Genome-Based Multi-Antigenic Epitopes Vaccine Construct Designing against Staphylococcus hominis Using Reverse Vaccinology and Biophysical Approaches

Staphylococcus hominis is a Gram-positive bacterium from the staphylococcus genus; it is also a member of coagulase-negative staphylococci because of its opportunistic nature and ability to cause life-threatening bloodstream infections in immunocompromised patients. Gram-positive and opportunistic bacteria have become a major concern for the medical community. It has also drawn the attention of scientists due to the evaluation of immune evasion tactics and the development of multidrug-resistant strains. This prompted the need to explore novel therapeutic approaches as an alternative to antibiotics. The current study aimed to develop a broad-spectrum, multi-epitope vaccine to control bacterial infections and reduce the burden on healthcare systems. A computational framework was designed to filter the immunogenic potent vaccine candidate. This framework consists of pan-genomics, subtractive proteomics, and immunoinformatics approaches to prioritize vaccine candidates. A total of 12,285 core proteins were obtained using a pan-genome analysis of all strains. The screening of the core proteins resulted in the selection of only two proteins for the next epitope prediction phase. Eleven B-cell derived T-cell epitopes were selected that met the criteria of different immunoinformatics approaches such as allergenicity, antigenicity, immunogenicity, and toxicity. A vaccine construct was formulated using EAAAK and GPGPG linkers and a cholera toxin B subunit. This formulated vaccine construct was further used for downward analysis. The vaccine was loop refined and improved for structure stability through disulfide engineering. For an efficient expression, the codons were optimized as per the usage pattern of the E coli (K12) expression system. The top three refined docked complexes of the vaccine that docked with the MHC-I, MHC-II, and TLR-4 receptors were selected, which proved the best binding potential of the vaccine with immune receptors; this was followed by molecular dynamic simulations. The results indicate the best intermolecular bonding between immune receptors and vaccine epitopes and that they are exposed to the host’s immune system. Finally, the binding energies were calculated to confirm the binding stability of the docked complexes. This work aimed to provide a manageable list of immunogenic and antigenic epitopes that could be used as potent vaccine candidates for experimental in vivo and in vitro studies.

. Schematic view of a designed computational framework for screening a broad-spectrum vaccine against S. hominis in the reference proteome followed by epitope prioritization, multiepitopes vaccine (MEV) construction, and MEV processing.

Pre-Screening
In this phase, we screened for the proteins that were conserved among all strains of the bacteria. We conducted a pan-genome analysis and screened the core genome sequence. The redundant proteins are not potent immunological proteins because of their poor conservation across the genomes of strains [12]; they are not part of the core genome. The non-redundant proteins in the core sequence were predicted using the CD-HIT web server with a sequence identity threshold of 90%, keeping all values default [13]. The nonredundant proteins were subjected to a subcellular localization check. Proteins that are localized at the surface of a pathogen or secretome are the vital proteins for a vaccine as they are in frequent contact with the host and cause infections [14]. The antigenicity of these proteins can be easily recognized by the host's immune system and trigger an immune response. Proteins localized at the inner and outer membrane, periplasmic spaces, and secretory proteins were selected for the next phase. Proteins with multiple and unknown localizations were discarded [15]. Figure 1. Schematic view of a designed computational framework for screening a broad-spectrum vaccine against S. hominis in the reference proteome followed by epitope prioritization, multi-epitopes vaccine (MEV) construction, and MEV processing.

Pre-Screening
In this phase, we screened for the proteins that were conserved among all strains of the bacteria. We conducted a pan-genome analysis and screened the core genome sequence. The redundant proteins are not potent immunological proteins because of their poor conservation across the genomes of strains [12]; they are not part of the core genome. The non-redundant proteins in the core sequence were predicted using the CD-HIT web server with a sequence identity threshold of 90%, keeping all values default [13]. The non-redundant proteins were subjected to a subcellular localization check. Proteins that are localized at the surface of a pathogen or secretome are the vital proteins for a vaccine as they are in frequent contact with the host and cause infections [14]. The antigenicity of these proteins can be easily recognized by the host's immune system and trigger an immune response. Proteins localized at the inner and outer membrane, periplasmic spaces, and secretory proteins were selected for the next phase. Proteins with multiple and unknown localizations were discarded [15].

