Immunoinformatics Approach to Design a Novel Epitope-Based Oral Vaccine Against Helicobacter pylori

Abstract

Helicobacter pylori is an infectious agent that colonizes the gastric mucosa of half of the population worldwide. This bacterium has been recognized as belonging to group 1 carcinogen by the World Health Organization for the role in development of gastritis, peptic ulcers, and cancer. Due to the increase in resistance to antibiotics used in the anti-H. pylori therapy, the development of an effective vaccine is an alternative of great interest, which remains a challenge. Therefore, a rational, strategic, and efficient vaccine design against H. pylori is necessary where the use of the most current bioinformatics tools could help achieve it. In this study, immunoinformatics approach was used to design a novel multiepitope oral vaccine against H. pylori. Our multiepitope vaccine is composed of cholera toxin subunit B (CTB) that is used as a mucosal adjuvant to enhance vaccine immunogenicity for oral immunization. CTB fused to 11 epitopes predicted of pathogenic (UreB_170–189, VacA_459–478, CagA_1103–1122, GGT_106–126, NapA_30–44, and OipA_211–230) and colonization (HpaA_33–52, FlaA_487–506, FecA_437–456, BabA_129–149, and SabA_540–559) proteins from H. pylori. CKS9 peptide (CKSTHPLSC) targets epithelial microfold cells to enhance vaccine uptake from the gut barrier. All sequences were joined to each other by proper linkers. The vaccine was modeled and validated to achieve a high-quality three-dimensional structure. The vaccine design was evaluated as nonallergenic, antigenic, soluble, and with an appropriate molecular weight and isoelectric point. Our results suggest that our newly designed vaccine could serve as a promising anti-H. pylori vaccine candidate.

1. Introduction

Helicobacter pylori infection is associated with gastrointestinal tract diseases, including chronic gastritis, peptic ulcer, mucosa-associated lymphoproliferative disorders, and gastric cancer (Marshall and Warren, 1984; IARC, 1994; Wroblewski et al., 2010). Its global prevalence is >50%, which may significantly vary within and among countries, according to geography, ethnicity, age, and socioeconomic factors (Leja et al., 2016; Burucoa and Axon, 2017). The conventional antimicrobial therapy against H. pylori is costly, causes serious side effects and potential reinfections, and eliminates <80% of H. pylori infection, which facilitate drug resistance (Thung et al., 2016). Therefore, considerable efforts have been made to develop a preventive strategy against H. pylori infection, such as recombinant vaccines, with limited effectiveness. Numerous studies of vaccination in animal models have shown that immunization with combinations of antigens confers a better protective immunity against this bacterium; however, the establishment of an immunological response that allows the eradication in gastric mucosa remains a challenge (Rappuoli et al., 2016; Sutton and Boag, 2018).

The technological advances in the fields of genomics, proteomics, human immunology, and structural biology have provided the molecular information for the discovery and prediction by bioinformatics tools of novel antigens, epitopes, and design of vaccines against pathogenic bacteria, such as meningococcus B (Serruto et al., 2012; O'Ryan et al., 2013; Rappuoli et al., 2016). Bioinformatics tools can predict sequences with binding affinity to MHC alleles and epitopes of T and B cells (Rosa et al., 2015). The focus on vaccine design and development has changed to the production of peptides composed of multiple epitopes (multiepitope vaccines), based on linear arrangements, as a novel alternative. Immunological studies have shown that H. pylori infection induces an activation of adaptive B and T cell responses and that a Th1/Th17-polarized immune response is required for its prevention (Müller and Hartung, 2016). Some recent studies have designed multiepitope vaccine of H. pylori using bioinformatics (Li et al., 2014; Meza et al., 2017). In addition, epitope-based vaccines have demonstrated various advantages, including safety, the opportunity to rationally engineer the epitopes for increased potency, breadth, and antigenicity, and the possibility to focus large repertories of immune responses on conserved epitope sequences (Livingston et al., 2002; Oyarzún and Kobe, 2015).

In the present study, a novel oral vaccine against H. pylori was in silico designed.

