Skip to main content

In-silico design and evaluation of an epitope-based serotype-independent promising vaccine candidate for highly cross-reactive regions of pneumococcal surface protein A

Abstract

Background

The pathogenicity of pneumococcus with high morbidity, mortality, and multi-drug resistance patterns has been increasing. The limited coverage of the licensed polysaccharide-based vaccines and the replacement of the non-vaccine serotypes are the main reasons for producing a successful serotype-independent vaccine. Pneumococcal surface protein A (PspA) is an extremely important virulence factor and an interesting candidate for conserved protein-based pneumococcal vaccine classified into two prominent families containing five clades. PspA family-elicited immunity is clade-dependent, and the level of the PspA cross-reactivity is restricted to the same family.

Methods

To cover and overcome the clade-dependent immunity of the PspAs in this study, we designed and tested a PspA1-5c+p vaccine candidate composed of the highest immunodominant coverage of B- and T-cell epitope truncated domain of each clade focusing on two cross-reactive B and C regions of the PspAs. The antigenicity, toxicity, physicochemical properties, 3D structure prediction, stability and flexibility of the designed protein using molecular dynamic (MD) simulation, molecular docking of the construct withHLADRB1*(01:01) and human lactoferrin N-lop, and immune simulation were assessed using immunoinformatics tools. In the experimental section, after intraperitoneal immunization of the mice with Alum adjuvanted recombinant PspA1-5c+p, we evaluated the immune response, cross-reactivity, and functionality of the Anti-PspA1-5c+p antibody using ELISA, Opsonophagocytic killing activity, and serum bactericidal assay.

Results

For the first time, this work suggested a novel PspA-based vaccine candidate using immunoinformatics tools. The designed PspA1-5c+p protein is predicted to be highly antigenic, non-toxic, soluble, stable with low flexibility in MD simulation, and able to stimulate both humoral and cellular immune responses. The designed protein also could interact strongly with HLADRB1*(01:01) and human lactoferrin N-lop in the docking study. Our immunoinformatics predictions were validated using experimental data. Results showed that the anti-PspA1-5c+p IgG not only had a high titer with strong and same cross-reactivity coverage against all pneumococcal serotypes used but also had high and effective bioactivity for pneumococcal clearance using complement system and phagocytic cells.

Conclusion

Our findings elucidated the potential application of the PspA1-5c+p vaccine candidate as a serotype-independent pneumococcal vaccine with a strong cross-reactivity feature. Further in-vitro and in-vivo investigations against other PspA clades should be performed to confirm the full protection of the PspA1-5c+p vaccine candidate.

Introduction

Streptococcus pneumoniae (pneumococcus) is an opportunistic pathogen and is a major cause of morbidity and mortality worldwide, with more than 98 serotypes based on their polysaccharide capsules [1, 2]. In 2018, the global pneumococcal burden was appraised to be 26.7 occurrences per 1,000 people, resulting in over 1,000,000 deaths [3, 4]. Effective treatment of pneumococcal diseases concerning antibiotic selection is a growing concern because of the increasing multi-drug resistance pattern of pneumococci [5, 6]. For the prevention of pneumococcal diseases, licensed vaccines are based on polysaccharide capsules of the most prevalent pneumococcal serotypes. The limited coverage of the licensed vaccines, broad geographical variation in circulating serotypes, non-vaccine serotype replacement, and the prevalence of non-encapsulated pneumococci from patients with invasive pneumococcal disease (IPD) are key reasons for an attempt to overcome the pneumococcal vaccine limitations and design the novel serotype-independent vaccines [7,8,9,10,11]. Pneumococcal protein-based vaccine (PPV) formulation is a cost-effective and promising candidate for serotype-independent vaccine development [12, 13]; and many pneumococcal conserved cell-surface proteins have already been identified as ideal antigens for PPV in recent years [6, 14, 15].

Pneumococcal surface protein A (PspA) is a very important virulence factor that has been widely studied and is present in all pneumococcal strains [3, 16, 17]. Various active or passive immunization studies using rPspAs demonstrated that animal models were protected against the lethal challenge of pneumococci [7]. Furthermore, the administration of PspA in early human adult clinical trials has been reported [13]. Another study demonstrated that PspA immunization provides more comprehensive protection than Prevnar pneumococcal conjugate vaccine [18].

The N-terminal end of PspA, which is more variable due to mutation accumulation [19], has protection-eliciting epitopes, that have been divided into three regions A, B, and C [20]. The B-region of PspA is serologically variable and forms the basis of classifying PspA into three families with six clades. This region is identified as a clade-defining region (CDR) and comprises two prominent families. Family 1 contains Clades 1 and 2, and Family 2 is made up of Clades 3, 4, and 5. These two families are exhibited in almost 100% of clinical isolates from adult IPD and non-IPD children. Finally, Family 3 is composed of Clade 6, which is extremely rare among pneumococci, and it has been reported that the percentage of Clade 6 in pneumococcal strains is less than 1%. So in many studies, this clade is excluded from the study [20]. Analysis of the CDR sequence showed that the sequences belonging to the same clades demonstrated a sequence identity of ≥ 90% and those of different families ≤ 55% sequence identity [21]. Previous studies have shown high levels of cross-reactivity between different PspA fragments within the B-region of PspA [7, 22]. The C region of PspA is the Proline-Rich Domain (PRD), characterized by the presence of repetitive motifs of proline residues, and highly conserved 22-amino acid immunogenic epitopes called the Non-Proline Block (NPB). Although this region has a partly variable sequence, it is serologically highly cross-reactive and elicits antibodies against the PRD region, which can passively protect mice from lethal pneumococcal disease [21, 23].

Increasing evidence strongly proposes that a single protein, especially PspA from one family or clade, will not be sufficient to stimulate protection against all pneumococcal strains [12, 24, 25]. Higher levels of cross-reactivity have been reported within the same family, not between families, and the family-elicited immunity is clade-dependent [7, 22]. Therefore, at least one fragment from each of two prominent families has been considered for PspA-based vaccines to extend protection [22]. Akbari et al. showed that immunization of mice with PspAB1-5 (B region of N-terminal from all PspA clades) led to higher protection than PspA4ABC (A, B, and C regions of PspA Clade 4) in pneumococcal challenges [7]. Other studies showed that the combined vaccine candidate composed of two segments of each PspA family exhibited varying degrees of cross-reactivity and protection. Piao et al. showed that in three constructs of the PspA, including N-terminal and proline-rich regions from PspA families 1 and 2, immunization with PspA2 + 4 and PspA2 + 5 exhibited no protection against pneumococcal challenge with two Clades 1 and 3. Also, the binding capacity of the anti-PspA3 + 2 specific IgG to the surface of pneumococci with PspA Clades 1–4 was high, but not for Clade5. Finally, they concluded that PspA3 + 2 has an advantage over PspA2 + 4 and PspA2 + 5 [7, 25]. Akbari et al. also suggested that all B and C regions of all clades should be used in PspA-based vaccine designs to achieve the full level of cross-reactivity and cross-protection against all pneumococci [7]. Therefore, an essential step for PspA-based vaccine design is to cover and overcome clade-dependent immunity against all pneumococcal strains expressing all PspA families by selecting immunodominant truncated domains of all PspA clades focusing on two cross-reactive B and C-regions. This can be achieved through cost and time-benefit approaches such as immunoinformatics tools in vaccine development. Many studies have reported that immunoinformatics, reverse vaccinomics, or computational immunological approaches are reliable, accurate, quick, and cost-effective methods, with a broad collection of available and powerful tools for epitope-based vaccine design and vaccine development [26,27,28,29,30]. Therefore, the present study is the first attempt to use immunoinformatics tools for epitope mapping analysis of the N-terminal sequence of all five PspA clades. Then we designed and constructed the PspA1-5c+p vaccine candidate and evaluated the PspA1-5c+p protection against pneumococcal infection by immunization of mice with recombinant PspA1-5c+p. We also evaluated the cross-reactivity ability of the anti-PspA1-5c+p antibody against pneumococcal strains representing both PspA families and the functional activity of the anti-PspA1-5c+p antibody. The findings suggest the potential use of this vaccine candidate as a novel serotype-independent PspA-based pneumococcal vaccine with a strong cross-reactivity response. The schematic procedure of this research has been shown in Fig. 1.

