results underscore robust binding interactions. These findings offer a foundational understanding for future structural and immunological studies aimed at combating dengue virus infections. Detailed MD trajectory results, including RMSD, RMSF, and Rg analyses of the vaccine construct-TLR2, TLR3, IgG heavy chain, and IgG light chain docked complexes (Fig. 6), are represented below. Additionally, the number of hydrogen bonds, SASA and number of secondary structures of the vaccine construct-TLR2 (i) TLR3 (ii) IgG heavy chain (iii), and IgG light chain (iv) docked complexes are provided in the supplementary figure (Supplementary Fig. 19). (Note: In case of the RMSD plot, the x-axis represents time in nanoseconds, while the y-axis represents RMSD values in nanometers; in the RMSF plot, the x-axis represents atom or residue number, while the y-axis represents RMSF values in nanometers; in the Rg plot, the x-axis denotes time in nanoseconds, whereas the y-axis represents Rg values in nanometers; in the hydrogen bonding graph, the x-axis denotes time in nanoseconds, whereas the y-axis denotes the number of hydrogen bonds; in the SASA plot, the x-axis denotes time in picoseconds, whereas the y-axis denotes SASA values in nanometer2; and in the secondary structure plot, the x-axis represents time in nanoseconds, while the y-axis represents the number of secondary structures). The Molecular Mechanics/Poisson-Boltzmann Surface Area (MM/PBSA) energy profile showed different binding free energy, which determined the complex’s stability and binding strength during the simulation. MM/PBSA calculations are represented in tabular form in the Supplementary Table (Supplementary Table 19). This findings suggest that the binding is stable as well as energetically favorable, which primarily driven by Van der Waals along with electrostatic interactions. Molecular dynamics trajectories of vaccine construct (DSVAX 1) - TLR2, TLR3, IgG heavy chain, and IgG light chain docked complexes. (A) RMSD plots showed structural stability, (B) RMSF plots indicated lower fluctuations of most residues, exhibiting structural rigidity, (C) Rg plots showed structural compactness and stability during the 100 ns simulation. Codon optimization along with in silico cloning The expression of any vaccine candidate in a proper expression vector is a key factor to produce recombinant proteins. So, codon optimization in the E. coli K12 strain was performed by the Java Codon Adaptation Tool. According to this tool, optimal gene expression and stability require a GC content of 30–70% and Codon Adaptation Index (CAI) value above 0.8 or closer to 1. A higher CAI value indicates stronger expression efficiency. The selected vaccine candidate DSVAX 1 of this work showed a 5432-nucleotide sequence with a CAI value of 1 and a GC content of 53.23%. These results indicate that the vaccine construct has good potential for expression in the E. coli K12 strain host system as the CAI value reflects its suitability for cDNA expression in host system. Additionally, restriction sites at the N- and C-terminal ends of the optimized codon sequence were added to enable in silico cloning in the pET-28a(+) vector with the help of Snapgene 6.2 software (Fig. 7). In silico restriction cloning was performed to insert the optimized vaccine sequence into the pET-28a(+) expression vector. The inserted optimized gene was highlighted in red at the top of the map between the forward and reverse primers within the MCS, while the backbone of the expression vector was highlighted by a black circle. Incorporation of restriction enzyme sites into the codon-optimized sequence accelerated the cloning. The optimized gene was effectively inserted between the forward and reverse primers within the multiple cloning sites (MCS). Specifically, the cloned insert was placed downstream of the T7 promoter, lac operator, ribosome-binding site (RBS), and ATG start codon. In conclusion, the vaccine construct gene was effectively inserted at the proposed cloning site within the MCS of the pET-28a(+) vector. In silico immune simulation Immunological simulation of the vaccine candidate (DSVAX 1) was performed by the C-ImmSim server to evaluate its immunogenic characteristics along with the potential immunological response within the immune system of human. At four-week intervals, corresponding to time steps of 1, 84, and 168, the vaccine was administered in three injections in this study. The primary immune response was characterized by elevated levels of IgM and IgG antibodies. In the subsequent secondary response, increased levels of IgM + IgG, IgG1 + IgG2, IgG1, and IgG2 were found (Fig. 8 A). Furthermore, highly stable B-cell isotypes were identified, indicating the potential for isotype switching along with the development of memory (Fig. 8 B and C). Both cytotoxic T lymphocytes (CTL) and helper T lymphocytes (HTL) showed strong immune responses after vaccination, supporting the immunogenicity of the incorporated T-cell epitopes (Fig. 8 D, E and F) Also, increased populations of macrophages (MA population) and dendritic cells (DC population) were observed after every time of exposure (Fig. 8 G and H). The immune simulation study also showed a marked elevation of IFN-γ, IL-10, IL-12, IL-23, and TGF-β, indicating a balanced, robust, and efficient immunological response triggered by the vaccine construct (Fig. 8 I). In silico simulation of the immunological response following vaccine administration. (A) Antigen as well as different immunoglobin (Ig) levels including IgM + IgG, IgM, IgG1 + IgG2, IgG1, and IgG2, was found. (B) B-cell population density. (C) B-cell population density across different states. (D) Cytotoxic T-cell (TC) population density. (E) Cytotoxic T-cell (TC) population density across different states. (F) Helper T-cell (TH) population density. (G) Macrophages (MA) population density across