Design an Efficient Multi-Epitope Peptide Vaccine Candidate Against SARS-CoV-2: An in silico Analysis

Introduction

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) which the cause of COVID-19 disease was first reported as a pneumonia epidemic in the Chinese city of Wuhan (Hubei province) on December 31, 2019, belongs to the Beta coronavirus genus.^1–3 Coronaviruses are positive-sense single-stranded RNA viruses. They belong to the order of Nidovirales and superfamily of Orthocoronaviridae. This superfamily has four genus including alpha, beta, gamma, and delta coronaviruses. Gamma and delta coronaviruses infect birds, while alpha and beta coronaviruses generally infect mammals such as humans. In immunocompetent individuals, they generally cause mild respiratory infections, such as common cold, while in some individuals, coronavirus infections cause serious diseases, such as SARS (Severe Acute Respiratory Syndrome) and MERS (Middle East Respiratory Syndrome) epidemics.

SARS-CoV-2 has a large mRNA genome, 26 to 32 kb in length, meaning a 5ʹ cap cover structure and a 3ʹ polyadenylated. This encodes several structural and nonstructural proteins. Among structural proteins, spike (S), envelope (E), membrane (M), and nucleocapsid (N) proteins are important in inducing immune responses.^3–5 Spike protein is the main tool for virus entry into the cells, which interacts with host cell receptor, angiotensin-converting enzyme 2 (ACE2). ACE2 is a metallopeptidase expressed in variant tissues, including alveolar epithelial cells, fibroblasts, endothelial cells, and enterocytes.^6–10 The N protein is an essential RNA-binding protein for transcription and replication of the viral RNA. It has significant roles in forming of the helical ribonucleoproteins during packaging the RNA genome, regulating of viral RNA synthesis during replication, transcription, and modulating of metabolism in the infected cells.^11–14 T cell responses against the S and N proteins of SARS-CoV virus are most dominant and long-lived than other structural proteins.¹⁵ E protein has an important role in the assembly of the viral genome.^16,17 The M protein plays a pivotal role in virus assembly, budding, and replication of virus particles in the host cells.¹⁸

SARS-CoV-2 is transmitted quickly and causes a considerable fatality rate, so that World Health Organization (WHO) reported over 14,562 550 cases globally and at least 9898 deaths because of this disease until July 21, 2020.¹⁹ It can be controlled when medical resources are sufficient but, there is currently no vaccine or approved treatment for this disease. One of the major challenges for scientists in dealing with the new coronavirus pandemic will be to get a useful and effective vaccine. For this reason, trying to develop an effective vaccine to control the virus has been the subject of research by many scientists around the world.^4,20-24 Previous studies on vaccine production against infectious diseases proved that achieving such a vaccine with conventional methods is time consuming and very expensive. As such, in silico prediction of the vaccine targets is very important because the targets can be selected with an open eye in a limited time and resources. Bioinformatic approaches are very helpful to identify the effective epitopes and developing vaccines. Several vaccine candidates for viral diseases have been reported based on this approach that include effective vaccines in Human papilloma viridea (HPV),²⁵ Ebola,²⁶ Zika²⁷ and MERS^28,29 viruses. There are few reports of COVID-19 vaccine.^30–32 They contained multiple cytotoxic T lymphocytes (CTL) and B-cell epitopes against several proteins of SARS-CoV-2 virus. However, those vaccines did not cover most of the immunodominant viral proteins. Therefore, to overwhelm these limitations, we designed a multi-epitope peptide vaccine consisted of S, M, N, and E viral proteins. The vaccine has appropriate physicochemical properties such as stability at room temperature, more antigenic capability, and no possibility of allergy. It consists of three epitopes from S protein and one epitope from each of M, N, and E proteins, respectively. Each of the epitopes has a high ability to stimulate humoral and cellular immune responses, especially the proper production of Interferon gamma (IFN-γ). The selected epitopes aligned between SARS-CoV-2 and SARS species. We used appropriate adjuvants in the vaccine structure to potentiate the immunogenicity of the antigens. Therefore, we also incorporated the potent adjuvants of N and C terminus of Flagellin of Salmonella enterica subsp. enterica serovar Dublin as a toll-like receptor (TLR) 5, a driven peptide from high mobility group box 1 (HMGB1) as HP-91, and human beta-defensin 3 (HBD-3) in the construct of the vaccine. The vaccine segments were connected to each other by appropriate linkers. Physicochemical properties, structural stability, and immunological characterizations of the vaccine were evaluated. Modeling, refinement, and validation were performed to access a high-quality three-dimensional (3D) structure of the vaccine protein. Docking evaluation showed an appropriate interaction between the vaccine and TLRs 3, 5, 8, and ACE-2. In silico cloning showed that the vaccine could be effectively expressed in E. coli. Totally, a potential vaccine candidate with proper immunological and stable physicochemical properties against SARS-CoV-2 was designed. It is expected the vaccine could be capable to protect humans from COVID-19 disease.

