Combating Streptococcus agalactiae infections by designing a multi-epitope vaccine using molecular dynamic simulations Authors: Jasmine Arya, Deepshikha Kaushik, Bhawna Berwal, Awadhesh Kumar, Rakesh Kumar, Shivani Budhwar, Navita Chaudhary, Himanshi Dahiya, Kirti Nain, Astha Chhillar, Mehak Dangi Published: 2026 ABSTRACT Graphical abstract Introduction Gram-positive bacteria Streptococcus agalactiae Group B (GBS) is a causative agent for various diseases and infections in neonates, pregnant women as well as elderly patients with significant underlying diseases such as diabetes mellitus, neurological impairment, cancer and cirrhosis of the liver[1], [2], [3]. GBS is found in the microbiome of healthy adults but can transition to a highly invasive pathogen and causes severe diseases in neonates, postpartum women and individuals with a weakened immune system4. These bacteria are generally present in the mucous membrane of the throat, gastrointestinal tract, vagina, and skin. Infants get infected from GBS-positive mothers and develop symptoms within 24 h of delivery5. Early onset of infection (within one week of birth) in newborns occurs when GBS is transmitted vertically through the ruptured membrane before or during birth and is prevented by giving intrapartum intravenous antibiotics to pregnant women through IV during delivery6 whereas, late onset of infection (before 90 days) transmission’s understanding is still incomplete but prematurity and maternal colonization increase the risk factor7. Identification of GBS infection is done by the CAMP test during the 35th and 37th week of pregnancy by the presence of group B Lancefield antigen and capability to hydrolyze sodium hippurateas as the characteristic feature8. Serotypes Ia, Ib, II, III and V of GBS are the most prevalent9. The World health organization determined 100,000 newborn deaths and at least 46,000 stillbirths associated with GBS. Various studies have shown that no licensed vaccine is available against meningitis caused by GBS whereas other bacterial pathogens have several developed vaccines based on their capsular type[1], [2]. Therefore, new therapeutic methods against GBS are desirable to be developed. Bioinformatics applications helped in discovering new avenues in the field of life sciences by providing efficient ways for analysis of big data, high-throughput genomic advancements and to identify antigenic proteins that are putative vaccine candidates for the development of multi-epitope vaccines10. With increase in the genome sequencing technologies and bioinformatics various vaccinology-related software, and databases have been developed to identify and predict different features in vaccinology[11], [12]. Hence, in this study, we developed a competent MEV against Streptococcus agalactiae by identifying virulent proteins by employing various bioinformatics approaches and tested its stability and efficacy through protein–protein docking. The designed MEV has the potential to form stable interactions with receptors present in human immune cells to trigger an effective immune response. Materials and methodology Protein sequence retrieval A complete proteome sequence file of Streptococcus agalactiae 2603 V/R strain was retrieved from UniprotKB “www.uniprot.org” (Taxon id: 208435). It belongs to serotype V sharing a similar proteome with pathogenic Streptococcus agalactiae strain ATCC BAA-611. Reverse vaccinology approach to select virulence-associated antigenic proteins Computational methods were used to identify potential drugs and vaccine candidates that are antigenic and undesirable proteins were discarded to be selected as potential vaccines targets13. CD-HIT algorithm, a sequence redundancy minimizing software was used for paralogous proteins identification and elimination14. Redundant proteins in bacteria are proteins duplicated during the process of evolution, transduction, transformation and conjugation15. Sequences which are paralogous in the proteome with sequence similarity of 80% were removed from the dataset14. The essential gene database DEG was utilized for determining essential proteins of the pathogen Streptococcus agalactiae strain 2603 V/R by scanning its proteome against the EGs (essential genes) database using blastp and non-essential proteins were discarded16. Further to determine the proper functioning of selected proteins subcellular localization prediction was performed using the Psortb server17. Subcellular localization of proteins selected was restricted to cell walls, extracellular regions, cell membranes and secreted substances as they can easily initiate interactions with immune cells18. Deep TMHMM was used for transmembrane helix prediction and proteins with more than 1 α-helix were chosen for further experimentations19. For signal peptide prediction Phobius was used to discard secreted proteins that are found in bacterial extracellular regions but do not contain the signal peptides20. Later allergenicity of proteins was checked by Algpred2.0 with a threshold of 0.321. Identification of antigenicity of non-allergenic