2. Methods

2.1. Selection and antigenic evaluation of proteins

H. pylori proteins were selected based on the following criteria: (1) reported antigenicity, (2) virulence, and (3) proteins related to the mechanisms of adhesion and colonization.

The entire protein sequences were retrieved from NCBI reference sequence database in FASTA format. For bacterial protein subcellular localization prediction, CELLO v2.5 was used (Yu et al., 2006). The database of Online GEne Essentiality (OGEE) was used to evaluate the essentiality of genes (Chen et al., 2017). PATRIC 3.5.16 database was used for study of the virulence role of proteins (Wattam et al., 2017). Proteins were screened by BLASTp to detect sequence homologs to Homo sapiens. To predict transmembrane (TM) helixes, TMHMM v2.0 was used (Krogh et al., 2001). Compute pI/Mw tool was used to calculate the estimated isoelectric point and molecular weight of all amino acid sequences (Wilkins et al., 1999).

2.2. Conservation, consensus sequence, and alignment

Conservation of selected proteins was analyzed using H. pylori strain 26695 as the reference. For proteins with high variability, a consensus sequence was generated using sequences of world H. pylori representative strains (85 complete genome sequences). The sequence alignment was performed using CLC Main Workbench v7.8 (QIAGEN Bioinformatics), Emboss Cons, and T-Coffee software.

2.3. T cell epitope prediction

To identify MHC-I binding epitopes, NetMHC 4.0 server was used (Andreatta and Nielsen, 2016). Eighty-one human leukocyte antigen (HLA) alleles (HLA-A, -B, -C, and -E) and six murine alleles (H-2) were evaluated. Predictions were calculated for nine-mers epitopes with a threshold for strong binders of 0.5% and a threshold for weak binders of 2%. The second was ProPred-I that implements matrices for 47 MHC Class-I alleles, proteasomal, and immunoproteasomal models. The threshold scoring peptides were 4% (Singh and Raghava, 2003). The third one was IEDB MHC-I Binding Predictions; the best-predicted binders for 27 HLA and 6 H-2 alleles were selected based on a percentile rank of ≤1%.

For MHC-II binding epitopes, ProPred was used; 51 HLA-DR alleles were evaluated using a default threshold of 3% (Singh and Raghava, 2001). The second was IEDB MHC-II Binding Predictions, a reference panel of 27 HLA alleles and 3 H-2-I alleles was tested; the selection was based on a consensus percentile rank of the top 10%. The third was NetMHCII 2.3 server; predictions were obtained for 25 HLA-DR alleles, 20 HLA-DQ, 9 HLA-DP, and 7 mouse H2 class II alleles using a threshold of −99.9, threshold for the strong binder of 2%, and threshold for the weak binder of 10%.

2.4. B cell epitope prediction

Linear B cell epitopes of 20-mers were predicted utilizing ABCpred with a threshold of 0.75. The second one was BcePred. The BCPred server was applied with a specificity threshold of 75%. For BepiPred server, only amino acids with score ≥1.3 were considered for the downstream analysis (Jespersen et al., 2017). DiscoTope 2.0 server was used for identification of discontinuous B cell epitopes with a threshold of −3.7; the sensitivity and specificity were 0.47 and 0.75, respectively (Kringelum et al., 2012). ElliPro server predicts linear and discontinuous epitopes. For prediction, a maximum distance (Å) of six and a minimum score of 0.5 were used (Ponomarenko et al., 2008).

2.5. Selection of epitopes

Epitopes were selected based on the following criteria: (1) 20-mer epitopes, (2) epitopes matching on all algorithms, if possible, and (3) potential to bind with the maximum number of MHC-I and MHC-II alleles. For selection, sequences were aligned and overlapped using Clustal Omega server.

