Innovative multi-epitope vaccine engineering for dual protection against Salmonella and Shigella leveraging FliC flagellin protein adjuvant properties: an immunoinformatics-based approach Authors: Elnaz Afshari, Neda Soleimani Published: 2026 ABSTRACT AbstractBackgroundShigellosis and salmonellosis are prominent food-borne illnesses associated with high mortality rates and morbidity, for which, as of now, no licensed vaccine is available. Multiple epitope vaccines offer promising prospects as alternatives to conventional vaccines by combining extremely immunogenic as well as conserved epitopes to overcome weaknesses related to weak antigens as well as serotype specificity.MethodsFor the first time, this study focused on developing a new protective multiepitope vaccine for dual protection against Salmonella and Shigella by integrating six conserved virulence factors: SpvB, SipD, RcK, SigA, IpaD, and FimA.ResultsWhen immunoinformatics tools were applied, highly conserved B- and T-cell epitopes with broad HLA allele coverage and high antigenicity were predicted, while allergenicity and toxicity assessments were used to assess safety. The selected epitopes were conjugated with suitable spacers to maintain structural integrity and increase immune processing. The conserved domain of the Salmonella flagellin protein (FliC) was used as both the scaffold and the TLR5 agonist adjuvant. Secondary and tertiary FliC-based multiepitope construct (FMEC) structures were predicted, refined, and validated; these structures exhibited desirable physicochemical properties, such as stability, hydrophilicity, and thermal resilience, with a molecular weight of 33 kDa, which is ideal for vaccine construction. In addition, ProSA and Ramachandran plot validation confirmed the reliability of the FMEC model. Immune simulation demonstrated powerful cellular and humoral immunity, whereas molecular docking and normal mode analysis confirmed the stable interaction of FMEC with the TLR4/5 or MHCI/MHCII receptors.ConclusionAlthough FMEC shows promise as a safe, nontoxic, nonallergen, and widely immunogenic multiepitope vaccine candidate to combat Salmonella and Shigella, further experimental validation needs to be carried out to ascertain its in vivo protective efficacy.Trial registrationNot applicable.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12967-026-08386-6.Keywords: Salmonella, Shigella, FliC, Vaccine, Foodborne diseases Background Foodborne diseases (FBDs) are perennial and pervasive global health hazards that affect health, community, and economic stability [1]. Approximately 10% of the global population falls ill with diseases caused by polluted food annually, generating approximately 600 million cases of illness, 420,000 fatalities, and a loss of 33 million healthy life years worldwide [2]. Shigella and Salmonella, which are members of the family Enterobacteriaceae, are major causes of these infections and are highly related to gastrointestinal disorders. For example, approximately 93.8 million cases of gastroenteritis are caused by Salmonella, which is responsible for approximately 155,000 annual fatalities at the global level and mostly impacts low-resource communities [3]. The prevalence of Shigella infections is comparable, with 160 million cases reported worldwide and 600,000 corresponding deaths recorded worldwide and an increasing prevalence of multidrug-resistant (MDR) strains; notably, of the Shigella isolates from Iran from 2008 to 2021, 63% were found to exhibit multidrug resistance [4]. Additionally, the prevalence rates of combined Shigella and Salmonella are 18% and 9%, respectively, where the most common isolates in Iran are Shigella sonnei (42%), Shigella flexneri (33%), and Salmonella enterica (22%). The minimal infectious dose and high virulence of these pathogens, especially Shigella, are great public health concerns worldwide and disproportionately affect high-risk groups such as children younger than five years and immunocompromised individuals [5]. Another serious challenge is the development of antimicrobial resistance, coupled with the substantial diversity of Salmonella serovars (over 2,600) and Shigella serotypes (more than 50). In addition to the problems and limitations of current serotype-specific vaccines, novel approaches are urgently needed to provide broad-spectrum protection against these infections [6]. For this purpose, multiepitope vaccines that leverage immunoinformatics and reverse vaccinology offer a promising approach. Immunoinformatics offers speedy advancements in vaccine research, such as in preclinical studies, whereas comparative structural and immunological analyses of antigens can facilitate the presentation of diverse epitopes from multiple antigens, potentially increasing the immunogenic efficacy of recombinant proteins and playing pivotal roles in immunization [5]. Furthermore, their strengths primarily include their high immunogenicity, safety, stability, low allergenicity, and ease of production [7]. Therefore, to capture conserved virulence factors that are important for the pathogenicity of these organisms, these vaccines may offer broad-spectrum immunity and overcome the limitations of existing vaccines. The present study aimed to develop a multiepitope vaccine containing six important antigens: SpvB, SipD, RcK, SigA, IpaD, and FimA. These virulence factors are vital in the colonization, survival, and invasion mechanisms of Salmonella and Shigella. Specifically, SpvB is an ADP-ribosylating toxin very important for the intracellular survival of Salmonella that induces host cell cytotoxicity [8]; SipD and IpaD are type III secretion system (T3SS) tip proteins and highly conserved virulence factors in Salmonella and Shigella, respectively, that assemble the translocon for effector translocation into host cells, mediating host cell invasion, and evoking inflammatory responses [5, 9, 10]; Salmonella RcK is involved in bacterial resistance to complement through lysis inhibition [11, 12]; SigA causes intestinal tight junction disruption, resulting in barrier dysfunction in Shigella infections [13, 14]; and FimA, as a fimbrial adhesin, promotes mucosal colonization and biofilm production of Shigella [15]. Selection is dependent on their cross-protection potential, conserved virulence mechanisms, and capacity to induce and evoke heterogeneous immune responses from the innate to adaptive immune systems. Structural homology between SipD and IpaD, for example, can enable cross-reactive antibody induction, while intracellular pathogenesis is targeted by SpvB [8]. Moreover, T3SS subcomponents are vital for epithelial invasion, whereas adherence factors such as FimA promote mucosal immunity. The construction of effective vaccine constructs is strongly dependent on the strategic use of adjuvants, whose role in immunomodulation toward desired responses is crucial [16]. Owing to their low immunogenicity, the use of adjuvants, which are unique to multiepitope peptide vaccines, can play a crucial role in augmenting immune responses. Immune system modeling is controlled mainly by the identification of pathogen-associated molecular patterns (PAMPs) through pattern recognition receptors (PRRs), i.e., Toll-like receptors (TLRs). When binding to specific PAMPs, the Toll-like receptor family of conserved receptors triggers immune responses within cascade processes [16, 17]. Bacterial flagellin, a TLR5 agonist, efficiently activates both the innate and adaptive immune systems through binding to TLR5, making it an efficient mucosal vaccine adjuvant [17, 18]. The structural composition of flagellin from Salmonella typhimurium (FliC) has been rationally deciphered and consists of four distinct regions, D0, D1, D2, and D3, in a boomerang-like structure. While D0 and D1 are conserved to a very high extent, D2 and D3 show more sequence variation as well as greater structural variation. Importantly, domain D1 physically interacts with TLR5, which consists of three alpha-helices that assemble to form a rod-like structure. Two of these proteins, αND1a and αND1b, are located in the N-terminal region of D1, whereas the third protein, αCD1, is located in the C-terminal region [17–19]. Therefore, the present study utilized bioinformatics tools to engineer a new multiepitope vaccine against pathogenic Salmonella and Shigella strains harboring multiple T- and B-cell epitopes to yield a broad-spectrum immune response and B-cell epitope-enriched conserved D1 regions of FliC as an adjuvant to provide added immunogenicity to the vaccine. Through the exploitation of the immunomodulatory capacity of the D1 domain, the vaccine is designed to evoke potent as well as long-lasting immunity against these pathogens. This strategy highlights the promise of judicious adjuvant inclusion in the augmentation of vaccine constructs, especially multiepitope peptide-based platforms. Materials and methods Sequence retrieval and initial analysis The amino acid reference sequences for SpvB, SipD, and Rck from Salmonella enterica subsp. enterica serovar Typhimurium str.LT2 and SigA, IpaD, and FimA from Shigella flexneri 2a srt.310 were retrieved from the National Center for Biotechnology Information (NCBI) server (https://www.ncbi.nlm.nih.gov/) in FASTA format. The accession numbers for these sequences are NP_490529.1, NP_461804.1, NP_490501.1, NP_708742.1, NP_858259.1, and NP_709920.2, respectively. Initial assessments of the antigenicity, toxicity, and allergenicity of these proteins were conducted via VaxiJen v2.0 (http://www.ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen.html) with thresholds of 0.4 for target organism of bacteria [20], ToxinPred (http://crdd.osdd.net/raghava/toxinpred/) [21], and AlgPred (http://crdd.osdd.net/raghava/algpred/) [22] servers, respectively. In accordance with the VaxiJen guidelines, sequences with a score higher than the threshold of 0.4 are considered antigens. The VaxiJen server, a pioneering tool for predicting protective antigens, leverages the physicochemical properties of proteins without relying on sequence alignment, achieving a prediction accuracy ranging from 70% to 89% [20]. ToxinPred, on the other hand, is a user-friendly web server that employs machine learning techniques and a quantitative matrix to predict peptide toxicity, with an accuracy of 94.50% [21]. The AlgPred server offers a comprehensive approach for predicting allergenic proteins with high accuracy by integrating multiple methods, including epitope similarity analysis, MEME/MAST allergen motif detection, IgE epitope mapping, amino acid composition analysis via SVM modules, and finally, a BLAST search against a database of 2890 allergen-representative peptides (ARPs) [22]. For further analysis of protein subcellular localization, the PSORTb v3.0 server (https://www.psort.org/psortb/) [23] was utilized. This server is distinguished as the most accurate tool available, offering extensive proteome-scale prediction coverage and introducing refined subcategories for protein localization [23]. Additionally, a similarity analysis between candidate virulence factors and the human proteome was conducted via PSI-BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi?CMD=Web%26PAGE=Proteins%26PROGRAM=blastp%26RUN_PSIBLAST=on). B-cell epitope prediction To predict linear B-cell epitopes from protein sequences, the Bepipred 2.0 and ElliPro tools available on the Immune Epitope Database (IEDB) server (http://tools.iedb.org/main/bcell/) were employed. These tools utilize a random forest algorithm trained on epitope and nonepitope data from sequence and crystal structures to generate predictions [24]. The antigenicity score of the predicted epitopes was determined via the VaxiJen v2.0 server (http://www.ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen.html) [20]. In accordance with the VaxiJen guidelines, epitopes with antigenicity scores approaching 1 are considered to exhibit higher probabilities of antigenicity and immunogenicity. In order to design a multi-epitope vaccine, a limited number of epitopes with the highest VaxiJen scores must be selected. For different B- and T-cell epitopes, VaxiJen score was justified depending on the type of protein and the maximum and minimum antigenicity scores of their epitopes, so that ultimately, epitopes with a limited number but the highest antigenicity scores were selected from each protein. In this regards, B-cell epitopes that met or exceeded a cutoff score of ≥ 0.8 were chosen for subsequent examination. To conduct the conservancy analysis, comprehensive amino acid sequences (without partial and repetitive sequences) of SpvB (n = 500), SipD (n = 500), and RcK (n = 336) derived from Salmonella spp., including Salmonella. enterica subsp. enterica, and Salmonella enterica subsp. enterica serovar Typhimurium along, with SigA (n = 506), IpaD (n = 506), and FimA (n = 98) from Shigella spp., including Shigella flexneri, Shigella boydii, Shigella dysenteriae, and Shigella sonnei, were retrieved from the NCBI server (https://www.ncbi.nlm.nih.gov/protein). Subsequently the conservancy score of predicted epitopes among all Salmonella and Shigella strains was assessed via the Epitope Conservancy Analysis tool available on the IEDB server (http://tools.iedb.org/conservancy/) [25]. Furthermore, the Emini surface accessibility prediction tool on the IEDB server (http://tools.iedb.org/bcell/) [26] was used to identify surface-accessible regions of the proteins, applying a threshold of 1. To predict conformational B-cell epitopes, the 3D structures of the SipD and IpaD proteins were obtained from the RCSB PDB database (https://www.rcsb.org/) with PDB IDs 2YM9 and 2J0O, respectively. For proteins such as SpvB, Rck, SigA, and FimA, where complete 3D structures were not available in the RCSB PDB database, models were predicted and generated via the I-TASSER server (https://zhanggroup.org/I-TASSER/) and refined through repeated structure perturbation and subsequent overall structural relaxation via molecular dynamics simulation on the Galaxy Refine server (https://galaxy.seoklab.org/cgi-bin/submit.cgi?type=REFINE) [27]. The best-refined model for each protein was validated via the ProSA server (https://prosa.services.came.sbg.ac.at/prosa.php) and Ramachandran plot on the Saves server (https://saves.mbi.ucla.edu/), with quality assessed via ProSA’s Z score [28] and residue distribution in the Ramachandran plot [29]. Finally, the validated 3D structure of each protein was applied to the ElliPro tools on the IEDB server (http://tools.iedb.org/ellipro/) [30] to predict conformational B-cell epitopes (with a cutoff value of ≥ 0.5), which were then visualized on the surface of each protein via PyMOL v.2.5 software [31, 32]. T-cell epitope prediction The TepiTool from the IEDB server (http://tools.iedb.org/tepitool/) was used to predict MHCII- and MHCI-binding epitopes. For MHCII, 15-mer epitopes were predicted for 26 of the most frequent human class II alleles (HLA-DPA1*01/DPB1*04:01, HLA-DPA1*01:03/DPB1*02:01, HLA-DPA1*02:01/DPB1*01:01, HLA-DPA1*02:01/DPB1*05:01, HLA-DPA1*03:01/DPB1*04:02, HLA-DQA1*01:01/DQB1*05:01, HLA-DQA1*01:02/DQB1*06:02, HLA-DQA1*03:01/DQB1*03:02, HLA-DQA1*04:01/DQB1*04:02, HLA-DQA1*05:01/DQB1*02:01, HLA-DQA1*05:01/DQB1*03:01, HLA-DRB1*01:01, HLA-DRB1*03:01, HLA-DRB1*04:01, HLA-DRB1*04:05, HLA-DRB1*07: For MHCI, 9-mer epitopes were predicted for 27 of the most frequent human class A and B alleles (A*01:01, A*02:01, A*02:03, A*02:06, A*03:01, A*11:01, A*23:01, A*24:02, A*26:01, A*30:01, A*30:02, A*31:01, A*32:01, A*33:01, A*68:01, A*68:02, B*07:02, B*08:01, B*15:01, B*35:01, B*40:01, B*44:02, B*44:03, B*51:01, B*53:01, B*57:01, and B*58:01) and six mouse class I alleles (H-2-Db, H-2-Dd, H-2-Kd, H-2-Kb, H-2-Kk, and H-2-Ld), with a median consensus percentile rank of ≤ 1. The TepiTool server employs a combination of IEDB-recommended and consensus methods, consisting of an artificial neural network (ANN_align), NetMHCpan, the stabilized matrix method (SMM_align), and the Sturniolo/Combinatorial Library, to predict high-affinity epitopes for multiple alleles of both MHC-I and MHC-II. These overlapping epitopes, recognized by multiple alleles, are crucial for vaccine design [33]. Additionally, NetChop and NetCTL predictors on the IEDB server (http://tools.iedb.org/netchop/) [34, 35] were used to predict cytotoxic T-cell epitopes with a threshold of 1. These web-based tools have enhanced the ability to predict proteasomal processing, including proteasomal cleavage, TAP transport efficiency, and MHC class I affinity, by utilizing neural networks to predict T-cell epitopes along protein sequences [34, 35]. Finally, the antigenicity and conservancy of all the predicted MHC-binding and cytotoxic T-cell epitopes across various Salmonella and Shigella strains were evaluated via VaxiJen v2.0 (http://www.ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen.html) [20] and the IEDB epitope conservancy analysis tool (http://tools.iedb.org/conservancy/) [25], respectively. Construction of the multiepitope platform To design a new multiepitope construct, B cell, MHC binding, and T-cell epitopes from each protein were prioritized on the basis of stringent criteria, including a VaxiJen antigenicity score of at least 0.8, a conservancy percentile of ≥ 90% across all Salmonella and Shigella strains, the highest amount of allelic coverage in both MHCI and MHCII in humans and mice that are common with predicted surface-available B-cell epitopes, and considerations of toxicity and allergenicity. To refine epitope selection in some proteins, the criteria were adjusted to a minimum conservation score of 70%, facilitating focused identification of potential targets. Linking epitopes together in multiepitope vaccine design involves the strategic use of KK and GGGGS linkers for B-cell epitopes, GPGPG linkers for helper T-cell (HTL) epitopes, and AAY linkers for cytotoxic T-cell (CTL) epitopes. Salmonella’s flagellum has a conserved D1 domain, which contains two N-terminal α-helical segments and a single C-terminal α-helical segment, and these, along with the others, enable interaction of the flagellum with Toll-like receptor 5 (TLR5). Among these invariant terminal ends lies the D2 and D3 domains, which have significant structural and immunological variability [17, 18]. In this study, regions of the conserved FliC domains containing B-cell epitopes were identified via the Bepipred 2.0 and ElliPro tools available on the IEDB server (http://tools.iedb.org/main/bcell/). These invariant N- and C-terminal ends containing B-cell epitopes and responsible for TLR5 binding were then selected for investigation. Afterward, the designed multiepitope construct was placed between these terminal ends to replace the highly variable D2 and D3 domains of the FliC. Ultimately, the final construct’s antigenicity, allergenicity, and toxicity were comprehensively evaluated via the VaxiJen v2.0 (http://www.ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen.html) [20], ToxinPred (http://crdd.osdd.net/raghava/toxinpred/) [21], and AlgPred (http://crdd.osdd.net/raghava/algpred/) [22] servers, respectively. Population coverage analysis To determine the population coverage of the selected epitopes, IEDB’s Population Coverage Analysis Tool (https://tools.iedb.org/population) was implemented. The significance of this analysis is to determine the number of populations that can be covered by the vaccine construct. Physicochemical properties and solubility prediction The physicochemical characteristics of the FliC-based multiepitope construct (FMEC), including molecular weight (MW), estimated half-life, theoretical isoelectric point (pI), instability and aliphatic indices, and grand average of hydropathicity (GRAVY), were examined via the ExPASy ProtParam Server (https://web.expasy.org/protparam/) [36, 37]. Additionally, the solubility of the construct was assessed through the SOLpro server (http://scratch.proteomics.ics.uci.edu/) [38]. A threshold of 0.45 was established for the experimental dataset, as any value above this threshold indicates a solubility higher than that of the average soluble E. coli protein. Prediction of secondary and tertiary structures The secondary structure of FMEC was predicted via PSIPRED 4.0 (http://bioinf.cs.ucl.ac.uk/psipred/) [39]. This server contains two feed-forward neural networks designed to accurately predict the transmembrane topology, transmembrane helix, folds, and domains. In addition, the I-TASSER server was used to predict the 3D structure of the FMEC. The I-TASSER server is recognized as one of the leading platforms for the design and prediction of three-dimensional protein structures. It employs multiple sequence alignments and structural elements sourced from the Protein Data Bank (PDB) to identify analogous structural patterns, subsequently simulating the assembly and arrangement of repetitive components to construct the protein’s 3D structure [40]. The server generates five proposed models, each accompanied by distinct C-scores, which serve as confidence indicators for assessing the quality of the predicted structural models. The optimal model is determined by the highest C score, with typical values ranging from − 5 to 2, as outlined in the server’s documentation [40]. The structural model with the highest C score was subsequently selected for refinement via the GalaxyRefine server (https://galaxy.seoklab.org/cgi-bin/submit.cgi?type=REFINE) [27]. This server begins the refinement process by reconstructing all side chains, placing the most likely rotamers starting from the core and progressing outward layer by layer. Through this method, GalaxyRefine enhances the protein structure by repacking the side chains. The server provides results quantified by several scores: the GDT-HA score assesses the accuracy of the backbone and overall structure; the GDC-SC score evaluates the accuracy and correctness of the side chain conformations; the MolProbity score measures the naturalness and physical correctness of the structure; and finally, the percentage of residues located in the favored region of the Ramachandran plot is reported as “Rama favored“ [27]. Following the refinement of the structure, the tertiary structures were validated through the ProSA server (https://prosa.services.came.sbg.ac.at/prosa.php), PROCHECK, and ERRAT tools available on the SAVES server (https://saves.mbi.ucla.edu/) [37, 38]. To present high-quality visual representations of the structures, PyMOL software version 2.5 was employed. Moreover, the final 3D structure of the predicted model for FMEC was utilized to identify B-cell conformational epitopes via the ElliPro server (http://tools.iedb.org/ellipro/). In silico immune simulation analysis The C-ImmSim server (http://150.146.2.1/C-IMMSIM/index.php) was employed to simulate the immune response, specifically T-cell or B-cell immune responses, for the FMEC. FMEC was administered at three time intervals: 1, 84, and 168 time steps, with the simulation volume set to 10. The default human HLA alleles utilized in the simulations included HLA-A*0201, HLA-B*0702, and HLA-DRB1*0101 [39, 41–43]. Cytokine-inducing epitope prediction To evaluate the potential of FMEC to induce the release of the cytokines IL-4 and IFN-γ, the online prediction tools IL-4Pred (https://webs.iiitd.edu.in/raghava/il4pred/scan.php) and IFNepitope (https://webs.iiitd.edu.in/raghava/ifnepitope/predict.php) were used. These servers utilize a support vector machine (SVM-based) model with default settings applied at the SVM threshold [44, 45]. Additionally, the antigenicity and conservancy of the predicted epitopes were examined via the VaxiJen v2.0 server (http://www.ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen.html) [20] and the Epitope Conservancy Analysis tool from the IEDB server (http://tools.iedb.org/conservancy/) [25], respectively. Molecular docking and normal mode analysis Molecular docking of the FMEC vaccine candidate with the TLR5 and TLR4 receptors was conducted via the ClusPro2 server (http://nrc.bu.edu/cluster/). To facilitate the docking process, the 3D structures of the receptors were obtained from the RCSB Protein Data Bank (www.rcsb.org) via the PDB IDs 3V47 and 3FxI. Subsequently, UCSF Chimera version 1.14 was applied to the preparation of proteins before the docking operation was performed. The best model was selected on the basis of energy levels and the number of clusters within the docked network. All intermolecular interactions within the docked complex were subsequently analyzed via the DimPlot program in LigPlot+ version 2.2.4 software, with a focus on the hydrogen and hydrophobic bonds present in the complex. Furthermore, to analyze the interactions between the FMEC vaccine candidate and MHC molecules (MHCI and MHCII), we first obtained 3D structures of MHCI (HLA A*0201) in complex with CD8 and MHCII (HLA DRB1*0101) in complex with CD4 from the RCSB PDB database (https://www.rcsb.org/) with PDB IDs 1AKJ and 3S4S, respectively. Next, the best-predicted MHCI- and MHCII-binding epitopes from FMEC, YLKVYEHAV and VQYGNATPA were used for docking analysis. After the PDB structure of MHC molecules was refined by eliminating ligands and water molecules via UCSF Chimera v.1.14 software, molecular docking was conducted between selected MHC-binding epitopes and their respective receptors. Additional docking analyses were performed as described above. Molecular dynamics simulations were performed via the iMOD server (iMODS, http://imods.chaconlab.org) [46] to assess the stability and dynamic behavior of the FMEC vaccine candidate, FMEC-TLR4, and FMEC-TLR5 docked complexes. This platform utilizes normal mode analysis (NMA) in internal dihedral coordinates, effectively capturing the collective functional motions of biological macromolecules and enabling the modeling of feasible conformational transitions between homologous structures, including those of large macromolecules. Additionally, the binding affinity and dissociation constant of the protein‒protein complexes were estimated via Prodigy (https://wenmr.science.uu.nl/prodigy/) [47], an online tool designed for predicting binding strength. In silico cloning To optimize the expression of the designed vaccine construct, codon adaptation was performed via the Java codon adaptation tool (JCat) (https://www.jcat.de/) [48]. Specifically, the algorithm was parameterized for Escherichia coli K12, guiding codon selection toward a GC content of 30–70% and a codon adaptation index (CAI) above 0.8, thereby facilitating efficient translation. The optimized sequence was subsequently subjected to the multiple cloning site (MCS) of the pET-28a plasmid vector via the NcoI and XhoI restriction sites via in silico cloning using SnapGene v7.2 software (https://www.snapgene.com/resources). Finally, the secondary structure of FMEC mRNA was analyzed via the mFold server (http://www.unafold.org/mfold/applications/rna-folding-form.php/) [62] and Visual Gene Developer software [63]. Results Protein sequence retrieval The initial investigation of the retrieved protein sequences, detailed in Supplementary Table 1, revealed promising characteristics for vaccine development. All the proteins exhibited antigenic properties (VaxiJen score > 0.4) and were predicted to be nonallergenic and nontoxic, except for FimA, which displayed toxicity in two short sequences (TGRDAHPCAIK and VIHLRECSTVVSE). Additionally, these proteins had no significant similarity to any sequence within the human proteome, easing fears of autoimmunity. Subcellular localization analysis revealed an extracellular distribution and hence enhanced their validity as possible targets for vaccines on the basis of greater access by the host immune system. Prediction of epitopes Defining B-cell epitopes The linear B-cell epitopes on the surface of each protein were predicted via the Bepipred 2.0 and ElliPro tools available on the IEDB server, and their potential antigenicity was subsequently assessed via VaxiJen, alongside their conservancy capacity across Salmonella and Shigella strains. In accordance with the VaxiJen guidelines, epitopes with antigenicity scores approaching 1 were considered to exhibit higher probabilities of antigenicity and immunogenicity. Linear B-cell epitopes demonstrating antigenicity scores exceeding 0.8 and conferring > 75% conservation across the analyzed strains are detailed in Table 1. This stringent selection process prioritized epitopes with robust potential for eliciting protective immune responses. Table 1Predicted linear B-cell epitopesProteinpositionPeptideAntigenicity ScoreConservancy among Shigella sp. / Salmonella sp. (%) Conservancy among S. flexneri / S. typhymurium (%) AlergenicityToxicitySpvB(Salmonella)135–141177–187433–444378–404180–187272–277YQPRTESARLSDPQAASHLSTNEPSDEEFKNSSRPKSSDPQAASHVDPQVP10.90.9211.1171.2275.2% (376/500)99.2% (496/500)99% (495/500)99.2% (496/500)99.8% (499/500)76% (380/500)100% (100/100)100% (100/100)100% (100/100)100% (100/100)100% (100/100)100% (100/100)NonNonNonNonNonNonNonNonNonNonNonNonSipD(Salmonella)36–41274–281224–242218–233STTEHRPGKDSKLEKAQEENMKTTQSGSGVKVATEAEARQ2.31.781.851.9898.6% (493/500)97.8% (489/500)80.2% (401/500)93.8% (469/500)89% (89/100)75% (75/100)60% (60/100)82% (82/100)NonNonNonNonNonNonNonNonRcK(Salmonella)68–7474–85QSGKRGEESGGIPGGMSWR2.851.0291.96% (460/500)93.75% (469/500)100% (100/100)97% (97/100)NonNonNonNonSigA(Shigella)521–5281070–10761140–1147293–301172–186845–853520–525TTQQHNNNTYTDSHAAGYRYHVTGAYSVTATHNGTQHHAITTQESADFSLSPEDAMKRLWRVNGQSANTTQQH10.941.221.130.90.871.9392% (466/506)91.7% (464/506)88.93% (450/506)58.1% (294/506)58.5% (296/506)97.63% (494/506)91.3% (462/506)100% (100/100)100% (100/100)100% (100/100)100% (100/100)100% (100/100)100% (100/100)100% (100/100)NonNonNonNonNonNonNonNonNonNonNonNonNonNonIpaD(Shigella)16–2431–37285–293214–225214–221172–193SPNNTNGSSDIKTTTSAEDETMKNNANNTVSQEQANKANNTVSQESSLAGWISPGGNDGNSVKLQVN2.141.711.531.311.530.8174% (370/500)76.6% (383/500)57.6% (288/500)45.4% (227/500)45.6% (228/500)72.40% (362/500)96% (96/100)99% (99/100)99% (99/100)90% (90/100)91% (91/100)93% (93/100)NonNonNonNonNonNonNonNonNonNonNonNonFimA(Shigella)79–86116–122EAGDKQMTCSTVVSE1.871.2696.94% (96/98)97.96% (95/98)100% (61/61)100% (61/61)NonNonNonNon Predicted linear B-cell epitopes SpvB (Salmonella) 135–141 177–187 433–444 378–404 180–187 272–277 YQPRTES ARLSDPQAASH LSTNEPSDEEFK NSSRPKS SDPQAASH VDPQVP 1 0.9 0.92 1 1.117 1.22 75.2% (376/500) 99.2% (496/500) 99% (495/500) 99.2% (496/500) 99.8% (499/500) 76% (380/500) 100% (100/100) 100% (100/100) 100% (100/100) 100% (100/100) 100% (100/100) 100% (100/100) Non Non Non Non Non Non Non Non Non Non Non Non SipD (Salmonella) 36–41 274–281 224–242 218–233 STTEHR PGKDSKLE KAQEENMKTT QSGSGVKVATEAEARQ 2.3 1.78 1.85 1.98 98.6% (493/500) 97.8% (489/500) 80.2% (401/500) 93.8% (469/500) 89% (89/100) 75% (75/100) 60% (60/100) 82% (82/100) Non Non Non Non Non Non Non Non RcK (Salmonella) 68–74 74–85 QSGKRGE ESGGIPGGMSWR 2.85 1.02 91.96% (460/500) 93.75% (469/500) 100% (100/100) 97% (97/100) Non Non Non Non SigA (Shigella) 521–528 1070–1076 1140–1147 293–301 172–186 845–853 520–525 TTQQHNNN TYTDSHA AGYRYHVT GAYSVTATHNGTQHHAITTQ ESADFSLSPEDAMKR LWRVNGQSA NTTQQH 1 0.94 1.22 1.13 0.9 0.87 1.93 92% (466/506) 91.7% (464/506) 88.93% (450/506) 58.1% (294/506) 58.5% (296/506) 97.63% (494/506) 91.3% (462/506) 100% (100/100) 100% (100/100) 100% (100/100) 100% (100/100) 100% (100/100) 100% (100/100) 100% (100/100) Non Non Non Non Non Non Non Non Non Non Non Non Non Non IpaD (Shigella) 16–24 31–37 285–293 214–225 214–221 172–193 SPNNTNGSS DIKTTTS AEDETMKNN ANNTVSQEQANK ANNTVSQE SSLAGWISPGGNDGNSVKLQVN 2.14 1.71 1.53 1.31 1.53 0.81 74% (370/500) 76.6% (383/500) 57.6% (288/500) 45.4% (227/500) 45.6% (228/500) 72.40% (362/500) 96% (96/100) 99% (99/100) 99% (99/100) 90% (90/100) 91% (91/100) 93% (93/100) Non Non Non Non Non Non Non Non Non Non Non Non FimA (Shigella) 79–86 116–122 EAGDKQMT CSTVVSE 1.87 1.26 96.94% (96/98) 97.96% (95/98) 100% (61/61) 100% (61/61) Non Non Non Non To predict conformational B-cell epitopes, three-dimensional structures of SigA, SpvB, RcK, and FimA proteins were initially generated via I-TASSER homology modeling. For each protein, the model with the highest C score (Model 1) was selected and subjected to refinement via the GalaxyRefine server. The accuracy of the refined structures was then assessed via the Saves and ProSA servers. The ProSA server, which employs X-ray and NMR crystallography data, assigned Z scores to the predicted protein structures. The Saves server was used to calculate ERRAT scores and generate Ramachandran plots, which were used to evaluate the distribution of amino acids within favored, allowed, and disallowed regions. The results of the initial and refined structure predictions, along with accuracy metrics, are summarized in Supplementary Tables 2 and Supplementary Fig. 1, demonstrating acceptable accuracy and precision for each protein’s final 3D structure. These refined structures, alongside the existing 3D structures of IpaD (PDBId:2J0o) and SipD (PDBId:2YM9), were subsequently used to identify conformational B-cell epitopes, with the results presented in Supplementary Tables 3 and a schematic representation in Supplementary Fig. 2. For each protein, a single surface B-cell epitope exhibiting the highest antigenicity and conservation across all Salmonella enterica spp. (especially Salmonella typhimurium) and Shigella spp. (especially Shigella flexneri) strains across multiple B-cell prediction servers was selected for constructing a vaccine candidate. The representative epitopes for the SpvB, SipD, RcK, SigA, IpaD, and