Methods

Protein Sequence Retrieval

The protein sequences of spike glycoprotein (accession number of QIC53213.1), nucleocapsid phosphoprotein (QHU79211.1), membrane glycoprotein (QIC53207.1) and envelope protein (QIC53215.1) were driven using NCBI (https://www.ncbi.nlm.nih.gov/) databases and saved in FASTA format for subsequent analysis.

Screening of Potential Epitopes

Screening of Major Histocompatibility Complex Class I (MHC-I) Epitopes

MHC-I humans alleles with 9 mer length were selected using IEDB (www.iedb.org) database through IEDB recommended 2.22 method and Net-MHC 4.0 online server (http://www.cbs.dtu.dk/services/NetMHC/). IEDB (Instructor/Evaluator Database) is a repository of web-based tools which predicts and analysis of immune epitopes. It uses a consensus method based on artificial neural network (ANN), stabilized matrix method (SMM), and Combinatorial Peptide Libraries (CombLib), if any corresponding predictor is available for the peptide sequence.^33–51 NetMHC 4.0 software is a free server for the prediction of peptide-MHCI binding affinity by gapped sequence alignment. Prediction methods of this server are based on alignments that include insertions and deletions and have significantly higher performance than those based on peptides of single lengths methods.³⁵

Screening of Cytotoxic T Lymphocytes (CTL) Epitopes

The prediction of CTL epitopes was done by an online server CTL Pred. This method is based on quantitative matrix (QM) and machine learning techniques, for example, artificial neural network (ANN) and support vector machine (SVM). The server also uses the consensus and combined prediction approaches. The consensus and combined prediction approaches are more specific and sensitive, respectively, in comparison with individual approaches such as ANN and SVM.⁵²

Screening of Human Leukocyte Antigen (HLA)-II Epitopes

IEDB database (IEDB recommended 2.22 method)^4353–56 (www.iedb.org) and RANKPEP online server (http://imed.med.ucm.es/Tools/rankpep.html) were employed for screening of HLA-II epitope. The database of IEDB was described previously. The RANKPEP server predicts MHC-II binding epitope by position-specific scoring matrices (PSSMs) that are structurally consistent with the binding mode of MHC-II ligands.⁵⁷

Screening of Linear B-Cell Epitopes

Prediction of linear B-cell epitope was done by IEDB⁵⁸ (Emini surface accessibility method) and BepiPred software. BepiPred 2.0 predicts the location of linear B-cell epitopes by a combination of a hidden Markov model and a propensity scale method. The epitope Threshold of this server was selected 0.5.⁵⁹

Selection of the Epitope Segments

The results of all predictions were collected and compared together, and the regions with the highest overlaps were determined. These immunodominant regions were employed for future analyses to finally select the most appropriate epitope domains for vaccine construct.