2.6. Vaccine design

The VaccineCAD (EpiVax Inc., RI) and EpiToolKit 2.0 were used to optimize epitope order. They link candidate epitopes into a string-of-beads design, minimizing nonspecific junctional epitopes that may be produced in the linking process (Schubert et al., 2015). In addition, Gly-Pro-Gly-Pro-Gly (GPGPG) and Lys-Lys (KK) spacer sequences were engineered at the interepitope position. Cholera toxin subunit B (CTB) and CKS9 (M cell homing peptide) sequences were added to the final vaccine model at the N-terminal and C-terminal, respectively.

2.7. Protein prediction and validation of secondary and tertiary structures

The secondary structure of the multiepitope antigen was predicted using PSIPRED v3.3 (McGuffin et al., 2000). The three-dimensional (3D) structure modeling was performed using I-TASSER server (Yang and Zhang, 2015). Molecular Operating Environment v2014.0901 and Jmol were used for visualizing 3D structures of proteins. For refinement of 3D model structure, GalaxyRefine and GalaxyLoop were applied (Park et al., 2011).

The best model was validated by the ProSA-web (Wiederstein and Sippl, 2007) and ERRAT (Colovos and Yeates, 1993). The residue-by-residue stereochemical qualities of models were validated by Ramachandran plot obtained from PROCHECK server (Laskowski et al., 2012). The best-refined model was selected.

2.8. Antigenicity, allergenicity, solubility, and physicochemical predictions of the vaccine

For antigenicity prediction, VaxiJen server was used. For allergenicity evaluation, AllerTOP v.2.0 and AlgPred servers were used. For solubility prediction, SOLpro server was used. Finally, ProtParam allowed the computation of various physical and chemical parameters (Wilkins et al., 1999).

2.9. Codon optimization

DNA coding sequence of the oral vaccine was optimized for high-level protein expression in Lactococcus lactis using Codon optimization On-Line server. The tool is based on several design parameters such as individual codon usage, codon pairing, codon adaptation index (CAI), and others (Chin et al., 2014).

3. Results and Discussion

3.1. Protein selection and evaluation

It has been shown that the combination of a long list of antigens can increase the effectiveness of vaccines (Rossi et al., 2004). Malfertheiner et al. (2008) evaluated a multicomponent vaccine in human volunteers; they observed that the vaccine was safe and highly immunogenic, inducing long-lasting humoral and cellular responses to the antigens. Therefore, 11 proteins were selected for our vaccine. UreB has been reported in many studies for antigenicity; VacA, CagA, GGT, NapA, and OipA are the main proteins involved in the pathogenic mechanisms. HpaA, FlaA, FecA, BabA, and SabA were selected for their participation in the mechanisms of adhesion and colonization.

Selected proteins were further screened in the next step for further parameters, including subcellular localization, essentiality, virulence, nonhuman homology, TM helixes, and molecular weight. Our predictions showed that UreB and NapA possessed a cytoplasmic location; CagA, FlaA, VacA, and BabA were extracellular, whereas OipA, SabA, HpaA, and FecA were predicted as outer membrane proteins. In addition, GGT was classified as a periplasmic protein, as shown in Table 1. Surface and extracellular proteins are good targets to develop a vaccine aiming toward prevention of bacterial infections and diseases (Dwivedi et al., 2016).

Table 1.

The Subcellular Localization, Gene Essentiality, Virulence, Human Homology, Transmembrane Helix, Isoelectric Point, and Molecular Weight Predictions of Helicobacter pylori Selected Proteins

Protein	Accession No.	Locus tag	Subcellular localization	Gene essentiality	Virulent protein	Human homology	TM helix	pI/MW (kDa)
UreB	NP_206872.1	HP0072	C	N-ES	V	N-H	0	5.64/61.68
NapA	NP_207041.1	HP0243	C	N-ES	V	N-H	0	5.59/16.93
CagA	NP_207343.1	HP0547	E	N-ES	V	N-H	0	8.82/132.4
FlaA	NP_207396.1	HP0601	E	N-ES	V	N-H	0	6.03/53.3
OipA	NP_207432.1	HP0638	OM	N-ES	V	N-H	0	9.78/34.2
SabA	NP_207519.1	HP0725	OM	N-ES	V	N-H	0	9.12/69.4
HpaA	NP_207590.1	HP0797	OM	N-ES	V	N-H	0	8.60/29.0
FecA	NP_207600.1	HP0807	OM	ES	N-V	N-H	0	9.66/88.9
VacA	NP_207680.1	HP0887	E	N-ES	V	N-H	1 (35–57 aa)	9.02/139.3
GGT	NP_207909.1	HP1118	P	ES	V	N-H	1 (5–27 aa)	9.27/61.2
BabA	NP_208035.1	HP1243	E	N-ES	V	N-H	0	8.95/79.1