Fig. 1
figure 1

The workflow of the PspA-based vaccine designing against pneumococcus

Materials and methods

Sequence retrieval and the structural characteristics

PspA sequence of the pneumococcal strain St 435/96 (Clade1), RX1 (Clade2), EF3296 (Clade3), EF5668 (Clade4), and strain ATCC 6303 (Clade5), which represented five different clade types of the PspA, were retrieved from the National Center for Biotechnology Information (NCBI) at https://www.ncbi.nlm.nih.gov with the accession number of the GenBank AAL92492.1, AAA27018.1, AAF27712.1, AAC62252.1, and AAF27715.1, respectively. To analyze the subcellular localization, transmembrane helices, and signal peptide of the proteins, online servers of CELLO v.2.5 (http://cello.life.nctu.edu.tw/) [31], TMHMM (http://www.cbs.dtu.dk/services/TMHMM-2.0/) [32], and signalP-5.0 (http://www.cbs.dtu.dk/services/SignalP-5.0/) [33] were used, respectively. The antigenicity of the proteins was also analyzed using the Vaxijen-v2 server (http://www.ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen.html) [34] with a threshold of 0.5.

Immunoinformatics analysis

Potential B-cell epitope prediction

For the prediction of linear B-cell epitopes, antigenicity, surface accessibility, flexibility, β-turn, and hydrophilicity of proteins, the amino acid sequences were analyzed using a collection of methods in the IEDB server (https://www.iedb.org/) [35] according to sequence characteristics of the antigen and BCpred (http://ailab.ist.psu.edu/bcpred/predict.html) [35] server. BCpred server uses support vector machine (SVM) classifiers for linear B-cell epitope prediction [35]. Since the three-dimensional (3D) structural template of PspA proteins did not exist in the Protein Data Bank (PDB) database, the prediction of the 3D structure of each PspA clade was performed using the I-TASSER server as a template-based fragment assembly simulator (https://zhanglab.ccmb.med.umich.edu/I-TASSER/) [36]. After prediction, the best-modeled structure for each clade was selected according to the model’s ranking scores. The acceptable confidence score (towards + 2), template modeling score (toward a score of 1), and a lower Root Mean Square Deviation score (RMSD) as a degree of similarity between the modeled protein and the other were considered. Then the selected model was refined using the Galaxy Refine server (http://galaxy.seoklab.org/cgi-bin/submit.cgi?type=REFINE) [37]. This structure refinement server carries out repeated structural disruptions by reconstruction of the side chains and subsequent overall structural relaxations through molecular dynamics simulation [37]. The final model was validated with a score of similarity to native proteins as a Z-score in the ProSA server (https://prosa.services.came.sbg.ac.at/prosa.php/) [38], distribution of residues in most favored, allowed, and disallowed regions of Ramachandran plot in the PROCHECK server, and quality percentage of structure in the ERRAT from Saves server (https://saves.mbi.ucla.edu/) [39]. Lastly, we used the refined and validated PspA clade 3D structure in the prediction of conformational B-cell epitopes using the Ellipro server (http://tools.iedb.org/ellipro/) [40].

Helper T-cell (HTL) epitope prediction

Several servers such as IEDB (https://www.iedb.org/) [41], RANKPEP (http://imed.med.ucm.es/Tools/rankpep.html) [42], and MHCPred (http://www.ddg-pharmfac.net/mhcpred/MHCPred/) [43] servers were used for predicting MHC-II binding epitopes. The prediction of (15 mer) peptide binding affinities to eight human HLA-II super type alleles HLA-DR B* (01:01-03:01-04:01-07:01-08:01-11:01-13:01-15:01), as well as mouse MHC-II H2 alleles I-Ad, I-Ab, and I-Ed, were applied to have strong T cell responses in human and BALB/c mice [44,45,46,47]. RANKPEP server utilizes protein sequence/s or sequence alignments by Position Specific Scoring Matrices (PSSMs) to predict peptide binders to MHCII molecules [42], and the MHCPred server employs the additive method for prediction of the MHCII binding affinity of molecules and to the Transporter associated with Processing (TAP) [43]. The predicted epitope sequences were analyzed for the presence or absence of similarity with the human proteome using the PIR peptide matching program (https://research.bioinformatics.udel.edu/peptidematch/index.jsp) [48].

Construction of chimeric PspA1-5c+p

The truncated domain from five PspA clades composing overlapping immunodominant B and T-cell epitopes in CDR sequence and the motif sequences from each of three PRD groups with the highly conserved and immunogenic region of NPB [21] were used in the formulation of chimeric PspA1-5c+p construct. The above regions were assessed to fuse as a multi-component vaccine using an EAAAK linker [49] to achieve the best three-dimensional structure with the least interaction between each part. The antigenicity and toxicity were predicted for the chimeric PspA1-5c+p construct using Vaxijen-v2 [25] with a threshold of 0.5 and ToxinPred servers (https://webs.iiitd.edu.in/raghava/toxinpred/index.html) [50], respectively.

Physicochemical properties and solubility prediction

The Protparam online server (https://web.expasy.org/protparam/) [51] was used to evaluate the physicochemical properties of the PspA1-5c+p construct. SOLpro server (http://scratch.proteomics.ics.uci.edu/explanation.html#SOLpro/) was applied to predict the solubility of PspA1-5c+p construct upon overexpression in E.coli [51, 52].

Prediction of secondary and tertiary structures

The prediction of the PspA1-5c+p secondary structure was performed by GORV (https://npsa-prabi.ibcp.fr/cgi-bin/npsa_automat.pl?page=/NPSA/npsa_gor4.html/), PSIPRED 4.0 (http://bioinf.cs.ucl.ac.uk/psipred/) [47], and RaptorX Property web servers (http://raptorx.uchicago.edu/StructurePropertyPred/predict/). Using I-TASSER (https://zhanggroup.org/I-TASSER/) server [36], the tertiary structure of the PspA1-5c+p was built. PyMol software v.2.5 was applied to visualize a high-quality image of the predicted model [53]. The best-modeled structure of the PspA1-5c+p construct was selected according to the C-score, TM-score, and RMSD-score.