Vaccine Candidate Prioritization
The pathogenic secretome and exoproteome were further filtered to find the proteins associated with pathogenesis. The screened proteins at this stage were inputted into the BLASTP software against the Virulent Factor Database (VFDB) for the screening of virulent proteins with a sequence identity ≥30% and bit score ≥100 [16]. The virulent proteins were further processed for physiochemical properties. The physiochemical characterization was performed through the ProtParam online tool. This tool predicts the computation of parameters such as instability index, molecular weight, estimated half-life, atomic composition, aliphatic index, theoretical PI, etc. [17]. The key factors to evaluate were the instability index and molecular weight. The proteins with instability index values of less than 40 and molecular weights of less than 100 kDa were considered the best choices for vaccine candidates. The next step was to examine the protein for transmembrane helices. Only proteins with 0 or 1 transmembrane helix were selected [18]. To eliminate the risk of inhibition of certain beneficial bacteria, the homology of proteins was checked against probiotics Lactobacillus species including L. rhamnosus (taxid: 47,715), L.casei (taxid: 1582), and L.johnsonii (taxid: 33,959) [19]. For this step, a BLASTP search was performed against them, and we only selected proteins that had no significant similarity [20].

Prediction of Immune Cells Epitopes
With respect to the IEDB data, the Bepipred linear epitope prediction method with a cut-off value of 0.5 was utilized to predict linear B-cell epitopes [21]. The B-cell epitopes were subsequently used for T-cell antigenic determinants using the IEDB MHC-I and MHC-II epitope predictors. A comparative analysis was performed to select subsequences that bound to both MHC-I and MHC-II alleles [22]. The method employed for this prediction was IEDB 2.22, where the peptides are sorted based on a percentile score, with the lower percentile score having a higher binding affinity [23].

Multi-Epitope Peptide Designing
Peptide vaccines have weak immunogenicity, which can be overcome by using immunodominant epitopes (adjuvants) [28]. The multi-epitope peptide vaccine has series of overlapping epitope protein sequences that can be assembled and combined using GPGPG linkers [29]. The 3D structure of the construct was simulated using the 3Dpro program from the Scratch Protein Predictor web server [30]. Furthermore, the structure was refined using the Galaxy Refine tool of the Galaxy web server [31].

Molecular Docking
Molecular docking plays a vital role in vaccine construct designing to interpret the binding affinity of the vaccine construct with the host's immune receptors [32]. An effective immune response will be activated when the MEV efficiently binds to the host's immune receptors. The molecular docking of the designed vaccine construct with immune receptors was performed using the web server PATCHDOCK, which allows for the docking of two interacting molecules based on the interrelation principles of the shapes [33]. A blind Vaccines 2022, 10, 1729 5 of 24 docking strategy was applied with the immune receptors, TLR4, MHC-I, and MHC-II molecules [34]. The input clustering for the RMSD and complex type were set as their defaults. The docked complexes were then refined using FireDock on the same web server. FireDock re-ranked the docking complex results by filtering out all clashes and molecular conformational errors [35]. The topmost selected complexes were then visualized using the UCSF Chimera 1.16 program.

In Silico Cloning and Codon Optimization
To obtain an efficient expression of the vaccine construct in the E. coli expression system, the sequence was reverse-translated and optimized for codon usage [36]. The vaccine sequence was reverse-translated and improved by utilizing the JCAT (Java Codon Adaptation Tool), which calculates the CAI (codon adaptation index) and GC content of the sequence. Ideally, the CAI should be one and the GC content should be around 30-70% [37]. The sequence was then cloned into the pET-28a (+) expression vector of E. coli (K12) using a software called SnapGene [38].

Disulfide Engineering
The MEV construct was engineered using a disulfide engineering approach for structure stability. The disulfide links were introduced into the structure using the bioinformatics tool Design2.0 [39].

