Development of a multi-epitope vaccine against Helicobacter pylori using a novel saRNA technology through an immunoinformatics approach

Abstract

The development of a vaccine against Helicobacter pylori remains a public health priority due to its role in various gastric and non-gastric diseases, compounded by the rise of antibiotic resistance. To address the urgent need for novel interventions, we designed an in-silico self-amplifying RNA (saRNA)-based multi-epitope peptide vaccine targeting critical stages of H. pylori pathogenesis: gastric acid neutralization, bacterial adhesion, and toxin-mediated tissue damage. We employed a comprehensive computational and immunoinformatics pipeline and tools such as Geptop 2.0, VaxiJen, PSORTb v3.0.2, VirulentPred 2.0, BLASTp, TMHMM, and Expasy, IEDB’s NetMHCpan, NetMHCIIpan, AutoDock Vina, GROMACS v2024, PSIPRED, I-TASSER, PROSA-web, ERRAT, ClusPro 2.0, iMODS to identify, prioritize and validate the designed multi-epitope peptide as an effective vaccine candidate. Five essential proteins in different stages (UreB, BabA, HpaA, CagA, and VacA) were selected based on their antigenicity, virulence, conserveness, human non-homologous, and transmembrane helix presence. Helper T-cell (HTL) and cytotoxic T-cell (CTL) epitopes were predicted to ensure broader HLA allele coverage. Predicted epitopes were assessed for immunogenicity, allergenicity, toxicity, and cytokine-inducing potential. The top 10 HTL and 10 CTL epitopes were incorporated into the multi-epitope Hp vaccine which showed strong binding affinities to MHC molecules and stable peptide-MHC interactions by in molecular docking and dynamics simulations. The designed Hp vaccine sequence was incorporated in standard saRNA model where second ORF (ORF-2) encoded the target vaccine peptide. Structural analyses of the translated antigen showed high structural reliability. Molecular docking and dynamic simulation of the Hp vaccine with Toll-like receptor 4 (TLR4) confirmed stable interactions, suggested effective innate immune activation. Finally, population coverage, discontinuous B-cell epitopes and post-translational modifications analysis confirmed immunogenic potential. Despite these promising in-silico findings, challenges remain in translating computational predictions into experimental efficacy, particularly regarding RNA stability, vaccine delivery, and immune response durability. Therefore, further in vitro and in vivo validation is warranted to confirm its efficacy and safety.

Introduction

Helicobacter pylori is one of the most common bacterial infections worldwide, affecting around 50% of the global population; in developing countries, that number soars to an astonishing 85–95%, while in western countries, it is between 30 and 50%¹. Chronic H. pylori infection is a major risk factor for several gastrointestinal diseases, including gastric atrophy, chronic gastritis, gastric mucosa-associated lymphoid tissue (MALT) lymphoma, intestinal metaplasia, gastric cancer, duodenal and gastric ulcer² as well as in various non-gastric conditions, such as coronary artery disease (CAD), iron deficiency anemia, idiopathic thrombocytopenic purpura, and potentially non-alcoholic fatty liver disease^3–5.

Upon encountering the host, the successful establishment of H. pylori infection and the subsequent disease outcomes are primarily affected by the interaction between the physiological condition of host and the bacterial virulence factors⁶. The bacteria sense the acidic environment through acid-responsive sensors, ArsRS or FlgS, activating the urease gene cluster (ureA, ureB, ureI, ureE-H, flgS) to deploy an acid acclimation mechanism to survive the hostile gut environment by regulating urease activity⁷ which may also help evade macrophage destruction by altering phagosome pH and promoting megasome formation⁸. Both intra-bacterial and extracellular urease activity protects the bacteria by rapid neutralization of both the protons entering the periplasm and the acidic micro-environment close to the bacteria, respectively^9,10. Flagellar motion then enables the bacteria to move towards a more neutral environment¹¹. To colonize the gastric epithelium, the bacteria adhere to the mucosal lining and attach to host cell receptors using various adhesins (including AlpA, AlpB, BabA, BabB, SabA, HopZ, HpaA, LabA, OipA, and the DNA starvation/stationary phase protection protein), helping the bacteria resist removal by stomach peristalsis and gastric emptying¹². After colonization, H. pylori unleash a slew of virulence molecules that disrupt host cell signaling and cause tissue damage¹³. The cytotoxin-associated gene A (CagA) of the Cag pathogenicity island (CagPAI) is a key player, injected into host cells via a type IV secretion system (T4SS)¹⁴. Once inside, CagA undergoes tyrosine phosphorylation, activating signaling pathways like MAPK and PI3K¹⁵, leading to morphological changes, disrupted cell polarity, and potential carcinogenesis¹⁶. Another critical virulence factor, Vaculating cytotoxin A (VacA), compromises the stomach epithelial barrier by forming channels in the cell membrane, increasing permeability¹⁷. VacA also regulates apoptotic pathways, promotes vacuole formation, and disrupts immune responses by altering T cell and other immune cell activities¹⁸. There is a strong association of gastric disease and the presence of CagPAI containing CagA gene¹⁹ and s1m1-genotype VacA²⁰ in the bacterial genome.

The standard treatment options of H. pylori typically involve antibiotics, with triple therapy including a proton pump inhibitor (PPI) and two antibiotics (clarithromycin and amoxicillin or metronidazole), and quadruple therapy comprising a PPI, bismuth, metronidazole, and tetracycline²¹. Unfortunately, the rising prevalence of antibiotic resistance has led to an increase in the failure rate of H. pylori eradication²² . As a result, developing a safe and effective vaccination has become a crucial approach for preventing H. pylori infection²³.

In the past, several vaccines have demonstrated strong immune protection in animal testing^24,25, but few have demonstrated good immune protection in clinical trials^26,27. A recombinant subunit vaccine using the Urease subunit B (UreB) is the only H. pylori vaccine to successfully complete a phase 3 clinical trial, initially showing 71.8% efficacy in preventing infection one-year post-vaccination²⁶. However, after two years, efficacy dropped to 55%, leading to the discontinuation of its development. Another recombinant vaccine containing VacA, CagA, and neutrophil-activating protein (NAP) failed to provide additional protection compared to placebo after being challenged with a CagA-positive strain during a phase 2 clinical trial²⁷.

More recently, bioinformatics has advanced significantly, offering a multitude of tools for the time-effective, accurate, economical and successful discovery and production of new vaccine candidates²⁸. Many in-silico studies have explored various vaccine strategies, including oral vaccines²⁹ and peptide-based vaccines³⁰ against H pylori, but following decades of research and development, no H. pylori vaccine has been on the market yet.

To address the challenges faced by those conventional vaccines, self-amplifying RNA (saRNA) vaccines present a promising approach. saRNA vaccines amplify themselves within host cells, leading to prolonged antigen expression and resulting in stronger and more durable immune responses even at a lower dose³¹. Moreover, saRNA vaccines can induce both humoral and cellular immune responses, a dual activation crucial for effective protection against H. pylori^32,33. The self-replicating RNA (srRNA) platform facilitates swift vaccine design and manufacturing, and its non-integrative nature ensures high safety³⁴. Additionally, advanced formulation techniques including 5’ cap modification, alteration of 5’and 3’ UTRs, codon optimization, adjustment of poly(A) tail length can improve the manufacturing efficiency, stability, translation, and biodistribution of saRNA vaccines³⁴.