FamA proteins were designated the epitopes SDPQAASH, STTEHR, QSGKRGE, LWRVNGQSA, DIKTTTS, and EAGDKQMT, respectively. Defining T-cell and MHCI/II binding epitopes The IEDB’s TepiTool was employed to predict binding affinities to human and mouse MHCII and MHC I molecules. For MHC II, 15-mer epitopes were predicted on the basis of their binding to the 26 most prevalent human class II alleles and three murine class II alleles. Similarly, for MHCI, 9-mer epitopes were predicted against the 27 most common human class A and B alleles, along with six murine class I alleles. Within each protein, conserved epitopes demonstrating strong binding affinity to both MHC class I and II molecules were predicted on the basis of the following criteria: percentile rank < 1, an antigenicity score exceeding 1, the number of interacting MHC class II alleles, broad coverage of human and murine alleles for both MHC classes, and a conservation score of > 90% across the analyzed Salmonella and Shigella strains. To refine epitope selection in some proteins, the criteria were adjusted to a minimum antigenicity score of 0.8 and a minimum conservation score of 70%, facilitating focused identification of potential targets. The prediction results are shown in Table 2. Finally, a consensus epitope of each protein was chosen on the basis of its ability to bind both the human and mouse MHC class I and II molecules while considering complete allele coverage, strong conservation, and high antigenicity for constructing a vaccine candidate. The representative epitopes for SpvB, SipD, RcK, SigA, IpaD, and FamA are VQYGNATPA, VVPSTTEHR, VAQADTHSV, IEAKNSQIV, YLKVYEHAV, and GVAFHGVADGKNPDV, respectively. Table 2Predicted helper T-cell epitopes using IEDB server (Percentile Rank ≤ 1)ProteinPositionPeptideAllelesAntigenicity ScoreConservancy in Shigella / Salmonella sp.Conservancy in S. flexneri/AlergenicityToxicity S. typhymurium SpvB(Salmonella)463–471ETDHRVVYRHLA-A*33:01, HLA-A*68:01, HLA-A*01:01, HLA-A*31:01, HLA-A*26:01, HLA-A*11:011.4298.8% (494/500)100% (100/100)NonNon299–307GFEIRLHRLHLA-B*08:012.7799.8% (499/500)99% (99/100)NonNon378–386NSSRPKSKWHLA-B*58:01,HLA-B*57:01,HLA-B*53:01 HLA-A*32:011.0199% (495/500)100% (100/100)NonNon298–306YGFEIRLHRHLA-A*33:01,HLA-A*31:01 HLA-A*68:012.7499.8% (499/500)99% (99/100)NonNon469–483LKLDKPALS16 (No. of MHCII)198.2% (491/500)100% (100/100)NonNon50–58FAPALALHYHLA-B*35:01, HLA-A*26:01, HLA-A*01:01, HLA-B*53:01, 22 (No. of MHCII)1.7199.2% (496/500)97% (97100)NonNon510–518NDTILNIYLH-2-Dd, H-2-Db, H2-IAb, H2-Ied, H-2-Kk1.1598% (490/500)99% (99/100)NonNon236–244VQYGNATPAHLA-A*02:06, H-2-Db, H2-IAb, H2-IAd, 13 (No. of MHCII)0.7299.8% (499/500)99% (99/100)NonNonSipD(Salmonella)70–78EENNDERTLHLA-B*44:03, HLA-B*44:02, HLA-B*40:01, H-2-Kk1.4598.6% (493/500)92% (92/100)NonNon30–44VVPSTTEHRHLA-A*68:01, HLA-A*31:01, HLA-A*33:01, H2-IEd, 19 (No. of MHCII)0.8997.8% (489/500)87% (87/100)NonNon202–216YNQINSNTVH-2-Db, 21(No. of MHCII)0.994.8% (474/500)86% (86/100)NonNon250–258SGYVVTVDLH-2-Kb, H-2-Dd, H-2-Ld, H-2-Db, H2-IAb1.5794.6% (473/500)73% (73/100)NonNon219–233SGSGVKVATEAEARQH2-Iad1.9594% (470/500)82% (82/100)NonNonRcK(Salmonella)98–106GPAVRVNELHLA-B*07:02, HLA-B*08:01, H-2-Ld, HLA-B*35:01, HLA-B*51:01, H2-Iab1.7194.05% (316/336)94.2% (65/69)NonNon95–103LMAGPAVRVHLA-A*02:03, HLA-A*02:01, HLA-A*02:06, HLA-A*68:02, H2-IAb1.2694.94% (319/336)95.65% (66/69)NonNon87–95DVKATYWSLHLA-B*08:01, HLA-A*26:01, HLA-A*68:02, HLA-B*51:011.3497.02% (326/336)98.55% (68/69)NonNon21–29VAQADTHSVHLA-B*51:01, HLA-A*02:06, H-2-Db, 19 (No. of MHCII)0.8870.24% (236/33678.26% (54/69)NonNonSigA(Shigella)542–550DISVRNAQLHLA-B*08:01, HLA-A*68:022.0597.63% (494/506)100% (100/100)NonNon585–593DQESTVNKRHLA-A*33:011.7599.21% (502/506)99% (99/100)NonNon611–619DWESRKFNFHLA-A*24:02, HLA-A*23:01, H2-IEd1.4299.6% (502/506)99% (99/100)NonNon668–676EGFGFRQQLHLA-A*68:02, HLA-B*51:01, H-2-Dd, HLA-B*08:01, H-2-Kb, H-2-Ld, H2-IAb, H2-IAd1.399.6% (504/506)99% (99/100)NonNon773–781GATLKAQNYHLA-A*30:02, HLA-B*15:011.1398.02% (496/506)99% (99/100)NonNon669–677GFGFRQQLRHLA-A*31:01, HLA-A*33:01, H2-IEd1.299.6% (504/506)99% (99/100)NonNon689–697GSLSAQNSKHLA-A*11:01, HLA-A*30:01, HLA-A*03:01, H2-IAb, H2-IAd1.23100% (506/506)100% (100/100)NonNon641–649IEAKNSQIVHLA-B*40:01, HLA-B*44:02,HLA-B*44:03, H-2-Kk, 18 (No.of MHCII)1.19100% (506/506)100% (100/100)NonNon665–673ITGEGFGFRHLA-A*68:01, HLA-A*31:01, HLA-A*33:012.1999.6% (504/506)99% (99/100)NonNon616–624KFNFRKLNLHLA-B*08:01, HLA-A*30:01, HLA-A*24:02, HLA-A*23:01, H2-IEd1.998.81% (500/506)96% (96/100)NonNon941–949KGSNKEIFKHLA-A*30:01, HLA-A*11:011.0494.86% (480/506)99% (99/100)NonNon621–629KLNLENATLHLA-A*02:01, HLA-A*32:01, HLA-A*02:031.0398.42% (498/506)95% (95/100)NonNon769–777MTEDGATLKHLA-A*11:01, HLA-A*68:01, HLA-A*01:01, HLA-A*03:011.598.42% (498/506)100% (100/100)NonNon664–672NITGEGFGFHLA-A*26:011.6299.6% (504/506)99% (99/100)NonNon636–644NVKGHIEAKHLA-A*68:01, HLA-A*30:01, HLA-A*33:01, HLA-A*11:01, HLA-A*03:011.71100% (506/506)100% (100/100)NonNon586–594QESTVNKRYHLA-B*44:02, HLA-B*44:03, HLA-A*01:01, HLA-B*40:01, HLA-A*30:02, H-2-Kk1.599.21% (502/506)99% (99/100)NonNon523–531QQHNNNANLHLA-B*15:01, HLA-B*40:01, H-2-Db1.0492.89% (470/506)100% (100/100)NonNon629–700SAQNSKITVHLA-B*51:01, H-2-Db1.29100% (506/506)100% (100/100)NonNon630–638SIGRDANVKHLA-A*11:011.94100% (506/506)100% (100/100)NonNon690–698SLSAQNSKIHLA-A*02:031.32100% (506/506)100% (100/100)NonNon917–925SLSVDIQKVHLA-A*02:03, HLA-A*02:01, H2-Iad1.2795.26% (482/506)100% (100/100)NonNon544–552SVRNAQLTLHLA-A*30:01, HLA-B*07:02, HLA-A*32:01, HLA-B*08:01, HLA-B*15:01, H-2-Ld1.1797.63% (494/506)100% (100/100)NonNon770–778TEDGATLKAHLA-B*40:011.8798.42% (498/506)100% (100/100)NonNon700–708VGDKSTVTMHLA-B*35:01, H-2-Dd, H-2-Db, H-2-Ld1.1999.21% (502/506)98% (98/100)NonNon1232–1246IKGEKDGRI18 (No. of MHCII)1.9786.56% (438/506)100% (100/100)NonNon977–991YKVAENTAS17 (No. of MHCII), H2-IAb, H2-IEd1.1892.09% (466/506)98% (98/100)NonNon1139–1153YRYHVTEDS19 (No. of MHCII)1.1188.54% (448/506)100% (100/100)NonNon543–551ISVRNAQLTH-2-Db1.5597.63% (494/506)100% (100/100)NonNonIpaD(Shigella)184–192DGNSVKLQVHLA-B*51:012.6586.6% (433/506)100% (100/100)NonNon81–89HSSQISMDVHLA-A*68:02, H2-Iab1.1487.4% (437/506)100% (100/100)NonNon12–026FSPNNTNGSH2-IAb, H2-IEd, 13(No. of MHCII)1.4374% (370/506)96% (96/100)NonNon145–159YLKVYEHAVHLA-B*08:01, HLA-A*02:03, HLA-A*02:01, HLA-A*02:06, H2-IAd, 19(No. of MHCII)0.6187% (435/506)100% (100/100)NonNonFimA(Shigella)29–37APVPFVIHLHLA-B*07:02, HLA-B*51:01, HLA-B*53:01, HLA-B*35:01, HLA-B*08:01, HLA-A*68:02, H2-IAb, H-2-Ld, H-2-Kb, H-2-Kk, H-2-Kd, H-2-Dd1.0792.86% (91/98)93.44% (57/61)NonNon79–87EAGDKQMTVHLA-A*68:02, HLA-B*51:01,HLA-B*08:01, H2-IAd,1.4695.92% (94/98)100% (61/61)NonNon115–123ECSTVVSERHLA-A*33:01, HLA-A*68:011.2593.88% (92/98)96.72% (59/61)NonNon106–114PVPFVIHLRHLA-A*68:01 HLA-A*33:01, H2-IAb, 2(No. of MHCII)1.0394.9% (93/98)93.44% (57/61)NonNon125–139GVAFHGVADGKNPDVH2-IAb, H2-IAd, 9(No. of MHCII)1.3595.92% (94/98)98.36% (60/61)NonNon Predicted helper T-cell epitopes using IEDB server (Percentile Rank ≤ 1) SpvB (Salmonella) H-2-Dd, H-2-Db, H2-IAb, H2-Ied, H-2-Kk SipD (Salmonella) RcK (Salmonella) SigA (Shigella) IpaD (Shigella) FimA (Shigella) Furthermore, cytotoxic T lymphocyte (CTL) epitopes were predicted via the NetChop and NetCTL tools within the IEDB database, considering proteasomal processing factors, including proteasomal cleavage, TAP transport efficiency, and MHC class I binding affinity. For each protein, a single epitope was selected on the basis of its highest predicted score and strong antigenicity (VaxiJen score > 0.7). The detailed results from this analysis are presented in Table 3. For the design of the vaccine construct, the representative epitopes selected from the Shigella proteins SigA, IpaD, and FimA were VTARAGLGY, VYEHAVSSY, and VSERVGVAF, respectively. Table 3Predicted cytotoxic T-cell epitopes by IEDB serverProteinStart positionPeptide sequencePredicted MHC binding affinityRescale binding affinityC terminal cleavage affinityTAP transport efficiencyPredictions scoreAntigenicity scoreSpvB14081327509179344569240197224259ESSFYRLEYRTSHGVPQYLVSRLLLEYINDTILNIYLSDPQAASHLCAARTLAYLSDDATADTNATPAADLYVTPAGEHIYGRDRSAMRYWLFTLVFDY0.57950.35070.32230.27600.28640.19780.20320.15910.14320.13480.10643.93592.38152.18871.87461.94551.34371.38021.08090.97280.91530.72270.98480.99880.99820.99420.10200.87970.14910.89030.87810.99800.99812.76703.27503.07302.5630-0.53502.9170-0.70502.94803.00402.93002.98004.22202.69512.49212.15191.93401.62151.36731.36181.25471.21151.02140.670.830.460.660.69-0.611.210.73-0.40.942.74SipD15631142129152141145160109279244VVAVYTDFYKYSNANSLYGTDIISLSQISDAEIWDMVYENVVAVYNISAIGDSYIGDSYLGVYYTDFYQAFSFSAPTSALFKLEMDNAKYCLKSYGSGY0.29160.25860.27510.26310.21990.20650.20040.23510.15900.15100.11671.98041.75661.86851.78681.49331.40241.36121.59701.08021.02560.79280.94040.97370.90610.36650.99640.66150.59300.02000.99770.99910.97693.28303.2430-0.35000.20203.07703.03702.6590-2.41602.65102.96102.95502.28562.06481.98691.85191.79661.65341.58311.47921.36241.32351.08700.270.56-0.210.370.370.360.47-0.34-0.240.980.68RcK8425102WRDDVKATYDTHSVSVGYRVNELVSLY0.21540.18960.16251.46301.28751.10350.99890.99810.99912.98202.77103.26001.76191.57581.41640.311.260.03SigA4971206112286104149598774997276310371255233257789352208107172640LTNTDAYMYVTARAGLGYATTSIGGAYSTNIAWTNYYSDNYTHVQLSLTNTDAYASSGAAKSYGTPTENNKYHTTWTLTGYMSDGYNMTEASGGYSDNYMNKIYSLKYFGDSGSGSYNLDWKNKSYDSGSGSYLYKSDKKASILLIFDKNQTYSTILNGKQYYTDSHASSNTVTAQGKMYILSSIYGSY0.54910.49840.43340.37720.40720.31150.27600.26590.24000.28990.18850.17760.17930.17720.15690.14830.12630.12510.16980.12810.10763.72913.38532.94362.56202.76552.11581.87461.80571.62991.96921.28001.20621.21741.20351.06551.00720.85770.84991.15310.87000.73100.97100.96870.92310.96250.33900.54080.84410.96660.99840.03700.90920.99900.99850.98920.99850.85140.99910.98940.04650.31050.99862.86703.08302.98102.9160-0.29102.99703.11502.71902.6090-1.70603.00002.81502.39602.70002.45600.93503.13903.0350-1.52303.16002.97204.01813.68473.23112.85212.80182.34682.15692.08661.91011.88951.56641.49681.48701.48691.33811.18171.16451.15011.08401.07451.02940.531.140.750.210.340.671.380.300.070.500.960.231.520.741.640.49-1.11-0.141.320.41-0.05IpaD15632415232223614129198AVSSYTQMYDTDKLFLHFVYEHAVSSYCTDTDKLFLKVSQKNGGYSINDINEQYNSDIKTTTSALEELKEKY0.42180.40010.27600.28320.23120.14510.17200.11572.86442.71751.87421.92371.57010.98571.16800.78570.99690.99530.98580.83870.92490.99890.25510.99693.18002.03103.20800.77003.13603.1530-2.29703.04503.17292.96842.18252.08801.86561.29321.09141.08750.220.240.740.281.200.201.750.43FimA181120203TSLHFIAKYVSERVGVAFAQAWFSLTY0.23960.15060.13271.62731.02270.90110.99880.64180.99923.01602.54003.30201.92791.24601.21610.250.760.45 Predicted cytotoxic T-cell epitopes by IEDB server 140 81 327 509 179 344 569 240 197 224 259 ESSFYRLEY RTSHGVPQY LVSRLLLEY INDTILNIY LSDPQAASH LCAARTLAY LSDDATADT NATPAADLY VTPAGEHIY GRDRSAMRY WLFTLVFDY 0.5795 0.3507 0.3223 0.2760 0.2864 0.1978 0.2032 0.1591 0.1432 0.1348 0.1064 3.9359 2.3815 2.1887 1.8746 1.9455 1.3437 1.3802 1.0809 0.9728 0.9153 0.7227 0.9848 0.9988 0.9982 0.9942 0.1020 0.8797 0.1491 0.8903 0.8781 0.9980 0.9981 2.7670 3.2750 3.0730 2.5630 -0.5350 2.9170 -0.7050 