Molecular Dynamic (MD) Simulations
An MD simulation is a computational framework to evaluate the structural dynamics of biomolecules at the atomic resolution with different environmental conditions. A software called Assisted Model Building and Energy Refinement (AMBER) v20 was used in our process [40]. This process has three phases: system preparation, preprocessing, and production. In the first phase, the top 3 docked complexes of the MEV were prepared using the antechamber program. The libraries and parameters were set. Afterward, the complexes were added to the solvation box (TIP3P, size 12 Å) by the leap program. For the neutralization of the system, counter ions were added. The energy of the system was minimized step-by-step during the preprocessing phase, including the energy minimization of hydrogen atoms, water box, and non-heavy atoms. The ensembles such as NVT (constant temperature, constant volume) and NPT (constant temperature, constant pressure) were used to increase the temperature and pressure, respectively. The hydrogen bond constraint was maintained using the SHAKE algorithm. The temperature was maintained using Langevin dynamics. For 1 ns, the system was allowed to equilibrate itself. In the production phase, the simulation was performed for 100 ns at a time scale of 2 fs using the Berendsen algorithm [41].

Binding Free Energies Calculation
The binding free energies of the top 3 docked complexes were calculated using the molecular mechanics/Poisson-Boltzmann surface area (MM/PBSA) and molecular mechanics/generalized Born surface area (MM/GBSA) approaches by employing the MMPBSA.py module of AMBER v20 [42]. A total of 1000 frames of the trajectory of simulation were selected to calculate the binding free energies. The main goal was to find the difference between the free energy of the solvated and non-solvated states of the complexes [43].

Proteins Sequence Retrieval
The sequences of the following bacterial strains were retrieved from the NCBI, as mentioned in Table 1.

Pan-Genome Analysis
The development of metagenomics and next-generation sequencing technologies has led to the focus shifting on the analysis and study of multiple genomes of different bacterial strains altogether rather than a few genome comparisons. The concept of a pan-genome is the outcome of such a multi-genome analysis [44,45]. This analysis provides an overview of the horizontal gene transfer and insights into the evolution of species [46]; it provides a detailed summary of the genomic diversity of the dataset, determining the core, accessory, and unique gene pool of the bacterial species [47]. Using the BPGA pipeline, we analyzed the pan-genome of all strains of S. hominis. This approach categorizes all the protein sequences based on the genetic conservation among them [48]. The core genome contains the sequences that are conserved among all the strains. The parts of the genome that are common in few strains are called accessory genes. Unique genes found only in a single strain are called singletons [49]. The output of the BPGA analysis is mentioned in Table 2.

CD-HIT Analysis
The core sequences were used further down the line. The core sequences are the pool of redundant and non-redundant proteins. The CD-HIT method was used to identify redundant and non-redundant proteins [50]; CD-HIT is a super-fast protein sequence clustering web server that can analyze millions of proteins and group together similar proteins into clusters based on sequence similarity. The core genome of S. hominis consists of 12,285 protein sequences that were analyzed with a sequence identity cut-off value of 90%. The proteins that are similar by up to 90% were discarded. Because of this filter, the redundant sequences (paralogous) were removed, leaving only the non-redundant sequences (orthologous), which included 1824 protein sequences [51].

Sub-Cellular Localization
In this phase, we localized the proteins of the orthologous sequence of S. hominis. This is the most important step in selecting effective vaccine candidates. Proteins found in the inner and outer membranes, periplasmic spaces, and secretory proteins are promising candidates for vaccine construct development [52]. These proteins play vital roles in the pathogenesis, invasion, and colonization of bacteria in a host's cells. Furthermore, the proteins are all surface proteins, and the antigenic epitopes of these proteins can efficiently be recognized by the host immune system and provoke innate immune responses [53]. Out of 1824 non-redundant sequences, only 22 extracellular protein sequences were opted for further processing, as presented by the different proteins and numbers in Figure 2. The online tool PSORTb was used in the categorization of the protein localities. This tool is broadly used for the determination of the sub-cellular localization of proteins [54]. In this phase, we localized the proteins of the orthologous sequence of S. hominis. This is the most important step in selecting effective vaccine candidates. Proteins found in the inner and outer membranes, periplasmic spaces, and secretory proteins are promising candidates for vaccine construct development [52]. These proteins play vital roles in the pathogenesis, invasion, and colonization of bacteria in a host's cells. Furthermore, the proteins are all surface proteins, and the antigenic epitopes of these proteins can efficiently be recognized by the host immune system and provoke innate immune responses [53]. Out of 1824 non-redundant sequences, only 22 extracellular protein sequences were opted for further processing, as presented by the different proteins and numbers in Figure 2. The online tool PSORTb was used in the categorization of the protein localities. This tool is broadly used for the determination of the sub-cellular localization of proteins [54].