In this study, we aimed to design a next generation saRNA vaccine against H. pylori, using a combination of bioinformatics tools and published experimental data. Our approach involved selecting and evaluating five potential vaccine candidates of H. pylori proteins associated with three critical stages of pathogenesis, developing a consensus sequence from diverse protein sequences covering different regions of the world, and predicting T-cell and B-cell epitopes.

Methodology

Selection and evaluation of Helicobacter pylori proteins as potential targets

Our strategy for developing an in-silico multi-epitope peptide vaccine against H. pylori, we considered potential vaccine candidates to block three critical stages of pathogenesis: (1) survival in acidic stomach conditions, (2) adhesion to cellular surface receptors via adhesins, and (3) toxin-mediated host tissue damage^{10,12,35–40}. To identify relevant target proteins with available RefSeq protein sequences, we conducted an extensive literature review in pubmed, google scholar databases using search terms such as “Helicobacter pylori” OR “Pathogenesis” OR “Virulence factors” OR "Acid neutralizing proteins" OR “Adhesins” OR "Toxin-associated proteins". These proteins underwent further screening based on parameters including essentiality (Geptop 2.0; cutoff score ≥ 0.24), subcellular localization (PSORTb v3.0.2 online server), virulence (virulentpred 2.0), antigenicity (Vaxijen v2.0; threshold of ≥ 0.4), nonhuman homology (BLASTp; exclusion of proteins with ≥ 35% similarity), TM helixes (TMHMM; ≤ 1 transmembrane helix), and molecular weight (Expasy tool; < 140 kDa).

Data retrieval and consensus sequence generation

From this pool, we selected 5 proteins as primary targets for vaccine design: UreB, BabA, HpaA, CagA & VacA. Numerous independent studies, including phase 3 clinical trials, phase 1/2 trials, and in vivo research, have shown these proteins to be effective in managing the H. pylori infection^25–27,41.

We used the H. pylori s1m1 strain NCTC 11637 as our reference, sourcing its whole genome sequence from NCBI (Refseq ID: GCF_900478295.1). From this genome, we extracted the amino acid sequences for specific proteins using the following accession IDs: WP_000724287.1 (UreB), WP_108169269.1 (BabA), WP_108169844.1 (HpaA), WP_108169893.1 (CagA), and WP_108169039.1 (VacA). Additionally, we gathered a total of 120 protein sequences (24 sequences per protein) from the Uniprot (reviewed), NCBI (refseq), and ViPR (refseq) databases, which included diverse H. pylori pathotypes from different WHO geographic regions including Western Pacific (Australia, China, South Korea, Vietnam), Americas (Colombia, USA, Argentina, Brazil), Europe (UK, Germany, Belarus, Belgium), South-East Asia (India, Bangladesh, Thailand, Indonesia), Eastern Mediterranean (Iran, Pakistan, Oman, Kuwait), and Africa (Angola, Cameroon, South Africa, Nigeria). This allowed us to analyze the conservation of each protein and develop a consensus sequence. The sequence alignment was performed using CLC Main Workbench v8 software (QIAGEN Bioinformatics).

Cytokine inducing T_H-cell epitope prediction

Primarily, CD4 + T-helper (T_H) cells mount an immunological response against H. pylori by identifying antigens presented by MHC class II molecules on antigen-presenting cells^42,43. In order to predict Helper T-cell lymphocytes for this investigation, we employed the 7-allele reference set, as described on the IEDB server. This set includes high-frequency alleles identified by Paul et al., namely HLA-DRB1*03:01, HLA-DRB1*07:01, HLA-DRB1*15:01, HLA-DRB3*01:01, HLA-DRB3*02:02, HLA-DRB4*01:01, and HLA-DRB5*01:01⁴⁴. We employed the IEDB’s NetMHCIIpan (EL) 4.1 v 2023.09 algorithm, and set the peptide length for screening at 15-mers. By considering extra elements involved in the normal processing and presentation of peptides, epitope likelihood (EL) predictions offer a sophisticated and all-encompassing approach to finding T cell epitopes⁴⁵. Epitopes with a percentile rank higher than 10% were considered weak binders and were excluded from further analysis.

The IFNγ induction potential of T_H epitopes were assessed using the IFNepitope tool, which employs a hybrid approach combining Motif and SVM techniques to accurately identify IFNγ-inducing peptides⁴⁶. Similarly, the IL-10Pred webserver (https://webs.iiitd.edu.in/raghava/il10pred/) was used to predict interleukin-10 (IL-10) inducing MHC II binders, with the threshold set to -0.3 and utilizing the SVM algorithm.

Immunogenic T_C-cell epitope prediction

Finding CD8 + cytotoxic T-cell (CTL) epitopes, in addition to CD4 + helper T-cell epitopes, is essential for a strong immune response against H. pylori^47,48. In order to achieve maximal population coverage, we concentrated on high-frequency HLA class I alleles present in at least 1% of the human population, or with an allele frequency of 1% or higher. Specifically, we selected the 7-allele reference set, which is designed to maximize the representation of HLA alleles across diverse populations. Those include HLA-A*01:01, HLA-A*02:01, HLA-A*02:03, HLA-A*03:01, HLA-A*11:01, HLA-A*23:01, HLA-A*24:02, HLA-A*26:01, HLA-A*30:01, HLA-A*31:01, HLA-A*32:01, HLA-A*33:01, HLA-A*68:01, HLA-A*68:02, HLA-B*07:02, HLA-B*08:01, HLA-B*15:01, HLA-B*35:01, HLA-B*40:01, HLA-B*44:02, HLA-B*44:03, HLA-B*51:01, HLA-B*53:01, HLA-B*57:01, and HLA-B*58:01 (http://tools.iedb.org/mhci/). We employed the IEDB’s NetMHCpan (EL) 4.1 v 2023.09 algorithm to forecast peptides with lengths of nine and ten-mers. Similar to the MHC class II selection criteria, epitopes with a percentile rank higher than 10% were considered weak binders and were excluded from further analysis.

Consequently, their immunogenicity was assessed using a Class I Immunogenicity prediction tool (http://tools.iedb.org/immunogenicity/).

Autoimmunity, antigenicity, allergenicity, and toxicity analysis of predicted epitopes

To ensure the safety and efficacy of the predicted epitopes, a comprehensive analysis was conducted. First, all predicted peptides were checked against the Homo sapiens protein database using the BLASTp algorithm (https://blast.ncbi.nlm.nih.gov/Blast.cgi?PAGE=Proteins) to minimize the risk of autoimmunity. Peptides with an E-value greater than 0.05 were considered non-homologous and thus less likely to cause autoimmune responses. For antigenicity assessment, the VaxiJen server (http://www.ddg-pharmfac.net/Vaxijen/VaxiJen/VaxiJen.html) was utilized with bacterial parameters and a threshold set at 0.4. This helped identify peptides likely to induce an immune response. Allergenicity was evaluated using the AllerTop v2.0 server (http://www.ddg-pharmfac.net/AllerTOP), with all settings kept at their default values to predict potential allergenic properties. Finally, the toxicity of the epitopes was assessed using the ToxinPred server (https://webs.iiitd.edu.in/raghava/toxinpred/multi_submit.php). This tool generated potential mutants of the peptides to evaluate their toxicity with default parameters.