Alignment of Selected Epitopes Between SARS-CoV-2 and SARS

The alignment of the selected epitopes in SARS-CoV-2 and SARS was performed to understand the selected epitopes that are conserved in two viruses. The amino acid sequences of structural proteins of SARS were retrieved from the NCBI (https://www.ncbi.nlm.nih.gov/) database and their epitopes blasted using CLC sequence viewer 8.0.⁶⁰

In silico Analyzing of IFN-γ Inducing Epitopes

IFN epitope server (http://crdd.osdd.net/raghava/ifnepitope) was employed to determine the ability of IFN-γ production in the selected epitopes. This web server classifies MHC binder epitopes into IFN-γ inducing (positive numbers) and non-inducing IFN-γ (negative numbers) using several methods such as; motifs-based search, machine learning technique, and a hybrid approach. Accuracy of Best prediction based on hybrid approach in this software is 82.10%.⁶¹

Evaluation of Antigenicity

Antigenic properties of the predicted-epitopes were assessed by ANTIGENpro (http://scratch.proteomics.ics.uci.edu/) and VaxiJen v2.0 (http://www.ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen.html) servers. Web server of ANTIGENpro is the first sequence-based, alignment-free, and pathogen independent predictor, using protein antigenicity microarray data for predicting the protein antigenicity.⁶² VaxiJen is also the first server for the prediction of antigens. This server applies a new alignment-independent approach that is based on auto-cross covariance (ACC) transformation of protein sequences into uniform vectors of principal amino acid properties. Depending on the target organism (bacteria, virus, or tumor) accuracy of this server varies from 70% to 89%. The Threshold of this server was selected at 0.5.^63,64

Allergenicity Assessment of Predicted Epitopes

To accurately predict the possibility of allergenicity, the software of AllergenFP v.1.0 (http://ddg-pharmfac.net/AllergenFP/) was used. This is an online server that identifies allergens based on amino acid principle properties such as hydrophobicity, helix, and β-strand forming with accuracy of about 88%.⁴⁰

Vaccine Engineering, Evaluation of Physicochemical Properties, Antigenicity, and Allergenicity

According to the results of the previous steps, three epitopes from S protein and one epitope from each protein of N, E, and M were selected to be incorporated in the vaccine construction. The multi-epitope vaccine construct was joined to the carboxyl and amino terminals of the flagellin adjuvant. In addition, HP-91 and HBD-3 synthetic peptide adjuvants were respectively incorporated at the N- and C-terminal of the epitopes. The felagellin molecule was derived from Salmonella enterica subsp. enterica serovar Dublin. Various segments of the designed-vaccine connected by suitable linkers.

After designing the vaccine, several physicochemical parameters including molecular weight, theoretical isoelectric point (pI), total number of positive and negative residues, extinction coefficient, Instability index, half-life, aliphatic index, and grand average hydropathy (GRAVY) were computed using Expasy’s ProtParam at http://us.expasy.org/tools/protparam.html.⁶⁵ Antigenicity of the final vaccine construct was evaluated using vaxijen v2.0 and ANTIGENpro and allergenicity were assayed by AllergenFP v.1.0 server.

Tertiary Structure Development

Three-dimensional (3D) structural modeling of the designed construct was predicted using Phyre2 server (http://www.sbg.bio.ic.ac.uk/phyre2). It is a suite of a free web server for prediction and analysis of protein structure, function, and mutations.⁶⁶ The predicted model by Phyre2 was refined using Galaxy Refine web services. It successfully tested in CASP10 (Critical assessment of techniques for protein structure prediction) experiments. This server refines the whole protein with gentle and aggressive relaxation methods. This first reconstructs all side-chain conformations and repeatedly relaxes the structure with short molecular dynamics simulations after side-chain repacking perturbations.⁶⁷ Finally, Galaxy Refine showed five structures as refined models of vaccine structure. The RAMPAGE (http://mordred.bioc.cam.ac.uk/~rapper/rampage.php), and ERRAT (http://services.mbi.ucla.edu/ERRAT/) servers were used to validate the refined 3D structures obtained from Galaxy Refine web service.^68,69 Finally, the refined and high-quality 3D structure of the vaccine was observed by PyMOL software v2.1.1.PyMOL is an open-source that widely used for bimolecular function.⁷⁰

