Initially, proteomic data for 121 strains of ten pathogens (Streptococcus mutans, Streptococcus viridans, Streptococcus pyogenes, Staphylococcus aureus, Enterococcus faecalis, Streptococcus agalactiae, Gemella morbillorum, Streptococcus pneumonia, Enterococcus faecium, and Streptococcus gallolyticus) that causes endocarditis were obtained from Genbank (18). The 121 highly pathogenic strains evaluated in the core proteome analysis are prevalent around the world and have been associated to infective endocarditis (IE) in various locales. Streptococcus mutans and Streptococcus viridans, members of the oral microbiota, have a role in community-acquired endocarditis worldwide, particularly in North America, Europe, and Asia (19, 20). In affluent countries, Streptococcus pyogenes and Streptococcus agalactiae are more common, and invasive soft tissue infections can cause secondary endocarditis (21, 22). Staphylococcus aureus is a major cause of endocarditis associated with healthcare and intravenous drug use, and it is widely distributed in North America, Europe, and Australia (23). Antibiotic resistance is becoming a major concern in Asia and Africa. Enterococcus faecalis and Enterococcus faecium are widespread, with high incidence of hospital-acquired infections in the United States, Europe, and Latin America (24, 25). Streptococcus pneumoniae, primarily a respiratory pathogen, has been linked to endocarditis cases in Africa and Asia, while Streptococcus gallolyticus, which is frequently linked to colorectal cancer and endocarditis, is more commonly reported in Western countries (26, 27). Gemella morbillorum is less common but has been documented in North America and Europe (28). The effect of endocarditis is highlighted by its diverse geographic distribution, which also highlights the need for regionally specific surveillance and preventative measures.

The proteomes underwent analysis through the Bacterial pan-genome analysis program (BPGA) to discern proteins that are universally shared among the pathogens and exhibit a degree of conservation (29). Core proteome was generated through BPGA, and this was done to prioritize prospective vaccine candidates.

Certain proteins may be necessary for cell survival. The essential proteins were identified by uploading the core proteome to the Geptop server and analyzing it with a threshold of 0.2 (30). Human homologs were eliminated by an examination of essential proteins, and non-essential proteins were eliminated as well. The human proteome was screened for essential proteins using NCBI’s BLASTp with a 10^-4 expectation value (31). The removal of human homolog proteins was implemented to prevent the potential activation of an autoimmune response within the host organism. In the process of vaccine design, it is imperative to take into account the detrimental impacts of virulent proteins. The Virulent Factor Database (VFDB) was utilized to discern pathogenic proteins (32). The VFDB homologs of pathogenic proteins exhibited identities over 30% and bit scores surpassing 100. The Vaxijen server was also used to prove that pathogenic proteins were antigenic (33). Antigenic proteins with high scores were chosen to be studied further in order to make vaccines. The protein’s allergenicity was evaluated using the AllerTOP server (34). Utilizing the ProtParam tool, the pathogen’s secretome and exoproteome were selected and physiochemically analyzed to facilitate experimental investigations of the vaccine (35). Proteins with a molecular weight (M.W) below 110 kDa are more favorable candidates for potential vaccine targets. The TMHMM server was utilized to assess the number of transmembrane helices in the selected proteins (36). Proteins possessing either one or no transmembrane helix were selected for subsequent analysis. The InterProScan server was used to anticipate the biological processes and molecular activities of the target proteins (37).

Epitope selection required B-cell and T-cell (HTL and CTL) prediction. To determine Cytotoxic T lymphocyte (CTL), Linear B lymphocyte (LBL), and Helper T lymphocyte (HTL) epitopes, the Immune Epitope Database (IEDB) was utilized, which includes MHC-I and MHC-II binding tools (38). The Immune Epitope Database service analyzed all selected proteins for MHC-I alleles, including 32 HLA-B, 18 HLA-A, and 20 HLA-C alleles. Regardless of MHC genotype, peptide epitopes might have varying length preferences. However, we focused on 9-mer epitopes since they are the best length for binding to MHC class I molecules, which play an important role in immune response activation. This study employed a consensus technique, 15-mer peptide length selection, and a reference set of 27 human leukocyte antigen (HLA) alleles to forecast HTL epitopes using the IEDB. To identify MHC-I and MHC-II epitopes with strong binding, we employed a threshold of adjusted rank < 2.

B-cell epitopes are important in vaccine development because they elicit a humoral immune response by attaching to B cell receptors. This activation causes B cells to generate antibodies that help destroy infections and give immunity. The ABCpred server predicted linear B-cell epitopes using antigen analysis (39). Epitope detection was conducted utilizing a recurrent neural network characterized by a 16-amino-acid window length and a threshold value set at 0.51.