Molecular docking analysis of T-cell epitopes with HLA alleles

Molecular docking simulations was performed to evaluate the binding affinity of T-cell epitopes to their corresponding MHC alleles. We selected HLA-DRB101:01, a class II allele associated with susceptibility and present in 95% of the population⁴⁹, and HLA-A02:01, one of the most common class I alleles worldwide⁵⁰. Using AutoDock Vina for the docking process, we obtained the 3D structures of these MHC alleles from the RCSB PDB database. The selected epitopes were folded into three-dimensional structures using the PEP-FOLD 3.5 server. MGL tools were used to purify and minimize the energy of MHC molecules prior to docking. The docking poses and interactions between the epitopes and MHC alleles were analyzed using ChimeraX, providing insights into their binding affinities and potential immunogenicity. To further analyze the stability of peptide-MHC complexes, molecular dynamics simulations (MDS) were performed using GROMACS v2024. The CHARMM36 force field was applied, and the system was solvated in a TIP3P water model within a cubic simulation box. Counterions (Na + /Cl-) were added to neutralize the system. Energy minimization was conducted using the steepest descent algorithm, followed by NVT (constant volume and temperature) and NPT (constant pressure and temperature) equilibration at 300 K and 1 atm pressure. A 100 ns simulation was carried out for the peptide-MHC complex. The stability of the complex was assessed by analyzing root-mean-square deviation (RMSD), root-mean-square fluctuations (RMSF), hydrogen bonding patterns, and the total energy of the bonds.

To validate the docking simulations, two control peptides, which are known to bind with high affinity to the HLA-A02:01 and DRB-104:01 alleles, were included in the analysis. The peptide TSKGLFRAAVPSGAS (alpha-enolase peptide 26–40) was used as a positive control for MHC-I binding⁵¹, while NLVPMVATV (derived from cytomegalovirus protein pp65) was used for MHC-II binding⁵². Both peptides are well-established as strong binders to their respective HLA alleles.

To evaluate the population coverage for the chosen T cell epitopes, we employed the Population Coverage tool available in the IEDB database (http://tools.iedb.org/population/). Following the analysis, ArcGIS was employed to visually map the geographical distribution of the population coverage.

Designing a saRNA-Hp vaccine

The sequential arrangement of the self-amplifying RNA-Hp vaccine construct from the 5’ cap to the poly(A) tail is depicted as Fig. 1:

Fig. 1.

Schematic of the saRNA-based vaccine design against H. pylori. (A) The structural diagram of the saRNA vaccine construct from the 5′ cap to the poly-A tail. (B) Translation of the construct into nsP1-4 proteins forms an RdRP complex that amplifies both genomic and subgenomic RNA, leading to the accumulation of the antigen (Hp Vaccine) within the cell (visualized using ChimeraX version 1.8rc202406072045).

The saRNA vaccine construct has two open reading frames (ORFs) linked by a sub genomic promoter (SGP) from the Venezuelan equine encephalitis virus. The first ORF encodes proteins for the RNA dependent RNA polymerase (RdRp) complex (NCBI Gene ID: 2652925; Location: 1–7526), and the SGP guarantees robust expression of downstream elements, which is the second ORF that encodes the desired vaccine⁵³. The design begins with a 5’m₂^7,2′−OGpp_SpG cap analog to enhance RNA stability and translation efficiency⁵⁴. Following the cap, the 5’ untranslated region (5' UTR) consists of human β-globin (Gene ID: 3043) sourced from NCBI, coupled with a Kozak sequence (GCCRCCATGG) to further optimize translation initiation^55,56. The inclusion of a tissue plasminogen activator (tPA) (UniProt ID: P0 process 0750) signal peptide directs the nascent protein to the secretory pathway⁵⁷, connected to nsP1-4 genes via a flexible GGGS linker (Gly-Gly-Gly-Ser) to ensure proper separation and function⁵⁸. An EAAAK linker (Glu-Ala-Ala-Ala-Lys) provides structural rigidity⁵⁹ and separates the replicase genes from the adjuvant (50S ribosomal protein L7/L12). GPGPG and AAY linkers were used to combine intra- HTL and CTL epitopes. Another GGGS linker separates the CTL epitopes from the MITD sequence (UniProt ID: Q8WV92), which aids in intracellular trafficking of CTL epitopes toward the MHC-I compartment of the endoplasmic reticulum⁶⁰. The sequence terminates with a stop codon TGA, followed by a 3’ UTR (NCBI Gene ID: 3039) to enhance mRNA stability and translation, and a poly(A) tail of approximately 120 adenine nucleotides for further stabilization.

Codon optimization and in-silico cloning of saRNA-Hp vaccine

To ensure effective expression of the saRNA vaccine in the host organism (Homo sapiens), we employed the Vectorbuilder Online server (https://en.vectorbuilder.com/tool/codon-optimization.html) for codon optimization of both ORF-1 and ORF-2. This optimization process enhances the DNA sequence by improving the codon adaptation index (CAI) and CG content. The VectorBuilder algorithm, in line with established codon optimization principles, targets a CAI value close to 1.0, indicating optimal adaptation to human codon usage bias⁶¹. In addition, the GC content range of 40–60% was chosen to ensure optimal mRNA stability and prevent the formation of stable secondary structures that could impede translation. The amino acid sequences were converted into a nucleotide sequence using the EMBOSS Backtranseq web server (https://www.ebi.ac.uk/Tools/st/emboss_backtranseq/), providing a streamlined and efficient workflow for generating a DNA sequence optimized for host expression. NheI and bstB17I restriction sites were inserted to the construct’s N- and C-terminals, respectively, after it had been reassembled and the saRNA-Hp vaccine DNA sequence was retrieved. These sites flank the desired insertion region within the vector, facilitating directional cloning. Both NheI and BstB17I are unique within the vector, ensuring that the insert is precisely inserted at a specific location, avoiding unintended cuts elsewhere in the vector backbone. The construct was computationally cloned into the pcDNA3 vector using SnapGene, a widely used in-silico tool for molecular cloning simulations. To assess the successful integration of the insert, a virtual agarose gel electrophoresis experiment was performed using a 1.2% agarose gel simulation, enabling visualization of the expected fragment sizes. The migration of the DNA fragments was analyzed based on their expected molecular weight. The band patterns were compared with the molecular weight markers to confirm the size of each fragment and the intensity of the bands was considered as an indicator of the fragment concentration.

Modelling of sgRNA and translated peptide structures

The sub-genomic mRNA (ORF-2) secondary structure was predicted using the RNAfold (http://rna.tbi.univie.ac.at/cgi-bin/RNAWebSuite/RNAfold.cgi) web server. This site estimates the thermodynamic minimum free energy (MFE) that the query RNA structures will produce. The alpha helix, beta-turn, random coil, and extended strand forms were examined by predicting the secondary structure of the translated antigenic peptide using the PSIPRED Workbench (PSIPRED 4.0) tool. The I-TASSER server was utilized to produce many 3D models of the vaccine. This server generates tens of thousands of conformations based on pair-wise structural similarity and reports up to five models that match to the five largest structure clusters. By refining the loop or terminal area in the 3D structure, the protein’s tertiary structure was further optimized through ab initio computation using the Galaxy-Refine modelling (https://galaxy.seoklab.org/cgi-bin/submit.cgi?type=REFINE).

Characterization and validation of the designed model

For validation of the improved tertiary structure from Galaxy refine server, we used two open-source online tools, PROSA-web for summarizing error scores and PROCHECK (UCLA-DOE LAB) for analyzing the Ramachandran plot. Generally, the evaluation of the translated vaccine centered on the ORF-2 peptide form, emphasizing adjuvants, selected epitopes, and their associated linkers. Some sequences were deliberately excluded from the analysis: Early protein translation and processing cleaves ORF-1(ns1-4), while the secretory and MHC-I pathways cleave tPA and MITD, respectively^62,63 (Ref.). Physicochemical properties, including molecular weight, aliphatic index (AI), instability index (II), solubility, amino acid composition, theoretical isoelectric point (pI), and grand average of hydropathicity (GRAVY), were assessed using the ExPASy-ProtParam tool and SOLpro server. Subsequent evaluations focused on biological (toxicity) and immunological (antigenicity, allergenicity) properties using the ToxinPred 2.0, VaxiJen 2.0, and AllerTOP 2.0 servers, respectively.