different states. (H) Dendritic cell (DC) population density across different states. (I) Production levels of various cytokines, including IFN-γ, TNF-α, and TGF-β as well as interleukins including IL-2, IL-4, IL-6, IL-10, IL-12, IL-18, and IL-23. Discussion The alignment of DENV envelope (E) protein sequences of all serotypes revealed significant sequence divergence due to gaps and mismatches in amino acids. This low sequence conservation indicates substantial variation in infection rates, particularly in functional domains, which exhibited large alignment gaps. Such diversity suggests differences in functional properties or antigenic structures among serotypes78. The E protein plays a crucial role in viral invasion, membrane fusion, as well as immune evasion, and these sequence differences influence serotype-specific biological properties. Amino acid composition analysis supported the hypothesis of distinct evolutionary pressures acting on different serotypes89, especially evident in DENV E protein structures. DENV-1, in particular, demonstrated structural flexibility, which aids in membrane fusion and host cell interaction, potentially leading to prolonged and severe infections90. In patients’ sera, IgM antibodies are found ~ 3–5 days after the onset of symptoms, continue to enhance for over 2 weeks, and can remain present for ~ 179 days in case of primary infections and ~ 139 days in case of secondary infections. Conversely, during the acute phase of secondary dengue infections, elevated levels of cross-reactive IgG antibodies are found, frequently rising before or together with the response of IgM. A speedy increase in IgG levels during the early stage of secondary infection is an essential indicator of dengue infection91. In this study, docking of the vaccine constructs with human IgG were performed for several purposes, such as the identification of main amino acids involved in recognition, and the prediction of the intensity of immune interactions along with immune escape. Despite extensive molecular insights gained from in silico analyses, further genotype-based field validation is needed. Computational tools have enhanced the identification of neutralizing epitopes and antigen–antibody interactions, significantly aiding antibody development92. Antigen–antibody docking has highlighted key interaction residues involved in immune recognition. Human Toll-like receptors (TLRs) are present on epithelial surfaces, including the skin, gastrointestinal, urogenital, respiratory tracts along with innate immune cells. After recognizing foreign molecules, TLRs initiate signaling cascades to activate the host immune response51,93,94. In our study, docking of the vaccine constructs with TLR2 and TLR3 also exhibited strong binding affinity. It signifies vaccine candidate’s efficacy. The E protein remains a highly immunogenic and functionally important viral component66,67. In our study, conservation analysis across all four DENV serotypes identified six key peptide regions spanning amino acid positions 45–65, 110–140, 190–225, 255–290, 300–345, and 346–390 that may serve as interaction domains on the E protein. These regions are likely under evolutionary pressure, balancing functional necessity with immune escape78. These findings reinforce the E protein’s significance as a therapeutic target. Understanding the structural and antigenic diversity among DENV serotypes is vital for designing cross-protective vaccines and antiviral agents, given dengue’s global health impact95,96. Developing a safe and effective dengue vaccine has been a global challenge for over eight decades. Despite multiple strategies, like inactivated vaccines, virus-like particle (VLP)-based vaccines, and subunit vaccines, no universally approved vaccine has achieved broad and safe protection across all four DENV serotypes. Advanced candidates like Dengvaxia and TV003/TV005 have progressed to clinical trials but have revealed notable limitations. Dengvaxia has been shown to induce suboptimal humoral immune responses, especially against DENV-2, and has raised serious safety risks for seronegative individuals, leading to severe adverse outcomes and fatalities. TV003/TV005 also has not consistently induced balanced immunity across all serotypes96–98 and has raised concerns about potential reversion or recombination occurring in vivo. In response to these concerns, research has increasingly shifted toward rational vaccine design using in silico approaches. Immunoinformatics provides a reliable platform for screening viral proteins and selecting immunodominant, non-allergic, non-toxic CTL, HTL, and B cell epitopes99–101 which are critical for developing cross-protective strategies to combat all four DENV serotypes102,103. In the present study, epitopes of the DENV E protein located within previously mentioned amino acid ranges were selected using the IEDB server78,79,2. A comprehensive study of toxicity, antigenicity along with allergenicity was conducted for all candidates. Finally, based on a favorable antigenicity score, other analyses, high binding energy, and a maximum Ramachandran favored region score, one novel epitope-based vaccine candidate, DSVAX 1, was selected. Molecular docking and 100 ns MD simulations validated the conformational behavior as well as the stability of the TLR2-vaccine construct, TLR3-vaccine construct, and IgG-vaccine construct complexes. The RMSD study depicted structural stability of the protein over the 100 ns simulation period. RMSF analysis identified flexible regions within the protein involved in immune recognition104,105. In silico simulations predicted that the vaccine candidate could stimulate innate immune components, like natural killer cells, dendritic cells, and macrophages, thereby promoting robust adaptive immunity, including both humoral and cell-mediated pathways through cytokine and chemokine