The predicted epitopes (MHC-I and B-cell) were validated using the Vaxijen (v2.0), ToxinPred, Allergen FP1.0, and the MHC class I immunogenicity servers (33, 39–41). The HTL epitopes chosen from the IEDB server underwent analysis to assess their capacity to stimulate IFN-γ in comparison to other cytokines, utilizing the IFNepitope web server for this evaluation (42). The capacity for IL-4 induction was analyzed through the IL4Pred web server, while the evaluation of IL-10 induction capability was conducted using the IL10Pred web server (43, 44). The IFNepitope, IL4pred, and IL-10Pred servers utilize SVM (Support Vector Machine) based prediction.

The diverse ethnic and regional composition influences the distribution and manifestation of HLA alleles, subsequently impacting the effectiveness of vaccines. The assessment of population coverage was conducted through the application of the IEDB population coverage tool, which also incorporated the HLA-binding alleles pertinent to the respective MHC Class-I and MHC Class-II epitopes (45). This tool assesses the demographic representation of each epitope across diverse global regions through a meticulous examination of the distribution of HLA-binding alleles.

The vaccine design was developed by integrating the CTL, HTL, and B-cell epitopes with suitable adjuvants and linkers. An adjuvant is a component of the vaccine that significantly enhances the capacity to elicit an immune response and requires careful selection. The adjuvant utilized in the vaccine construct was Cholera enterotoxin subunit B, identified by the accession number P01556. The EAAAK linker was used to join the additives together because it makes the whole structure more stable. After that, the AAY linker, the KK linker, and the GPGPG linker were used to connect the Cytotoxic T lymphocyte (CTL), Linear B lymphocyte (LBL), and Helper T lymphocyte (HTL) epitopes.

An essential element of the molecular composition of peptides and proteins involves the formation of disulphide bridges between cysteine residues. Such interactions may contribute to the stabilization of proteins by elevating the free energy associated with the denatured state and diminishing the conformational entropy. In this study, the disulphide bond was identified using the online Disulphide by Design 2.0 (DbD2) server (56). The server can produce three-dimensional vaccine structures to identify any residual pairs that are capable of forming disulphide linkages.

C-ImmSim, an in-silico approach, presented the immune response profile of vaccine construct, allowing the characterization of both cellular and humoral responses of the mammalian immune system caused by the chimeric vaccine (57). The vaccine was administered over four weeks, with three separate instances, using the default settings for all simulations. The time intervals used were 1 hour, 82 hours, and 126 hours, which were chosen based on the concept of 8-hour cell division cycles in real life. The simulations were conducted with a random seed of 12,345. The vaccine injection lacked lipopolysaccharide (LPS), with the simulation volume established at 50µl and the number of steps fixed at 1000.

The vaccine prompts a robust immune response by engaging with the host’s immune cells. The possible interaction of the vaccine with human TLR4 and MHC molecules was investigated by means of protein-protein docking. The vaccine was docked with MHCI (PDB ID: 1I1Y), TLR4 (PDB ID: 4G8A), and MHCII (PDB ID: 1KG0) by the HADDOCK v.2.2 server (58). HADDOCK, a sophisticated docking technique grounded in information analysis, serves to model biomolecular complexes. The residues participating in interactions were identified by employing the PDBsum online database (59), and PyMOL was employed to visualize the docked complex (60).

A molecular dynamics simulation lasting 100 ns was conducted to assess the stability of docked complexes. The solvation of the complex was conducted within a periodic box measuring 10 Å, which was filled with TIP3P water molecules (61). The system was designed to reach neutralization by including counter ions of Na+ and Cl-. Using the steepest descend approach of 5000 steps after neutralization, steric conflicts were eliminated, so minimizing the system. After minimization, the systems were equilibrated for 50,000 and 100,000 steps, respectively, at 310 K inside the NPT and NVT ensembles (62). The Parrinello-Rahman algorithms and the Berendson thermostat were employed to maintain the temperature and pressure at 310 K and 1 abm, respectively, during the simulation. The system was relaxed by adjusting the duration to τ P = 2.0 ps and τ T = 0.1 ps. Then, Verlet computed the non-bonded interactions (63) while the hydrogen atom bond lengths were kept at their ideal lengths by means of the LINCS technique (64). The particle mesh Ewald method was applied to estimate the electrostatic interactions beyond the short-range limit (65). Applications of periodic boundary conditions in the x, y, and z directions preceded a system production run. Every 10 picoseconds, the trajectory of the production run was recorded and analyzed utilizing the GROMACS commands and R BIO3D package (66). The simulation was run with the CHARMM36 forcefield and the Gromacs simulation tool (67).

Optimization of codons is crucial for the efficient expression of a foreign gene within a host organism while also taking into account the host’s distinctive characteristics. Codon optimization was conducted utilizing the VectorBuilder platform. The altered sequence of codons was introduced into the PET28a (+) E. coli expression vector by positioning it between the HindIII and XhoI restriction sites utilizing GenSmart (68).

Core proteins are currently recognized as significant components in vaccine formulations, as they are found in all or most strains of the target pathogen. Their incorporation into vaccines provides immune protection against a wider array of pathogenic species. To develop a vaccine against endocarditis, 121 pathogenic strains of ten pathogens were evaluated ( Additional File 1 ). Upon completion of the core proteome analysis, the total protein count was 12285.