proteins was done by VaxiJen server and proteins with antigenicity of 0.5 or more were selected as putative vaccine candidates22. Virulent proteins play a crucial role in vaccine design protocol due to their host-invasive nature and pathogenesis therefore, selected proteins were recognized as virulent proteins by subjecting them to a virulence factor database (VFDB) and integrated blastp with default cut-off bit score > 100 and E-value 1e − 0423. To minimize the probability of cross-reactivity with a human host, isolated virulent proteins were further screened for homologous sequences using blastp and human non-homologous proteins were considered suitable vaccine candidates24. Lastly, ProtParam tool checked the molecular weight of the target proteins25 and their structures were predicted through Swiss model26. Structures obtained were further evaluated qualitatively through ERRAT server27. Epitope prediction Cytotoxic T-lymphocytes epitopes (CTL) prediction Cytotoxic T cell receptors have an important role in the recognition of antigenic proteins28. Therefore, CTL epitopes play a vital role in designing putative vaccine and the prediction of 12-mer epitopes was done by the human consensus method of MCH-1 binding tool of the Immune Epitope Database (IEDB)29. Homo sapiens were selected as source species for HLA alleles and a percentile score of 2 was considered because a lower rank has a greater affinity. The epitopes should be devoid of allergenicity and toxicity and was checked by employing the Allertop 2.0 and ToxinPred server[30], [31]. The antigenic characteristics of epitopes were assessed by Vaxijen v2.022 and the immunogenicity tool of IEDB was employed to examine the immunogenic nature of epitopes32. Helper T-cell lymphocytes (HTL) epitope prediction and assessment Helper T cells (HTL) influence CTL to kill parasitized target cells, stimulate the macrophages to inundate the pathogen and secrete antibodies and therefore to generate a better immune response it is crucial to add HTL epitopes33. Thus, the selected proteins were assessed to predict their 15-mer HTL epitopes using the consensus approach of the IEDB MHCII binding tool34. Helper T- cells induce different cytokinesis like interferon-gamma (IFN-γ), interleukin-4 (IL-4), and interleukin-10 (IL-10) which stimulate T-cells so, the IFN-epitope server was used to forecast HTL epitopes generating IFN-γ35. Moreover, IL-4 pred server and IL-10 pred server were used to understand IL-4 and IL-10 inducing properties at threshold values of 0.2 and − 0.3 respectively[36], [37]. Prediction and assessment of B- cell epitopes (Linear B- Lymphocytes: LBL) B-cell epitopes detected by B-cell receptor plays a vital function in adaptive immunity as they produce antigen-specific immunoglobin. Therefore, Linear B-cell was identified by utilizing ABCpred online38 which employs neural networking-based methodology. The amino acid length and threshold were fixed at 16 and 0.5 respectively. The allergenicity, toxicity, and antigenicity of the predicted B-cell epitopes were further verified by employing AllerTop v1.0, ToxinPred, and VaxiJen v2.0 servers respectively[24], [32], [33]. World population coverage analysis Different HLA are believed to be dispersed at different frequencies among different ethnic communities and regions throughout the globe and aid to improve the multiple epitope vaccine developments39. Distribution HLA alleles is essential for designing an effectual multi-epitope vaccine40. In order to select the most suitable vaccine candidate for large groups of individuals a software, IEDB-AR v2.20 was employed in this study which predicted the HTL and CTL population coverage analysis of chosen epitopes and their specific alleles that bind to HLA41. MEV construction and validation By connecting all epitopes that were identified in the above procedure with an appropriate adjuvant and linkers, a MEV construct was designed. An adjuvant’s role is to enhance the immunogenicity and stability of the vaccine; therefore, we used Cholera enterotoxin subunit B (Accession no: P01556) and attached it through EAAAK linker to the N-terminal of the first CTL epitope. Subsequently, AAY linkers were used to link CTL epitopes, HTL epitopes were linked with the help of GPGPG linkers and finally, KK linkers connected LBL epitopes42. Analysis of primary and secondary structure of MEV construct The physicochemical properties of the MEV construct which include an alphabetic index, theoretical isoelectric Point, Molecular Weight (MW) and Grand Average of Hydropathicity(GRAVY), were evaluated via Protparam43. Further, the VaxiJen v2.0 server22 was utilized to analyse the antigenicity of the vaccine construct and the immunogenicity was assessed by means of the IEDB immunogenicity tool32. To secure the MEV construct from an allergic response AllerTop online server was employed to analyse the allergenic property of the designed vaccine30. The SOPMA program was used to predict its secondary structure as it is a key element in protein folding.44. Tertiary structure prediction The linear sequence of the construct was converted to the 3-D structure of the vaccine construct for performing further analysis by an online server AlphaFold. AlphaFold predicts the structure and offers a distinguished confidence score for the forecast of the accuracy of the model45. 