2.9480 3.0040 2.9300 2.9800 4.2220 2.6951 2.4921 2.1519 1.9340 1.6215 1.3673 1.3618 1.2547 1.2115 1.0214 0.67 0.83 0.46 0.66 0.69 -0.61 1.21 0.73 -0.4 0.94 2.74 156 311 42 129 152 141 145 160 109 279 244 VVAVYTDFY KYSNANSLY GTDIISLSQ ISDAEIWDM VYENVVAVY NISAIGDSY IGDSYLGVY YTDFYQAFS FSAPTSALF KLEMDNAKY CLKSYGSGY 0.2916 0.2586 0.2751 0.2631 0.2199 0.2065 0.2004 0.2351 0.1590 0.1510 0.1167 1.9804 1.7566 1.8685 1.7868 1.4933 1.4024 1.3612 1.5970 1.0802 1.0256 0.7928 0.9404 0.9737 0.9061 0.3665 0.9964 0.6615 0.5930 0.0200 0.9977 0.9991 0.9769 3.2830 3.2430 -0.3500 0.2020 3.0770 3.0370 2.6590 -2.4160 2.6510 2.9610 2.9550 2.2856 2.0648 1.9869 1.8519 1.7966 1.6534 1.5831 1.4792 1.3624 1.3235 1.0870 0.27 0.56 -0.21 0.37 0.37 0.36 0.47 -0.34 -0.24 0.98 0.68 84 25 102 WRDDVKATY DTHSVSVGY RVNELVSLY 0.2154 0.1896 0.1625 1.4630 1.2875 1.1035 0.9989 0.9981 0.9991 2.9820 2.7710 3.2600 1.7619 1.5758 1.4164 0.31 1.26 0.03 497 1206 112 286 1041 495 987 749 972 763 1037 1 255 233 257 789 352 208 1071 726 40 LTNTDAYMY VTARAGLGY ATTSIGGAY STNIAWTNY YSDNYTHVQ LSLTNTDAY ASSGAAKSY GTPTENNKY HTTWTLTGY MSDGYNMTE ASGGYSDNY MNKIYSLKY FGDSGSGSY NLDWKNKSY DSGSGSYLY KSDKKASIL LIFDKNQTY STILNGKQY YTDSHASSN TVTAQGKMY ILSSIYGSY 0.5491 0.4984 0.4334 0.3772 0.4072 0.3115 0.2760 0.2659 0.2400 0.2899 0.1885 0.1776 0.1793 0.1772 0.1569 0.1483 0.1263 0.1251 0.1698 0.1281 0.1076 3.7291 3.3853 2.9436 2.5620 2.7655 2.1158 1.8746 1.8057 1.6299 1.9692 1.2800 1.2062 1.2174 1.2035 1.0655 1.0072 0.8577 0.8499 1.1531 0.8700 0.7310 0.9710 0.9687 0.9231 0.9625 0.3390 0.5408 0.8441 0.9666 0.9984 0.0370 0.9092 0.9990 0.9985 0.9892 0.9985 0.8514 0.9991 0.9894 0.0465 0.3105 0.9986 2.8670 3.0830 2.9810 2.9160 -0.2910 2.9970 3.1150 2.7190 2.6090 -1.7060 3.0000 2.8150 2.3960 2.7000 2.4560 0.9350 3.1390 3.0350 -1.5230 3.1600 2.9720 4.0181 3.6847 3.2311 2.8521 2.8018 2.3468 2.1569 2.0866 1.9101 1.8895 1.5664 1.4968 1.4870 1.4869 1.3381 1.1817 1.1645 1.1501 1.0840 1.0745 1.0294 0.53 1.14 0.75 0.21 0.34 0.67 1.38 0.30 0.07 0.50 0.96 0.23 1.52 0.74 1.64 0.49 -1.11 -0.14 1.32 0.41 -0.05 156 324 152 322 236 141 29 198 AVSSYTQMY DTDKLFLHF VYEHAVSSY CTDTDKLFL KVSQKNGGY SINDINEQY NSDIKTTTS ALEELKEKY 0.4218 0.4001 0.2760 0.2832 0.2312 0.1451 0.1720 0.1157 2.8644 2.7175 1.8742 1.9237 1.5701 0.9857 1.1680 0.7857 0.9969 0.9953 0.9858 0.8387 0.9249 0.9989 0.2551 0.9969 3.1800 2.0310 3.2080 0.7700 3.1360 3.1530 -2.2970 3.0450 3.1729 2.9684 2.1825 2.0880 1.8656 1.2932 1.0914 1.0875 0.22 0.24 0.74 0.28 1.20 0.20 1.75 0.43 181 120 203 TSLHFIAKY VSERVGVAF AQAWFSLTY 0.2396 0.1506 0.1327 1.6273 1.0227 0.9011 0.9988 0.6418 0.9992 3.0160 2.5400 3.3020 1.9279 1.2460 1.2161 0.25 0.76 0.45 In addition, the population coverage analysis of selected MHCI/MHCII binding epitopes in FMEC revealed a global coverage of 80.57 ± 12.15%, supporting its potential for broad immunization. The details of global population coverage analysis are shown in Table 4. Table 4Global population coverage analysis of selected MHCI/MHCII epitopesPopulation/areaClass combinedcoverageaaverage_hitbpc90cCentral Africa75.01%4.640.4Central America49.91%3.290.8East Africa80.35%5.20.51East Asia93.85%7.771.58Europe96.75%7.872.08North Africa88.1%5.950.84North America96.62%8.12.42Northeast Asia80.22%4.940.51Oceania76.08%4.740.42South Africa59.23%1.950.25South America81.62%4.680.54South Asia87.16%6.180.78Southeast Asia80.59%4.680.52Southwest Asia74.17%3.460.39West Africa80.09%5.140.5West Indies89.34%5.920.94Average80.575.280.84Standard deviation12.151.630.61a projected population coverageb average number of epitope hits / HLA combinations recognized by the populationc minimum number of epitope hits / HLA combinations recognized by 90% of the population Global population coverage analysis of selected MHCI/MHCII epitopes a projected population coverage b average number of epitope hits / HLA combinations recognized by the population c minimum number of epitope hits / HLA combinations recognized by 90% of the population Multiepitope construct design The final multiepitope construct was designed by assembling predicted epitopes through rational linker selection to preserve structural flexibility and immunogenic synergy. Flexible KK and GGGGS linkers were used to connect B-cell epitopes, GPGPG linkers were employed between helper T-cell epitopes, and AAY linkers were applied to join cytotoxic T-lymphocyte epitopes. The flagellin protein FliC from Salmonella typhimurium (UniProt ID: P06179) was incorporated as an adjuvant to potentiate innate immune activation. To define suitable regions of FliC for integration, sequence segments rich in both linear and conformational B-cell epitopes were identified through the IEDB database. The antigenicity of the predicted regions was further evaluated via the VaxiJen server, and epitopes with an antigenicity score greater than 1 were selected from conserved regions of the FliC structure (see Fig. 1; Tables 5 and 6). The engineered construct was strategically inserted into the FliC scaffold, replacing the hypervariable domains that exhibit limited conservation among Salmonella strains. This insertion was confined between the highly conserved α-helical segments (D1 domain) in the N- and C-terminal regions (residues 60–130 and 400–460, respectively). Furthermore, these segments encompass key residues involved in TLR5 recognition. The complete amino acid sequence and structural layout of the designed chimera are illustrated in Fig. 2. 3D structure of Salmonella typhimurium FliC, showing conserved and hypervariable domains, and B-cell epitope prediction. (a) Predicted linear B-cell epitopes and (b) predicted conformational B-cell epitopes. Epitope-containing regions are highlighted in yellow, with the protein backbone shown in gray Table 5Predicted linear B-cell epitopes of FliCStartEndLinear B-cell Epitope SequencelengthAntigenicity score6368QASRNA61.0489111QRVRELAVQSANSTNSQSDLDSI230.86100111ANSTNSQSDLDS121.17117411TQRLNEIDRVSGQTQFNGVKVLAQDNTLTIQVGANDGETIDIDLKQINSQTLGLDTLNVQQKYKVSDTAATVTGYADTTIALDNSTFKASATGLGGTDQKIDGDLKFDDTTGKYYAKVTVTGGTGKDGYYEVSVDKTNGEVTLAGGATSPLTGGLPATATEDVKNVQVANADLTEAKAALTAAGVTGTASVVKMSYTDNNGKTIDGGLAVKVGDDYYSATQNKDGSISINTTKYTADDGTSKTALNKLGGADGKTEVVSIGGKTYAASKAEGHNFKAQPDLAEAAATTTENPLQK2951142156TLTIQVGANDGETID151.41173183NVQQKYKVSDT111.06196221ALDNSTFKASATGLGGTDQKIDGDLK261.36230279YAKVTVTGGTGKDGYYEVSVDKTNGEVTLAGGATSPLTGGLPATATEDVK501.04310320SYTDNNGKTID111.53401412ATTTENPLQKID120.86438458TNLGNTVNNLTSARSRIEDSD210.52450460ARSRIEDSDYA110.52 Predicted linear B-cell epitopes of FliC Table 6Predicted conformational B-cell epitopes of FliCNo.ResiduesNumber of residuesScore1A: N56, A: I57, A: K58, A: G59, A: L60, A: T61, A: Q62, A: A63, A: S64, A: R65, A: N66, A: A67, A: N68, A: D69, A: G70, A: I71, A: S72, A: Q75, A: Q130, A: F131, A: N132, A: G133, A: V134, A: T142, A: L143, A: T144, A: I145, A: Q146, A: V147, A: G148, A: A149, A: N150, A: D151, A: G152, A: E153, A: T154, A: I155, A: D156, A: A427, A: V428, A: Q429, A: N430, A: R431, A: F432, A: N433, A: S434, A: A435, A: I436, A: T437, A: N438, A: L439, A: G440, A: N441, A: T442, A: V443, A: N444, A: N445, A: L446, A: T447, A: S448, A: A449, A: R450620.772A: A196, A: L197, A: D198, A: N199, A: S200, A: T201, A: F202, A: K203, A: A204, A: S205, A: A206, A: T207, A: G208, A: L209, A: G210, A: G211, A: T212, A: D213, A: Q214, A: K215, A: I216, A: D217, A: G218, A: D219, A: L220, A: K221, A: D223, A: D224, A: T225, A: T226, A: K228, A: Y230, A: A231, A: K232, A: V233, A: T234, A: V235, A: T236, A: G237, A: G238, A: T239, A: G240, A: K241, A: D242, A: G243, A: Y244, A: Y245, A: E246, A: V247, A: S248, A: D250, A: K251, A: T252, A: N253, A: G254, A: E255, A: V256, A: T257, A: L258, A: A259, A: G260, A: G261, A: A262, A: T263, A: S264, A: P265, A: L266, A: T267, A: G268, A: G269, A: L270, A: P271, A: A272, A: T273, A: A274, A: T275, A: E276, A: D277, A: V278, A: K279800.743A: Q89, A: R90, A: R92, A: E93, A: L94, A: A95, A: V96, A: Q97, A: S98, A: A99, A: N100, A: S101, A: T102, A: N103, A: S104, A: Q105, A: S106, A: D107, A: L108, A: S110, A: I111, A: G168, A: L172, A: N173, A: V174, A: Q175, A: Q176, A: K177, A: Y178, A: K179, A: V180, A: S181, A: D182, A: T183, A: S310, A: T312, A: D313, A: N314, A: N315, A: G316, A: K317, A: T318, A: I319, A: D320, A: A366, A: D367, A: A400, A: A401, A: T402, A: T403, A: T404, A: E405, A: N406, A: P407, A: L408, A: Q409, A: K410570.66 Predicted conformational B-cell epitopes of FliC New designed FliC-based multiepitope construct (FMEC) Physicochemical properties and solubility prediction The physicochemical properties of the designed FMEC were assessed via the ProtParam server. Analysis revealed a molecular weight of nearly 33.6 kDa and an isoelectric point (pI) of 7.2, confirming that the protein is neutral in nature. It is made up of 325 amino acid residues, 29 of which are positive (Arg + Lys) charges and 29 of which are negative (Asp + Glu) charges. An instability index of 32.64 suggests that the protein is stable (II < 40). The aliphatic index of 61.62 implies favorable thermal stability, whereas a GRAVY score of -0.7 indicates hydrophilicity and good water solubility. The predicted half-life of the construct is 30 h in mammalian reticulocytes, > 20 h in yeast, and > 10 h in E. coli under laboratory conditions. Solpro analysis suggested high solubility (score of 0.8) under high expression conditions in E. coli. Furthermore, VaxiJen, ToxinPred, and AlgPred analyses predict that the designed FMEC construct is antigenic (score of 1.07), nontoxic, and nonallergenic, supporting its potential for further development. Secondary and tertiary structure prediction The PSIPRED results, indicating the positions and amino acid compositions of 32.3% alpha-helices, 31.9% beta-sheets, and 35.8% random coil structures, are depicted in Fig. 3a. The I-TASSER server predicted five 3D structure models with C-scores between − 3.41 and − 1.94, within the permissible range of -5 to 2. The model with the best C score of -1.94, a TM score of 0.48 ± 0.15 and a root-mean-square deviation (RMSD) of 10 ± 4.6 angstroms was chosen for further refinement. A TM score above the threshold of 0.5 indicates a reliable topology, and the RMSD is the average of the residue pair distances. The GalaxyRefine server subsequently predicted five refined models, of which Model 2 was deemed optimal on the basis of the GDT-HA (0.99), RMSD (0.24), and MolProbity (2.43) scores. Ramachandran plot analysis of this model revealed that 90.8% of the residues were in favored regions, 7.3% were in allowed regions, and only 1.9% were in disallowed regions, indicating a high-quality structure. The refined structure achieved an overall quality factor of 90.14% and a ProSA Z score of -11.4, aligning closely with scores observed in native proteins of similar size. A comprehensive summary of the 3D protein structure prediction and the analysis of the ProSA and Ramachandran plots before and after refinement are presented in Fig. 3b and c. Finally, using the refined 3D structure, the EliPro server predicted conformational B-cell epitopes, identifying 4 epitopes involving 325 residues, with antigenicity scores ranging from 0.523 to 0.799 and epitope lengths ranging from 4 to 104 amino acids. These epitope prediction results are