Reverse vaccinology approach to design a vaccine targeting membrane lipoproteins of Salmonella typhi

Abstract Typhoid fever caused by Salmonella is one of the major health issues worldwide, resulting in millions of cases and has very high rates of morbidities. The therapeutic approaches need to be updated for the effective elimination of the bacterial pathogen. The designing of the multiepitope vaccine against Salmonella using comparative proteomics and reverse vaccinology has covered up all the epitopes that induce sufficient immune responses in the host body. Out of the 4293 proteins, 15 outer membrane proteins have been selected based on their antigenicity, low transmembrane helix (<1), and virulence-associated factors. With the help of the reverse vaccinology approach, the epitopes of MHC Class I, Class II, and B-cell with antigenic, low toxicity, and that have the potential to generate immunogenic response have been identified. Based on the comparative analysis of all the epitopes, a multiepitope-based construct has been designed. Based on physicochemical properties and docking scores for HLA and TLR4, the VC5 construct has been selected, and the molecular dynamic simulation studies have confirmed their interaction. The dissociation constant of the VC5 and TLR4 was found to be 3.1 x 10−9. Different immune cell activation has been analyzed, representing the potentiality of the VC5 construct as an effective vaccine target. In silico cloning of VC5 in pET28a has also been performed, which requires experimental validation. Therefore, the present study designs a multi-epitope vaccine VC5 targeted to the membrane lipoproteins of Salmonella typhi. Communicated by Ramaswamy H. Sarma


Introduction
With the advent of technologies and applications, the postgenomics era offers in-silico screening of proteome information for vaccine designing. The causative organism no longer needs to be grown as the most probable antigenic agents could be predicted using immunoinformatic tools. It also renders a very cost-effective and time-saving approach to predict the antigenic agents. Salmonella, Gram-negative bacteria, is the globally leading causative agent of food and water-related infections with high mortality rates. Typhoid fever has become a severe health problem worldwide as it is known to create 33 million cases and results in 500,000 deaths annually (Garmory et al., 2002). In humans, the Salmonella is related to two types of infections, including gastroenteritis and typhoid fever (Johnson et al., 2018). It has gained multi-drug resistance (MDR) against most antibiotics and drugs. The resistance mechanism could be attributed to the MDR proteins that efflux almost all of the antibiotics/drugs out of the bacterial system (Jebastin & Narayanan, 2019). Various efforts have been made for designing the vaccine against Salmonella, including attenuated strains, whole-cell killed, subunit vaccines (Garmory et al., 2002) and targeting two cytoplasmic proteins DnaK and GroEL (Verma et al., 2018). The focus of the present study lies in exploring the lipoproteins of the Salmonella typhi.
The LPS associated vaccines have been successfully tested in various animals, but the severe side effects associated with them raises the need for some alternative options (Solanki et al., 2019). The protein-based vaccines focus on the immunogenic proteins, including the lipoproteins present on the surface and outer membrane proteins (OMPs).
To develop the multiepitope-based vaccine against Salmonella, reverse vaccinology and subtractive proteomics approach were used. For the present study, the proteome of Salmonella typhi (ATCC-700931/Ty2, organism ID-90370) was explored to determine the antigenic proteins of the bacteria. The focus was on the lipoproteins that are secreted out or destined for the outer membrane of the bacteria. Those proteins have been screened for predicting the B-cell and Tcell epitopes with their MHC alleles. The epitopes of the bacterial proteins were screened for their antigenicity, global coverage of the MHC alleles (Tahir Ul Qamar et al., 2020). The most potential epitopes were used to design a multiepitope-based vaccine with suitable linkers and adjuvants. Multi-epitope-based vaccines are beneficial compared to single-epitope-based vaccines because they are cost-effective, time-saving, and show high specificity (Tahir Ul Qamar et al., 2020). Furthermore, they are known to induce both the humoral and cellular immune responses due to the presence of T-cell and B-cell epitopes . The designed vaccine construct has been studied using docking and molecular dynamics simulation. The vaccine has also been tested for being efficient in in-silico cloning for ensuring the expression of the final vaccine construct. The binding energy and the dissociation constant of the VC5 and TLR4 were predicted. On the other hand, the immune simulation was also performed for the vaccine construct, when administered at different time intervals showed very effective antigen clearance.