3D structure refinement and validation of vaccine ProSa-web was used to verify the average quality score of the input structure, while ERRAT server evaluated the interaction of non-bonded atoms and lastly, Ramachandran plot analysis was conducted for evaluation of disallowed phi (φ) and permissible dihedral psi (ψ) angles of amino acid residues[46], [47], [48]. B-cell epitopes prediction of MEV constructs Linear and conformational prediction of designed MEV B-cell epitopes was performed by the ABCpred and IEDB-AR v.2.22 Ellipro tool[38], [49]. The input of amino acid sequence was given to ABCpred server at a threshold of 0.5 and the length of peptide sequence was set to 16. However, Ellipro took three-dimensional structure of the MEV construct as input at default parameters. Molecular docking and molecular dynamics simulations (MEV + TLR2, MHCI and MHCII) To initiate an effective immune response MEV should have the potential to interact and bind with the host cells. Therefore protein–protein docking was performed to perceive the thermo-stability, binding affinities, and interactions of MEV construct with TLR2, MHC-I and MHC-II molecules of humans. To design a bimolecular complex, an online server PyDock was used to perform blind docking50. MEV was docked with MHC I, TLR2, and MHC II (Pdb Id: 1I1Y, Pdb Id: 2Z7X, Pdb Id: 1 KG0 respectively). The interaction was studied by utilizing the online database COCOMAPS 2.051 and visualization of the docked complex was done by Pymol52. MD simulations were used to verify the stability of the docked complex structure. Using GROMACS 2023.2 version, all systems for MD simulations were constructed in triclinic periodic boundary conditions (PBC) under OPLS AA/L all-atom force field. The docked complexes were positioned in the centre of the PBC box, and systems were solved using an explicit simple point charge (SPC) water model at a threshold distance of one nanometer from the protein surface. After inserting the Na+ ions to neutralize the systems, the steepest descent method was used to save energy further and eliminate the system's poor connections. Using the steepest descent minimization method, the system molecules were optimized over 50,000 cycles. Systems were progressively heated to 300 K for 10 ns in the NVT53 ensemble and equilibrated at 300 K and subjected to 1 atm pressure for 10 ns in the NPT54 ensemble using a leap-frog integrator. All of the simulated systems underwent their final production runs within a 100 ns time window. GROMACS tools were used to analyse the generated simulated trajectories, and GRACE tool was used to view the graphs. The results were analysed for the root mean square deviation (RMSD) plots that denote stability, the smaller the fluctuation, the more stable the protein, root mean square fluctuation (RMSF) that denoted flexibility, larger the fluctuation, more is the flexibility, and radius of gyration (Rg) that exhibited the compactness of vaccine bound to the receptor[55], [56], [57]. Results and discussion Proteome screening Streptococcus agalactiae serotype V (strain ATCC BAA-611 / 2603 V/R) proteome was downloaded from Uniprot (Accession no: UP000000821) having 2105 proteins. The proteome was filtered by CD-HIT for removal of paralogs at (0.8) 80% threshold. The CD-HIT suite removed thirteen redundant proteins, whereas the 2092 non-redundant proteins were considered for analysis in further steps. The proteins that are non-redundant were searched for essential genes on DEG and only 1336 essential proteins were identified. To figure out the function of their essential proteins and their position in the cells, subcellular localization was done by the Psortb server which identified 372 proteins to be extracellular membrane and total 408 proteins were selected including extracellular region, cell wall and secreted. Elimination of the proteins located in cytoplasm was done, and the residual proteins were chosen for analyses. Further this, 175 proteins were discovered having 0–1 TMHs out of which 11 proteins (Table 3.1) were discovered having signal peptides.Table 3.1Summary of proteome screening and selection of epitopes using immunoinformatics approach.Sr NoMethodologyResults1Number of proteins in proteome21052Non-paralogous proteins20923Proteins Selected through Subcellular localization4084Protein with 0–1 trans-membrane helices1755Protein With Signal Peptide (Phobius)116Non-allergen proteins87Antigenic Proteins58Molecular weight > 100 kDa59Virulent Proteins410Selected epitopes (CTL)1711Selected epitopes (HTL)212Selected epitopes (LBL)813Amino acid residues in MEV construct565 Table 3.1Summary of proteome screening and selection of epitopes using immunoinformatics approach.Sr NoMethodologyResults1Number of proteins in proteome21052Non-paralogous proteins20923Proteins