3D structure-based B-cell epitopes analysis

The humoral immune response against H. pylori has been shown to lower bacterial loads in colonized C57BL/6 mice⁶⁴. To identify potential discontinuous B-cell epitopes, we used the ElliPro server, which predicts epitopes based on 3D structural features such as protrusion, solvent accessibility, and flexibility. The protein structure, provided in PDB format, was analyzed using default parameters. ElliPro calculates the Protrusion Index (PI) for each residue based on its position relative to the largest ellipsoid encompassing the protein, with higher PI values indicating greater solvent accessibility. Discontinuous epitopes were identified and clustered based on the distance R between residue centers of mass. Larger R values indicate the prediction of larger, more discontinuous epitopes.

Post‐translational modification analysis

To assess the post-translational modifications (PTMs) of the designed vaccine construct, various analyses were performed. Glycosylation, phosphorylation, and acetylation were predicted using the NetNGlyc 1.0, NetPhos 3.0, and NetAct 1.0 tools, respectively, accessible at http://www.cbs.dtu.dk/services/. Additionally, lipid modifications, including N-terminal glycine myristoylation and GPI-anchor attachment, were evaluated using the MyrPS/NMT server (https://mendel.imp.ac.at/myristate/) and the big-PI/GPI animals server (https://mendel.imp.ac.at/gpi/gpi_server.html).

Molecular docking of TLR-4 and vaccine peptide

Toll-like receptors (TLRs) play a crucial role in host innate and adaptive immune responses to microbial pathogens and their products⁶⁵. The ClusPro 2.0 server (https://cluspro.bu.edu/) was utilized for molecular docking between the designed vaccine peptide and human Toll-like receptor 4 (hTLR-4) (PDB ID: 4G8A). The hTLR-4 structure was prepared by removing heteroatoms, water molecules, and ligands, followed by energy minimization using BIOVIA Discovery Studio. Docking calculations involved uploading the prepared hTLR-4 as the receptor and the potential vaccine candidates as the ligand with default settings. The top ten docked complexes were selected based on clustering properties and binding energies. PDBsum was used to visualize the bonds formed between the vaccine construct residues and TLR4.

Molecular dynamic simulation

The molecular dynamic simulation of the HpVac (proposed H. pylori Vaccine) and TLR-4 docked complex was performed using the iMODS tool to assess the stability and dynamic behavior of the interaction. The iMODS tool utilizes normal mode analysis (NMA) in internal (dihedral) coordinates to predict the collective motions of proteins. This approach provides insights into the conformational flexibility and stability of the docked complex. Key parameters calculated by iMODS include deformability, which indicates the potential for structural change at each residue; eigenvalues, which represent the stiffness of the motion; and variance, which measures the extent of fluctuation. Additionally, the covariance map visualizes the correlated movements between residue pairs, the B-factor reflects the atomic displacement parameters, and the elastic network models the interactions contributing to the overall flexibility of the complex. These analyses collectively enhance our understanding of the molecular dynamics and stability of the vaccine-receptor complex.

Result

Evaluation and selection of candidate proteins

Based on our literature review, we identified 21 H. pylori proteins which are associated with critical pathogenic stages as our preliminary candidates for vaccine development (Table 1). These proteins, involved in bacterial survival in the acidic environment, cellular adhesion, and toxin-mediated tissue damage, were carefully curated to target essential processes in H. pylori pathogenesis.

Table 1.

Potential protein targets for Helicobacter pylori and their characteristics.

Phases	Proteins	Essen	Accession	SL	Virulence	Ag	HH	TH	M.W
1	UreA	Y	WP_000779233.1	C	Virulent	N	N	0	26,567.58
	UreB	Y	WP_000724308.1	C	non	Y	N	0	61,713.72
	UreE	Y	WP_000583090.1	C	Virulent	N	N	0	19,440.59
	UreH	N	BDO44070.1	C	non	N	N	0	29,505.97
	UreG	N	WP_000238762.1	C	non	N	N	0	21,955.32
	UreI	Y	WP_000901274.1	CM	Virulent	N	N	6	21,691.53
	FlgS	N	WP_077645049.1	CM	Virulent	N	N	0	44,238.74
2	SabA	N	WP_209611253.1	OM	Virulent	Y	N	0	80,587.91
	HopZ	N	WP_000842054.1	OM	Virulent	Y	N	0	74,269.9
	AlpA	Y	WP_212872089.1	OM	Virulent	Y	N	0	56,636.89
	AlpB	Y	WP_000812509.1	OM	Virulent	Y	N	0	56,984.62
	BabA	Y	WP_187888179.1	OM	Virulent	Y	N	0	82,279.73
	LabA	Y	WP_033587791.1	OM	Virulent	Y	N	0	78,628.67
	DNA starvation/ stationary phase protection protein	N	WP_000846487.1	C	Virulent	Y	N	0	16,877.33
	BabB	Y	WP_172417390.1	OM	Virulent	Y	N	0	76,083.54
	OipA	Y	WP_000709684.1	OM	Virulent	Y	N	0	34,281.84
	HpaA	Y	WP_000855929.1	OM	Virulent	Y	N	0	29,137.48
3	CagA	Y	WP_209612238.1	EC	Virulent	Y	N	0	131,860.72
	CagL	Y	WP_209612230.1	C	Virulent	N	N	0	26,847.06
	CagY	Y	WP_181229842.1	C	Virulent	Y	N	1	204,746.96
	VacA	N	WP_209611245.1	EC	Virulent	Y	N	1	139,270

Essen. = Essentiality, SL = Subcellular localization, Ag = Antigenicity, HH = Human homology, TH = Transmembrane helix, M.W. = Molecular Weight, Y = Yes, N = No, C = Cytoplasmic, CM = Cytoplasmic membrane, OM = Outer membrane, EC = Extracellular.

Following computational analysis and referencing published experimental data, five proteins—UreB, BabA, HpaA, VacA, and CagA—were prioritized for their proven immunogenic potential and critical roles in pathogenesis. Firstly, we analyzed seven proteins, dedicated to acid neutralization, and determined that UreA, UreB, and UreE were essential, but UreB was only antigenic protein and recognized as particularly critical for the bacterium’s viability in acidic environments. Secondly, we assessed proteins linked to bacterial adhesion to the gastric epithelium in order to initiate another critical stage of pathogenesis. A total of seven, out of 10 outer membrane (OM) proteins—AlpA, AlpB, BabA, BabB, HpaA, LabA, and OipA—were identified as virulent, antigenic, and essential for bacterial survival. However further scrutiny of the available experimental data revealed consistent evidence regarding their pathogenic roles in adhesion and colonization, antigenic properties, well-established immunogenicity, and expression variability, leading to the selection of BabA and HpaA as the most promising candidates. Finaly, we focused on proteins associated with toxin-mediated tissue damage. Here, both CagA and VacA were distinguished as extracellular proteins with notable antigenicity and pathogenic potential, reinforcing their candidacy for vaccine formulation. Importantly, none of these selected proteins share homology with human proteins, minimizing the risk of autoimmunity. Moreover, the TMHMM-predicted protein topology prediction showed that only VacA has a single transmembrane helix that complies with permissible bounds.