Prediction of Discontinuous B-Cell Epitope

Discontinuous B-cell epitopes were predicted from the 3D vaccine structure using ElliPro in IEDB database (http://tools.immuneepitope.org/tools/ElliPro). ElliPro is a useful research tool for identifying immune epitopes in protein antigens which implements Thornton’s method and with a residue clustering algorithm. The MODELLER program and the Jmol viewer allow the prediction and visualization of immune epitopes in a given protein sequence or structure. In comparison with six other structure-based methods that are using for epitope prediction, ElliPro performs the best prediction and gave an AUC value of 0.732, when the most significant prediction is considered for each protein.⁷¹

Molecular Docking with TLR 3, 5, 8, and ACE-2

The first, tertiary structure of the human TLR 3, 5, 8 and ACE-2 were obtained from PDB database (www.rcsb.org) with codes of 3J0A, 5GS0, 3w3g, and 1R42, respectively. Next, protein–protein docking of the vaccine structure (as a ligand) and each receptor was performed by CLUSPRO 2.0 online server (cluspro.bu.edu/login.php). CLUSPRO 2.0 is a fully automated and fast rigid-body protein–protein docking server. This server evaluates docking of two interacting proteins based on three techniques: first, the Fast Fourier Transform (FFT) correlation approach, second, clustering of the best energy conformations, third, refining the obtained model using short Monte Carlo simulations and the medium-range optimization method SDU.^72–74 The interaction between molecules was observed using Discovery studio 4.5.

Codon Optimization, in silico Cloning and Prediction of the mRNA Secondary Structure

The reverse translation of the designed-vaccine gene sequence was performed by Sequence Manipulation Suite (https://www.bioinformatics.org/sms2/rev_trans.html) to prepare a suitable vaccine sequence for cloning and expression in an appropriate expression system. The properties of sequence genes such as Codon Adaptation Index (CAI), GC content, and Codon Frequency and Distribution (CFD) have the key roles in attaining a high level of protein expression in the host were evaluated using GenScript online server (https://www.genscript.com/tools/rare-codon-analysis).²⁴ Finally, the restriction sites of NcoI and XhoI were, respectively, added to the N- and C-terminus of the vaccine DNA sequence using CLC Sequence viewer v8.0 (http://www.cacbio.com) to facilitate the cloning in E. coli expression system.

To predict the mRNA folding and secondary structure of the vaccine, the optimized DNA sequence was converted to possible RNA sequence using DNA<->RNA<->Protein tool (http://biomodel.uah.es/en/lab/cybertory/analysis/trans.htm). Then, the mRNA secondary structure was predicted using RNAfold (http://rna.tbi.univie.ac.at/RNAWebSuite/help.html) server. This server predicts the mRNA secondary structure thermodynamically and provides minimum free energy (ΔG Kcal/mol) for structure.⁷⁵

Results

Immunoinformatic Analysis

Protein S, M, N, and E were used to predict HLA-І binding epitopes (HLA-A, B, and C) using IEDB and NetMHC 4. CTL Pred server was used to predict CTL epitopes of SARS-CoV-2 structural proteins. HLA-II binding epitope (DP, DQ, and DR) from these proteins were predicted using RANKEP and IEDB servers. Prediction of linear B-cell epitopes was performed by BepiPred and IEDB. Finally, the obtained epitopes from the comparison of the above analyses were applied to future selection (Table 1).

Table 1 The Final Immunodominant Epitopes Were Selected from S, M, N, and E Proteins of SARS-CoV-2

Antigenic Ability Screening and Avoid Allergenicity

All selected peptides were submitted to the IFN epitope server for evaluation of their ability to induce IFN-γ. Allergenicity probability of epitopes evaluated by AllergenFPE valuation of epitope antigenicity was performed by ANTIGEN pro and VaxiJen v2.0.The six epitopes (three from S and one from each of E, M, N proteins) that had the highest ability in IFN-γ induction, low allergenicity, and potent antigenic ability were chosen to be used in the vaccine structure (Table 1).

Alignment of the Selected Epitopes Between SARS-CoV-2 and SARS Viruses

The alignment of the selected epitopes between SARS-CoV-2 and SARS viruses showed the selected epitopes were conserved between two species (Table S1).