Selected through Subcellular localization4084Protein with 0–1 trans-membrane helices1755Protein With Signal Peptide (Phobius)116Non-allergen proteins87Antigenic Proteins58Molecular weight > 100 kDa59Virulent Proteins410Selected epitopes (CTL)1711Selected epitopes (HTL)212Selected epitopes (LBL)813Amino acid residues in MEV construct565 Summary of proteome screening and selection of epitopes using immunoinformatics approach. Virulent and antigenic proteins prediction Out of 11 signal peptide proteins a total of eight proteins were evaluated to have no allergic reaction. These eight proteins were then analysed for their antigenicity and a total of 5 proteins were revealed to be highly antigenic thus being prioritized for putative vaccine candidates. Furthermore, these proteins have a molecular weight of more than 100 kDa. 5 proteins were checked for virulent factors by Blastp against VFDB, at precise criteria and 4 qualified to be considered as virulent. To prevent an autoimmune response, proteins homologous to human should be avoided and therefore these 4 proteins were subjected to blastp against human proteome. These 4 proteins are Zinc ABC transporter, zinc-binding adhesion liprotein, 2 Cell wall surface anchor family protein and C5a peptidase. The best predicted structure for 4 proteins were selected from Swiss Model and evaluated using ERRAT. (Table 3.2) enlists details of 4 target proteins.Table 3.2Brief description of 4 selected antigenic proteins.Uniport IDProteinSubcellular LocalizationMolecular WeightTMHsAntigenic score3-D structureERRAT ScoreQ8E128ABC transporter, zinc-binding adhesion liproteinCell Membrane57621.6500.562193.40Q8DYR9Cell wall surface anchor family proteinMembrane34298.6810.610990.43Q8E0S8Cell wall surface anchor family proteinMembrane33928.0810.614197.10Q8E1E1C5a peptidaseMembrane135883.2800.539594.05 Table 3.2Brief description of 4 selected antigenic proteins.Uniport IDProteinSubcellular LocalizationMolecular WeightTMHsAntigenic score3-D structureERRAT ScoreQ8E128ABC transporter, zinc-binding adhesion liproteinCell Membrane57621.6500.562193.40Q8DYR9Cell wall surface anchor family proteinMembrane34298.6810.610990.43Q8E0S8Cell wall surface anchor family proteinMembrane33928.0810.614197.10Q8E1E1C5a peptidaseMembrane135883.2800.539594.05 Brief description of 4 selected antigenic proteins. In our previous research we did genomic comparison of pathogenic Streptococcus agalactiae 2603 V/R with non-meningitis causing Streptococcus dysgalactiae strains NCTC7136 and ATCC12394 to identify unique and virulent genes of pathogenic strains responsible for neonatal meningitis54. Three out of four identified proteins of Streptococcus agalactiae 2603 V/R in the current study are also the part of the findings in our above-mentioned research incorporating genome comparisons. The targeted proteins in the current study that includes C5a peptidase enzyme (Q8E1E1), that helps in immune inactivation in host and cell wall surface anchor family proteins (Q8DYR9, Q8E0S8) are also specific to the pathogenic strain Streptococcus agalactiae 2603 V/R.These proteins have no reported homologs in E. coli isolate K1 and Klebsiella pneumoniae isolate K158. However, ABC transporter zinc-binding adhesion lipoprotein (Q8E128) is not a part of previous findings but in this study, it has shown good antigenicity score of 0.56, no allergenicity in addition to no transmembrane helix and therefore has been considered as a good target protein for epitope prediction and vaccine construction. Selection of predicted T-cell & B-cell epitopes MEV vaccine construct should mirror the natural immunity triggered by infection as well as generate a good adaptive immunity. Therefore, T-cell and B-cell epitopes of 4 shortlisted immunogenic proteins were identified. Overall predicted CTL epitopes were 69 in total and after analysis the best 17 non-toxic and non-allergic epitopes which had high antigenic and immunogenic score were selected for the MEV construct (Table 3.3).Table 3.3List of selected CTL epitopes predicted using IEDB.ProteinAlleleStart PositionPeptideConsensus RankAntigenicityImmunogenicityQ8E128ABC transporter, zinc-binding adhesion lipoproteinHLA-A*30:02 HLA-A*29:0231KLKVVTTFYPVY0.550.76260.21262HLA-A*02:0110SFVAMFAAWQLV0.810.46870.17624HLA-A*24:02246IKKYNINYIYFE1.41.02920.23024Q8E0S8Cell wall surface anchor family proteinHLA-A*02:01167TKKPLAGVVFEL0.980.62200.2155HLA-A*01:0157GTPIEGVLYQLY1.60.65880.1825Q8DYR9Cell wall surface anchor family proteinHLA-A*30:02159KKRLSGVIFVLY0.290.57360.1859HLA-B*15:0144LQPNTTEEGISY0.790.98660.30805HLA-A*01:01201EIEVEGLLPGKY0.850.72970.0751Q8E1E1C5a peptidaseHLA-A*30:02429KGKAYDVVYANY0.370.72880.09353HLA-A*01:01 HLA-A*30:02556TASGFEIYSSTY0.490.63020.04523HLA-A*29:02965DGFYTYRLRYTP0.890.85530.17636HLA-A*02:01709PQALLDTNWQKV10.57870.02859HLA-A*01:01300IDSDKFGEAYAK1.10.92740.05969HLA-A*02:01668RMGDKFDITVTI1.21.64490.18258HLA-A*01:01 HLA-A*29:021038DEEYGDETSYHY1.41.14920.07873HLA-A*01:01 HLA-A*30:02557ASGFEIYSSTYN1.50.81240.06607HLA-B*57:01769MSIPFVGFNGDF1.50.54690.36871 Table 3.3List of