Proteome retrieval for Salmonella
To identify potential vaccine candidates, the ATCC-700931/ Ty2 (organism ID-90370) strain of Salmonella typhi was selected. The whole proteome for this strain was downloaded from the UniProt server. These proteins have been subjected to different servers to identify the potential antigenic targets that have a role in the virulence of the bacteria. The different methods used in the present study are summarized in Figure 1.

Identification of the signal peptide
In the present study, bacterial proteins have been targeted to analyze the potential targets, and to screen this, the SignalP 4.1 software was used (Petersen et al., 2011). This software predicts the presence of signal peptide cleavage sites in the proteins. The neural network-based combination efficiently predicts the cleavage sites and a signal/non-signal peptide in the proteins under study. The proteins with a signal peptide would form a basis for further analysis for the vaccine targets.

Analysis of the lipoproteins of Salmonella typhi
The lipoproteins are the most immune-stimulating sites in the bacteria. To shortlist the lipoproteins out of the 412 exported/secretory proteins, we used the LipoP 1.0 server (Juncker et al., 2003). This server distinguishes between lipoprotein signal peptides, other signal peptides, and N-terminal membrane helices, especially in Gram-negative bacteria. These lipoproteins would be taken up for further analysis of antigenicity and immunogenicity.

Screening for the transmembrane helices in the proteins
The transmembrane helix of a protein explains the nature and anchoring strength of the protein in the bacterial cell membrane (Solanki et al., 2019). The ease of purifying a protein for in-vivo studies increases with the decrease in transmembrane segments. Hence, the proteins with the minimum amount of alpha-helix were the main targets. The 76 shortlisted lipoproteins were further analyzed for the trans-membrane helices present in their structure. This was achieved via TMHMM v.2.0 (Trans-membrane helices hidden Markov Model for Topology Prediction) (Krogh et al., 2001), and HMMTOP 2.0 server (Tusn ady & Simon, 1998).

Determining the location of the lipoproteins
The Gram-negative bacteria have some major protein locations: proteins of extracellular space, outer membrane proteins, cytoplasmic proteins, periplasmic proteins, and proteins of the inner membrane (C. S. Yu et al., 2004). For the construction of vaccine, membrane proteins/extracellular proteins were considered as a potential target. For this, CELLO v.2.5 (C. S. Yu et al., 2004) and PSORTb v.3.0 analysis (N. Y. Yu et al., 2010) were performed. Only those proteins were taken up for further analysis, which shows localization towards the outer membrane or is in the extracellular region.

Analyzing the adhesion probability and antigenicity of the proteins
Targeting the adhesion probability of bacteria will interfere with its interaction with the host. Hence the adhesion probability of the suitable protein candidates was analyzed using the Vaxign server. This server predicts the vaccine targets based on the protein sequences (He et al., 2010). This server analyzes the protein localization, adhesion probability, transmembrane part, conservation to pig/mouse/humans, and binding of the epitopes to the MHCI and MHCII. For analyzing the antigenicity of the proteins, the VaxiJen v2.0 server was used. This server predicts the antigenicity of a pathogen based on the virulence and the ability of the pathogen to produce defensive responses (Doytchinova & Flower, 2007).

Analysis of the virulence factors and essential genes for the survival of the bacteria
Identification of the genes responsible for the survival of the bacteria and having significance in virulence is essential. The virulence factor database (VFDB) is a source that provides information about the virulence factors of bacterial pathogens. It measures the pathogenicity and the probability of a bacteria to cause disease (Liu et al., 2019). Database of essential genes (DEG) is a server with all the information about the essential genes among prokaryotes and eukaryotes (Luo et al., 2014). These genes are requisite for the survival of a pathogen. Hence for the prediction of vaccine candidates, the OMPs were also shortlisted using VFDB and DEG.

Prediction of the epitopes of T-Cell MHC Class I
After short-listing genes involved in the virulence and survival of the bacteria, the next step was to analyze these genes' ability to stimulate the host's immune system. Thus, the 15 proteins associated with virulence were subjected to the different software to identify their MHC-I interacting epitopes. The proteins were subjected to NetMHCpan3.0 (Nielsen & Andreatta, 2016), IEDB MHC-I prediction server (Reynisson et al., 2020), and the NetCTLpan (Stranzl et al., 2010). The software predicts the MHC I HLA interacting alleles by analyzing the affinity scores, IC50 value, and percentile rank.