detailed in Fig. 3d and Table 7. Secondary and tertiary structure analysis of the designed FMEC vaccine with conformational B-cell epitope mapping. (a) Secondary structure of the designed protein, including alpha-helices (pink), beta-sheets (orange), and random coils (gray). Three-dimensional structural validation analysis of FMEC via the ProSA server and Ramachandran plot: (b) pre- and (c) postrefinement, including a local model quality plot showing energy as a function of amino acid sequence position; ProSA analysis yielded scores of -3.23 and − 4.11 before and after refinement, respectively, indicating a shift toward the structural characteristics of native proteins (based on X-ray crystallography and NMR data). The local model quality plot shows a reduction in positive energy values (above the threshold), suggesting correction of problematic regions and increased structural stability postrefinement. (d) Homology model of the FMEC generated by the I-TASSER server, visualized via PyMol v.2.5, and three-dimensional representation of conformational B-cell epitopes on the designed FMEC structure, as identified via the EliPro server. Epitope-containing regions are highlighted in yellow, with the protein backbone shown in gray Table 7Characterization of conformational B-cell epitopes within the vaccine constructNo.ResiduesNumber of residuesScore1A: M1, A: H2, A: H3, A: H5, A: H6, A: H7, A: G8, A: L9, A: T10, A: Q11, A: R14, A: E92, A: R94, A: K96, A: Q97, A: S98, A: G99, A: K100, A: R101, A: G102, A: E103, A: K104, A: N295, A: N298, A: S299, A: A300, A: I301, A: T302, A: N303, A: L304, A: G305, A: N306, A: T307, A: V308, A: N309, A: N310, A: L311, A: T312, A: S313, A: A314, A: R315, A: S316, A: R317, A: I318, A: E319, A: D320, A: S321, A: D322, A: Y323, A: A324, A: T325510.7992A: E42, A: L43, A: V45, A: Q46, A: A48, A: N49, A: S50, A: T51, A: N52, A: S53, A: S55, A: D56, A: S59, A: I60, A: A62, A: E63160.7363A: T124, A: T125, A: S126, A: G127, A: G128, A: G129, A: G130, A: S131, A: E132, A: A133, A: G134, A: D135, A: K136, A: Q137, A: M138, A: T139, A: K140, A: K141, A: V142, A: Q143, A: G145, A: N146, A: A147, A: T148, A: P149, A: A150, A: G151, A: P152, A: G153, A: P154, A: G155, A: V156, A: V157, A: P158, A: S159, A: T160, A: T161, A: E162, A: H163, A: R164, A: G165, A: P166, A: G167, A: P168, A: G169, A: V170, A: A171, A: Q172, A: A173, A: D174, A: T175, A: H176, A: S177, A: V178, A: G179, A: P180, A: Q190, A: G193, A: P194, A: G195, A: P196, A: A205, A: V206, A: G207, A: P208, A: G209, A: G211, A: G212, A: V218, A: A219, A: D220, A: G221, A: K222, A: N223, A: P224, A: D225, A: V226, A: G227, A: P228, A: G229, A: P230, A: G231, A: V232, A: T233, A: A234, A: R235, A: A236, A: G237, A: L238, A: G239, A: Y240, A: A241, A: A242, A: V244, A: Y245, A: E246, A: H247, A: A248, A: V249, A: S250, A: S251, A: Y252, A: A253, A: A2541040.644A: D80, A: P81, A: Q82, A: A8340.52 Characterization of conformational B-cell epitopes within the vaccine construct In silico immune simulation analysis The C-ImmSim server was used to simulate in silico humoral and cellular immune responses in mammalian cells following immunization with the designed FMEC (Fig. 4). The simulations further forecasted early IgM response following the first dose administration and signs of early immune activation. The administration of subsequent increasing doses was correlated with a strong secondary immune response represented by B lymphocyte proliferation and differentiation into plasma cells, isotype switching (to IgG + IgM, IgG1, and IgG2), and the development of memory B cells. Moreover, expansions of T-cell populations (helper, cytotoxic, regulatory, and memory T cells), natural killer cells, and dendritic cells were expected. In addition, the upregulation of humoral and cell-mediated immune cytokines (e.g., IL-4, IL-6, IL-10, IL-2, IL-12, IL-18, IL-23, IFN-γ, IFN-β, TNFα, and TGFβ) was observed. In silico immune response simulation. C-ImmSim analysis predicts that immunization with the FMEC-designed construct elicits a robust, multifaceted immune response in mammalian cells. (a) Simulated immunoglobulin production shows an initial IgM response followed by IgG isotype switching and the production of IgG1 and IgG2 antibodies. (b) Predicted cytokine and chemokine profiles reveal elevated levels of key mediators of both humoral and cellular immunity. (c, d) The model predicts the proliferation and maturation of B lymphocytes, plasma cells (c), and T lymphocyte subsets (helper, cytotoxic, and regulatory) (d). (e-g) Simulations also revealed increased populations of natural killer cells (e), macrophages (f), and dendritic cells (g), suggesting the activation of both the innate and adaptive immune pathways Moreover, epitope prediction servers were employed to comprehensively scan the designed FMEC for potential 15-mer epitopes capable of stimulating cytokine production. Notably, assays using both the IL-4 and IFN-γ cytokine induction prediction tools revealed the ability of multiple high-scoring epitope regions for both cytokines within the construct sequence, highlighted in red in the main sequence, to elicit a balanced immune reaction involving both humoral and cell-mediated routes (Table 8). Table 8Location of predicted cytokine-inducing epitopes in the designed vaccine constructIL-4 epitopes: Red positionMHHHHHHGLTQASRNANDGISIAQTTEGALNEINNNLQRVRELAVQSANSTNSQSDLDSIQAEITQRLNEIDRVSGQTSDPQAASHKKSTTEHRKKQSGKRGEKKLWRVNGQSAGGGGSDIKTTTSGGGGSEAGDKQMTKKVQYGNATPAGPGPGVVPSTTEHRGPGPGVAQADTHSVGPGPGIEAKNSQIVGPGPGYLKVYEHAVGPGPGGVAFHGVADGKNPDVGPGPGVTARAGLGYAAYVYEHAVSSYAAYSSSNVGQTYAATTTENPLQKIDAALAQVDTLRSDLGAVQNRFNSAITNLGNTVNNLTSARSRIEDSDYAIFN-γ epitopes: Red positionMHHHHHHGLTQASRNANDGISIAQTTEGALNEINNNLQRVRELAVQSANSTNSQSDLDSIQAEITQRLNEIDRVSGQTSDPQAASHKKSTTEHRKKQSGKRGEKKLWRVNGQSAGGGGSDIKTTTSGGGGSEAGDKQMTKKVQYGNATPAGPGPGVVPSTTEHRGPGPGVAQADTHSVGPGPGIEAKNSQIVGPGPGYLKVYEHAVGPGPGGVAFHGVADGKNPDVGPGPGVTARAGLGYAAYVYEHAVSSYAAYSSSNVGQTYAATTTENPLQKIDAALAQVDTLRSDLGAVQNRFNSAITNLGNTVNNLTSARSRIEDSDYA Location of predicted cytokine-inducing epitopes in the designed vaccine construct Molecular docking and normal mode analysis of FMEC Molecular docking experiments indicated that the engineered FMEC construct established stable and energetically favorable interactions with both TLR4 and TLR5. In the TLR4 complex, the dominant configuration (cluster 0; weighted score of -868.1 kcal mol⁻¹; 39 members) exhibited multiple stabilizing contacts. Five and twenty residues of the vaccine interacted with four and twenty-two residues of TLR4 chains A and B, respectively, forming thirty-nine hydrogen bonds complemented by a range of hydrophobic associations (Fig. 5a and b; Supplementary Table S4a). A trend comparable to that of TLR5 was observed. The lowest-weighted score cluster (cluster 5; -767.8 kcal mol⁻¹; 33 members) represented the most favorable binding orientation, characterized by specific contacts between one and eleven residues of the peptide and the corresponding positions on chains A and B of TLR5. Eleven hydrogen bonds, supported by nonpolar interactions at the interface, contributed to overall stabilization (Fig. 5c and d; Supplementary Table S4b). These findings collectively point to the dual-receptor compatibility of the FMEC vaccine construct, suggesting its potential to engage in innate immune signaling through both TLR pathways. Analysis of FMEC-TLR4/5 interactions via molecular docking. (a, c) Representative poses from molecular docking simulations illustrating stable binding of FMEC to TLR4 (a) and TLR5 (c). (b, d) Two-dimensional interaction maps (Dimplot) detail residue-specific contacts between the designed FMEC and TLR4 (b) and TLR5 (d), highlighting key amino acids involved in complex stabilization, hydrogen bonding, and hydrophobic interactions. These interactions suggest that the designed FMEC has the capacity to engage TLR4 and TLR5, potentially initiating downstream immune signaling pathways. The corresponding lowest scores (-8.767 and − 1.868 kcal/mol for TLR5 and TLR4, respectively) and interacting residues are presented Independent docking of the YLKVYEHAV epitope of FMEC with the MHCI (HLA-A*0201)/CD8+ complex produced clear evidence of strong affinity. The most favorable cluster (cluster 0; 473 members; weighted score of -1485.5 kcal mol⁻¹) was examined via LigPlot + v.2.2.4, which revealed a dense network of interactions. Nine residues from the epitope engaged ten residues located within the α1 and α2 domains of MHCI, forming twelve hydrogen bonds and several hydrophobic contacts (Fig. 6a and b; Supplementary Table S5a). Visualization via PyMOL confirmed a compact interface extending across the central binding groove, which was consistent with stable anchoring of the peptide within the MHCI/CD8+ assembly. Docking analysis of the VQYGNATPA epitope of FMEC with the MHCII (HLA-DRB1*0101)/CD4+ complex yielded similarly robust results. The principal cluster (cluster 0; 525 members) displayed a minimum weighted score of -978.5 kcal mol⁻¹, accompanied by numerous hydrogen bonds and hydrophobic contacts distributed along the MHCII binding cleft. The overall topology suggested a securely accommodated peptide orientation, capable of efficient interaction with CD4+ T-cell receptors and subsequent immune activation. Intermolecular interactions within this complex were characterized via LigPlot + v.2.2.4, which demonstrated the involvement of 3 VQYGNATPA residues in interactions with 3 and 4 residues from the MHCII α and β chains, respectively. A total of ten hydrogen bonds, alongside extensive hydrophobic contacts, contributed to the stability of the peptide-MHC II interface. These data indicate a robust association between the VQYGNATPA peptide and the MHC II complex. The key interacting residues, hydrogen bond distances, and overall binding architecture are illustrated in Fig. 7a and b and detailed in Supplementary Table 5b. Normal mode analysis (NMA) was subsequently performed to investigate the intrinsic dynamics and structural stability of the FMEC vaccine both in its unbound state and when it was bound to TLR4 and TLR5 (Fig. 8). The standalone FMEC construct exhibited a compact and dynamically balanced architecture, defined by a relatively rigid core surrounded by flexible linkers and terminal loops. These flexible surfaces, corresponding mainly to inter-epitope connections, may facilitate receptor accommodation. In contrast, conserved epitope-bearing segments displayed low mobility, suggesting a stable scaffold that supports consistent antigen presentation. Covariance mapping further indicated coordinated motions among epitope-rich regions, which was consistent with domain-level coupling favorable for immune recognition. These flexible regions are balanced by more rigid domains, as reflected by the B factor and elastic network analysis, supporting well-organized structural integrity. Docking to TLR4 produced a pronounced stabilizing effect on the receptor ectodomain. The FMEC-TLR4 complex showed restricted fluctuations across the binding interface, particularly within regions involved in receptor dimerization and activation. B-factor and covariance analyses revealed correlated motions linking the vaccine contact zones with