Virulence Check
Virulence is the pathogen's ability to cause an infection in a host cell and initiate immune responses [55]. The core redundant exoproteome/secretome was checked for virulence. This filtration of proteins was performed through the VFDB database, with specific criteria of selection as mentioned in the methodology of [56]. A BLASTP search was performed to shortlist the proteins. Out of the 22 non-redundant core sequences, only six (27.3%) protein sequences were found to be virulent, and they are listed in Table 3.

Transmembrane Helices and Physiochemical Properties Analysis
Different physiochemical properties and transmembrane helices were evaluated to filter out the stable virulent proteins. The proteins with transmembrane helices of zero or

Virulence Check
Virulence is the pathogen's ability to cause an infection in a host cell and initiate immune responses [55]. The core redundant exoproteome/secretome was checked for virulence. This filtration of proteins was performed through the VFDB database, with specific criteria of selection as mentioned in the methodology of [56]. A BLASTP search was performed to shortlist the proteins. Out of the 22 non-redundant core sequences, only six (27.3%) protein sequences were found to be virulent, and they are listed in Table 3.

Transmembrane Helices and Physiochemical Properties Analysis
Different physiochemical properties and transmembrane helices were evaluated to filter out the stable virulent proteins. The proteins with transmembrane helices of zero or one were selected. Such choices were made because of the failure of poor protein expression in in vitro systems such as E. coli. Having more than two transmembrane helices makes it difficult to conduct studies and cloning [57]. In the physiochemical properties, the prime factor to evaluate is molecular weight. The isolation and purification of proteins with low molecular weight are easy to perform. The other factor to analyze is the instability index. The instability index computes the stability of proteins by analyzing the disulfide bonds in proteomic sequences. Out of the six virulent proteins, three (50%) proteins were screened to be stable proteins. Table 4 summarizes the output results of the physiochemical properties and transmembrane helices. A threshold value of 40 was used for the instability index. Proteins having an instability index value greater than 40 were considered unstable and discarded from the study. Proteins having an instability index of less than 40 were considered stable and subjected to further analysis.

Homology Check against Probiotics
In the NCBI database, a BLASTP search was performed against probiotic lactobacillus bacteria species. These bacteria are commensals in the human gut and help to prevent diarrhea, improve gut health, lipid metabolism, and modulate immune and inflammatory responses [58]. Any type of homology between the probiotic and pathogen proteomic sequence could disturb gastrointestinal health and worsen the situation. In this filtration, only two proteins were found to have no significant sequence similarity [59]. For the comparison purpose, the non-homologous proteins must have ≤30% of sequence identity with the probiotic bacteria and must have an E-value of 10−4.

B and T Cells Epitope Prediction
Specificity in killing pathogens is the foundation of adaptive immunity [60]. B and T cells are the fundamental units of adaptive immunity. These cells recognize and kill the pathogens by a series of chemical reactions and store the data in their memory cells to prevent the same infection in the future [61]. The process of memorizing the pathogen's identity and immunological data becomes the fundamental principle of vaccination. The major cells involved in adaptive/acquired immunity are the T and B lymphocytes [62]. They initiate two types of responses: humoral and cell-mediated immune responses against specific pathogens. The filtered two proteins were used to predict the B cells' epitopes, and then these epitopes were subsequently used for the T cell epitope mapping. B cell epitope prediction is important because these epitopes could activate responses such as agglutination, neutralization, opsonization, complement system, and cell-mediated cytotoxicity administrated by antibody lymphocytes (natural killer cells, eosinophil, etc.) which can destroy the invader pathogen [63]. These epitopes were further used for T cell epitope mapping, where important sites for the MHC class I and II molecules were predicted. The MHC-I molecules are present on the surface of all nucleated body cells and present the pathogenic proteins to the cytotoxic T cells. In the same way, MHC-II molecules are expressed on the surface of special antigen-presenting cells such as dendritic cells, monocytes, and B lymphocytes [64]. This displaying of the pathogenic proteins leads to the full-force production of antibodies [65]. The predicted B and T cells' epitopes are presented in Tables 5 and 6, respectively.

Epitope Prioritization
Using different immunoinformatics approaches to prioritize the epitope selection, we determined the successful vaccine candidate to be water-soluble, non-toxic, non-allergenic, and antigenic proteins [66]. The antigenicity checks measure the ability of the antigen to bind to antibodies and other lymphocytic products [67]. The non-toxic, non-allergenic, water soluble, and antigenic epitopes are presented in Figure 3.