Tertiary structure refinement and validation

The selected 3D structure model of the PspA1-5c+p was refined using the 3DRefine (http://sysbio.rnet.missouri.edu/3Drefine/index.html/) [54] and then Galaxy Refine (http://galaxy.seoklab.org/cgi-bin/submit.cgi?type=REFINE/) [37]. 3Drefine has refined the 3D model to optimize the hydrogen-bonding network and minimize the atomic energy of the model. Galaxy Refine server refined the model by molecular dynamics simulation. The final model of the PspA1-5c+p was validated with ProSA (https://prosa.services.came.sbg.ac.at/prosa.php/) [38], PROCHECK, and ERRAT from the Saves server (https://saves.mbi.ucla.edu/) [39], to recognize the errors in the generated 3D model. We also used the refined and validated 3D structure of the PspA1-5c+p to predict conformational B-cell epitopes using the Ellipro server (http://tools.iedb.org/ellipro/) [55].

Molecular dynamic simulation

The molecular dynamic (MD) simulation was carried out to study the stability of the PspA1-5c+p protein model using GROMACS 2018 [56, 57]. The structure was simulated for 85 ns with optimized potential for liquid simulations (OPLS) force field in a 10 A° cubic simulation box of the simple point charge (SPC) as a water molecule type. The neutralization of the system charge was done by adding Na+‏ and Cl ions. Then simulation system was minimized by the steepest descent minimization integrator and the maximum force was less than 100 kJ.mol−1 nm−1 with 5000 minimization steps. Afterward, the equilibration of the simulation system was performed with NVT (constant number of particles, volume, and temperature) and NPT (constant number of particles, pressure, and temperature with a leap-frog integrator for 100 picoseconds (ps). All bond constraints were organized with the linear constraint solver (LINCS) algorithm. The electrostatic interaction of the particle mesh Ewald (PME) algorithm was calculated through a 1.0 nm radius cut-off and grid spacing of 0.16 nm. The equilibrated system was subjected to MD simulation with two femtoseconds (fs) time steps, which continued until the system was stable. The output trajectory analysis includes Root Mean Square Deviation (RMSD) and Root Mean Square Fluctuation (RMSF) [57].

Molecular docking

For performing docking simulations, the 3D structure of HLA-DR1 (DRB1*0101) Human class II histocompatibility protein (PDB id: 1AQD) [58] was retrieved from RCSB (www.rcsb.org) server. The PDB structure of 1AQD was refined by removing the present ligand from the crystal structure using UCSF Chimera v.1.14 software. The ClusPro 2.0 server (http://nrc.bu.edu/cluster/) was used for PspA1-5c+p-HLA-DRB1 docking [46]. Finally, the model of the ClusPro with the largest cluster size and the lowest binding free energy was checked for the interaction of the amino acids using the DimPlot tool in LigPlot + v.2.2.4 software and PYMOLv. 2.5 software [46]. Because the CDR region of the PspA molecule as a lactoferrin binding domain can attach to the N-lobe of the human lactoferrin to block surface accessibility of this bactericidal peptide [59], we performed a docking simulation between the PspA1-5c+p construct and human lactoferrin N-lobe (PDB id: 2PMS) to furthermore validation of the 3D structure of the modeled PspA1-5c+p. The details of the interaction of the control PspA (Clade 2) with lactoferrin will help us to compare our designed construct with it in order to find out whether this designed PspA1-5c+p is able to bind to lactoferrin like the original structure of PspA and maintain its 3D structure as in the natural state or not? The mentioned docking steps were repeated for molecular docking and further analysis of the PspA1-5c+p construct and human lactoferrin N-lobe.

In-silico immune response simulation

IL-4, IL-10, and IFN-γ inducing peptide construction of PspA1-5c+p were predicted via the IL-4pred server (https://webs.iiitd.edu.in/raghava/il4pred/scan.php), IL-10Pred server (https://webs.iiitd.edu.in/raghava/il10pred/predict3.php), and IFNepitope server (https://webs.iiitd.edu.in/raghava/ifnepitope/predict.php), respectively. These servers can generate overlapping peptides of the query protein/antigen sequence and predict the cytokine-inducting ability of these antigenic regions. Furthermore, in-silico mammalian immune simulations against PspA1-5c+p were designed utilizing the C-ImmSim server (http://150.146.2.1/C-IMMSIM/index.php). This immune response simulator server uses machine learning methods and a position-specific score matrix (PSSM) to predict immune interactions. According to the literature, three injections of the PspA1-5c+p construct were administered at intervals of 4 weeks, on days 1, 30, and day 60. For this purpose, the simulation parameters were set as follows: vaccine injection not containing LPS, time steps at 1, 84, and 168; the entire simulation ran 1400 time steps (about 15 months), the random seed of 12,345, and the simulation volume of 10. Each step is 8 h. HLA alleles of parameters were also set based on predominant human HLA alleles (HLA-A*1101, HLA-B*3501, and HLA-DRB1*0101) [55, 60].

Codon adaptation, in-silico cloning, and RNA structure

Using the Genscript (https://www.genscript.com/tools/rare-codon-analysis/) and NovoPro servers (https://www.novoprolabs.com/tools/codon-optimization/) [61], the codon adaptation of the PspA1-5c+p sequence was carried out according to E. coli K12 codon usage. We used the mFold server (http://www.unafold.org/mfold/applications/rna-folding-form.php/) [62] and Visual Gene Developer software [63] to analyze the mRNA secondary PspA1-5c+p structure mainly by using thermodynamic methods (the Gibbs free energy). In-silico cloning of the designed PspA1-5c+p sequence was carried out in pET-28a using the SnapGene 6.0 software. Finally, Biomatik Corporation (Cambridge, Ont., Canada) synthesized the optimized sequence of PspA1-5c+p [53, 60].

Expression and purification of recombinant PspA1-5c+p

The recombinant (PspA1-5c+p -pET28a) vector was transformed into the chemically prepared competent E. coli BL21 (DE3) cell via heat shock transformation. Positive clones were recognized by restriction enzyme digestion and colony PCR with universal T7 primers (https://www.addgene.org/mol-bio-reference/sequencing-primers/). Expression of the recombinant PspA1-5c+p was induced by adding 1 mM Isopropyl-β-D, Thiogalactopyranoside (IPTG) (Thermo Fisher Scientific, USA) in Luria–Bertani Broth (LB) medium (Sigma Aldrich, USA) at 37 °C and incubated for 16 h, then was evaluated by 12% SDS-PAGE, and using HRP-conjugated anti-His tag antibody (Sigma, USA) confirmed by western blot analysis. By the manufacturer's instructions (Qiagen, Hilden, Germany), the recombinant PspA1-5c+p was purified using a Ni–NTA column (Qiagen, Hilden, Germany) under native conditions. The purified PspA1-5c+p was dialyzed with dialysis tubing (cutoff 12KDa) overnight at 4 °C against PBS (Sigma, USA) and measured by the Bradford protein assay [47]. A Limulus amebocyte lysate (LAL) test was done to assess LPS contamination using the LAL kit (Lonza QCL-1000 ®, Basel, Switzerland).