residues essential for downstream signaling. This dynamic pattern implies that binding not only stabilizes the receptor but also preserves the limited conformational adaptability required for signal initiation. Consistent with these structural observations, the PRODIGY server predicted a binding free energy of -19.6 kcal mol⁻¹ and an equilibrium dissociation constant (Kd) of 4.2 × 10⁻¹⁵ M at 25 °C. The interface comprises 205 intermolecular contacts, including numerous charged-charged (41) and charged-polar (45) interactions, indicating a highly complementary electrostatic network reinforced by hydrophobic contributions from apolar residues (~ 33.7% of the non-interacting surface). The FMEC-TLR5 complex displayed comparable stability, albeit with slightly greater global flexibility. NMA revealed reduced motion in the central leucine-rich repeat (LRR) domains upon ligand engagement, whereas peripheral loops retained modest flexibility. These concerted yet localized motions reflect a stable but dynamically responsive complex, which is consistent with the functional behavior of TLR5. Correspondingly, PRODIGY estimated a binding free energy of -18.3 kcal mol⁻¹ and a Kd of 4.1 × 10⁻¹⁴ M, indicating strong but marginally weaker affinity than that observed for TLR4. The interface contained 151 contacts, dominated by charged-apolar and apolar-polar interactions, with a balanced distribution of charged (26.3%) and apolar (33.5%) residues across the surface. The results of both analyses demonstrate that the designed FMEC vaccine forms dynamically stable and energetically favorable complexes with TLR4 and TLR5. The TLR4 complex exhibits increased stiffness and denser interfacial networking, which is consistent with stronger electrostatic coupling and reduced large-scale motion, whereas the TLR5 complex retains slightly greater plasticity within its signaling domain. Together, these features depict an FMEC construct capable of engaging innate immune receptors through structurally robust yet functionally adaptable interactions, a dynamic profile likely conducive to efficient immune activation (Fig. 8). Analysis of VQYGNATPA peptide-MHCII (HLA-DRB1*0101)/CD4+ interactions via molecular docking. (5a) Representative poses from molecular docking simulations illustrate stable binding of the VQYGNATPA peptide to MHCII (HLA-DRB1*0101)/CD4+. (5b) Dim plot detailing residue-specific contacts between the VQYGNATPA peptide and MHCII (HLA-DRB1*0101)/CD4+. The corresponding lowest score (-1485.5 kcal/mol) and interacting residues are presented Analysis of YLKVYEHAV peptide-MHCI (HLA*A0201)/CD8+ interactions via molecular docking. (6a) Representative poses from molecular docking simulations illustrate stable binding of the YLKVYEHAV peptide to MHCI (HLA*A0201)/CD8+. (6b) Dimplot detail residue-specific contacts between the YLKVYEHAV peptide and MHCI (HLA*A0201)/CD8+. The corresponding lowest score (-978.5 kcal/mol) and interacting residues are presented Normal mode analysis (NMA) of standalone FMEC (a) and in complex with TLR4 (b) and TLR5 (c). In the deformability plot, the ability of the FMEC molecule to deform at each residue is represented by the linker regions between the epitopes and nonstructural regions, and the rigid regions contain structural epitopes and the overall core of the protein. Mobility (B-factor), characterized by flexibility at the head and tail and relative stability in the middle of the structure, indicates the behavior of the FMEC vaccine in the free environment, which is in the form of a modular protein with surface mobile parts and is favorable for interaction with the immune system receptors TLR4 and TLR5. The presence of dominant collective motions with appropriate eigenvalues allows FMEC to undergo the gentle deformations necessary for receptor binding in the physiological environment without structural collapse. The eigenvalue plot associated with each normal mode represents the motion stiffness. Its value is directly related to the energy required to deform the structure. The lower the eigenvalue is, the easier the deformation. Furthermore, the variance plot for each normal mode is inversely related to the eigenvalue. The colored bars show the individual (red/purple) and cumulative (green) variances. The covariance map shows pairs of residues moving in the same direction (red), opposite directions (blue), or almost independently (white). Continuous red blocks on and around the major diameter indicate that parts of the vaccine (e.g., a domain or an epitope block) move together in a domain; if the epitopes are located in such correlated blocks, then this part can act as a structural–dynamic unit. The presence of large blue spots/blocks between two distant regions means that these two parts move together in opposite directions; it can be interpreted that the movement of one part may be compensated by bending or shifting another part, which contributes to the overall flexibility of the FMEC molecule. In addition, the motion correlation patterns indicate that epitope-containing segments move in coordinated bands and are likely to be exposed to the immune system. Finally, the elastic network (dark and light dots) shows which residues are mechanically connected by stronger “springs”; regions with dense, dark dots form the rigid, connected core of the FMEC vaccine. Therefore, according to the plots, the FMEC design is considered to be elastically balanced: a stable core and flexible margins Codon optimization, mRNA secondary structure, and in silico cloning Following codon optimization, employing the E. coli K12 parameters, the gene construct presented a CAI of 0.95 and a GC content of 51%. Assessment of the secondary structure of mRNAs via Mfold and Visual Gene Developer and the application of thermodynamic principles revealed an absence of hairpin or pseudoknot formations in the 5’ untranslated region. This structural characteristic is anticipated to promote efficient ribosomal engagement and subsequent translational efficacy. The predicted Gibbs free energy (ΔG) for the optimized mRNA was − 367.9 kcal/mol, which is indicative of a stable and energetically favorable conformation (Fig. 9a). Finally, in silico cloning into the pET28a vector, which was mediated by the NcoI and XhoI restriction enzymes, was digitally confirmed via SnapGene 6.0 (Fig. 9b). Analysis of the designed mRNA structure and cloning. (a) Predicted mRNA secondary structure of the designed construct. The arrowhead indicates the unobstructed 5’ end, with minimal secondary structure formation, resulting in a stable conformation (ΔG = -367.9 kcal/mol). (b) Schematic representation of the in silico cloning of the gene construct into the pET28a expression vector. The green region denotes the inserted gene construct within the vector backbone (black circle) Discussion Over the last few years, multiepitope vaccines have demonstrated substantial advantages over traditional vaccine platforms [49]. By incorporating epitopes with high immunogenicity and broad conservation, these vaccine candidates effectively address the limitations of conventional or serotype-specific vaccines, which often fail to elicit robust immune responses owing to the presence of weak antigens [50]. In previous studies, multiepitope vaccine candidates against different Salmonella serovars were prepared using Salmonella virulence factors alone or in combination formulations such as TolA [51], UPF0721 [52], Fimbriae [53], GroEL [54], DnaK [55], Curlin major subunit A (CsgA) [56], (OmpA, OmpD and enterotoxin) [57], (OmpC, OmpF and pgtE) [58], and secretion type IV systems (T4SSs), including VirB4, VirB5, VirB6 [59], (OmpC, OmpF, OmpA, and FliC) [60], and (OmpS1, OmpL, OmpF, PaL, and Hem) [61]. With respect to Shigella, most studies have used combinations of several virulence factors, including IpaB and IpaD (OmpA) [16] (IpaA, IpaB, IpaC, IpaD, OmpC, OmpF, and VirG) [62], (FepA and OmpW) [63], (IpaD, IpaB, VirG, GuaB, Stx1 and Stx2) [64], (OmpA, OmpD and Stn) [57], and (LptF and LolE) [65], to design multiepitope vaccines. These studies have mostly been designed as multiepitope vaccines against a single bacterium to increase coverage of different serotypes of that bacterium. However, to date, no study has designed a conserved multiepitope vaccine that provides dual protection against all serotypes of both bacteria via the use of different proteins from each bacterium. Notably, chimeric vaccines consisting of fragments of proteins from both bacteria have been designed, such as studies on chimeric vaccine candidates (rIpaB-T2544) [66], (IpaD-InvH) [67], and (SigA-Pic-Sap-GroEL) [68], but a multivalent multiepitope vaccine comprising conserved epitopes that target the core virulence factors shared among all serotypes in a candidate vaccine formulation has not been designed. Strategically chosen epitopes of diverse virulence factors of both bacteria enable the development of strong and durable immune memory against diverse stages of pathogen pathogenesis and hence improved protective efficacy. Moreover, the selection of both B-cell and T-cell epitopes and those eliciting cytokines such as IL-4 or IFN-γ expands the panel of immune activation, promoting complete humoral and cellular immunity. Importantly, the intentional exclusion of allergenic or toxin-associated regions during rationally designed vaccines prevents the possibility of undesirable autoimmune reactions or hypersensitivity, especially in subunit or chimeric vaccine candidates. Overall, these features make multiepitope vaccines promising and versatile tools for combating infectious disorders with improved protection, safety, and versatility compared with traditional approaches [69]. For the first time, in this study, we used six virulence factors of Salmonella and Shigella, which are involved in different stages of pathogenicity and could be more effective than the multiepitope vaccines designed against Salmonella or Shigella alone. We utilized MHC-binding predictive programs (e.g., NetMHC and IEDB tools) to ensure widespread HLA allele coverage of CTL and HTL responses, whereas B-cell epitopes were predicted via tools such as ElliPro and BepiPred to elicit neutralizing antibodies. Furthermore, we selected epitopes on the basis of their ability to pass through multiple immune filters, including a high antigenicity score, surface accessibility, and high conservation score across all Salmonella and Shigella strains, the highest coverage of human and murine alleles for both MHC classes, and a lack of toxicity and allergenic properties. Those epitopes that did not pass this screening process were not included. Various linkers with distinct roles in antigen processing were employed to connect the epitopes in the current FMEC construct. The solubility and flexibility of FMEC were enhanced by the incorporation of flexible linkers KK and GGGGS, which also ensured that the B-cell epitopes were accessible for antibody recognition. The GPGPG linker, which is frequently employed in the development of multiepitope vaccines, was used to attach helper T-cell epitopes. This linker is responsible for the separation of HTL epitopes, the maintenance of proper processing in the MHCII pathway, and the enhancement