Prediction of the epitopes of T-Cell MHC Class II
The next step was to identify the potential epitopes of proteins that could activate the MHC Class II. For this, the IEDB server was used to predict the epitopes interacting with Tcell MHC Class II (Jensen et al., 2018). This software also predicts the MHC II HLA interacting alleles by analyzing the affinity scores, IC50 value, and percentile rank.

MHC cluster analysis
MHC cluster analysis has been performed by using the MHCcluster v2.0 server (Thomsen et al., 2013). The functional relationship between the epitopes and the HLA alleles is predicted as a heat map. It was important to study as it indicates the frequency of alleles that show maximum binding with the epitopes. The filtered epitopes would have the potential to interact with different HLA alleles, hence enhancing the host's immune response.

B-cell epitope prediction
Using subtractive proteomics, the proteins were filtered based on their capability to induce the humoral immunity of the host. This was tested by analyzing the interaction of the epitopes with B-cells. For this purpose, ABCPred (Saha & Raghava, 2006) and IEDB BepiPred 1.0 server was used. Only those epitopes were taken into consideration which was predicted by both servers.
2.12. Analyzing the allergenicity, antigenicity, and toxicity of the epitopes The epitopes predicted for MHC Class I, Class II, and B-cells were further screened for their allergenicity, antigenicity, and toxicity. For allergenicity, the epitopes were subjected to AllergenFP v.1.0, and their antigenicity was analyzed using Vaxijen 2.0 server (Doytchinova & Flower, 2007). For toxicity prediction, the ToxinPred server was used (Gupta et al., 2013). Only those epitopes were considered, which were predicted as non-allergenic, antigenic, and showed nontoxic behavior.

Comparative analysis of epitopes and constructing model vaccine
As per the comparative analysis of the MHC I, MHC II, and Bcell interacting epitopes, the multi-epitope-based vaccine construct was designed. Non-overlapping epitopes were taken for designing the vaccine construct, and the epitopes were joined using different linkers (HEYGAEALERAG, GGGS, and EAAAK) (Solanki et al., 2019). The vaccine constructs were joined randomly with the help of these linkers. The vaccine constructs were modeled using the Phyre2 server. This server is a good tool to predict and analyze the 3D structure of protein (Kelley et al., 2015). The tertiary structure of the VC5 construct was predicted after refining the structure with GalaxyWeb (Mirela-Bota et al., 2021) and analyzed by the Ramachandran plot of the construct.

Physicochemical behavior analysis
Physico-chemical properties of the construct have a significant change in the stability and immunogenicity of the vaccine (Mahdevar et al., 2021). The toxicity, allergenicity, antigenicity, and hydropathicity of the vaccine construct were analyzed using the ToxinPred (Gupta et al., 2013), AllergenFP, Vaxijen, and Expasy ProtParam server (Solanki et al., 2019). It analyzes the pI value, molecular weight, solubility of the protein, and hydropathicity GRAVY values (Wilkins et al., 1999).

Molecular docking and molecular dynamics simulation
Molecular docking of the vaccine constructs with different alleles, and TLRs was performed using the PatchDock server (Schneidman-Duhovny et al., 2005). Finally, the binding energy and the dissociation constants were also predicted using the PRODIGY software (Xue et al., 2016). This software also identified the list of amino acid residues that interact in the complex. Molecular dynamics simulation (MDS) was performed for 50 ns using Gromacs software, and 10 ns using Desmond module of the Schrodinger 2019 as per published methods (V. Solanki et al., 2019;Tiwari, 2021).