aa, amino acid; C, cytoplasmic; E, essential; ES, extracellular; N-ES, nonessential; N-H, nonhomology; N-V, nonvirulent; OM, outer membrane; P, periplasmic; TM, transmembrane; V, virulent.

In the analysis of essential genes by OGEE, HP1118 (GGT) and HP0807 (FecA) genes were identified. Essential genes are those genes of an organism that are critical for its survival; essential genes are of particular importance because of their theoretical and practical applications such as studying the robustness of a biological system, defining a minimal genome/organism and identifying effective therapeutic targets in pathogens (Chen et al., 2017). Furthermore, 9 of the 11 proteins were found in the database of virulence factors of PATRIC; FecA and GGT were not identified in this category. Homology analysis of the 11 prioritized proteins using BLASTp revealed ≤30% identity, which was significant to declare the sequences as nonhuman homologs; Vaccine targets should not be human homologues to avoid autoimmunity. The prediction of the topology of proteins by TMHMM showed that VacA and GGT had one TM helix, whereas other proteins do not unveil the presence of any such topology. VacA and GGT showed the presence of a TM helix located at 35–57 and at 5–27 amino acid positions, respectively. Finally, molecular weights calculated by Compute pI/Mw tool of nine proteins resulted to weigh <110 kDa, whereas VacA and CagA have molecular weights of 139.3 and 132.4 kDa, respectively (Table 1).

Good vaccine candidates are considered as those that do not present homology with human proteins to avoid the generation of a potential autoimmune response; these candidates must also lack TM regions, to facilitate their expression. Another characteristic a good vaccine candidate should have is to possess good antigenic properties, which are important for the pathogenesis of the microorganism and for protection against the disease (Monterrubio-López et al., 2015). Some authors have used these approaches in the selection of candidate proteins for the in silico design of H. pylori vaccines (Naz et al., 2015; Nezafat et al., 2017).

3.2. Consensus antigen

To develop a universal H. pylori vaccine, a conservation analysis of the selected antigens was performed. The amino acid sequences of each protein were compared and aligned with 85 sequences corresponding to fully annotated genomes of H. pylori from different geographical regions at the time of the study. A high variability in the proteins CagA, VacA, FecA, BabA, and SabA was observed, as shown in Figure 1. Consequently, to reduce the variability between H. pylori strains, consensus sequences were obtained. This approach has been used against infectious agents with a high variability rate such as HIV (Thomson et al., 2005; Grimm and Ackerman, 2013). In contrast, the proteins with the highest percentage (≥95%) of conservation were UreB, NapA, GGT, FlaA, HpaA, and OipA (Fig. 2).

FIG. 1.

Analysis of conservation of selected proteins. Entire protein sequences among the 85 annotated genomes of Helicobacter pylori were aligned using CLC Main workbench, and conservation scores were represented in a 2D line chart. (a) CagA, (b) SabA, (c) BabA, (d) VacA, and (e) FecA. 2D, two-dimensional.

FIG. 2.

Analysis of conservation of selected proteins. Entire protein sequences among the 85 annotated genomes of Helicobacter pylori were aligned using CLC Main workbench, and conservation scores were represented in a 2D line chart. (a) UreB, (b) GGT, (c) NapA, (d) FlaA, (e) OipA, and (f) HpaA.