Designing a Multi-Epitopes Vaccine Construct

Six epitopes with high scores which selected to be incorporated in the final vaccine construct were the sequences of 20–49, 418–446, and 505–534 from S, 51–71 from E, 360–388 from N and 169–198 from M protein. Two epitopes, sequences 418–446 and 505–534, of S protein were located in RBD portion. Besides, three immunopotent adjuvants Flagellin of Salmonella enterica subsp. enterica serovar Dublin, HP-91, and HBD-3 were also added to the vaccine structure. The vaccine pieces were linked to each other using a short linker sequence of LE dipeptide repeats. The final multi-epitope peptide vaccine was 485 amino acid residues (Figure 1). Figure S1 .

Figure 1 Schematic representation of the designed multi-epitope peptide based-vaccine. The vaccine consist of ten parts: Epitopes from structural proteins S, E, N, and M, adjuvants: Flagellin (in N-and C-terminus), HP-91 and HBD-3 that join to each other by linkers of repeat sequence of LE (A). Tertiary structure of the modeled multi-epitope vaccine construct (B). The 3D structure of the designed vaccine was predicted via homology modeling by Phyre2, then the best-predicted model was refined by Galaxy Refine and visualized using Discovery studio 4.5 software. N-and C-terminus of Flagellin is shown in green, HP-91 in gray, S epitopes in blue, E epitopes in red, N epitope in purple, M epitope in orange, HBD-3 in green, and linkers are shown in light pink.

Immunological and Physicochemical Properties of the Vaccine Structure

To evaluate the antigenicity of the whole protein vaccine, the “virus” option was chosen as a target organism. The antigenicity of 0.5585 (probable antigen) was estimated at 0.5% threshold for the virus model. Assessment of antigenicity by Antigen pro showed this vaccine is high antigenic with antigenicity of 0.927820. AllergenFP showed that the T-cell epitopes in the vaccine protein are non-allergen. The molecular weight (MW) and theoretical isoelectric point (pI) of the vaccine protein were computed as 53.102 kDa and 7.92, respectively. The predicted half-lives were calculated as 30 hours (h) (mammalian reticulocytes, in vitro), 20 h (yeast, in vivo), and 10 h (E. coli, in vivo). The instability index was 36.41, indicating that the protein vaccine was stable enough (Table 2)

Table 2 Analysis of the Physicochemical and Immunological Properties of the SARS-CoV-2 Vaccine

Tertiary Structure Modeling, Refinement, and Validation

The primary 3D model of the protein vaccine construct was predicted using Phyre2. Then, the selected Phyre2 model was refined by GalaxyRefine software. The quality of the designed-vaccine was evaluated using the Ramachandran plot in the RAMPAGE server and characteristic atomic interaction in the ERRAT server. The RAMPAGE results of the final model showed that the majority of residues (95.9%) are located in the favored region and 2.1% are allowed, and only 2.1% of residues are the outlier. The ERRAT result showed that the refined model has the ERRAT score 89.362 (Figure S2). The outputs of the Ramachandran plot and ERRAT indicated that the refined 3D structure is good and therefore, can be utilized as a reliable model for further evaluations.

Conformational B-Cell Epitope Identification

The tertiary structure of the designed vaccine was used as an input for conformational (discontinuous) B-cell epitope prediction via ElliPro in IEDB. From 485 amino acid residues, 246 were defined as discontinued B-cell epitope (Table 3).

Table 3 Conformational B-Cell Epitopes Identified in the Refined Tertiary Structure of the Multi-Epitope Vaccine Using ElliPro

Identifying Binding Sites and Protein–Protein Docking

ClusPro server has been used to perform the docking of the vaccine with TLR3, 5, 8, and ACE-2 molecules. Cluspro generated 30 models for each interaction. The docked model was selected based on most atoms interacted in the vaccine and each of the receptors (Figures 2–5).

Figure 2 Docking model (cartoon representation) of human TLR3 in complex with the vaccine obtained using Cluspro server. TLR3 protein is shown in chocolate and vaccine was colored from N-to C-terminals (N-flagellin is shown in blue and C-flagellin is shown in red). As it shows some parts of two S, M, N, and E epitopes, HBD-3, and HP-91 interacted with TLR3. To more visualized interaction points, some of the interacting residues of the vaccine and TLR3 are magnified in 20 Angstrom. Docked model was visualized via Discovery studio 4.5 software.