In the initial step, essential proteins were retrieved from core proteome. After uploading the data to the Geptop server, 48 proteins were estimated to be essential. To identify essential proteins that are distinct from their human homologs, Blast P was implemented. Among the 48 essential proteins, there were 16 non-homologous proteins. After evaluating these proteins as virulence factors, it was determined that seven of these proteins are pathogenic. In addition, their antigenicity, allergenicity, immunogenicity, stability, transmembrane helices, and biological and molecular processes were evaluated. Three proteins with high immunogenicity, non-allergenicity, antigenicity, stability, and 0 transmembrane helices were selected as targets for the vaccine. Their information is presented in Table 1 .

The methodology for vaccine design involved selecting seven cytotoxic T lymphocyte (CTL) epitopes from Table 2 , five helper T lymphocyte (HTL) epitopes from Table 3A , and seven B lymphocyte (LBL) epitopes from Table 3B . The selection of these epitopes (B-cell and CTL) was based on immunogenicity, non-toxicity, antigenicity, and non-allergenicity. HTL epitopes was selected on the basis of their Ifn, IL4, and IL10 inducing ability.

The vaccine we have developed is expected to provide widespread coverage across a significant proportion of the human population. We employed the IEDB population coverage analysis tool to evaluate the extent to which the vaccine is effective in protecting the entire global human population. The inclusion of 54 HLA alleles for T-cell epitope prediction ( Figure 1 ) achieved a population coverage of roughly 89.55% for the worldwide human population.

The design of the multi-epitope vaccine comprised five linear B cell epitopes alongside twelve T cell epitopes, which including seven MHC-I and five MHC-II. The epitopes were interconnected through the use of KK, GPGPG, and AAY linkers. Each anticipated epitope was determined to be significantly antigenic, devoid of toxicity, and free from allergens following a meticulous selection process. An EAAAK linker was used to attach the 236 amino acid cholera enterotoxin subunit B to the vaccine’s N-terminus ( Figure 2 ).

The Disulfide by Design 2.13 server was employed to analyze the vaccine sequence, and a total of thirty residue pairs that could potentially produce disulfide bonds were identified. By considering the X3 parameters and bond energy, three pairs of residues (23PRO-28ASP, 104GLU-123ALA, and 210LEU-303ALA) were selected based on their findings meeting the predefined requirements ( Figure 4 ).

The C-ImmSim server was used to increase the half-life of immune cells in order to study their robust and efficient immune responses. This technique demonstrated that some cells had a significantly longer half-life than others, resulting in their prolonged survival. The immune simulation outcomes generated by the C-ImmSim server aligned closely with empirical immunological responses observed in real-world scenarios. The response was primarily defined by elevated levels of IgM. Furthermore, there was a rise in the expression of several immunoglobulins, including IgG1+IgG2, IgM, and IgG+IgM, leading to a reduction in serum antigen levels ( Figure 5A ). This rise in B-cell population was followed by memory extension ( Figure 5B ) and a substantial increase in the number of Th cells ( Figure 5C ). An increase in IFN-γ production was also observed as a result of vaccination ( Figure 5D ). The results pertaining to the T cell population closely aligned with the expected patterns as memory evolved, whereas the other immune cell populations exhibited a degree of stability.

MHC molecules and toll-like receptors are examples of immunological receptors that a vaccine must attach to firmly in order to successfully elicit a potent immune response. Using HADDOCK version 2.4, the relationships between vaccines and human TLR4, MHCI, and MHCII were investigated. The vaccine binding scores to TLR4, MHCI, and MHCII were 70.5 +/- 51.0 kcal/mol, 97.3 +/- 12.5 kcal/mol, and -133.9 +/- 22.5 kcal/mol, respectively. These values were within the range of the standard deviation. Data about docking is presented in Table 6 .

Figures 6A–C depicts docked complexes with the vaccine (in green) and receptor molecule (in yellow), as well as MHCI (in blue) and MHCII (in purple), respectively. The PDBsum server was utilized in order to produce a map of connections between docked complexes as well as discoveries regarding the binding interactions that occur between vaccine components and receptor molecules. As part of the research, a schematic depiction of the interactions between docked proteins was presented. These interactions included both H-bond and non-bonded combinations. Upon analysis, it was discovered that the vaccine and TLR4, MHCI, and MHCII possessed 24, 23, and 14 hydrogen bonding interactions, respectively, with a range of 3.11 Å, 3.23 Å, and 3.27 Å.

Different codon usage patterns prevent foreign genes from being translated, hence the only effective technique to boost translational efficacy is codon adaptation. Using vectorBuilder to optimize codon usage, we were able to successfully express the vaccine we created against the E. coli K12 strain in the E. coli host. The vaccine exhibits a GC content of 49.77% and a CAI of 0.94%, which substantiates this assertion. Subsequently, the altered codon sequence of the proposed vaccine was integrated into the PET28a (+) E. coli expression vector within the segment located between the BamHI and XhoI restriction sites ( Figure 10 ).