3.3. T and B cell epitopes

The prediction analysis by different bioinformatics servers for T and B cells (using MHC-I/-II alleles for human and mouse BALB/c) allowed the selection of 11 epitopes based on their score, number of alleles, and agreement between the servers used. Epitopes obtained were NapA_30–49, CagA_1103–1122, FlaA_487–506, OipA_211–230, GGT_106–126, HpaA_33–52, BabA_129–149, FecA_437–456, SabA_540–559, VacA_459–478, and UreB_170–189 (Table 2). HpaA_33–52 epitope has been previously reported by Hu et al. (2016); they evaluated the response of antigen-specific CD4⁺ T lymphocytes stimulated with recombinant HpaA from 101 H. pylori-infected subjects, a significant increase of the interferon-γ producing cells was observed. In addition, nine residues of UreB_170–189 epitope predicted in this study match with UreB_181–195 epitope reported by Qiu et al.; they observed the positive reactivity with sera from H. pylori-infected patients by enzyme-linked immunosorbent assay; mice immunized with glutathione S-transferase-UreB_181–195 fusion peptide showed that epitope-specific antibodies were capable of inhibiting urease enzymatic activity. In another study, Guo et al. (2017) designed and evaluated a multivalent epitope-based vaccine (named CWAE) that includes UreB_181–195 epitope. Oral therapeutic immunization with CWAE significantly reduced the number of H. pylori colonies in the stomach of Mongolian gerbils. The protection of CWAE was associated with higher levels of mixed CD4⁺ T cell response, immunoglobulin G, and secretory immunoglobulin A (Guo et al., 2017).

Table 2.

Potential Antigenic Epitopes Predicted by Different Servers

										T cell epitopes
				B cell epitopes						IEDB		ProPred		NetMHC
Order	Protein	Position	Sequence	BepiPred	BCPRED	ABCpred	DiscoTope	ElliPro	BcePred	I	II	I	II	I	II
1	NapA	30–49	GTDFFNVHKATEEIYEEFAD	0.520	0.573	0.31	−7.806	0.511	0.826	6H 2M	5H 0M	17H 3M	10H	31H 1M	12H 0M
2	CagA	1103–1122	IVAHNVGSVPLSEYDKIGFN	0.087	0.45	0.86	−0.852	0.928	0.953	8H 3M	6H 0M	24H 6M	16H	34H 1M	7H 0M
3	FlaA	487–506	LAQSGSYAMSQANTVQQNIL	0.136	0.363	0.82	−4.984	0.916	0.935	10H 3M	5H 1M	19H 5M	10H	46H 2M	13H 1M
4	OipA	211–230	ASTWFASNNLTPFNQVKSHT	0.557	0.892	0.35	−2.603	0.564	0.714	10H 2M	2H 0M	27H 3M	3H	46H 2M	11H 0M
5	GGT	106–126	APLKATKNMFLDKQGNVVPKL	0.553	0.697	0.42	−5.047	0.705	0.925	8H 2M	4H 1M	29H 6M	11H	43H 2M	14H 0M
6	HpaA	33–52	IETNEVALKLNYHPASEKVQ	0.627	0.21	0.8	−5.593	0.401	0.812	9H 1M	7H 0M	23H 2M	21H	37H 3M	7H 0M
7	BabA	129–149	NANGQKSTSSTTIFNNEPGYR	1.243	0.969	0.67	−11.203	0.678	0.812	13H 2M	3H 0M	16H 4M	0H	35H 2M	5H 0M
8	FecA	437–456	NLVVNTGKVKQTFNMGMRFL	−0.398	0.617	0.85	−3.091	0.681	0.842	9H 2M	8H 0M	26H 5M	34H	32H 2M	15H 1M
9	SabA	540–559	WLNSQYVNLTAFNNPYSAKV	0.546	0.621	0.76	−4.667	0.907	0.885	14H 3M	7H 0M	26H 5M	28H	67H 3M	16H 0M
10	VacA	459–478	NQVGGYALAGSSANFEFKAG	0.563	0.558	0.85	−13.843	0.261	0.893	6H 2M	5H 2M	15H 4M	11H	38H 2M	11H 1M
11	UreB	170–189	TTITPGRRNLKWMLRAAEEY	0.649	0.845	0.87	−3.764	0.584	0.790	9H 1M	6H 1M	21H 3M	15H	39H 2M	13H 0M

H, human; I, MHC class I; II, MHC class II; M, mouse.