Animal and immunization

Six to eight-week-old male BALB/c mice were purchased from the Pasteur Institute of Iran (Karaj, Iran) and immunized intraperitoneally three times at 14-day intervals with 10 μg of recombinant PspA1-5c+p construct in PBS solution plus the Alum adjuvant (Imject TM Alum, Thermo Fisher Scientific, USA) at 1:1 (v/v) in a final volume 200 μl per mouse. The control group was injected with PBS and Alum. Before injections and two weeks after the last injection, sera from blood samples in each group were collected and stored at − 20 °C [47, 64].

Ethics

All animal experiments were done in accordance with the Institutional Animal Care and Use Committee’s guidelines Animals (Scientific Procedures) Act of Pasteur Institute of Iran and Islamic Azad University-Science and Research Branch. Ethical approval was obtained from the Institutional Research Ethics Committee, Islamic Azad University-Science and Research Branch (approval ID: IR.IAU.SRB.REC.1398.065).

Assessment of immune response

The presence of specific IgG antibodies was analyzed in the experimental sera using indirect ELISA. Briefly, the 96-well ELISA plate (Nanc MaxiSorp, Thermo Fisher, USA) was coated with 100 μl of the recombinant PspA1-5c+p (1 μg/well) in coating buffer (0.05 M carbonate bicarbonate buffer, pH 9.6) overnight at 4 °C, then blocked with 5% bovine serum albumin (BSA; Sigma, USA) in PBST (PBS containing 0.05% Tween20). Following, three times wash with PBST; 100 μl of 0.001 diluted sera in blocking buffer was added to the plate and incubated for one hour at 37 °C. Afterward, the 1:10,000 dilution of HRP-conjugated anti-mouse total IgG (Sigma, USA) was used and incubated for one hour at 37 °C. After washing, the plate was incubated with the tetramethylbenzidine (TMB) substrate (Thermo Fisher Scientific, USA) to evaluate antibody reactivity at 450 nm using an Epoch absorbance microplate reader (BioTek Company) [47, 65].

Cross-reactivity analysis of anti-PspA1-5c+p using whole cell ELISA

The sera were analyzed for investigation of the cross-reactivity of the anti-PspA1-5c+p IgG against three strains of the pneumococcus, representing two families of the PspA including strains ATCC 49619 (Clade 1), ATCC 6305 (Clade 2), and ATCC 700678 (Clade 5) using the whole-cell ELISA test according to the method described by Ahmadi et al. [7]. Briefly, the 96-well ELISA plate was coated with 100 μl of the whole cell of bacterial suspensions overnight at 4 °C, which were grown to log phase in BHI broth (107 cells/well) and then blocked with 10% BSA in PBST. The practical steps continued as described above (indirect ELISA).

Serum bactericidal assay (SBA)

The SBA assay was performed to evaluate the complement-mediated killing features of the anti-PspA1-5c+p antibody against three strains of the pneumococcus expressing two families of the PspA. For this purpose, Thermo Scientific Nunc™ 96-Well Polystyrene Round Bottom microwell plates were coated with 12.5 μl of the three strains of pneumococcus at 105 CFU/ml (based on the standard of 0.5 McFarland) separately, and 12.5 μl of diluted inactivated serum sample at 56 °C for 30 min (1:2 to 1:64). Afterward, fresh infant rabbit serum (4%) was added to each well as a source of the complement. At two intervals (0 and 2 h), the sample from each well was cultured in blood agar media. After 18–24 h incubation at 37 °C in 5% CO2, the colony-forming unit of the bacteria was counted. The wells containing bacteria and rabbit complement were used as a negative control [64, 65].

Opsonophagocytic killing activity (OPK)

The serum of immunized BALB/C with the PspA1-5c+p construct was evaluated for Opsonophagocytic killing (OPK) activity of the anti-PspA1-5c+p antibody using phagocyte cells against three strains of the pneumococcus, strain ATCC 49619, ATCC 6305, and ATCC 700678. Pneumococcus strains were prepared at 107 CFU/ml (based on the standard of 0.5 McFarland). For macrophage cell collection from the peritoneal cavity of the naïve mice, 10 ml of the RPMI and FBS 10% were inoculated intraperitoneally in anesthetized mice. Afterward, the aspirated contents of the peritoneum were washed with RPMI and 10% FBS, and finally, live phagocyte cells were measured by the Neubauer slide. For OPK assay, the 100 μl of inactivated sera were incubated with 100 μl of the pneumococcus strains, and then 100 μl of the phagocyte cells (1 × 106 cells/ml) and infant rabbit serum (4%) were added. Subsequently, 25 μl of the sample was cultured on a blood agar plate at two intervals (0 and 90 min). After 18–24 h incubation at 37 °C in 5% CO2, the colony-forming unit of bacteria was counted. The assessment of the opsonic activity of the anti-PspA antibody against the three pneumococcus strains compared to the PBS group was measured using the following formula:

Percentage of killed bacteria = [1 − (CFU of immune serum/CFU of pre-immune serum)] × 100 [47, 65].

Statistical analysis

Statistical analysis was conducted using GraphPad Prism 6 software. One and two-way analysis of the variances (ANOVA) followed by Tukey’s multiple comparison test was performed for the analysis of immune responses. All experiments were performed in triplicate and expressed as the average ± S.D. P-values of less than 0.05 were considered statistically significant.

Results

Sequence retrieval

Additional file 1: Table S1 showed the results of sequence retrieval of five PspA clade types from the NCBI server, the subcellular localization, transmembrane helices, and signal peptide of the proteins. The results of the subcellular localization showed PspA proteins in five clades are extracellular or cytoplasmic localization. The results of the online server TMHMM showed the PspA proteins have a maximum of one transmembrane helices. The transmembrane helices prediction can help us to predict the state of the cloning, expression, and purification of the recombinant protein. The higher antigenic proteins with zero or one transmembrane helices were selected for the development of a vaccine, and proteins with multiple transmembrane helices should be eliminated due to their difficult cloning, expression, or purification [32]. The online server signalP-5.0 also showed PspA proteins have one signal peptide at positions 31 and 32.

Immunoinformatics analysis

Defining B‑cell epitopes

The schematic results of the IEDB server are shown in Additional file 1: Fig. S1. The immunodominant overlapped predicted linear B-cell epitopes with high antigenicity, surface accessibility, flexibility, and hydrophilicity in CDR regions of each PspA were selected using BCPred and IEDB servers and provided in Additional File1 Table S2. According to the B-cell epitope prediction servers, the cross-reactive regions of each clade had at least five B-cell epitope sequences of ~ 6–25 mer in length with VaxiJen scores of 0.5 to 1. Some of the predicted epitopes had antigenicity scores of 2 to 3.3. According to the results of the 3D structure predictions of the PspA clade using the I-TASSER server, Model 1 with the highest C-score was chosen for refinement with the GalaxyRefine server. After the refinement of the predicted structure of each clade, Ramachandran plot analysis and ProSA validation are shown in Additional File1 Figure S2. Analysis of the predicted models for each PspA clade with C-score values, TM-score, Galaxy-refine scores, Rama favored score, ERRAT score, MolProbity score, and ProSA Z-score are shown in Additional file 1: Table S3. The continuous predicted B-cell epitopes in each PspA clade are also shown in Additional file 1: Fig. S1 and Table S4.