Evaluating the immunogenicity of the vaccine construct
The C-IMMSIM (http://kraken.iac.rm.cnr.it/C-IMMSIM/) software was used to analyze the immune responses in the host after administering the vaccine. This server is developed based on the Celada-Seiden model, which predicts the immune simulation after administering the vaccine construct. The server simulates bone marrow, thymus, and lymph nodes, the three major components of the mammalian immune system (Rapin et al., 2010). It simulates various types of immune cells such as HTL, B-cells, NK cells, CTL, dendritic cells, and immunoglobulins (Tahir Ul Qamar et al., 2020). The immune simulation was performed using the default parameters as per our published protocol . Four injections were administered using the C-ImmSim server at an interval of 1, 84, 336, and 504 time-step where each timestep is 8 hours, and the bacteria with selected proteins were injected at 504. The different immune responses in the host after injection of the VC5 were analyzed from different plots. The 15 shortlisted lipoproteins were also individually analyzed for the immune simulation to identify the 'true-multiepitope' nature of the vaccine construct.
2.17. In-silico cloning of the designed vaccine construct To improve the heterologous protein production in E. coli strains Java codon Adaptation Tool (JCAT) was used (Grote et al., 2005). The vaccine construct was reverse translated and adapted for codon usage. Some of the sites, including rho-independent transcription terminators, prokaryotic ribosome binding sites, and few restrictions were avoided. The gene sequence of the final vaccine construct was cloned in E. coli Pet28a vector by Snapgene server to confirm the vaccine construct's expression. The reason for selecting pET-28a is that it is a very efficient vector to express diverse peptides, presence of the T7 promoter and the appropriate restriction enzymes (Safavi et al., 2019) that help in high yields of the required peptide construct. It has also been reported that the proteins ranging from 14 to 51 kDa are highly expressed in the pET-28a vector (Safavi et al., 2020).

Collection of proteome data
The ATCC-700931/Ty2 strain of Salmonella typhi (organism uniport ID-90370) was retrieved from the central repository of the protein data, UniProt. This proteome has 4293 proteins, and all these proteins were used for further analysis.

Signal peptide detection for selecting the secretary proteins
The SignalP 4.1 software predicted the presence of the signal peptide in the proteins. Out of the 4293 proteins of Salmonella, 412 proteins showed the presence of the signal peptide. This signal peptide present on the N-terminal of the proteins is essential for the transport of the proteins to the outer membrane/periplasmic surface. This subtractive proteomics approach was used to shortlist the 412 proteins for further analysis.

Analysis of the lipoproteins of Salmonella typhi
The LipoP 1.0 server was used to analyze the lipoproteins among the secretory proteins. Out of 412 secretory proteins, 76 were found to be lipoproteins. The reason for focusing on bacterial lipoproteins is that they are very well known for their role in bacterial physiology and virulence-associated factors (Z€ uckert, 2014). These 76 lipoproteins are screened for their location, whether they are present on the membrane or periplasmic space.

Analysis of the alpha-helical content in the lipoproteins
Out of the 76 proteins, after analyzing with TMHMM and HMMTOP servers, only those proteins were selected, which showed a score of 1. This score indicates the presence of 0 or 1 trans-membrane part present in the structure of the protein. A total of 72 proteins with a score of 1 was shortlisted for further analysis (Table 1). This analysis was necessary to perform to have an idea about the ease of extracting the potential vaccine protein for in-vivo validation of the findings.

Confirming the localization of proteins to shortlist the outer membrane proteins
Out of the major localization sites of the proteins, the target proteins were those which are localized on the outer membrane space/in the extracellular region. CELLO and PSORTb tools were used to predict the location of these proteins. Out of 72 proteins shortlisted for the minimum number of alpha-helical regions, only 65 proteins showed localization at the outer membrane or extracellular region (Table 1). These 65 proteins were the primary candidates to analyze the antigenic and immunogenic properties.

Analysis of the antigenicity and adhesion probability of various lipoproteins
The proteins that were localized in the outer membrane and the periplasmic space were subjected to their antigenicity analysis. Based on the analysis of the Vaxign and VaxiJen v2.0 server, 30 proteins were shortlisted. Only those proteins were taken into consideration whose adhesion probability was greater than 0.5, and that showed no similarity with humans, mice, and pig protein. These proteins also had 0-1 transmembrane helices in their structure (Table 1). Based on the results of these two servers, the 30 shortlisted proteins were used for further analysis.
3.7. Confirmation of the essential genes and the virulence-associated factors 15 proteins were shortlisted after the analysis of virulenceassociated factors and the essential genes. According to the DEG server, the protein candidates with minimum e-value indicate the genes essential for the survival of the bacteria. Similarly, the VFDB server predicted the proteins that can induce an immune response in the host (Table 1). The proteins that were predicted by both the servers were taken for further analysis.