In addition, the residue-to-residue conservation of the selected epitopes was evaluated; H. pylori 26695 was used as the reference. CagA_1103–1122, BabA_129–149, and FecA_437–456 showed one amino acid change; OipaA_211–230 showed two changes (Fig. 3).

FIG. 3.

Conservation of selected epitopes among strains of Helicobacter pylori. Individual proteins among the 85 annotated genomes were aligned using CLC Main workbench, and consensus sequence for each epitope was obtained and demonstrated in the black boxes. The percentage of conservation for each residue is shown in a bar chart. (a) NapA_30–49, (b) CagA_1103–1122, (c) FlaA_487–506, (d) OipA_211–230, (e) GGT_106–126, (f) HpaA_33–52, (g) BabA_129–149, (h) FecA_437–456, (i) SabA_540–559, (j) VacA_459–478, and (k) UreB_170–189.

3.4. Multiepitope antigen design

A step of the design pipeline is concerned with the delivery of the selected epitopes. A common strategy concatenates the epitopes into a so-called string-of-beads polypeptide. The epitope order within a string-of-beads plays a crucial role especially in degradation (Schubert et al., 2015). Therefore it was necessary to optimize the ordering such that the recovery probability of the epitopes was maximal, as shown in Figure 4.

FIG. 4.

Schematic diagram of vaccine construct consists of adjuvant in N-terminal joined to 11 epitopes fused together by proper linkers (GPGPG or KK) and CKS9 M cell homing peptide in C-terminal. CTB, cholera toxin subunit B; GPGPG, Gly-Pro-Gly-Pro-Gly; KK, Lys-Lys.

It has been reported that by coupling the antigen to CTB, a stronger response is achieved through the oral administration route. Moreover, the adjuvant potential of CTB has also been shown in large animal models, indicating that the adjuvant potential would be scalable to higher species (Baldauf et al., 2015). To increase the immunogenic capacity of our vaccine design, the CTB was added in N-terminal.

Oral immunization has ample advantages compared to other administration routes, such as safe and easy administration, high patient compliance without pain, no need of trained medical personnel, easy soak up, and immune responses in systemic and mucosal sites. In the intestine, M cells are antigen-collecting portals located on the follicle-associated epithelium of Peyer's patches, which facilitate to transport antigens from the gut lumen to underneath mucosal immune system (Lelouard et al., 2012). Thus, an M cell targeting peptide ligand (CKS9), consisting of nine amino acids with cyclic conformation, was introduced in C-terminal. Yoo et al. (2010) demonstrated that CKS9 is an efficient target coupled with chitosan nanoparticles by in vitro and in vivo assays.

The final vaccine consists of the adjuvant CTB (CAA53976.1) in the N-terminal fused to 11 epitopes of H. pylori separated by linkers (DPRVPSS, GPGPG, and KK) and fused with CKS9 peptide in the C-terminal. Schematic diagram of the vaccine is displayed in Figure 4.

3.5. Protein structure prediction and validation

The vaccine is composed of 373 amino acids, and prediction of secondary structure showed that it contains 35% α helixes, 14.0% β sheets, and 49.0% others (random coil and β-turn), as shown in Figure 5.

FIG. 5.

The secondary structure prediction of vaccine by PSIPRED. The protein vaccine consists of 35% α helix (H, cylinder), 14% β strand (E, arrow), and 49% coil (C, line) secondary structural elements. The bar chart represents the percentage of confidence.

Five 3D models of the protein vaccine were generated among which the model with the highest c-score = −0.62 was selected for further refinement; the c-score range is typically from −5 to 2, the higher the value, the higher the confidence.