selected CTL epitopes predicted using IEDB.ProteinAlleleStart PositionPeptideConsensus RankAntigenicityImmunogenicityQ8E128ABC transporter, zinc-binding adhesion lipoproteinHLA-A*30:02 HLA-A*29:0231KLKVVTTFYPVY0.550.76260.21262HLA-A*02:0110SFVAMFAAWQLV0.810.46870.17624HLA-A*24:02246IKKYNINYIYFE1.41.02920.23024Q8E0S8Cell wall surface anchor family proteinHLA-A*02:01167TKKPLAGVVFEL0.980.62200.2155HLA-A*01:0157GTPIEGVLYQLY1.60.65880.1825Q8DYR9Cell wall surface anchor family proteinHLA-A*30:02159KKRLSGVIFVLY0.290.57360.1859HLA-B*15:0144LQPNTTEEGISY0.790.98660.30805HLA-A*01:01201EIEVEGLLPGKY0.850.72970.0751Q8E1E1C5a peptidaseHLA-A*30:02429KGKAYDVVYANY0.370.72880.09353HLA-A*01:01 HLA-A*30:02556TASGFEIYSSTY0.490.63020.04523HLA-A*29:02965DGFYTYRLRYTP0.890.85530.17636HLA-A*02:01709PQALLDTNWQKV10.57870.02859HLA-A*01:01300IDSDKFGEAYAK1.10.92740.05969HLA-A*02:01668RMGDKFDITVTI1.21.64490.18258HLA-A*01:01 HLA-A*29:021038DEEYGDETSYHY1.41.14920.07873HLA-A*01:01 HLA-A*30:02557ASGFEIYSSTYN1.50.81240.06607HLA-B*57:01769MSIPFVGFNGDF1.50.54690.36871 List of selected CTL epitopes predicted using IEDB. Also, 253 unique HTL epitopes were identified as they are found to be crucial for developing an effective immune response and the best two epitopes were chosen for further study which were non-allergenic, extremely antigenic, IL-10 and IL-4 inducing and were either IFN-γ inducing or not (Table 3.4).Table 3.4List of Selected HTL epitopes predicted using IEDB.ProteinPredicted EpitopesAlleles (HLA)IFN-inducerIL-10IL-4Q8E1E1C5a peptidaseIALNDKVKLALKLASHLA-DRB1*11:01Non-InducerPositiveInducerSDFKVQVSTKSPNLPHLA-DRB1*04:01InducerPositiveInducer Table 3.4List of Selected HTL epitopes predicted using IEDB.ProteinPredicted EpitopesAlleles (HLA)IFN-inducerIL-10IL-4Q8E1E1C5a peptidaseIALNDKVKLALKLASHLA-DRB1*11:01Non-InducerPositiveInducerSDFKVQVSTKSPNLPHLA-DRB1*04:01InducerPositiveInducer List of Selected HTL epitopes predicted using IEDB. Also, 48 exclusive B-cell epitopes were predicated out of which best 8 epitopes were selected for MEV was constructed based on non-toxic, non-allergenic and high antigenic nature of the epitopes (Table 3.5).Table 3.5List of Selected B cell epitopes predicted using ABCpred.Uniprot IdProtein NameB-cell EpitopeStart PositionScoreAntigenicityQ8E128ABC transporter, zinc-binding adhesion lipoproteinEHEGHGHEGHHHELDP2300.881.9977TRQEQESTRAINCACT10.851.2055Q8DYR9Cell wall surface anchor family proteinPFYIELPDDKLSNQLQ2180.821.5653DYRQDYRSTRACELLW40.741.2496Q8E0S8Cell wall surface anchor family proteinYGLAVKAGEKNRNVSA2150.821.2924EKNGRTPIRVKNGVHS2760.81.2533Q8E1E1C5a peptidaseAQYYITGNDGKAKINL7270.941.4113YRLRYTPVAEGANSQE10460.931.3522 Table 3.5List of Selected B cell epitopes predicted using ABCpred.Uniprot IdProtein NameB-cell EpitopeStart PositionScoreAntigenicityQ8E128ABC transporter, zinc-binding adhesion lipoproteinEHEGHGHEGHHHELDP2300.881.9977TRQEQESTRAINCACT10.851.2055Q8DYR9Cell wall surface anchor family proteinPFYIELPDDKLSNQLQ2180.821.5653DYRQDYRSTRACELLW40.741.2496Q8E0S8Cell wall surface anchor family proteinYGLAVKAGEKNRNVSA2150.821.2924EKNGRTPIRVKNGVHS2760.81.2533Q8E1E1C5a peptidaseAQYYITGNDGKAKINL7270.941.4113YRLRYTPVAEGANSQE10460.931.3522 List of Selected B cell epitopes predicted using ABCpred. Population coverage analysis HLA alleles distribution might vary around the world amongst several ethnic groups and geographical constituencies hence; population coverage is a necessary parameter for development of successful and effective MEV construct. To check the global effectiveness ofthe vaccine, cumulative population coverage of selected HTL and CTL epitopes along with their HLA alleles was carried out. The statistical outcomes for the global population demonstrated a combined distribution of 60.72% for selected epitope. Central America has revealed the lowermost population coverage (4.22%) on the contrary, the peak population coverage analysed was in Europe and North America at 85.48% and 82.15% respectively. Conclusively, selected epitopes have potential to be good subunit for designing a multi-epitope vaccine. Multi-epitope vaccine (MEV) construct Validation of designed vaccine construct The generated MEV vaccine construct linked 17 CTL, 2 HTL and 8B-cell epitopes with appropriate adjuvants and linkers. EAAAK linker was used to join the adjuvant to the N-terminal of the construct. Later, CTL, HTL and LBL epitopes were joined by AAY, GPGPG and KK linkers respectively, shown in figure 1a. The final linear MEV construct designed was of the size 565 amino acids. Furthermore, the physiochemical properties were assessed beforehand. The molecular weight of the vaccine was 63410.12 kDa and no Transmembrane helices were present and using protparam tool, it was noted that the construct has aliphatic index of 75.08, the instability index (II) of 34.56 and hydropathicity was − 0.392 (negative sign shows hydrophilic nature). Outcomes of MEV physiochemical properties represented that protein upon expression has a stable nature therefore the usage capacity are further enhanced. The predicted mean half-life of MEV was 