Multi-Epitope Vaccine Construct Processing
As the MEV is the combination of many epitopes, these epitope domains were combined using GPGPG linkers [16]. Because protease enzymes can easily degrade the antigenic peptides and epitopes, the B cell receptor (BCR) and T cell receptor (TCR) cannot recognize them and thus have a poor immunogenic effect [68]. In order to avoid this, the intermolecular adjuvants were used to boost the immune responses of the MEV and enable efficient transmission inside the body [69]. This adjuvant molecule was linked with the epitopes constructed with the help of the EAAK linker. The cholera toxin B subunit adjuvant was used in the process [70]. The 3D structure of the construct was modeled using the 3Dpro program of Scratch protein predictor. The structure visualized using Chimera 1.16 is presented in Figure 4, whereas the schematic representation is mentioned in Figure 5.

Multi-Epitope Vaccine Construct Processing
As the MEV is the combination of many epitopes, these epitope domains were combined using GPGPG linkers [16]. Because protease enzymes can easily degrade the antigenic peptides and epitopes, the B cell receptor (BCR) and T cell receptor (TCR) cannot recognize them and thus have a poor immunogenic effect [68]. In order to avoid this, the intermolecular adjuvants were used to boost the immune responses of the MEV and enable efficient transmission inside the body [69]. This adjuvant molecule was linked with the epitopes constructed with the help of the EAAK linker. The cholera toxin B subunit adjuvant was used in the process [70]. The 3D structure of the construct was modeled using the 3Dpro program of Scratch protein predictor. The structure visualized using Chimera 1.16 is presented in Figure 4, whereas the schematic representation is mentioned in Figure 5. intermolecular adjuvants were used to boost the immune responses of the MEV and enable efficient transmission inside the body [69]. This adjuvant molecule was linked with the epitopes constructed with the help of the EAAK linker. The cholera toxin B subunit adjuvant was used in the process [70]. The 3D structure of the construct was modeled using the 3Dpro program of Scratch protein predictor. The structure visualized using Chimera 1.16 is presented in Figure 4, whereas the schematic representation is mentioned in Figure 5.

Codon Optimization
The codons are the complimentary nucleotide triplets of the genetic codes enc in DNA [71]. They decide which amino acid should be added to the growing protein in protein synthesis. There are 64 codons for 20 different types of amino acids. Diff organisms use different codons for the same amino acid [72]. To obtain the maxi expression in the host cells, we must optimize the codons accordingly. The prote quence of the MEV was reverse-translated into DNA using a codon optimization nique, which optimizes the given sequence according to the host's codon usage pa In this case, we selected E. coli (K12) as a host for our sequence expression [73]. Th content of the sequence is 52%, and the codon adaptation index is 0.95%. These valu considered ideal for the expression process. At last, the optimized MEV was cloned the pET-28a (+) expression vector. The cloned vector is presented in the Figure 6.

Codon Optimization
The codons are the complimentary nucleotide triplets of the genetic codes encoded in DNA [71]. They decide which amino acid should be added to the growing protein chain in protein synthesis. There are 64 codons for 20 different types of amino acids. Different organisms use different codons for the same amino acid [72]. To obtain the maximum expression in the host cells, we must optimize the codons accordingly. The protein sequence of the MEV was reverse-translated into DNA using a codon optimization technique, which optimizes the given sequence according to the host's codon usage pattern. In this case, we selected E. coli (K12) as a host for our sequence expression [73]. The GC content of the sequence is 52%, and the codon adaptation index is 0.95%. These values are considered ideal for the expression process. At last, the optimized MEV was cloned into the pET-28a (+) expression vector. The cloned vector is presented in the Figure 6. Vaccines 2022, 10, x FOR PEER REVIEW 13 of 24 Figure 6. The cloned pET 28a (+) Vector.

Disulfide Engineering
In protein engineering, improving the stability of proteins is one of the main goals. The best logical approach is to improve the stabilizing molecular interactions that are naturally found in proteins [74]. Disulfide bridges are covalent bonds that provide considerable structural stability. Disulfide engineering is a method of directing disulfide bonds into the vaccine construct to make it structurally stable. Many residues are enzyme degradable, so they are replaced with cysteine residues in the vaccine construct [75]. The mutated residues are tabulated in Table 7 while the mutant and original structure of the designed vaccine is mentioned in Figure 7.