of the CD4+ T-cell response. Finally, the AAY linker, which is a known linker that improves proteasomal cleavage and the presentation of epitopes in the MHCI pathway and enhances the CTL response, was also used to link cytotoxic T-cell epitopes together [51, 70]. The protein backbone structure, which can also act as an agonist of the body’s receptors, has been employed in recent studies to establish stable secondary and tertiary protein structures and further stimulate the immune system. The designed multiepitope structure has been replaced with surface-available epitopes in the protein backbone. For example, in 2024 [71], the epitopes of the three virulence factors IpaB, IpaD, and OmpA in the backbone structure of the Salmonella enterica flagellum were used as TLR5 agonists to develop a multiepitope vaccine candidate encapsulated in alginate nanoparticles against Shigella soneii; alternatively, Li et al. in 2023 employed the IpaD protein backbone to design a multiepitope polyvalent protein vaccine candidate consisting of epitopes of the conserved antigens IpaB, IpaD, VirG, GuaB, and Shigella shiga toxin (Stx1 and Stx2), called MEFA, against Shigella flexneri, which are lethal pulmonary challenges [64]. In line with previous research, we employed this feature and the backbone structure of the flagellin of Salmonella typhimurium (FliC) as a mucosal adjuvant to increase the innate immune response in our new construct. The N- and C-termini of the flagellin protein contain conserved alpha-helical regions that can be recognized by TLR5 receptors in host cells. This recognition triggers innate immune responses and signaling cascades that are dependent on MID88, followed by the expression of intracellular proinflammatory cytokines. Furthermore, the middle portion of this protein has a highly variable conserved region, and the sequence of this region varies among Salmonella strains [17, 18]. Therefore, to improve the protective immune response in all Salmonella strains, the middle highly variable region was eliminated in this study and replaced with the sequence of FMEC designed. To achieve this goal, the flagellum structure was initially analyzed, and regions that were abundant in linear and conformational B-cell epitopes with antigenicity scores above 0.5 among the conserved alpha-helical regions of the flagellum structure were selected (residues 60–130 and 400–460 of FliC). These regions are conserved across all Salmonella serotypes and serve as binding sites for TLR5 (αND1a and αND1 domains) [18]. Finally, a FliC-based multiepitope construct (FMEC) was designed. As demonstrated in previous studies, the shortest form of FliC, a peptide known as pFlg (residues 85–111), contains key points (hot spots) that bind to TLR5 and can be expressed on the surface of liposomes. This peptide can be used as a liposome delivery system to deliver tumor antigens to dendritic cells [72]. This finding could support the accurate prediction and selection of the TLR5-binding region of FliC as a mucosal adjuvant. Then, the best FMEC’s 3D structure was modeled using the I-TASSER server with a C-score of -1.94, which was within the acceptable range of -5 to 2. This server is one of the best and most popular ones for designing three-dimensional protein structures. The I-TASSER server finds structural templates from the PDB’s multiple threading alignments and builds 3D structures by simulating repeated fragment assembly [73]. In numerous recent papers that referenced the I-TASSER server for protein 3D structure predictions [16, 74–76], the I-TASSER server was ranked as the top server for protein structure prediction in recent community-wide CASP7, CASP8, CASP9, CASP10, CASP11, CASP12, CASP13, and CASP14 experiments. Additionally, it was ranked as the best function prediction system in CASP9. Utilizing cutting-edge algorithms, the server is currently under active development with the objective of delivering the most accurate predictions of protein structure and function [73]. Nevertheless, any possible structural inaccuracies in this predictive model could have an impact on the downstream 3D analysis, such as binding affinity calculations for the TLR4 and TLR5 docking simulations. Therefore, we proceeded to optimize the hydrogen-bonding network, reduce the model’s atomic energy, and enhance the 3D structure through molecular dynamics simulation using the GalaxyRefine server for structural refinement. The ProSA Z score of -11.4 and an overall quality factor of 90.14% were achieved by the refined FMEC structure, which is in close agreement with the scores observed in native proteins of comparable size. Furthermore, molecular docking and NMA analyses validated the stable interactions of FMEC with TLR4/5 and indicated that the FliC-based multiepitope vaccine construct was dynamically optimized for epitope presentation and interaction with both innate immune receptors. The combination of a structurally stable core with controlled flexibility in loop and linker regions provides sufficient conformational adaptability to engage TLR4 and TLR5 without compromising the overall integrity of the macromolecule. Nevertheless, NMA remains an approximate, elastic-network–based description of intrinsic dynamics and does not explicitly capture solvent, membrane, or long-timescale effects, underscoring the need to complement these findings with atomistic molecular dynamics simulations in explicit solvents. Next, the physicochemical characteristics of the designed FMEC were assessed. This construct was expected to be neutral in nature, depending on the theoretical isoelectric point. A negative GRAVY value means that the protein is hydrophilic and may interact with water molecules. An instability index of 32.64 suggests that the protein is stable (II < 40). The aliphatic index of 61.66 implies favorable thermal stability, whereas a GRAVY score of -0.7 indicates hydrophilicity and good water solubility. The molecular weight of the FMEC construct was 33 kDa. Proteins with a molecular weight of less than 100 kDa are suitable for vaccine design because of their easy expression and purification steps [43]. Therefore, this designed protein is an acceptable vaccine candidate. Additionally, the results of the codon adaptation index (CAI) and GC content of 0.95 and 51%, respectively, revealed good efficiency of the final vaccine transcription and translation in the E. coli host. Therefore, the Gibbs free energy after sequence optimization for FMEC construct mRNAs was − 367.9 kcal/mol, indicating the lowest free energy and stable structure, and no unsuitable pseudoknot or loop at the 5’ end of the mRNA was observed. Furthermore, the immunological properties of FMEC designed via VaxiJen, ToxinPred, and AlgPred analyses indicated that the designed vaccine is a potent immunogen with an antigenic score of 1.07, which is nontoxic and nonallergenic. The C-ImmSim simulation results also demonstrated the capacity of FMEC to elicit both innate and adaptive (both cellular and humoral) immune responses. The population coverage analysis also demonstrated that 80.57 ± 12.15% of global populations can be covered by this FMEC vaccine construct. Despite successful bioinformatics results of FMEC construct design, in vitro and in vivo assays are essential for validating the protective efficacy of the FMEC vaccine candidate. For example, further confirmatory analyses should concentrate on expression of the FMEC construct in the E. coli BL21 expression host. Upon the purification of the recombinant FMEC protein, animal models (e.g., mice, rabbits, etc.) can be immunized with different doses of the recombinant FMEC via various injection routes (intranasal, oral, intramuscular, and intraperitoneal), along with different adjuvants (alum, Freund, various nanoparticles, etc.). Finally, FMEC’s efficacy, safety, and potential for protection can be evaluated by employing various bacterial challenge tests with both Salmonella and Shigella strains and analyzing the immune responses of the immunized animal by ELISA. Conclusion In summary, the purpose of this study was to design a protective multiepitope vaccine against both Salmonella and Shigella infections for the first time by integrating immunodominant B- and helper T-cell epitopes from six highly conserved and protective antigens: SpvB, SipD, RcK, SigA, IpaD, and FimA. Computational methods identify these epitopes, which are linked via suitable and optimized spacers to preserve structural and immunological functionality. A novel approach employed Salmonella FliC as both a structural scaffold and intrinsic adjuvant to increase immunogenicity. The engineered FMEC construct underwent rigorous evaluation, including analysis of physicochemical properties, antigenicity, and allergenicity. Three-dimensional structure modeling was performed via computational tools, followed by validation of structural integrity and binding affinity to TLR4/5 receptors. Finally, the design was further optimized for E. coli expression via codon adaptation and in silico cloning into the pET28a(+) vector. To move this candidate closer to preclinical development, experimental confirmation of the vaccine’s safety and effectiveness remains essential, even though computational analyses demonstrate promising antigenic and structural profiles of FMEC. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Acknowledgements The authors thank the Iran National Science Foundation (INSF) and Shahid Beheshti University for their funding (project No.4033138), assistance and cooperation with this project. Abbreviations FliC-based Multi-Epitope Construct Food-Borne Diseases Multi-Drug-Resistant Type III Secretion System Pathogen-Associated Molecular Pattern Pattern Recognition Receptors Toll-Like Receptors Flagellin from Salmonella typhimurium National Center for Biotechnology Information Allergen-Representative Peptides Immune Epitope Database Protein Data Bank Helper T-cell Cytotoxic T-cell Molecular Weight Theoretical Isoelectric Point Grand Average of Hydropathicity Support Vector Machine Java Codon Adaptation Tool Codon Adaptation Index Multiple Cloning Site Root-Mean-Square Deviation Author contributions Elnaz Afshari: Conceptualization, methodology, validation, formal analysis, investigation, writing-original draft preparation, writing-review and editing, visualization, project administration, Resources, Supervision. Neda Soleimani: Conceptualization, validation, review, project administration, Resources, Supervision. Funding This work is based upon research funded by Iran National Science Foundation (INSF) under project No.4033138. Data availability All data generated or analysed during this study are included in this published article and its supplementary information files. Declarations Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Competing interests The authors declare that they have no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Elnaz Afshari, Email: E_Afshari@sbu.ac.ir. Neda Soleimani, Email: N_Soleimani@sbu.ac.ir. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Data Availability Statement All data generated or analysed during this study are included in this published article and its supplementary information files.