Prediction of the cytotoxic T cells epitopes with potential antigenicity, no-toxicity, and no allergenicity
Identifying epitopes that show an interaction with MHC Class I is important because only those epitopes would be effective in designing a suitable vaccine. The prediction of MHC Class I was done using NetMHCpan3.0, IEDB MHC-I prediction server, and NetCTLpan. Only those epitopes were shortlisted, showing a high binding affinity with an IC50 value of <50nM, and the overall percentile rank <0.1. These epitopes were used for vaccine designing, which were predicted at least by two servers. The antigenicity, allergenicity, and toxicity of these epitopes were also tested ( Table 2). Further analysis was only performed with the epitopes that showed antigenicity >0.5, non-allergic and non-toxic, respectively.

Prediction of helper T cells epitopes with their antigenicity, no allergenicity, and no toxicity
The interaction of different epitopes with T-Cell MHC Class II was also analyzed using the IEDB server. Epitopes whose IC50 value was <50 nM and the percentile rank was below one were analyzed further. The IC50 value indicates the binding affinity of the epitope with alleles. The lesser the IC50 value greater is the epitope's binding affinity with the allele (Table 3). Hence, the more effectively the epitope would be able to initiate the host immune response.

MHC cluster analysis of the selected epitopes
The MHC I and MHC II finalized epitopes were further analyzed by the interaction of MHC alleles. The resultant heat maps showed the interaction efficiency of the alleles and the  epitopes where red and yellow color represents the strong and weak interaction, respectively ( Figure 2). The epitopes that strongly bind to the MHC cluster would have the capability to induce a strong immune response when interacting with the host.

Prediction of B-cell epitopes with antigenicity, notoxicity, and no allergenicity
A vaccine must have the ability to activate both humoral and cell-mediated immunity after its administration. Hence, The Outer membrane lipoproteins were shortlisted using IEDB software. The HLA alleles selected for this analysis were those which cover the maximum population. The epitopes with their interacting HLA alleles along with the antigenicity and toxicity scores have been mentioned in the table. The Outer membrane lipoproteins were shortlisted using IEDB, NetMHC and NetCTLpan software. The HLA alleles selected for this analysis were those which cover the maximum population. The epitopes with their interacting HLA alleles along with the antigenicity of the respective epitope has been listed below. the epitopes were screened for their interaction with the Bcell epitopes. The ABCPred and IEDB BepiPred server was used for screening. Only those epitopes were taken for further analysis, which was predicted by both the servers (Table 4). Figure 3 represents the epitopes of all the 15 proteins that were predicted by the software. The Outer membrane lipoproteins were shortlisted using IEDB software. The HLA alleles selected for this analysis were those which cover the maximum population. The epitopes with their interacting HLA alleles along with the antigenicity and toxicity scores have been mentioned in the table. Figure 3. Identification of B cell epitopes of antigenic Proteins (UNIPROT ID) using Bepipred linear epitope server. The probable epitopes of the 15 outer membrane lipoproteins have been predicted. The yellow region represents the amino acid range for the probable epitope, and the green region represents the non-epitopes.  The Epitopes predicted in the MHC Class I/II and B-cell analysis were further analyzed for their hydrophobicity. This helps in understanding the location of a particular epitope in the structure of the protein. Six vaccine constructs were designed by combining the epitopes randomly and those constructs were analyzed for their physico-chemical properties. The molecular weight, pI value, and GRAVY analysis have been represented in the table.
3.12. Determining the allergenicity, antigenicity, and toxicity of the epitopes AllergenFP v.1.0, Vaxijen, and ToxinPred were used for determining allergenicity, antigenicity, and toxicity, respectively, for the epitopes known to interact with the MHC Class I, Class II, and B-cells. It is important to study all these aspects that make potential vaccine construct non-allergenic, highly antigenic, and non-toxic. Only those epitopes were taken to build up the final vaccine construct, which showed an antigenicity of >0.5 and were non-allergic and non-toxic.