Generation of consensus sequences

A conserved sequence analysis of five selected proteins were performed. A total of 120 protein sequences (24 sequences per protein) derived from different H. pylori strains across various geographic regions, including the Western Pacific, Americas, Europe, South-East Asia, Eastern Mediterranean, and Africa (Supplementary Table S1).

The analysis identified regions of high conservation within these proteins. The resulting consensus sequence (≥ 90% of conservation) and its specific location is given in Supplementary Table 2. Figure 2 depicts the conservation percentage for each amino acid position within the proteins, where conservation levels are illustrated as columns.

Fig. 2.

Conserved regions of selected proteins generated using CLC Main workbench where A indicates conserved regions for UreB, similarly B for BabA, C for HpaA, D for CagA, E for VacA. The red box highlights mutation site.

T-cell epitope prediction

Based on their lowest percentile scores, we evaluated the top 10 MHC-II epitopes for each target protein, generated by the IEBD server. In addition, toxicity, allergenicity, antigenicity, and the capacity to produce IL-10 and IFN-γ cytokines were evaluated. For UreB, seven antigenic epitopes were identified. Out of them, five were predicted to induce IFN-γ and were both non-allergenic and non-toxic. Of these, MGIFSITSSDSQAMG and IFSITSSDSQAMGRV are particularly notable for being highly antigenic peptides with antigenicity score 0.8163 and 0.6482 respectively (Table 2). For BabA, we found only two antigenic epitopes, VWTYGFGADALYNFI and WTYGFGADALYNFIN, which are safe and can induce IFN-γ. All epitopes of HpaA were antigenic, but only two—KGTDNSNDAIKSALN and LSELDIQEKFLKTTH—were non-allergenic, non-toxic, and capable of inducing both IFN-γ and IL-10. In case of CagA, eight antigenic epitopes were identified where five were non-allergenic and safe. Only two epitopes, YDKIGFNQKNMKDYS and DKIGFNQKNMKDYSD, were predicted to induce both IFN-γ and IL-10. Furthermore, among nine antigenic epitopes of VacA, two—RNALVLKPSVGVSYN and FFRNALVLKPSVGVS—were non-allergenic, safe, and capable of inducing IFN-γ, though they did not induce IL-10.

Table 2.

Characteristics of top-ranked linear MHC-II epitopes for each protein predicted using the IEDB server.

Protein	Peptide	PR	Ag	Al	Tx	IFN-Y	IL-10
UreB	IFSITSSDSQAMGRV	3.4	0.8163	N	N	P	N
	MGIFSITSSDSQAMG	5.3	0.6482	N	N	P	N
BabA	WTYGFGADALYNFIN	2.8	0.5096	N	N	P	N
	VWTYGFGADALYNFI	4	0.8222	N	N	P	N
HpaA	KGTDNSNDAIKSALN	8.4	0.7378	N	N	P	P
	LSELDIQEKFLKTTH	8.9	0.6876	N	N	P	P
CagA	YDKIGFNQKNMKDYS	2.7	0.4752	N	N	P	P
	DKIGFNQKNMKDYSD	4.9	0.4425	N	N	P	P
VacA	RNALVLKPSVGVSYN	1.2	1.003	N	N	P	N
	FFRNALVLKPSVGVS	3.3	0.7343	N	N	P	N

PR = Percentile rank, Ag = Antigenicity score, Al = Allergenicity, Tx = Toxicity, INF-Y = Interferon gamma, IL-10 = Interleukin 10.

Subsequently, MHC-I epitopes were predicted for each protein, and the top 10 epitopes were selected based on their lowest percentile ranks for further evaluation, including assessments of antigenicity, allergenicity, toxicity, and immunogenicity. Among the top 10 UreB epitopes, only two (RVGEVITRTW, TSSDSQAMGR) were identified as antigenic (with scores ranging from 0.5366 to 1.4911), non-allergenic, safe, and immunogenic (Table 3). Similarly, epitope analysis of BabA and CagA revealed two antigenic peptides for each protein that were also safe, non-allergenic, and immunogenic. For the HpaA protein, six antigenic peptides were identified, five of which were non-allergenic, but only two (FLKTTHSSH, SELDIQEKF) were non-toxic and immunogenic. All VacA epitopes were predicted to be highly antigenic (with scores ranging from 0.7194 to 1.2124). However, half of them were excluded due to their predicted allergenicity. Of the remaining epitopes, only SYNHLGSTNF and ASYGYDFAF were both safe and immunogenic.

Table 3.

Characteristics of top-ranked linear MHC-I-1epitopes for each protein predicted using the IEDB server.

Protein	Peptide	PR	Ag	Al	Tx	Immuno
UreB	RVGEVITRTW	0.03	0.5366	N	N	0.3844
	TSSDSQAMGR	0.21	1.4911	Y	N	0.2569
BabA	NSASDVWTY	0.06	0.5199	N	N	0.1163
	YGFGADALY	0.18	0.7533	N	N	0.1576
HpaA	FLKTTHSSH	0.62	0.4372	N	N	0.1989
	SELDIQEKF	0.01	0.7679	N	Y	0.3968
CagA	KIGFNQKNMK	0.12	1.0979	N	N	0.1718
	GFNQKNMKDY	0.58	0.5331	N	N	0.0741
VacA	SYNHLGSTNF	0.05	0.7194	N	N	0.3013
	ASYGYDFAF	0.05	1.2124	N	N	0.6693

Here, PR = Percentile Rank, Ag = Antigenicity Score, Al = Allergenicity Immuno = Immunogenicity.

T-cell epitope docking with MHC-alleles and MDS analysis

Twenty characterized T-cell epitopes from the previous step were subjected to a molecular docking investigation using Autodock Vina to determine their binding affinity and pose with corresponding HLA alleles. Control peptides, serving as benchmarks, yielded binding affinities of -8.9 kcal/mol for MHC-I and − 7.5 kcal/mol for MHC-II. Twenty predicted T-cell epitopes were subsequently docked, demonstrating a range of binding affinities. For MHC-II epitopes docked with HLA-DRB101:01, scores ranged from − 6.9 to − 7.7 kcal/mol, closely aligning with the control. Notably, MHC-I epitopes docked with HLA-A02:01 exhibited a broader range, from − 7.4 to − 10.3 kcal/mol, indicating that some epitopes displayed superior predicted binding compared to the control. (Table 4). Among the MHC-I epitopes, ASYGYDFAF (VacA) showed highest binding affinity − 10.3 kcal/mol and molecular visualization revealed that the peptide skillfully fitted into the binding groove of the HLA-A*02:01allele and formed 3 hydrogen bonds between ASP77, ARG97, TRP147 residues of MHC-I and SER2, GLY4 and ALA4 residues of the peptide as well as 100 non-bonded contacts between the interacting atoms (Fig. 3A). Moreover, molecular dynamic simulation of 100 ns shed light into the atom’s movement and overall stability of the complex. The RMSD value indicates the complex remained stable throughout the simulation with occasional fluctuation of approx. 0.5 nm (Fig. 3B). We used the Ligand Root Mean Square Fluctuation (L-RMSF) to assess the peptide’s local adaptability. Increases in RMSF values suggest more positional flexibility in certain amino acids. ASYGYDFAF- HLA-A*02:01 showed high RMSF value ranges from 0.1 to 0.5 nm (Fig. 3D). During this 100 ns simulation, the ASYGYDFAF formed the maximum number of hydrogen bonds with HLA-A*02:01molecule (Fig. 3C). Moreover, the energy values range from approximately 5000 to 5750 kJ/mol suggesting the strength of the interactions, with lower values generally indicating more stable interactions (Fig. 3E).