30 h > 20 h and > 10 h in reticulocytes (mammalian), yeast cell, and E. coli respectively. Upon evaluating immunogenic properties of MEV construct it was concluded that vaccine construct was non-toxic and non-allergic and highly antigenic with antigenicity score of 0.79 (threshold at 0.5).MEV structure modelling and validation MEV structure modelling and validation MEV structure modelling and validation SOPMA utilized the amino acid sequence to analyse and predict the secondary structure of the designed MEV construct. In accordance with results of SOPMA, α-helix was formed by 236 (41.77%) amino acids, 1 β-strand had 28 (22.65%) amino acids and 152 (26.90%) amino acids form coils, conclusively these properties exhibited the potential of designed MEV to be considered as a candidate vaccine construct. Fig. 1b represented the 3D structure of MEV generated using the amino acid sequence as input on the AlphaFold3 server. Fig. 1c exhibited the Ramachandran plot of the designed MEV construct with 92.9% residues in the favoured region and 1.6% in disallowed region, and 5.5% in allowed region. Only 9 residues were found to be in disallowed region that included LEU4, SER16, TYR141, GLY181, LEU242, ASN260, GLY376, ASN480 AND GLY481 as per the report generated by Ramachandran plot server. The constructed MEV structure has 87.149 quality score as predicted by ERRAT which can be seen in Fig. 1d. Qmean score of the vaccine construct was reported as −5.88 using Swiss model to estimate the quality of predicted protein structure.Fig. 1(a) Sequence amino acid residues in the MEV where black represents aminoacid residues, orange represents adjuvant and blue and purple represents linkers (b) Tertiary structure of vaccine construct (c) Ramachandran plot analysis(d)ERRAT results for MEV. (a) Sequence amino acid residues in the MEV where black represents aminoacid residues, orange represents adjuvant and blue and purple represents linkers (b) Tertiary structure of vaccine construct (c) Ramachandran plot analysis(d)ERRAT results for MEV. B-cell epitopes prediction of MEV-construct Several kinds of immunoglobins are secreted when B cell epitopes initiate humoral immunity which is why MEV preferably ought to have B cell epitopes within its domains. The epitopes present in the planned construct is vital in order to start a sturdy immune response. Ellipro tool predicted 17 conformational B-cell epitopes with length between 5 and 15aa, and 9 linear B-cell epitopes from the proposed vaccine construct using default parameters. Molecular docking analysis of MEV with TLR2, MHCI and MHCII The cell receptors of MHC I, MHC II and TLR must need to interact with MEV so as to initiate a persistent immune response. TLRs play an important part by sensing conserved molecular pattern that is pathogen associated thus being a crucial part to trigger innate immunity. In this analysis, the online server pyDock performed protein–protein docking (Table 3.9) between MEV and MHCI, MHCII molecules, and TLR2. The docking results displayed impressive binding interactions of MEV with TLR2, and both MHC molecule. The binding score of MEV with TLR2, MHCI and MHCII were – 40.862 kcal/mol, −43.866 kcal/mol and −56.942 kcal/mol, (Table 3.9) respectively. Negative value implies a better docking and therefore a strong binding affinity with MHC molecules and TLR2. Non covalent interactions stabilizing the interface between vaccine and receptor complex using CoCoMaps2 web server areshown below in Table 3.6, Table 3.7, Table 3.8 respectively.•MEV-MHCI: The interaction distribution among the various chains of MEV-MHC1 was found having 26.2 percent in the start and it gradually increases as the simulation proceeds, further the hydrogen bonding increases from 3.3 percent to 6.4 percent.•MEV-MHCII: The interaction distribution among the various chains of MEV-MHC1 was found having 16.5 percent in the start and it gradually increases to 46.8% as the simulation proceeds, further the hydrogen bonding increases from 2.4 percent to 6.4 percent.•MEV-TLR2: The interaction distribution among the various chains of MEV-MHC1 was found having 26.2 percent in the start and it gradually increases to 50.3% as the simulation proceeds, further the hydrogen bonding increases from 2 percent to 3 percent.Table 3.6Atomic interactions(pie chart, heatmap and 3D contact map) for MEV- MHC I represented by chain A and chain B respectively.0 ns25 ns50 ns75 ns100 nsAtomic Interactions DistributionHeat Map3D Contact MapTable 3.7Atomic interactions (pie chart, heatmap and 3D contact map) for MEV- MHC II represented by chain A and chain B respectively.0 ns25 ns50 ns75 ns100 nsAtomic Interactions DistributionHeat Map3D Contact MapTable 3.8Atomic interactions (pie chart, heatmap and 3D contact map) for MEV- TLR 2 represented by chain A and chain B respectively.0 ns25 ns50 ns75 ns100 nsAtomic Interactions DistributionHeat Map3D Contact MapTable 3.9Molecular Docking Analysis Results of MEV with TLR2, MHCI and MHCII.ParametersMEV TLR2MEV MHCIMEV MHCIIElectrostatic Energy(kcal/mol)−21.961–23.277−19.126Desolvation