secretion. Peptides of the SARS-CoV-2 spike protein capable of triggering the anti-inflammatory cytokine IL-4 are considered to possess anti-inflammatory properties, which help to prevent excessive immune-related damage51. Cross-reactive non-neutralizing antibodies promote Fcγ receptor–mediated viral entry in dengue virus infection, leading to antibody-dependent enhancement (ADE) as well as amplified illness severity. Contrary to, EDIII-specific neutralizing antibodies target envelope protein’s domain III, block receptor-binding sites, manifest strong serotype specificity, diminish enrichment, and show promising candidates for safer vaccine design along with effective immunotherapeutic approaches106–109. The E protein and its active domains help to elicit T-cell responses and neutralizing antibodies. By designing a vaccine composed of highly immunogenic, highly antigenic, non-toxic, and non-allergenic epitopes, the candidate offers a promising strategy to overcome the serotype-specific challenges of previous dengue vaccines110,111. Additionally, human β-defensin 2 was added at the N-terminal end of the epitopes by the linker EAAAK to optimize immunogenicity and stability51. An in silico cloning strategy confirmed the successful insertion and high-level expression of the vaccine construct in an Escherichia coli expression system, enabling large-scale production. Expression efficiency was further improved by codon optimization. The selected vaccine candidate DSVAX 1 of this work showed a CAI value of 1 and a GC content of 53.23% which indicate that the vaccine construct has good potential for expression in the E. coli K12 strain host system. In silico immune simulations exhibited enhanced immune memory related to B and T cells, along with production of IFN-γ, IL-10, IL-12, IL-23, and TGF-β, indicating a balanced, robust, and efficient immunological response triggered by the vaccine construct51. The construct shows promise to elicit protective immunity against DENV, which needs experimental validation through in vitro and in vivo studies112,113,77. Our final vaccine construct, DSVAX 1, was highly antigenic (0.7991), non-allergenic, and non-toxic. Molecular docking and molecular dynamics simulations further confirmed the stability and high binding energy of the vaccine construct with TLR2 (ΔG = -13.9 kcal/mol), TLR3 (ΔG = -12.4 kcal/mol), human IgG heavy chain (ΔG = -11.7 kcal/mol and light chain (ΔG = -11.3 kcal/mol). According to our immunoinformatics study, the addition of an adjuvant and specific linker increases the structural integrity and bioavailability of the epitope-based peptide vaccine candidate, which should be validated through in vivo experiments. Limitations Our study is based on immunoinformatics and structural bioinformatics approaches. Although all these in silico analyses, including epitope prediction, antigenicity, allergenicity, toxicity, BLAST analysis, physicochemical property assessment, molecular docking, molecular dynamics simulations, in silico cloning, and immune simulation, provide important insights, the designed epitope based peptide vaccine also shows promising potential; however, further in vitro and in vivo experimental validation is essential to authenticate the proposed vaccine’s effectiveness and safety. Experimental validation verifies whether the in silico outcomes actually occur in real biological settings. Conclusions This present research work reveals significant sequence and structural differences in the envelope proteins of DENV-1 to DENV-4, particularly within regions important for viral entry, immune evasion along with host interaction. These variations point out distinct evolutionary pressures, affecting each serotype’s pathogenicity as well as immune response. The findings highlight the necessity of serotype-specific data for vaccine development and also call for ongoing surveillance along with molecular characterization of DENV. Using immunoinformatics approaches, a peptide-based vaccine candidate was considered in this work, with predicted stability, safety, immunogenicity, as well as the potential to offer wide-ranging protection against all four DENV serotypes. Stable interaction between the vaccine candidate DSVAX 1 with TLR2, TLR3, and human IgG was confirmed by molecular docking and MD simulations. But in vivo along with in vitro validation should be conducted before it goes to the market. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Acknowledgements We gratefully acknowledge Swami Vivekananda University, Barrackpore along with Oriental Institute of Science & Technology, Purba Bardhaman campus for providing the computational facilities. Author contributions The Study and Method development By A.S.P and B.B, material preparation, analyzing data by A.C, experiments performed A.C, S.A, A.S and A.B, interpreted the results B.M, S.G. and A.S.P and manuscript is written by S.G and A.S.P. Data availability All data analyzed during this study are included in this article which is publicly available in supplementary materials of this article. Declarations Competing interests The authors declare no competing interests. Compliance with ethical statement. This article contains information and analysis of in-silico data only. This article does not involve studies with animal or human subjects performed by any of the authors, which should be approved by Ethics Committee. Ethical approval This article does not contain studies with animal or human subjects performed by any of the authors. So, requirement of ethical approval is inapplicable for this article. Generative AI disclosure No Generative AI was used in the preparation of this manuscript. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Data Availability Statement All data analyzed during this study are included in this article which is publicly available in supplementary materials of this article.