Table 4.

Predicted HTL and CTL peptides with MHC-I and MHC-II binding affinities for Helicobacter pylori vaccine candidate proteins.

Prot	HTL peptides	MHC-II allele	Binding affinity (kcal/mol)	CTL peptides	MHC-I allele	Binding Affinity (kcal/mol)
UreB	IFSITSSDSQAMGRV	HLA-DRB1* 01:01	− 7.2	RVGEVITRTW	HLA-A*02:01	− 7.9
	MGIFSITSSDSQAMG		− 7.0	TSSDSQAMGR		− 7.4
BabA	WTYGFGADALYNFIN		− 7.6	NSASDVWTY		− 8.6
	VWTYGFGADALYNFI		− 7.7	YGFGADALY		− 9.6
HpaA	KGTDNSNDAIKSALN		− 7.1	FLKTTHSSH		− 9.9
	LSELDIQEKFLKTTH		− 6.9	SELDIQEKF		− 8.7
CagA	NQKNMKDYSDSFKFS		− 7.4	KIGFNQKNMK		− 7.4
	KIGFNQKNMKDYSDS		− 8.1	GFNQKNMKDY		− 7.9
VacA	RNALVLKPSVGVSYN		− 6.5	SYNHLGSTNF		− 9.4
	FFRNALVLKPSVGVS		− 7.3	ASYGYDFAF		− 10.3

Fig. 3.

Investigations of HLA and epitope interaction using molecular dynamics (MD) simulation. (A) Complex interaction between ASYGYDFAF (in green) and HLA-A*02:01 (in purple and golden), visualized using ChimeraX v1.8rc202406072045, with interaction analysis performed using PDBsum server. (B) the complex’s RMSD; (C) the H bonds that are generated between the ligand and the HLA allele; (D) the ligand’s RMSF; and (E) the bonds’ total energy.

The global population coverage for the 20 T cell epitopes selected for the Hp-Vaccine candidates was assessed using the IEDB Population Coverage tool, as shown in Fig. 4. The analysis revealed a robust global population coverage of 99.09%, with an average of 28 HLA allele hits per individual and a pc90 value of 19.14. The uniform standard deviation of 0 across all metrics indicates consistent epitope performance globally. Region-specific analysis showed high coverage in Europe (99.85%), Northern America (99.41%), and the West Indies (98.90%). Strong coverage was also observed in Asia (95.60%), Oceania (95.19%), Africa (92.55%), and South America (90.08%).

Fig. 4.

Global population coverage of 20 epitopes, selected as the Hp-Vaccine candidates. The population coverage was predicted using the IEDB tool.

saRNA-Hp vaccine design

The saRNA vaccine for H. pylori is designed with a structured arrangement to optimize expression and immune response. The construct begins with a 5’ m27,2′ - OGppSpG cap, which enhances RNA stability, followed by a 5’ untranslated region (5' UTR) derived from human β-globin (Supplementary table S3) (supplementary file). This is coupled with a Kozak sequence (GCCRCCATGG) to facilitate efficient translation initiation. A tissue plasminogen activator (tPA) signal peptide directs the nascent protein towards the secretory pathway.

The vaccine harnesses the power of the viral replicase complex RdRp (RNA-dependent RNA polymerase), composed of the non-structural proteins (nsP1-4) linked by a GGGS linker. This facilitates self-amplification and high-level antigen production under the control of a subgenomic promoter (SGP). The SGP drives expression of the antigenic ORF-2 sequence, which includes our selected 10 MHC-II specific cytotoxic T-lymphocyte (CTL) epitopes and 10 MHC-I specific helper T-lymphocyte (HTL) epitopes, together. These combined epitopes are linked by GPGPG and AAY linkers for MHC-II specific epitopes and MHC-I specific epitopes, respectively, to facilitate optimal epitope presentation and stimulate broad immune responses. Additionally, the first part of the vaccine construct incorporates the immunostimulatory 50S ribosomal protein L7/L12 (adjuvant), enhancing antigen-specific immunity. The antigenic multiepitope ORF-2 sequence is:

(EAAAKMAISKEELLDYIGGLSVLELSELVKAFEEKFGVSAAPTVVAGAGGGVAAEAVEEKTEF

SVVLAETGAEKIKVIKVVREITGLGLKEAKEATEKTPHVLKEGVNKEEAESIKKKLEEVGAKAEI

K GPGPG IFSITSSDSQAMGRV GPGPG MGIFSITSSDSQAMG GPGPG WTYGFGADALYNFIN G

PGPG VWTYGFGADALYNFI GPGPG KGTDNSNDAIKSALN GPGPG LSELDIQEKFLKTTH GPG

PGYDKIGFNQKNMKDYSGPGPGDKIGFNQKNMKDYSDGRNALVLKPSVGVSYNPGPGG

PGPG FFRNALVLKPSVGVS AAY RVGEVITRTW AAY TSSDSQAMGR AAY NSASDVWTY AAY YG

FGADALY AAY FLKTTHSSH AAY SELDIQEKF AAY KIGFNQKNMK AAY GFNQKNMKDY AAY S

YNHLGSTNFAAYASYGYDFAF).

The construct concludes with a GGGS linker leading to the MITD sequence for CTL epitope trafficking, followed by a stop codon, a 3’ UTR from human α-globin for mRNA stability, and a poly(A) tail of approximately 120 adenine nucleotides to further enhance stability and translation efficiency.

Codon optimization, and in-silico cloning

To ensure optimal expression of the saRNA-Hp vaccine in humans, we conducted codon optimization for both ORF-1 and ORF-2 using the Vectorbuilder Online server. This process increased the Codon Adaptation Index (CAI) to 0.92 and 0.91, respectively, and adjusted the GC content to 58.13% and 59.49%, thereby enhancing translation efficiency in the host organism. We used pcDNA3.1 eukaryotic expression vector with the size of 5,428-bp to clone our 11,343-bp saRNA-Hp vaccine (Fig. 5A). NheI and BstB17I restriction sites were incorporated at the 5’ and 3’ ends of the saRNA-Hp vaccine construct. The entire 11,343 bp construct was then inserted into the multiple cloning site (MCS) of the pcDNA3 expression vector (Fig. 5B). This resulted in a recombinant DNA molecule of 14,137 bp (Fig. 5C). Using a 1.2% agarose gel for the simulated gel electrophoresis experiment, it was possible to visually confirm the size and integrity of the DNA constructs involved in the cloning process. This experiment demonstrated the presence and correct assembly of the saRNA-Hp vaccine, pcDNA3 vector, recombinant DNA construct, and ORF-2 which will be translated in antigenic multiepitope Hp vaccine candidate. (Fig. 5D).

Fig. 5.

Construction and in-silico cloning of the saRNA-Hp vaccine into the pcDNA3 Vector. (A) Schematic representation of the complete saRNA-Hp vaccine cDNA construct. (B) Map of the pcDNA3 eukaryotic expression vector. (C) Recombinant DNA construct formed by the insertion of the self-amplifying Hp vaccine construct (in red) into the pcDNA3 vector. (D) Simulated gel electrophoresis, where lane 1 represents the saRNA-Hp vaccine with vector (recombinant construct), lane 2 represents the pcDNA3 vector, lane 3 represents the complete saRNA-Hp vaccine cDNA construct, and lane 4 represents ORF-2.