Figure 3 Docking model (cartoon representation) of human TLR5 in complex with the vaccine obtained using Cluspro server. TLR5 is shown in chocolate and vaccine was colored from N-to C-terminals (N-flagellin is shown in blue and C-flagellin is shown in red). As the figure shows some parts of Flagellin, S-epitope and HP-91 interacted with TLR5. In order to more visualized interaction points, some of the interacting residues of the vaccine and TLR5 are magnified in 20 Angstrom. Docked model was visualized by Discovery studio 4.5.

Figure 4 Docking model (cartoon representation) of human TLR8 in complex with the vaccine obtained using Cluspro server. TLR8 protein is shown in chocolate and vaccine was colored from N-to C-terminals (N-flagellin is shown in blue and C-flagellin is shown in red. As the figure shows some parts of S, M, N, and E epitopes, HBD-3 and HP-91 interacted with TLR8). In order to more visualized interaction points, some of the interacting residues of the vaccine and TLR8 are magnified in 20 Angstrom. Docked model was visualized using Discovery studio 4.5.

Figure 5 Docking model (cartoon representation) of human ACE-2 in complex with the vaccine obtained using Cluspro server. ACE-2 protein is shown in chocolate and vaccine was shown in purple. As the figure shows RBD epitopes of vaccine interacted with ACE-2. In order to more visualized interaction points, some of the interacting residues of the vaccine and ACE-2 are magnified in 20 Angstrom. Docked model was visualized using Discovery studio 4.5.

In silico Codon Optimization of the Vaccine Construct and Prediction of the mRNA Secondary Structure

The reverse translation of the protein vaccine into a nucleotide sequence was performed simultaneously using the Sequence Manipulation Suite server to express high-level protein in E. coli. Codon Adaptation Index (CAI), GC content, and Codon Frequency and Distribution (CFD) were evaluated using GenScript online server. A CAI of the vaccine optimized nucleotide sequence was 1. The GC content of the structure was in the ideal range of 58.73% (30–70%), and CFD 100 indicating the effective expression of the protein in the host. To clone the gene into E. coli vectors, NcoI and XhoI restriction sites were added into the N- and C-terminals of the sequence using CLC Sequence viewer v.8. Finally, the vaccine gene was inserted into the PET-28 vector (Figure 6). Minimum free energy of the vaccine was determined using RNA fold server, the optimized mRNA construct showed ΔG value of −520 kcal/mol. Figure S2 illustrates the predicted secondary structure of the vaccine.

Figure 6 Evaluation of the three important parameters of the codon-optimized gene to express high-level protein in E. coli. (A) CAI of the gene sequence was 1. It is noted that a CAI of > 0.8 will be considered as good for expression in the selected host. (B) Average GC content of the sequence was 58.73%. (C) Codon frequency distribution (CFD) value of the gene sequence is 100. A CFD equal to100 supports maximum protein expression in the desired host. (D) Insertion of the vaccine gene in the PET28 vector by restriction enzymes of NcoI and XhoI.

Discussion

Today, there is an urgent need to design effective vaccines to stop the indescribable spread of COVID-19. In light of recent advances in bioinformatic approaches, we can identify immunodominant T- and B-cell epitopes and design a potential vaccine to prevent the disease. We used an in silico analysis to design a potent multi-epitope peptide vaccine against SARS-CoV-2. The vaccine contains 485 amino acids which constructed from three specific epitopes from S (two epitopes from RBD section (and one epitope from each of the structural proteins of E, M, and N)). These proteins have essential roles in the infection of host cells and host immune modulation. Therefore, it is expected that the vaccine will have a high ability to induce the production of neutralizing antibodies by B lymphocytes and the production of IFN-γ by (T helper) Th and CTL cells.