Helper T-cell (HTL) epitopes

The results of predicted helper T-cell epitopes from PspA CDRs using IEDB (percentile rank < 20), RANKPEP server, and MHCPred server (IC50 ≤ 100 nM) are shown in Additional file 1: Tables S5–S7. Eight human HLA-DRB1 alleles (DRB1*01:01, DRB1*03:01, DRB1*04:01, DRB1*07:01, DRB1*08:01, DRB1*11:01, DRB1*13:01, and DRB1*15:01) and three mouse alleles (H2-IAb, H2-IAd, and H2-IEd) had been considered for predictions [44,45,46,47]. The results of the PIR peptide-matching program showed that the predicted epitope sequences had no similarity with the human proteome. Some predicted B-cell epitopes were also predicted as HTL epitopes.

Subunit PspA1-5c+p vaccine construction

The chosen sequences from each PspA clade contain overlapped immunodominant regions of the surface and high antigenic epitopes of B and T-cells, as the truncated domain of PspA residues from each clade. Position 193 to 294 AA from Clade 1 and position 223 to 318 AA of Clade2 PspA were located in the N-terminal of the PspA1-5c+p construct as a representative of the truncated domains of PspA Family1. Based on Mukerji’s study [21], we selected sequences representing each PRD group’s repetitive motif. The new region of the proline-rich domain with the highly conserved and immunogenic region of NPB and PR epitopes was located in the center of the PspA1-5c+p construct. Finally, the truncated domain of PspA from Clades 3, 4, and 5 was located in the C-terminal of the PspA1-5c+p construct, representing the PspA Family2. Position 346 to 444 AA of Clade 3, position 276 to 374 AA of Clade 4, and position 273 to 392 AA of Clade 5 were chosen. The above regions were fused with the EAAAK linker. In addition, a 6xHis tag was added to the N and C terminus for easy protein purification (Fig. 2). The final PspA1-5c+p consisting of 614 amino acid residues was antigen and non-toxic. The amino acid sequence of the final PspA1-5c+p construct has been shown in Table 1.

Fig. 2
figure 2

Schematic representation of the final PspA1-5c+p vaccine candidate. To cover the maximum cross-reactivity between PspA two families and the diversity of all PRD groups, the 614 amino acid long peptide sequence contains immunodominant B-cell and T-cell epitopes as the truncated domain of the CDRs and the highly conserved region of NPB and repetitive motifs of PRD group. The CDR regions of the PspA family1 (orange) at the amino-terminal end are connected to the multi-epitope sequence of the PRD (green) via an EAAAK linker (yellow). PspA family 2 CDR regions (blue) are also linked using an EAAAK linker (yellow) together with PRD in the carboxy-terminal of the construct. Two 6 × His tags are added to the amino and carboxyl terminus of the construct for purification and identification purposes. For the cloning of the PspA1-5c+p construct into the pET28a vector, the restriction enzyme sites (NcoI and XhoI) are considered

Table 1 The amino acid sequence of the final PspA1-5c+p construct

Physicochemical properties and solubility prediction

Using the ProtParam server, the molecular weight (MW) of the final PspA1-5c+p construct was predicted to be 67.93 kDa. The theoretical isoelectric point value (pI) was 4.39. The instability index (II) was calculated to be 40.12. The aliphatic index and grand average of the hydropathicity (GRAVY) were estimated to be 82.23 and -0.997, respectively. The half-life was assessed to be 30 h in mammalian reticulocytes in vitro, > 20 h in yeast, and > 10 h in E. coli in-vivo [51]. Using the Solpro server, the PspA1-5c+p construct was predicted to be soluble upon overexpression in E. coli with a solubility probability score of 0.905.

Prediction of the secondary tertiary structures of PspA1-5c+p

The GOR V prediction server reported that the final PspA1-5c+p secondary structure contains 83.22% alpha-helix, 0.49% extended strand, and 16.28% random coil. The secondary structure of the final PspA1-5c+p using PSIPRED prediction is shown in Additional file 1: Fig. S3. RaptorX Property server reported 199 residue positions (32%) as disordered. The I-TASSER server predicted five models of tertiary structure for the PspA1-5c+p construct based on 10 threading templates. The five predicted models for the PspA1-5c+p construct had C-score values between − 3.65 and − 0.58. Model 1, with the highest C-score of − 0.58, was chosen for further refinement (Fig. 3a). This model had an estimated TM-score and RMSD of 0.64 ± 0.13 and 9.1 ± 4.6 Å, respectively.

Fig. 3
figure 3

PspA1-5c+p modeling, refinement, validation, and prediction of the conformational B-cell epitopes. a The I-Tasser 3D homology modeling of the PspA1-5c+p construct before and after refinement was displayed by PyMol v.2.5 software. Each domain is indicated in color. Validation of the model before and after refinement using b ProSA Web, c Ramachandran plot, and d ERRAT plot. The ProSA Web analysis shows the z-scores of − 4.41 and − 5.14 before and after refinement, respectively, and the plot of the residue scores showing local model quality by plotting energies as a function of amino acid sequence position is also shown. Ramachandran plot analysis after refinement showed 92.4%, 6.4%, and 1.2% of PspA1-5c+p protein residues were in preferred, valid, and non-valid (outlier) regions, respectively. In general, positive values correspond to problematic or erroneous parts of the input structure. The overall quality score of the selected model before and after refinement is 89.66% and 98.14%, respectively, using the ERRAT2 server. These values are expressed as the percentage of the protein for which the estimated error value falls below the 95% rejection limit. Two lines in the error axis reveal the confidence with which it is possible to eliminate areas that exceed this error value. Good high-resolution structures generally produce values around 95% or higher. e The conformational B-cell epitopes using the Elipro server on a refined and validated final 3D PspA1-5c+p model were predicted to be located in seven conformational B-cell epitopes. The conformational B-cell epitopes are shown in yellow and the gray parts are the rest of the residues

Tertiary structure refinement and validation

We used two servers for refinement. The optimal final model from Galaxy Refine was found to be based on various parameters including Ramachandran plot score (94.6%), RMSD (0.244), MolProbity (1.892), GDT-HA score (0.9926), clash score (10), and poor rotamers score (0.6). This model was selected as the final PspA1-5c+p vaccine candidate model for further investigation. The quality and potential errors in the 3D model were verified by ProSA-web and ERRAT. The PspA1-5c+p protein fell close to the range of scores commonly found in native proteins of comparable size with the ProSA-web Z-score of − 5.14 (Fig. 3b). The Ramachandran plot analysis of the Saves server revealed that, among the 608 residues, 526 (92.4%) and 36 (6.4%) in the protein were in the most favored and allowed regions, respectively. There were only seven residues (1.2%) in the disallowed region, indicating that the predicted model is acceptable (Fig. 3c). The overall quality factor of the chosen model after the last refinement was 98.14% (Fig. 3d). Then, using the Elipro server, the prediction of conformational B-cell epitopes was performed on a refined and validated 3D final PspA1-5c+p model (Fig. 3e). The results revealed that a total of 608 residues were situated in seven conformational B-cell epitopes, with scores ranging from 0.514 to 0.818. The conformation epitopes ranged in size from three to 91 residues. The details of conformational B-cell epitope prediction are presented in Table 2. The PspA1-5c+p final construct was also analyzed to be predicted as non-toxic and immunogenic, with an antigenicity score of 0.77.