Energy (kcal/mol)–23.077−19.748−43.124Van der Waals Energy(kcal/mol)41.759−8.40853.076Total Binding Energy(kcal/mol)−40.862−43.866−56.942Docked Structures of MESV with selected receptors •MEV-MHCI: The interaction distribution among the various chains of MEV-MHC1 was found having 26.2 percent in the start and it gradually increases as the simulation proceeds, further the hydrogen bonding increases from 3.3 percent to 6.4 percent. MEV-MHCI: The interaction distribution among the various chains of MEV-MHC1 was found having 26.2 percent in the start and it gradually increases as the simulation proceeds, further the hydrogen bonding increases from 3.3 percent to 6.4 percent. •MEV-MHCII: The interaction distribution among the various chains of MEV-MHC1 was found having 16.5 percent in the start and it gradually increases to 46.8% as the simulation proceeds, further the hydrogen bonding increases from 2.4 percent to 6.4 percent. MEV-MHCII: The interaction distribution among the various chains of MEV-MHC1 was found having 16.5 percent in the start and it gradually increases to 46.8% as the simulation proceeds, further the hydrogen bonding increases from 2.4 percent to 6.4 percent. •MEV-TLR2: The interaction distribution among the various chains of MEV-MHC1 was found having 26.2 percent in the start and it gradually increases to 50.3% as the simulation proceeds, further the hydrogen bonding increases from 2 percent to 3 percent. MEV-TLR2: The interaction distribution among the various chains of MEV-MHC1 was found having 26.2 percent in the start and it gradually increases to 50.3% as the simulation proceeds, further the hydrogen bonding increases from 2 percent to 3 percent. Table 3.6Atomic interactions(pie chart, heatmap and 3D contact map) for MEV- MHC I represented by chain A and chain B respectively.0 ns25 ns50 ns75 ns100 nsAtomic Interactions DistributionHeat Map3D Contact Map Atomic interactions(pie chart, heatmap and 3D contact map) for MEV- MHC I represented by chain A and chain B respectively. Table 3.7Atomic interactions (pie chart, heatmap and 3D contact map) for MEV- MHC II represented by chain A and chain B respectively.0 ns25 ns50 ns75 ns100 nsAtomic Interactions DistributionHeat Map3D Contact Map Atomic interactions (pie chart, heatmap and 3D contact map) for MEV- MHC II represented by chain A and chain B respectively. Table 3.8Atomic interactions (pie chart, heatmap and 3D contact map) for MEV- TLR 2 represented by chain A and chain B respectively.0 ns25 ns50 ns75 ns100 nsAtomic Interactions DistributionHeat Map3D Contact Map Atomic interactions (pie chart, heatmap and 3D contact map) for MEV- TLR 2 represented by chain A and chain B respectively. Table 3.9Molecular Docking Analysis Results of MEV with TLR2, MHCI and MHCII.ParametersMEV TLR2MEV MHCIMEV MHCIIElectrostatic Energy(kcal/mol)−21.961–23.277−19.126Desolvation Energy (kcal/mol)–23.077−19.748−43.124Van der Waals Energy(kcal/mol)41.759−8.40853.076Total Binding Energy(kcal/mol)−40.862−43.866−56.942Docked Structures of MESV with selected receptors Molecular Docking Analysis Results of MEV with TLR2, MHCI and MHCII. Molecular dynamics simulations The statistical parameters were analysed for all the three docked complexes using MD simulations (Fig. 2). MD simulations are frequently used to forecast how complexes would behave functionally through atomic movements in a predefined environment. The MD trajectories of the MHC1-MEV, MHC2-MEV and TLR2-MEV complexes were examined and contrasted in terms of solvent accessible surface area (SASA), radius of gyration (Rg), root mean square fluctuation (RMSF), and root mean square deviation (RMSD), H-bonds, etc.Fig. 2Analysis of MD simulations (a) RMSD, (b) RMSF, (c) Radius of gyration and (d) SASA. Analysis of MD simulations (a) RMSD, (b) RMSF, (c) Radius of gyration and (d) SASA. RMSD observed the structural deviation occurring in the protein–protein binding complexes (MHC1-MEV, MHC2-MEV, and TLR2-MEV). The structural deviation was recorded at different fluctuations ranging from 1.5 nm to 2.5 nm, throughout the 100 ns simulations run. Comparative RMSD analysis reveals that TLR2-MEV complex exhibited the stable structural motions with slight fluctuations as compared to MHC1-MEV, but the MHC2-MEVcomplex was found to exhibit the lowest RMSD throughout without any significant jumps or instability. The structural motions in MHC1-MEV receptor was found higher and it went up to 2.5 nm till the end of the MD simulation. The RMSD fluctuations in TLR2-MEV initially starting from 2.0 and went up to nearly 2.2 nm. The calculation of RMSD of protein–protein complex exhibited thatall the three complexes exhibited stable confirmations throughout without any significant peaks observed. The low degree of fluctuations demonstrated the structural integrity and uninterrupted interactions between all the three of the receptors and the designed vaccine construct. The simulation converges due to the stabilization of RMSD values around the fixed value and no large change in RMSD was observed that suggested the system has