Disulfide Engineering
In protein engineering, improving the stability of proteins is one of the main goals. The best logical approach is to improve the stabilizing molecular interactions that are naturally found in proteins [74]. Disulfide bridges are covalent bonds that provide considerable structural stability. Disulfide engineering is a method of directing disulfide bonds into the vaccine construct to make it structurally stable. Many residues are enzyme degradable, so they are replaced with cysteine residues in the vaccine construct [75]. The mutated residues are tabulated in Table 7 while the mutant and original structure of the designed vaccine is mentioned in Figure 7.

Molecular Docking
Molecular docking analysis approaches were used to elucidate the interaction between the vaccine construct and the host's innate and adaptive immune cells [76]. To provoke immune responses, the MEV should have good interactions with the host's immune cells receptors such as TLR4, MHC-I and MHC-II. For this purpose, a blind docking strategy was used to predict the interaction of the MEV with the host's immune receptors [77]. The results are tabulated in Tables 8-10. The server generated docking solutions, which were ranked on their binding energy score. The docked complexes having the best binding energy score were considered to be stable complexes. In each case, 20 solutions were generated, and the one with the highest binding energy score was selected. For instance, Solution 1 was chosen in all three receptors, and the vaccine-MHC-I complex had a best score of 19,690; the vaccine-MHC-II had a score of 19,030, and vaccine-TLR-4 had a score of 20,864.

Docking Refinement
The PATCHDOCK results were further refined using FireDock. The top 10 docked complexes were subjected to refinement [78]. FireDock refinement ranks the results with the lowest binding energy complex at the top and proceeds downward. The docked complexes having the lowest binding energies were selected for the simulation [79]. The FireDock results are tabulated in Tables 11-13, and the docked confirmation is presented in Figure 8.

MD Simulation
An MD simulation is a method for understanding molecular behavior and properties at the atomic level; it provides useful information about the structure, interaction energies, and movement of atoms (dynamics), which complements the experimental data [80]. In this process, the interaction of two molecules is evaluated at different conditions of temperature and pressure for a short period of time by using different algorithms and ensembles. The RMSD value was calculated based on the alpha carbon atom. The top three complexes were analyzed. The graph shows a steady line, although the vaccine and TLR4 complex show some deviations, seen by the line fluctuating [81] around the value 12-14Å .

MD Simulation
An MD simulation is a method for understanding molecular behavior and properties at the atomic level; it provides useful information about the structure, interaction energies, and movement of atoms (dynamics), which complements the experimental data [80]. In this process, the interaction of two molecules is evaluated at different conditions of temperature and pressure for a short period of time by using different algorithms and ensembles. The RMSD value was calculated based on the alpha carbon atom. The top three complexes were analyzed. The graph shows a steady line, although the vaccine and TLR4 complex show some deviations, seen by the line fluctuating [81] around the value 12-14Å. The main goal here was to investigate the vaccine binding interaction with the host's immune cells and initiate immune responses. A graph of the RMSD is presented in Figure 9. The main goal here was to investigate the vaccine binding interaction with the host's immune cells and initiate immune responses. A graph of the RMSD is presented in Figure 9.

Binding Free Energies Calculations
The binding free energies of the top three docked complexes were calculated using the MM/GBSA approach [82]. The total binding free energies of the TLR4, MHC-I and MHC-II complexes were −74.47 kcal/mol, −62.25 kcal/mol, and −71.95 kcal/mol, respectively. The estimated values are tabulated in Table 14. As can be seen in the table, both the van der Waals and electrostatic energies majorly contributed to the intermolecular binding and showed a stable binding conformation. Key: VDWAALS (van der Waals), EEL (electrostatic), Delta G gas (net gas phase energy), Delta G solv (net solvation energy), Delta Total (net energy of system).

Discussion
The current study is an in silico based-model of a multi-epitope vaccine against S. hominis. We used a computational framework to evaluate the core proteomes of S. hominis for non-homologous, antigenic, and highly conserved proteomes among all the strains of S. hominins. These proteins were further used to screen antigenic, non-allergenic, nontoxic, and water-soluble epitopes for B and T cell alleles. Further analysis (such as analyzing the docking, MD simulation, and binding energies calculation results) revealed great immunogenic effects and stable molecular configurations [83].