Table 2 The conformational B-cell epitope prediction for the final PspA1-5c+p construct using EliPro server

Molecular dynamic simulation

To assess the stability and dynamics of the designed PspA1-5c+p vaccine candidate, molecular dynamic simulation was performed until the protein structure reached the stability state (Fig. 4). To find PspA1-5c+p conformational changes or stability against the initial structure, the root mean square deviation (RMSD) based on the structure of the backbone was applied. The RMSD plot analysis revealed that the protein structure deviated until 50 ns and then reached a plateau with a maximum RMSD value of 2.45 nm. To evaluate the fluctuated residues of PspA1-5c+p protein, root mean square fluctuation (RMSF) was measured. The plot of RMSF showed the RMSF values less than 0.35 nm for almost residues, indicating low changes in the structure of PspA1-5c+p protein. But the C-terminal residues of PspA1-5c+p showed more flexibility with a RMSF value of 0.7 nm (Fig. 4b).

Fig. 4 
figure 4

Molecular Dynamics simulation of the PspA1-5c+p protein. a The RMSD plot of the PspA1-5c+p showed the steady state of the model at 50 ns. b The RMSF plot of the PspA1-5c+p protein revealed the low flexibility for residues, except for the C-terminal region

Molecular docking results

ClusPro online server performed molecular protein–protein docking between refined PspA1-5c+p and HLADRB1*01:01 (the most common binding allele in the Iran population). Cluster No. 0.00 of PspA1-5c+p-HLADRB1 docked complex with 64 members having the lowest energy of -744.3 kcal.mol−1 were selected for further analysis. The interaction surface residues of the docked complex were checked with Dimpolt tools in LigPlot+ software and visualized using PyMol software (Fig. 5). A total of 7 and 6 PspA1-5c+p residues coupled with 6 and 5 residues of A and B chains from HLADRB1*01:01 molecule, respectively. Altogether, a number of 16 hydrogen bonds and 6 salt bridges, and many hydrophobic bonds, were formed between the PspA1-5c+p residues and HLADRB1*01:01 molecule (Fig. 5 and Table 3). We docked the PspA1-5c+p construct with Human Lactoferrin N-lobe (HLF) to furthermore 3D structure validation of modeled PspA1-5c+p and analysis of 3D structure conformation preserving of each CDR region in the vaccine formulation. The structure of the PspA (clade2)-HLF docked complex with PDB id: 2PMS was used as a control. The output result of the ClusPro server exhibit 30 clusters for the docked complex ranked (0–29) according to cluster members with weighted scores of the cluster energies. Cluster No. 0.00 and 2.00 of the PspA1-5c+p-HLF docked complexes were chosen, which had different CDR binding positions to HLF, and maximum cluster members of 80 and 58 with the lowest energy of -987.2 and -1128.9 kcal.mol−1, respectively. The interaction surface residues of the PspA1-5c+p-HLF docked complex and PspA (clade2)-HLF control docked complex were analyzed with Dimpolt tools in LigPlot + software and visualized using PyMol software (Fig. 6 and Table 4). The results of comparisons between the PspA1-5c+p-HLF docked complex with PspA (clade2)-HLF control docked complex showed that PspA1-5c+p construct could be connected to HLF molecule via both regions representing PspA families 1 and 2 in PspA1-5c+p construct same as in a control docked complex. In coordination with the docked control molecule, in two models of the PspA1-5c+p-HLF docked complex, the most residues of the HLF that have been in contact with CDR residues include Arg4, Arg5, Arg25, Arg28, Arg31, Arg40, Gln14, Gln24, and lys39. The details of the number of hydrogen bonds and salt bridges, the name and number of residues involved in the interactions, and altogether the details of interaction-docked complexes are shown in Figs. 7 and 8, and Table 4. These results can be shown the good 3D structure conformation preserving of the CDR region in the PspA1-5c+p vaccine formulation.

Fig. 5
figure 5

Molecular docking of the PspA1-5c+p and HLADRB1*01:01 (Chains A and B). a The 3D structure of the PspA1-5c+p before docking. b The 3D structure of the HLADRB1*01:01 (Chains A and B) (PDB ID: 1AQD) before docking. c Docked complex of the PspA1-5c+p with HLADRB1*01:01. The cartoon representation of the PspA1-5c+p-HLADRB1*01:01 complex is illustrated using the PyMOL software. The PspA1-5c+p, chains A and B of the HLADRB1*01:01 are shown in magenta, green, and cyan, respectively. The lowest energy value of this complex model was -744.3 kcal.mol−1, indicating good binding affinity. d Dimplot interaction diagram between PspA1-5c+p residues and the HLADRB1*01:01 molecule in the docked complex. A number of 7 and 6 residues of PspA1-5c+p were paired with 6 residues from chain A and 5 residues from chain B of the HLA-DRB1_01:01 molecule, respectively. PspA1-5c+p residues, HLADRB1*01:01 residues, hydrogen bonds, salt bridges, and unbound residues are exhibited in green, blue, blue dashed lines, green dashed lines, and red/pink eyelashes, respectively

Table 3 Analysis of Dimplot 2D-interaction plot between PspA1-5c+p residues and HLADRB1*01:01 molecule in the docked complex
Fig. 6
figure 6

a Molecular docking of the PspA (clade2) and HLF as control of docking. Two 3D structures of the docked complex of the PspA with HLF have been shown. The surface depictions of the PspA-HLF complex are illustrated using the PyMOL software. The PspA and HLF have been shown in red and green, respectively. b Dimplot interaction plot between PspA (clade 2) residues and HLF molecule in the docked complex as a PDB:2PMS control complex. PspA (Clade2) residues, HLA residues, hydrogen bonds, salt bridge, and non-bonded residues are exhibited in green, blue, blue dashed lines, green dashed lines, and red/pink eyelashes, respectively

Table 4 Dimplot interaction plot analysis between PspA1-5c+p residues and HLF molecule in the docked complex and control. a The interaction detail between PspA (clade2) residues and HLF A and B chains (as a docked control model). b The interatin detail between model 0 of PspA1-5c+p residues (Family 2 Domain) and HLF A and B chains. And c The interaction detail between model 2 of PspA1-5c+p residues (Family 1Domain) and HLF A and B chains
Fig. 7
figure 7

a Molecular docking of the PspA1-5c+p (Family1 domain) with HLF (Chain A and B). The surface depictions of the PspA1-5c+p-HLF complex are illustrated using the PyMOL software. The PspA1-5c+p, chains A and B of the HLF, have been shown in red, green, and cyan, respectively. b Dimplot interaction plot between PspA1-5c+p residues (Family1 domain) with HLF molecule in the docked complex (Cluster No. 0.0). PspA1-5c+p residues, HLA residues, hydrogen bonds, salt bridge, and non-bonded residues are exhibited in green, blue, blue dashed lines, green dashed lines, and red/pink eyelashes, respectively