The quality and potential errors in the best model were analyzed. The initial input model z-score was −4.03, which falls within those commonly observed in similar size-native proteins (Fig. 6a). ProSA-web indicated that the preliminary model requires refinement processes. Hence, the crude model was subjected to loop refinement and energy minimization. After all refinement procedures, ERRAT factor was improved from 68.8 to 83.3. The z-score of the final model reached a value of −4.9 (Fig. 6b). The starting and refined models were superimposed (Fig. 7).

FIG. 6.

The z-score plot of unrefined and refined 3D structure of vaccine by ProSA-web. (a) The z-score of the starting model is −4.03, (b) The z-score of model after refinement steps is −4.9. The z-score indicates overall model quality and is depicted as a black spot. The z-score plot contains the z-scores of all experimentally determined protein chains in current protein data bank (PDB) from NMR spectroscopy (charcoal) and X-ray crystallography (silver). 3D, three-dimensional.

FIG. 7.

Comparison of initial and refined 3D protein structure conformations of vaccine by Molecular Operating Environment software. The initial and refined models are colored in silver and charcoal, respectively.

To validate the 3D models, Ramachandran plot analysis was performed before and after refinement processes. The Ramachandran plots of the unrefined model indicated that 63.5% of residues were located in most-favored regions, 28.6% in the additional allowed region, 5.7% in generously allowed regions, and 2.2% in disallowed regions of the plot (Fig. 8a). The refined model showed that 85.1% of residues were located in most-favored regions, 12.1% in additional allowed regions, 0.6% in generously allowed regions, and only 2.2% in disallowed regions (Fig. 8b). Our results indicated that the quality and stability of the final refined model were remarkably improved based on Ramachandran plot predictions.

FIG. 8.

Validation of vaccine 3D model using Ramachandran plot. The Ramachandran plots of (a) the unrefined model and (b) the refined model. The most-favored (A, B, and L) and additional allowed (a, b, l, and p) regions were demonstrated with charcoal and silver gray colors, respectively. The generously allowed regions (∼a, ∼b, ∼l, and ∼p) are indicated in silver, and the disallowed regions are in white color. Glycine residues are shown in black triangles, and other residues of protein are shown in black squares.

3.6. Antigenicity, allergenicity, solubility, and physicochemical parameters of the vaccine

An antigenicity score of 0.5547 was obtained. The allergenicity prediction showed that the vaccine is not allergenic. The molecular weight and theoretical pI of protein were 40.7 kDa and 9.36, respectively. The recombinant protein vaccine solubility upon overexpression in Escherichia coli was 0.821589. Half-life was estimated to be 30 hours in mammalian reticulocytes, >20 hours in yeast and >10 hours in E. coli. The vaccine was founded as stable with instability index of 32.58. GRAVY and aliphatic index were assessed as −0.485 and 67.77, respectively.

3.7. Codon optimization for expression in L. lactis

CAI of the gene sequence was 0.9464; a CAI >0.8 is acceptable for high-level expression in the L. lactis host (Fig. 9). The guanine-cytosine (GC) content fitness, individual codon fitness, codon autocorrelation fitness, and 5′ RNA folding instability were −0.0062602, −0.1577, −0.099844, and −1.9, respectively. The overall GC content was 31.63%; the ideal percentage range of GC content ranged from 30% to 70%. Negative CIS elements and repeat sequences were not found in the optimized gene. A total of 78 hidden stop codons were found; Shine-Dalgarno mRNA sequences were excluded. To clone the gene in E. coli-L. lactis pNZ8048 shuttle vector, NcoI and HindIII restriction sites were introduced to 5′ and 3′ ends of DNA sequence, respectively.

FIG. 9.

Comparison of the individual codon usage between host and optimized sequences. The host and optimized sequences are shown in charcoal and black lines, respectively.

4. Conclusion

In this study, we designed a multiepitope oral vaccine against H. pylori based on bioinformatics analysis to produce a robust humoral and cellular immune response for H. pylori prevention. Our novel oral vaccine design could be a good vaccine candidate against H. pylori. However, to validate the prophylactic and therapeutic effects of our oral vaccine design, in vitro and in vivo immunological studies are required.

Author Disclosure Statement

The authors declare there are no competing financial interests.