Comparative analysis of the epitopes and construction of a model vaccine
The model vaccine construct was designed by randomly joining the MHC I, MHC II, and B-cell epitopes using different linkers (HEYGAEALERAG, GGGS, and EAAAK). Table 5 represents the comparative analysis of various proteins for their epitopes. Only non-overlapping epitopes were taken for designing the vaccine. A total of 6 vaccine constructs were designed (Table 6), and these constructs were subjected to different screening (antigenicity, allergenicity, physiological properties, and toxicity). Figure  4 represents the secondary structure of the final vaccine construct. On the other hand, the 3D structure of the vaccine construct was predicted using the Phyre2 server. The Phyre 2 server predicted the protein folding and different ligand binding sites for the vaccine construct ( Figure 5A). Out of the six constructs, VC5 showed the most suitable folding and stability. The Ramachandran plot of the vaccine construct showed 90.5% of the structure in the allowed and favored region ( Figure 5B).

Analysis of different physicochemical behavior of VC5 construct
The toxicity, allergenicity, antigenicity, and hydropathicity value of the vaccine construct was analyzed using ToxinPred, Figure 4. Prediction of the secondary structure of vaccine construct VC5 using PESIPRED server. The yellow, pink, and grey colours represent the strand, helix, and coiled part in the VC5 construct, respectively. AllergenFP, Vaxijen, and Expasy ProtParam server, respectively. The VC5 construct showed a molecular weight of 23,805Da, a pI of 4.6, and an instability index of 50. Vaxijen analysis predicted it as an antigen with a score of 1.6518. The À0.786 hydropathy value of the VC5 indicates the hydrophilic nature of the construct.

Molecular docking and molecular dynamics simulation confirms the interaction of VC5 and TLR4
The molecular docking analysis of the vaccine construct was performed with the help of the PatchDock server. The docking results were analyzed for their different HLA alleles, and the complex with the lowest global energy was taken for further analysis. Here, the global energy represents the binding affinity of the VC5 construct with the alleles. In this case, the lowest global energy was represented by the interaction of the VC5 and the 2Z63-TLR4 with a value of À47.17 ( Figure  6). This effective binding would result in the initiation of both the cell-mediated and humoral immunity of the host. The binding energy and the dissociation constant of the VC5 and TLR4 were predicted to be À11.6 kcal/mol and 3.1 Â 10 À09 M at 25 C by PRODIGY server. Negative binding energy indicates favorable binding of the vaccine and our TLR receptor. Table 3, supplementary material, represents the interacting amino acid residues of the VC5 and TLR4. The molecular dynamics simulation till 50 ns showed that the VC5 has a very stable interaction with TLR4 (Figure 7).

Immune simulation of the vaccine construct
The immune simulation results carried out using C-ImmSim cover up the interaction of the construct with both the B-cell epitopes and HLA class I and class II with T-cell receptors.
The C-ImmSim model illustrates the responses of the host immune system in the presence of the antigen. Injecting the VC5 vaccine showed an elevated level of IgM ( Figure 8A) after the first dose. The second and third doses showed the clearance of the antigen within a very short span. After the injection, the IgM titer was observed to be more as compared to the IgG titer. Similarly, for B-cells ( Figure 8B), the memory B-cell titer was more than the other B-cell population. Among the T-cell population, the T-helper ( Figure 8D) response was more dominant. The level of the cytotoxic Tcells ( Figure 8D) showed a significant increase in the population and remained constant during further doses. The booster doses were also given, showing a rise in the IgG and IgM concentrations. It was also observed that when the bacteria were injected at six months (504 time step) it was eliminated in a very short span. IgG and IgM's titer values were found to be elevated along with the presence of the Thelper memory cells and the T-cytotoxic memory cells. On the other hand, the antigen elimination was highly enhanced in the following booster doses as well. All the 15 lipoproteins were also individually analyzed to determine their immunestimulating properties. This analysis confirmed the 'true multiepitope' nature of the vaccine construct VC5 because the proteins, when injected individually, showed an effective stimulation of the immune system. The immune simulation responses of the 15 lipoproteins with their protein ID have been shown in Figure 1, supplementary material. Hence, the final vaccine construct VC5 effectively induces a sufficient immune response in the host cells.

In-silico cloning of VC5 vaccine construct
In in-silico cloning of the VC5 construct, JCAT was used as the construct was translated and adapted for codon usage.
Codon optimization analysis showed a total of 61.32% of GC content. The CAI value of 1.0 indicates the high expression of the clone in the E. coli cells. Figure 9 represents the clone Figure 6. The docked complex of vaccine constructs VC5 with the human TLR4-complex. The docking has been performed using the Patch dock server. The blue structure represents human TLR4, and the red one represents the VC5 vaccine construct.   in between the HindIII and BamHI restriction sites at 3 0 and 5 0 respectively.