Fig. 8
figure 8

a Molecular docking of the PspA1-5c+p (Family2 domain) with HLF (Chains A and B). The surface depictions of the PspA1-5c+p-HLF complex are illustrated using the PyMOL software. The PspA1-5c+p, chains A and B of the HLF, have been shown in red, green, and cyan, respectively. b Dimplot interaction plot between PspA1-5c+p residues (Family2 domain) with HLF molecule in docked complex (Cluster No. 2.0). PspA1-5c+p residues, HLA residues, hydrogen bonds, salt bridge, and non-bonded residues are exhibited in green, blue, blue dashed lines, green dashed lines, and red/pink eyelashes, respectively

Codon adaptation, in-silico cloning, and RNA structure

Codon Adaptation Index (CAI) and GC content of the optimized codon sequence of PspA1-5c+p with a length of 1826 bp in E. coli (strain K12) were 0.84 and 42.97%, respectively. These results showed good efficiency of the final vaccine candidate transcription and translation in the E. coli host. After the codon optimization, we analyzed the PspA1-5c+p mRNA secondary structure (Fig. 9a). There was no observed unsuitable pseudoknot or loop at 5’ for transcription. The Gibbs free energy after sequence optimization for PspA1-5c+p construct mRNA was − 445.5 kcal.mol−1 showing the lowest free energy and stable structure. Finally, using SnapGene 6.0 software, we inserted the optimized codon sequence into the pET28a ( +) vector between NcoI (1978) and XhoI (158), forming a clone with a total length of 7051 bp (Fig. 9b).

Fig. 9
figure 9

The mRNA folding analysis and PspA1-5c+p construct in-silico cloning into the pET28a( +) expression vector. a The mRNA secondary structure of the PspA1-5c+p construct with Gibbs free energy of -445.5 kcal.mol−1. b In-silico cloning of the final PspA1-5c+p construct sequence into the pET28a ( +) expression vector. The yellow part represents the gene encoding PspA1-5c+p, and the black circle represents the vector backbone

In-silico immune response simulation

Using the IL-4pred server, overlapping peptides of the query PspA1-5c+p sequence were generated, and antigenic regions of PspA1-5c+p that have a potential for inducing IL-4 were predicted and shown in Additional file 1: Table S8. The results of the IL-10pred web server showed that the PspA1-5c+p construct with a score of 0.99, was predicted as an IL-10 inducer. Using the IFNepitope server, the PspA1-5c+p construct was scanned and predicted to have many IFN-γ inducing MHC class II binder peptides throughout its sequence. The maximum and minimum scores of IFN-γ inducer peptides were 2.2 and 0.3, respectively. Furthermore, the graph of the humoral and cellular response of the mammalian immune system against the PspA1-5c+p vaccine administrations using the C-ImmSim server showed an increase in the IgM titer characterizing a primary response. Following injection of the booster dose of PspA1-5c+p, an increase in B cell populations, isotype switching, and the formation of memory cells as well as an increase in immunoglobulin expression (IgG1 + IgG2, IgM, and IgG + IgM) were observed as secondary and tertiary reactions. In addition, an increase in Th (helper) and TC (cytotoxic) cells with memory development, natural killer cells, and dendritic cell responses was found. High levels of macrophage activity are also identified. In parallel, the immune simulation also showed that IFN-γ and IL-2 production were stimulated after immunization, resulting in increased macrophage activity. IL-10 as an immunosuppressive cytokine and the secreted factor by established Th2 cells was also found (Fig. 10). The results of the IL-4pred, IL-10pred, and IFNepitope predictions were consistent with the results of the C-ImmSim simulation and showed that the PspA1-5c+p construct potentially increases the production of both cytokines in both humoral and cellular immune pathways.

Fig. 10
figure 10

In-silico simulation of the immune response with PspA1-5c+p vaccine. A Diversity of the immunoglobulin production in response to PspA1-5c+p injections (PspA1-5c+p antigen, and IgM, IgG1, and IgG2 subclasses are shown as black and colored peaks, respectively). B Cytokines and interleukin levels. The inset graph shows the level of IL-2 with the Simpson index, D shown by the dotted line, as a measure of diversity and danger signal along with the leukocyte growth factor IL-2. C The evolution of B-cell and plasma cell populations after the three administrations. D The evolution of T-helper, T-cytotoxic cell, and T regulatory cell populations after the injections. The cells not presented with the antigen were defined as being in the resting state, while the anergic state indicates the tolerance of the T-cells to the PspA due to repeated exposure. The evolution of the Natural killer cell population (E), Macrophage population (F), and Dendritic cell population (G)

Expression, purification, and confirmation of recombinant PspA1-5c+p

The positive transformed E. coli BL21 clones containing recombinant plasmid were confirmed using restriction enzyme digestion with NcoI and XhoI (resulting in two bands with sizes of approximately 1826 bp and 5369 bp) and colony PCR with universal T7 primers (resulting in a single band with the size of about 1826 bp) (Fig. 11a, b). Then, the expression of the recombinant PspA1-5c+p protein was carried out with IPTG and analyzed with 12% SDS-PAGE. The results of the SDS-PAGE revealed the presence of a 67 kDa recombinant PspA1-5c+p band, as expected by MW calculations. The purification of the recombinant PspA1-5c+p was applied by Ni–NTA affinity chromatography under native conditions (Fig. 11c). In accordance with the solubility bioinformatics analysis of the PspA1-5c+p construct, the experimental analysis showed that the PspA1-5c+p construct was soluble. The expression of the PspA1-5c+p construct was verified by the western blot on PspA1-5c+p using an anti-His tag antibody (Fig. 11d). The LAL test showed an imperceptible level of LPS (< 0.5 EU/ml) in PspA1-5c+p solution.

Fig. 11
figure 11

a Double digestion of recombinant plasmid pET28a-PspA1-5c+p with NcoI and XhoI restriction enzymes. Lane1. 1kB DNA Ladder marker, 2. Undigested plasmid, 3. Double digestion of the plasmid with NcoI and XhoI, resulting in two bands of the PspA1-5c+p (~ 1826 bp) and pET28a (~ 5639 bp). 3. Monodigestion of plasmid with NcoI. b Colony PCR on transformed E. coli BL21 colonies with universal T7 primers. Lane1. 1kB DNA Ladder marker, lanes 2 to 6. Positive colonies with a band size of the PspA1-5c+p (~ 1826 bp). c Purification of recombinant PspA1-5c+p protein by Ni–NTA chromatography. d Western blot analysis of the purified recombinant PspA1-5c+p

Assessment of immune responses

Two weeks after the last immunization, the specific IgG level against the PspA1-5c+p construct was analyzed using ELISA. The group of mice that were immunized with the PspA1-5c+p construct and Alum revealed that the specific IgG level significantly increased compared to the control group (p < 0.0001) at different times of administration (Fig. 12). The results showed that the immunization of mice with this construct could stimulate the immune system response.