Modelling of sgRNA and translated peptide

The secondary structure of the sub-genomic mRNA (ORF-2) and its associated free energies were predicted using the RNAfold web server. The minimum free energy (MFE) of the most stable secondary structure was calculated to be − 588.50 kcal/mol (Fig. 6A), while the centroid secondary structure had an MFE of − 522.00 kcal/mol (Fig. 6B). These findings indicate that the mRNA is likely to maintain stability post-manufacturing, as lower energy values correspond to greater structural stability. Following the prediction of the mRNA’s secondary structure, the 3D structure of the translated ORF-2 peptide was modeled and refined to gain further insights into its potential conformation and functionality (Fig. 7A, B). The predicted normalized B-factor for the target protein showed a trend consistent with the expected B-factor profile (Fig. 7C), reinforcing the reliability of the model and providing valuable insights into the protein’s thermal mobility and flexibility. Furthermore, the Ramachandran plot indicates that 83.1% of the residues are located in the most preferred areas, 13.6% in further permitted regions, 0.5% in generously permitted regions, and 2.7% in disallowed regions (Fig. 7D). The protein’s quality assessment showed an ERRAT value of 88.308 (Fig. 7E), and the Z-score calculated by the ProSA-web server was − 4.64 (Fig. 7F), indicating a high-quality model. The secondary structure of the protein comprises 2 sheets, 2 beta hairpins, 4 strands, 25 helices, 28 helix-helix interactions, 49 beta turns, and 35 gamma turns, contributing to the overall stability and folding pattern of the protein, with proportions consistent with the expected conformation of the tertiary structure (Supplementary Figure S1).

Fig. 6.

Modeling of sub-Genomic mRNA (ORF-2) secondary structure, illustrating the vaccine’s base pair probabilities with the (A) minimum free energy (B) Centroid secondary structure.

Fig. 7.

Structural modeling and validation of the translated Hp-Vaccine. (A) Refined 3D model of the Hp-vaccine shown in cartoon representation with CTL epitopes in yellow, HTL epitopes in green, and the adjuvant in cyan. (B) Surface model of the Hp-vaccine visualized using ChimeraX. (C) B-factor analysis indicating thermal mobility and flexibility by I-TASSER server. (D) Ramachandran plot illustrating the distribution of residues in favored and disallowed regions using PRO-CHECK server. (E) ERRAT plot assessing the overall quality of the model PROCHECK (Structure Validation Server). (F) Z-score plot from ProSA-web, evaluating the structural reliability of the Hp-vaccine model.

Characterization and validation of the designed model

The designed Hp-vaccine model was further characterized and validated, revealing promising attributes. It has a high antigenicity score of 0.7848, is predicted to be non-allergenic and non-toxic, and consists of 454 amino acids with a molecular weight of 47,815.58 Da and a theoretical pI of 5.90. The protein’s formula is C2156H3282N556O661S9, comprising 6,664 atoms. Extinction coefficients were measured at 60,740 M⁻¹ cm⁻¹ at 280 nm in water. The estimated half-life varies across systems, with 1 h in mammalian reticulocytes, 30 min in yeast, and over 10 h in E. coli. The model is stable, with an instability index of 14.83, an aliphatic index of 67.38, and a GRAVY score of − 0.297, indicating moderate hydrophobicity. These findings support the vaccine’s potential for further development.

Tertiary structure-based B-cell epitopes analysis

Discontinuous B-cell epitopes of the designed Hp vaccine were predicted using the Ellipro server, with a score threshold of ≥ 0.5 considered significant for continuous epitopes (Fig. 8). The analysis identified seven B-cell epitopes based on the protrusion index of the tertiary structure. These epitopes, with scores ranging from 0.507 to 0.867, are highlighted in different colors-orange red (highest score), orange, yellow, green, blue, purple, pink (lowest score)- corresponding to their respective scores.

Fig. 8.

Prediction of discontinuous B-cell epitopes in the designed vaccine using Ellipro server. Epitopes are visualized using ChimeraX v1.8rc202406072045 and color-coded based on their scores: orange-red for the highest score, followed by orange, yellow, green, blue, purple, and pink for the lowest score.

Post‐translational modification analysis

The post-translational modification (PTM) analysis revealed that the vaccine construct lacks N-glycosylation and N-acetylation sites, as indicated by the NetNGlyc-1.0 and NetAcet-1.0 servers, respectively. However, the NetPhos-3.0 server identified 54 phosphorylation sites. Lipid PTM analysis predicted no N-terminal glycine myristoylation or GPI-anchor modifications using the MyrPS/NMT and big-PI/GPI animals servers, respectively.

Molecular docking of TLR-4 and vaccine peptide

ClusPro 2.0 server was employed to perform molecular docking, confirming potential interactions between TLR-4 and the proposed Hp vaccine (Fig. 9A). The docking process involved 70,000 ligand rotations, with the best translation in x, y, and z selected for each. The top 1000 lowest-scoring poses were clustered within a 9-angstrom radius, focusing on ligand positions with the most neighbors. The best docking pose, illustrated in Fig. 9B, had the lowest weighted center score of − 1119.4 and the lowest energy value of − 1355.2. Using the PDBsum server, interactions between the hTLR-4 receptor and the vaccine peptide were visualized, revealing that Chains B (pink) and D (light orange) formed 23 hydrogen bonds, one salt bridge, and 222 non-bonded contacts with the vaccine (Fig. 9C, D).

Fig. 9.

(A) Molecular docking of hTLR-4 (depicted in blue, green, pink, and light-orange) with the Hp-Vaccine peptide (shown in red) using Cluspro 2.0. (B) Visualization of the docked hTLR-4-Hp Vaccine complex using chimera v1.8rc202406072045. (C) & (D) Detailed receptor (chain B & D)-ligand (chain E) interaction analysis performed using the PDBsum web server.

Molecular dynamic simulation

The stability and dynamic behavior of the interaction were evaluated by performing a molecular dynamic simulation of the Hp-Vaccine (HpVac) and TLR-4 docked complex (Fig. 10A) using the iMODS tool. The analysis, based on normal mode analysis (NMA) in internal coordinates, provided insights into the conformational flexibility and overall stability of the complex. The TLR4-Hp-vaccine complex appears to be significantly stable based on the residue interactions seen in the elastic network model (Fig. 10B), where darker regions denote stiffer, more stable connections. The correlated movements between residue pairs are shown in the covariance map (Fig. 10C), where positive correlations (in red) denote synchronized movement and negative correlations (in blue) denote opposing motions. Peaks in the deformability plot (Fig. 10D) highlight residues with high deformability, potentially acting as hinges in the structure. The eigenvalue analysis (Fig. 10E) revealed a low eigenvalue of 2.009426e⁻⁰⁵, signifying a stable and crucial vibrational mode for the complex. The degree of atomic mobility inside the protein is shown by the B-factor plot, which displays the atomic displacement values (Fig. 10F). Greater mobility or flexibility is indicated by higher B-factors in certain places, whereas more rigid, stable regions are suggested by lower B-factors. Early modes (up to Mode index 4) show reduced variance, indicating better stability, whereas later modes contribute more to the complex’s mobility and flexibility, according to variance analysis (Fig. 10G).

Fig. 10.

Molecular dynamic simulation analysis of the Hp-Vaccine (HpVac) and TLR-4 docked complex using iMODS. (A) 3D structure of the HpVac-TLR-4 docked complex visualized by ChimeraX version 1.8rc202406072045. (B) Elastic network model displaying the interactions between residues (C) Covariance map showing correlated movements between residue pairs (D) Deformability plot (E) Eigenvalue analysis plot (F) B-factor plot reflecting atomic displacement parameters. (G) Variance analysis highlighting stability and flexibility of different modes.