Adjuvants are essential components of multi-epitope vaccines because they increase the immunologic properties of the vaccine structures. We used Flagellin of Salmonella enterica subsp. enterica serovar Dublin, a TLR5 agonist, in two N and C terminus of the vaccine construct. Flagellin induces innate immune effectors such as cytokines and nitric oxide,^76,77 activates TLR5-positive dendritic cells (DCs), neutrophils, and epithelial cells^78–81 and stimulates the activation of adaptive immune responses mainly Th2-type and IgA production.^{78,79,82–85} Intranasal administration of Flagellin stimulates the signaling of TLR5 in lung epithelial cells and pneumonocytes.⁸⁶ It was used broadly as an adjuvant in vaccine structure designed against viral infections such as influenza^85,87,88 and HPV.^25,89 In addition, we incorporated HP-91 and HBD-3 in the final vaccine construct. The peptide of HP-91 derived from B-box domain amino acids of 89–108 from HMGB1 protein. This peptide induces high levels of IL-2 and IL-15, increases secretion of IL-12 and IFN-α in human DCs and augmented T cell activation.⁹⁰ Also, it activates DCs independent of TLR2, 4, and 9, and MyD88 pathways.⁹¹ This is a potent immune adjuvant for inducing cellular and humoral immune responses and has been used in vaccines against viral infections such as HIV and HPV.^90–93 HBD-3 is the third adjuvant in our vaccine, used as adjuvant to viruses such as influenza and MERS-CoV.^28,94 This peptide blocks viral fusion using creating a protective barricade of immobilized surface proteins.⁹⁴ It activates APCs via TLR1 and TLR2⁹⁵ and induces IL-22,⁹⁶ TGF-α^97,98 and IFN-γ.^99,100 It has been implicated in the chemotaxis of immature DCs and T cells via its interaction with chemokine receptor 6 (CCR6) and the chemotaxis of monocytes via interaction with CCR2,¹⁰¹ As well as this peptide activates natural killer (NK), promotes and activates myeloid DCs directly and dependent NK cells activity.^95,100 To link different pieces of the vaccine, a repeated of a dipeptide linker of LE was used. This linker improves the expression of vaccine protein via the decreasing of a pI.^102,103

An efficient vaccine should not only have stimulating ability immune response but also possess good physicochemical properties during production, formulation, storage, and consumption. According to the results of bioinformatic predictions, the designed vaccine was stable with a stability index of 36.41 and had a pI of 7.92. Conformational B-cell epitopes have a central role in induction humoral responses. The accessibility of a significant number of B-cell epitopes in the vaccine molecule indicates the high ability of this structure to stimulate B lymphocytes`. The binding affinity of the vaccine with the immune receptors (TLR3, 5, and 8) is necessary to effectively transport vaccine protein into antigen-presenting cells.

Moreover, it has been reported that TLR3 induces a strong type I IFN-dependent antiviral response in murine Coronavirus Infection¹⁰⁴ and TLR8 has the inhibitory roles against viruses.^105,106 These three TLRs were used in designing in previous bioinformatic-based vaccines against SARS-CoV-2.^30,107,108 The results of docking analysis showed the vaccine protein properly interacts with TLR3, 5, and 8. In addition, docking results of the vaccine and ACE-2 reveal the RBD epitopes of the vaccine can efficiently interact with ACE-2. This interaction indicates the vaccine will be attached to ACE-2 and therefore, induces humoral and cellular responses against SARS-CoV-2.

The codon optimization and in silico cloning illustrated the vaccine can efficiently transcript and translate in E. coli. Alignment of the selected epitopes of structural proteins showed these epitopes are conserved between SARS-CoV-2 and SARS, so it is also possible that the vaccine may induce neutralizing antibodies against SARS coronavirus.

Conclusion

In this study, using a variety of bioinformatic methods, we developed a multi-epitope subunit vaccine against the SARS-Cov-2 virus. The results of this study showed that it could be possible to predict vaccine candidates against new emerging viral diseases such as COVID-19 with the help of reverse vaccinology. The present study, with its effective vaccine design against SARS-CoV-2, showed that bioinformatic approaches could help to develop effective treatments for other emerging infectious diseases in a short time and at low cost. However, the in silico results of this study need to be verified using laboratory and animal models.