Binding Free Energies Calculations
The binding free energies of the top three docked complexes were calculated using the MM/GBSA approach [82]. The total binding free energies of the TLR4, MHC-I and MHC-II complexes were −74.47 kcal/mol, −62.25 kcal/mol, and −71.95 kcal/mol, respectively. The estimated values are tabulated in Table 14. As can be seen in the table, both the van der Waals and electrostatic energies majorly contributed to the intermolecular binding and showed a stable binding conformation.

Discussion
The current study is an in silico based-model of a multi-epitope vaccine against S. hominis. We used a computational framework to evaluate the core proteomes of S. hominis for non-homologous, antigenic, and highly conserved proteomes among all the strains of S. hominins. These proteins were further used to screen antigenic, non-allergenic, non-toxic, and water-soluble epitopes for B and T cell alleles. Further analysis (such as analyzing the docking, MD simulation, and binding energies calculation results) revealed great immunogenic effects and stable molecular configurations [83].
In recent history, the exponential growth of antimicrobial resistance has been observed because of the overuse of antibiotics that have often deviated from prescription. Once the strain emerges with antibiotic resistance, it can quickly spread and acquire resistance to other classes of drugs as well. These multidrug-resistant bacteria limit the choice of choosing specific antibiotics, increasing the mortality and morbidity rates. It is estimated that by 2050, 10 million lives a year may be lost to AMR, exceeding the 8.2 million lives a year currently lost to cancer. To put this number into perspective, at least 700,000 people die of resistant infections every year worldwide, more than the combined number of deaths caused by tetanus, cholera, and measles [84]. Furthermore, it is a big loss to the economy. It has been estimated that, if the AMR trend continues, the cumulative loss to world economies might be as high as USD 100 trillion by 2050 [85]. The development of new drugs cannot keep pace with the increasing risk of antimicrobial resistance. However, the immunization of individuals through vaccination can affect this both directly and indirectly [86]. As the development of vaccines greatly reduce the intake of antibiotics, this in turn decreases the chances of new antimicrobial-resistant strains emerging. Traditional vaccinology approaches have several defects, i.e., they generate inaccurate immune responses, and have lengthy processing times, high cost, less specificity, less safety, hypersensitivity, and less stability [87]. Computational vaccine designing strategies are exponentially growing, mainly because of the huge growth of genomic data that has allowed scientists to move beyond Pasteur's rule of vaccinology and instead use computational approaches, tools, and software to design vaccines without the need for any wet-lab techniques. A previous study conducted by Aldakheel et al. designed an MEV using proteome-wide mapping and reverse vaccinology and used the same computational framework to model highly antigenic and potent MEV targets against Clostridium perfringens [70]. The same computational approach was used by Al-Megrin et al. They designed a novel MEV against Staphylococcus auricularis using immunoinformatics and biophysics approaches. They prioritized the vaccine candidate on the tight criteria provided in the literature. The vaccine candidates are non-homologic, conserved, and immunogenic. The epitopes derived from these proteins were also immunogenic, non-allergic, non-toxic, and water-soluble. These epitopes were derived from B and T cell alleles. This MEV construct has a higher molecular stability and efficient immune response, according to the simulation and binding energies calculations [70].

Conclusions
Our work presents an in silico design of a broad-spectrum multi-epitope vaccine against S. hominis. The vaccine construct is composed of antigenic, non-allergic, non-toxic multi-epitopes extracted from highly virulent proteins that are part of a pathogen's core exoproteome [88]. Many immunoinformatics, biophysical, and subtractive proteomic techniques were used in the process [89]. During the selection of a vaccine candidate for vaccine development, tough selection criteria were used, such as that proteins should be from the core proteome, should be present on cell surface (or be excretory), should be nonhomologous to the host, and should be probiotics [90]. The epitopes that were prioritized were those with non-allergenic, non-toxic, antigenic, and water-soluble properties that had a higher binding affinity to B and T cell alleles. The results of the MD simulation and binding energies calculations elaborated the stable molecular configuration and minimize system energies. Although the pan-genome-based reverse vaccinology is an effective method to develop a multi-epitope vaccine, the potency of the vaccine should be studied and confirmed by in vivo and in vitro immunological methods [91].