Discussion
The health issues caused by Salmonella are one of the major health problems resulting in mortality and morbidity globally. The current vaccines that contain attenuated and heatkilled Salmonella strains need to be updated about the postgenomics era. A study has shown that the current live-attenuated vaccine shows a very low immunogenic response in young children. On the other hand, the emerging multi-drug resistant strain is also a serious concern. Multiepitope vaccines cover up all the antigenic epitopes of the bacterial proteins. The proteins have been analyzed for their essential role in the survival of the bacteria, their role in the virulence of the bacteria, antigenicity, and the content of alpha-helical segments of the proteins. A total of 15 proteins qualified as potential targets for designing the vaccine construct. These proteins were shortlisted based on the reverse vaccinology and subtractive proteomics approach. Only those epitopes of the proteins were selected for constructing the final vaccine which showed a very strong binding efficiency with the MHC class I/II and B-cells. These epitopes were analyzed for their antigenicity, allergenicity, and toxicity. Based on this analysis, the final vaccine VC5 was constructed and its model was generated using Phyre2 server, and this was also subjected to different servers such as VaxiJen, Vaxign for its different validations. The VC5 was also analyzed for its physicochemical, antigenic, and immune system simulation. The construct is also easy to clone in the E. coli expression vector. When analyzed for the immune responses, the VC5 construct showed a very effective induction of the immune responses in the host with a sufficient rise in the T-helper and IgG and IgM molecules. Studies showed that a well-functioning cellmediated immune response is the key factor to clear the Salmonella infection (Galen et al., 2016). After the administration of the vaccine at 0 and 30 days, the immune simulation responses show very sharp clearance of the antigens. The current Vi capsule-based vaccine prevents antibodies binding with the O antigen of the Salmonella, making the vaccination inefficient (Garmory et al., 2002). PRODIGY server result showed that the binding energy (À11.6 kcal/mol) of TLR4 and VC5 indicates favorable interaction (ST-1). Table 3, supplementary material, contains the list of amino acids that interact between the VC5 and TLR4. The BLAST results of the VC5 against gut microbiome and the human proteome indicate only very few similarities between the two, hence eliminating the risk of cross-reactivity represented in Tables 4 and 5 (ST-2 and ST-3), supplementary material. Apart from that, the immune simulation of the proteins whose epitopes were used to construct the final vaccine has also been analyzed individually (Figure 1, supplementary material). This analysis was important to understand the immune responses when only the protein is used as a vaccine. Effective antigen clearance has also been seen in the booster doses on the 28th day, three months and six months shows that even after six months, the memory of the immune system is effective in clearing the antigens that too at a very higher rate. The rise in the B, T-helper, and Tcytotoxic cells indicates the stimulation of both the host's humoral and cell-mediated immune systems. From these results, it could be concluded that the VC5 construct may be a potent candidate to use for vaccination to reduce the high rates of Salmonella infections. The present study only targets the outer membrane proteins for designing the multi epitope-based vaccine as the outer membrane proteins have the highest probability of interacting with the host at initial levels. Before using the VC5 vaccine, further studies, including the in-vivo validation of the VC5 in animal models and humans, are necessary to analyze the multiepitope-based vaccine's behavior under physiological conditions.

Conclusion
The post-genomics era offers the ability to computationally approach the proteome of the pathogenic bacteria eliminating the need to grow them. The data included here represent the stepwise elimination of the non-antigenic proteins and prioritize only those proteins with antigenic and immune-stimulating properties. The drawbacks of the in-vivo design of the vaccines have been eliminated by taking further only those proteins that show non-toxic, non-allergic, and antigenic properties. Validating these epitopes in a stepwise manner and the immune simulation results suggest that the final vaccine construct VC5 could be used for further studies.

Funding
The author(s) reported there is no funding associated with the work featured in this article.

Consent for publication
All the data are available in the manuscript, and all the authors agree to publish it. No third-party data is used hence this section is not applicable in current manuscript.

Ethics approval
The present study does not involve human and animal samples.