been equilibrated (Fig. 2a). The higher RMSD of MHCI complex might be the result of flexible ligand that is MEV. Lower RMSD of MHCII-MEV complex suggested fewer internal arrangements and a more rigid complex. The degree of atomic motion in the protein–protein complex structures was analyzed by root mean square fluctuation (RMSF). The RMS fluctuation in Cα of backbone residues is measured and depicted fluctuations were recorded in RMSF plot Fig. 2b. RMSF plot observed that protein–protein complexes are dynamic in nature and fluctuation is depicted in the range of 0.25 to near 2.3 nm. MHC2-MEV exhibited comparatively lower residual fluctuations as compared to MHC1-MEV and TLR2-MEV despite its longer size. MHC1-MEV complex exhibited highest fluctuations specifically between residues 640 to 750, as peak exceeds 2 nm. These peaks could be contributed by flexible linker regions and epitope boundaries. These peaks might be the result of epitope repositioning that has also contributed to higher RMSD of this complex. The TLR2-MEV complex has exhibited localized fluctuations around mid-region, but comparatively lower fluctuations in terminal region, showing that observed motions were restricted to a certain region only. Compactness in protein–protein structure complexes can be measured in terms of the radius of gyration (Rg). The packing of secondary structures from their original to compressed 3D forms is predicted by Rg. Analysis of Rg were observed that the gyration curve of complexes is stable throughout the simulation. The Rg in the TLR2-MEV complex initially found to be 4.2 nm with decrease upto ∼ 3.5 nm in initial 15 ns, once stabilized it has exhibited very narrow fluctuation range. This exhibited that despite having high RMSF, MEV remained globally compact when bound with TLR. Whereas the MHC1-MEV initially exhibited slight variations and maintained stability at 3.4 nm after 20 ns of the simulation. MHC2-MEV complex structure also maintained stability at 4.2 nm at the later stages of the simulation i.e. after 40 ns. It confirmed that the protein complexes were compact and the binding of MHC1, MHC2, and TLR2 didn’t cause any structural deformities like unfolding or structural breakage (Fig. 2c). The compactness in protein–protein complexes can be validated with the SASA test, the more compactness (lower SASA) shows the less contact with the surrounding water or more buried surface. The decrease in SASA scores indicated that AA residues were less accessible to solvent interactions with time in all three complexes (Fig. 2d). Comparative analysis of SASA showed that there was a larger accessible area available for the MHC2-MEV complex as compared to TLR2-MEV and MHC1-MEVcomplexes. This may be contributed to the large MEV regions that remained exposed outside the MEV grooves in the MHC2-MEV complex. MHC1-MEV complex exhibiting lowest SASA values depicted a more compact conformational state and less solvent exposure. Whereas in the TLR2-MEV complex the SASA exhibited a steep decrease by 10 ns and then it has exhibited minor fluctuations suggesting tightening of the TLR2-MEV interface with time and strong surface burial. Conclusion The in-silico techniques allow researchers to construct the novel vaccines in less time and at less expenditure. Deaths due to GBS causing meningitis and other severe infections caused by this pathogen are becoming uncontrollable. As we have stated earlier, there is no vaccine available for GBS infections and an increase in resistance against antimicrobials, consequently necessitates the effective medical preventive measures to be discovered for the patients including vaccines to recover at a faster pace. In this study, we targeted the most suitable antigenic and virulent proteins required by Streptococcus agalactiae to infect their host, providing insights into crucial targets for vaccine designing and development of computational-based MEV. The epitopes used for construct are antigenic, non-allergenic and non-toxic. Hence, the designed MEV construct is efficient, highly specific, and safe. The anticipated vaccine construct will potentially provide a long-term immunity to majority of people against Streptococcus agalactiae with population coverage of more than 85% in North America where serotype V is more prevalent pathogen for neonatal meningitis. To conclude, we believe that the MEV construct is ready for additional experimentation nevertheless, because of limitation of computational tools and servers and for the determination of biological potency immunological assays are recommended. CRediT authorship contribution statement Jasmine Arya: Writing – original draft. Deepshikha Kaushik: Validation, Formal analysis. Bhawna Berwal: Writing – original draft, Data curation. Awadhesh Kumar: Formal analysis. Rakesh Kumar: Resources. Shivani Budhwar: Resources. Navita Chaudhary: Resources. Himanshi Dahiya: Resources. Kirti Nain: Resources. Astha Chhillar: Resources, Investigation. Mehak Dangi: Supervision, Investigation, Conceptualization. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.