Discussion

This study addresses the pressing need for novel vaccine strategies against Helicobacter pylori, a pathogen that affects nearly half of the global population and is a major contributor to gastrointestinal and non-gastric diseases^66–69. The lack of a commercial vaccine, coupled with increasing antibiotic resistance, complicates existing treatment options^70–74. In light of this, we sought to develop a next-generation multiepitope vaccine by utilizing self-amplifying RNA (saRNA) technology, targeting conserved, immunogenic proteins associated with key stages of H. pylori pathogenesis.

In addressing the initial stage, survival in the acidic condition, of the bacterial pathogenesis, only UreB emerged as an antigenic candidate. Despite its primary cytoplasmic localization, surface-localized or extracellular presence has been documented both in vivo and in stationary-phase culture⁷⁵. It also has been previously demonstrated effective, safe, and immunogenic in H pylori reduction in H pylori-naive children during a phase 3 clinical trial²⁶. For the next stage, several proteins, including AlpA, AlpB, BabA, LabA, BabB, OipA, and HpaA, were evaluated based on their roles in H. pylori pathogenesis, outer membrane localization, virulence, antigenicity, and their divergence from human proteins. BabB and HpaA were selected as vaccine candidates due to their crucial roles in bacterial adhesion, colonization, and well-established immunogenicity^76–79. HpaA functions as a hemagglutinin, promoting initial attachment to gastric epithelial cells⁸⁰, while BabB facilitates binding to blood group antigens, aiding persistent colonization^81,82. Conversely, AlpA, AlpB, BabA, LabA, and OipA were excluded due to less consistent evidence regarding their pathogenic roles, antigenic properties, or variability in expression. For example, BabA, although a key adhesin, undergoes phase variation, reducing its reliability as a universal vaccine target^82,83. OipA, associated with inflammation, is similarly subject to variable expression⁸⁴, while the roles of AlpA, AlpB, and LabA in virulence are not as clearly defined or strongly supported ^85,86. Notably, H. pylori SS1 mutants deficient in AlpA and AlpB induce more severe inflammation compared to the isogenic wild-type strain in animal models like Mongolian gerbils complicating their suitability as vaccine targets⁸⁵. For the final stage, CagA and VacA were selected due to their extracellular presence, high virulence, strong antigenicity, and significant dissimilarity to human proteins. These proteins are well-supported by numerous studies as key factors in H. pylori pathogenesis^27,87. The vacA gene is present in all H. pylori strains and is associated with immune evasion and tissue damage, which exacerbate disease severity⁸⁸. Similarly, infections caused by CagA-positive H. pylori strains are linked to increased severity and poorer clinical outcomes in patients, as the CagA protein plays a critical role in disrupting host cell functions and promoting inflammation⁸⁹.

One of the key aspects of this work was the conservation analysis of the selected proteins. The identification of highly conserved regions (≥ 90%) among different Hp pathotypes across diverse geographic locations ensures that the proposed vaccine candidates can provide broad coverage, reducing the risk of strain-specific inefficacy. This finding is consistent with previous research emphasizing the importance of including conserved antigens to address the genetic variability of H. pylori^29,90. Moreover, the predicted consensus sequences within these conserved regions provide a strong foundation for designing an immune response capable of targeting multiple strains. The vaccine’s potential impact on public health in South Asia, East Asia, and Southeast Asia, regions with a high incidence of H. pylori infections and growing multidrug resistance (MDR), is highlighted by its high population coverage rates. With coverage rates of 97.49% in East Asia, 96.74% in South Asia, and 95.00% in Southeast Asia, this vaccine has the potential to drastically lower infection rates in areas where traditional antibiotic treatment is challenging owing to MDR strains^91,92.

The prediction and analysis of T-cell epitopes revealed the immunogenic potential of each protein candidate, with both MHC-I and MHC-II epitopes showing strong binding affinities with MHC alleles, surpassing similar findings in other bacterial vaccine studies^29,49,93. Additionally, molecular dynamics simulation validated the stability of the peptide-MHC complexes. The observed RMSD and RMSF values indicated minimal fluctuations, while a high number of hydrogen bonds and favorable total complex energy reinforced their stability throughout the simulation period. This observation suggests that these complexes are likely to persist in vivo, promoting sustained immune responses, which is crucial for vaccine efficacy. Notably, the antigenic epitopes of UreB, BabA, HpaA, CagA, and VacA were identified as capable of inducing IFN-γ responses, further strengthening the hypothesis of their utility in stimulating an effective cellular immune response. This aligns with earlier studies, such as those by Ottsjö et al. 2015, and Obonyo et al., 2002 which highlighted the importance of targeting epitopes that trigger IFN-γ production, as it plays a critical role in the clearance of intracellular pathogens like H. pylori^94,95.

The saRNA-based vaccine platform offers significant advantages, including the ability to rapidly induce potent immune responses while ensuring prolonged antigen expression⁹⁶. Codon optimization and the construction of the saRNA-Hp vaccine, followed by in-silico cloning, demonstrated enhanced expression and stability within human cells. This codon adaptation, with CAI values above 0.90, mirrors the findings of recent research in RNA vaccines for infectious diseases, where codon optimization significantly improved protein translation and immunogenicity. The 3D modeling and validation of the translated peptide from the saRNA vaccine construct yielded high structural quality, with 83.1% of the residues falling within the most favored regions, 13.6% in further permitted regions, 0.5% in generously permitted regions of the Ramachandran plot and a Z-score of -4.64, suggesting a reliable and functional protein model. These findings are consistent with structural models of other multiepitope vaccines, such as those targeting Escherichia coli and avian influenza H9N2, where similar quality metrics were observed for their immunogenic proteins^97,98.

The molecular docking of the designed vaccine with TLR-4 elucidated critical interactions that are expected to activate the innate immune response⁶⁵. The docking results revealed that the vaccine construct engages with the TLR-4 receptor, forming numerous hydrogen bonds and non-bonded interactions, indicative of a strong binding affinity. The best docking pose, characterized by favorable weighted center score and an energy value, supports the hypothesis that the vaccine can effectively engage TLR-4, a key player in initiating immune responses. Molecular dynamics simulations provided further insight into the stability and dynamic behavior of the TLR-4-vaccine complex. The elastic network model and covariance map analyses indicated a robust network of residue interactions, suggesting that the complex is stable during dynamic fluctuations. The low eigenvalue identified a stable vibrational mode, while the B-factor plot highlighted areas of flexibility within the protein, essential for potential conformational changes upon receptor engagement.

In conclusion, this study successfully identified and characterized multiple conserved, immunogenic proteins from H. pylori to develop a potential multiepitope vaccine candidate using saRNA-based vaccine development technology. The strong antigenic properties, coupled with the stability and binding affinities demonstrated in-silico, indicate the feasibility of this approach. The broad conservation of these antigens across various H. pylori strains also underscores their potential for global vaccine applicability. Future experimental validation, including in vitro and in vivo testing, will be essential to further confirm the immunogenicity and protective efficacy of the designed vaccine.

Author contributions

MTM and FA conceived and designed the study, reviewed manuscript. AR collected data and conducted all immunoinformatics analysis and first drafted the manuscript. MTM also involved data analysis. All authors were involved in the subsequent revision and approved the final manuscript.

Data availability

All data generated or analyzed during this study are included in this published article (and its Supplementary Information files).

Declarations

Competing interests

The authors declare no competing interests.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-99512-9.