Fig. 12
figure 12

Evaluation of immune response in immunized mice. The booster effect of PspA1-5c+p administration and specific IgG levels at different times of the injections were shown. A significant increase was observed in the mice immunized with PspA1-5c+p at a dilution of 1:1000 compared with the control group. The error bar is representative of the mean ± standard deviation

Whole-cell ELISA analysis

The results of whole-cell ELISA analyzing the cross-reactivity and binding ability of anti-PspA1-5c+p IgG showed that anti-PspA1-5c+p IgG reacted strongly to the surface of all three pneumococcal strains (Fig. 13). No significant difference was seen between anti-PspA1-5c+p antibody cross-reactivity optical densities against all three pneumococcus strains representing two PspA families (p-value = 0.2).

Fig. 13
figure 13

a Evaluation of cross-reactivity of anti-PspA1-5c+p antibody using the Whole-cell ELISA. b The binding ability of anti-PspA1-5c+p IgG against the surface of three pneumococcus strains, representing two families of PspA as cross-reactivity response. The error bar is representative of the mean ± standard deviation

SBA analysis

The complement-mediated killing feature of the anti-PspA1-5c+p antibody against three strains of pneumococcus, expressing two PspA families, was assessed using serial dilutions of the prepared serum up to 1:64. The results of the SBA are shown in Fig. 14a. The highest bactericidal activity was detected in 1:4 dilution. This dilution showed the ability of the anti-PspA1-5c+p antibody to kill more than 50% of pneumococci compared to the control group. No significant difference was seen between the complement-mediated killing feature of the anti-PspA1-5c+p antibody against three strains of pneumococcus expressing three clades of PspA (0.8 < p-value < 0.9). No bactericidal effects were reported in the negative controls.

Fig. 14
figure 14

Evaluation of the functional activity of the anti-PspA1-5c+p antibody. a The complement-mediated killing feature of anti-PspA1-5c+p antibody against pneumococcal strains. b Phagocytosis-mediated killing feature of anti-PspA1-5c+p antibody against pneumococcal strains. The error bar is representative of the mean ± standard deviation

OPA analysis

The phagocytic killing capacity of mouse peritoneal macrophages and phagocyte cells exposed to anti-PspA1-5c+p antibody revealed a significant increase compared to the control group (p < 0.0001) leading to a more than 50% reduction in the number of bacteria. In addition, no significant difference was found between the phagocytosis-mediated killing feature of the anti-PspA1-5c+p antibody against pneumococcus strains ATCC6305, ATCC700678, or ATCC49619 (p = 0.2). No opsonic killing activity was observed in the PBS group. The data indicated that antibodies raised against PspA1-5c+p act as a good opsonin for killing pneumococcus strains representing both PspA families (Fig. 14b).

Discussion

A successful serotype-independent PspA-based vaccine against pneumococcus is a vaccine that comprises multiple immunological surface components with high cross-reactivity feature including various N-terminal domains of the PspA families [24, 66]. So, the purpose of designing this study was first to improve the PspA-based vaccine potency and efficacy using immunoinformatics tools as the first line of vaccine design. Then make experimentally a new construct with highly conserved and variable regions with high antigenic binding epitopes of B- and T-cell, with emphasis on cross-reactive regions of PspA N-terminal to evaluate the covering immune response against pneumococcal PspA clades.

To our knowledge, this is the first study on the immunoinformatics-based design of PspA families-based vaccines. We first modeled, refined, and validated the 3D structure of five different PspA clades using computational approaches. Then, we also predicted linear and conformational B-cell, and T-cell epitopes, especially at cross-reactive regions of PspA families 1 and 2 using various databases. According to B-cell epitope prediction servers (BCPred, IEDB, and Ellipro), the cross-reactive regions of each clade were analyzed. These regions had at least five B-cell epitope sequences of ~ 6–25mer in length with VaxiJen scores of 0.5 to 1. Some of the predicted epitopes had antigenicity scores of 2 to 3.3. The antigenicity score, surface accessibility, flexibility, hydrophilicity, beta-turn features, and conformational B-cell epitopes of predicted immunodominant regions were also considered suitable for designing cross-reactive PspA-based vaccines. As not all antibodies against PspA are protective; therefore, understanding which epitopes can elicit a protective response is critical [67]. So epitope mapping of PspA is one of the most widely used methods for identifying these epitopes. McDaniel et al. (1994) showed that the protection-eliciting regions of PspA were localized at 192–260 amino acid regions of PspA from the strain Rx1 using four of the nine monoclonal antibodies [68]. In accordance with McDaniel, we analyzed the B-cell epitopes of strain Rx1 as a clade 2 using immunoinformatics databases, and our results showed that the most predicted epitopes were located in the predicted region by McDaniel, which has thirteen sequences of 7–104 mer in length with VaxiJen scores of 0.5 to 1. Findings from McDaniel’s experimental study have the potential to support our predictions. Therefore, we also used this immunoinformatics prediction method for epitope mapping of other PspA clades. In agreement with Singh et al. [69] we used MHC-II binding epitopes prediction servers. Then the predicted epitopes for strain 435/96 (clade1) were compared with the predicted epitopes by Singh and showed almost similar results that could be considered for inducing IFN-γ and IL-4 production. So, we analyzed other PspA clades for MHC-II binding epitope prediction. Some predicted HTL epitopes were also predicted as B-cell epitopes, so we selected the immunodominant truncated CDR regions of each clade.

Mukerji et al. classified the proline-rich domain (PRD) of the PspA into three relatively distinct groups [21]. On the other hand, these PRD regions, especially the PKPEQP motif and non-proline block (NPB) sequence, can elicit protection against pneumococcal infection. As Daniels et al. indicated that when mice were immunized by group 2 PRD, they have been shown protection against the challenge test by a pneumococcal strain with group 3 PRD. These results showed cross-protection against epitopes shared by different groups of PRD [21, 70]. PRD group’s motifs have also been reported to be linear epitopes, and human antibodies can recognize all three PRD groups [21]. In completing the study by Mukerji et al. [21], we used the repetitive motif sequences from all three PRD groups and NPR sequence as a highly conserved and immunogenic domain in PRD of PspA to cover all diversity and cross-protection of the PRD groups. Finally, the designed construct has been named PspA1-5c+p representing the CDR and proline regions of five PspA clades. We used the rigid linker EAAAK between each truncated domain to make the least interaction between domains and maintain the best three-dimensional structure and accessible B-cell conformational epitopes. According to the literature, many natural linkers have alpha-helical structures, which are stable and rigid spacers to keep a fixed distance that is used for separating the functional domains. Another advantage of rigid linkers compared to flexible linkers is that the flexible linkers lead to low expression yields with loss of biological activity[49].

The designed PspA1-5c+p construct was assessed for its physicochemical characteristics. The PspA1-5c+p construct was expected to be acidic in nature, depending on the theoretical isoelectric point. The aliphatic index (indicating thermostability) and grand average of hydropathicity (GRAVY) were estimated at 82.23 and -0.997, respectively. The negative GRAVY value means that the protein has a hydrophilic nature and may interact with water molecules. The in-vivo half-life, as an estimation of time for destroying half the amount of protein after synthesis in the cell, was estimated at 30, 20, and 10 h in mammalian, yeast, and E. coli, respectively. Although the instability index was computed at 40.12, which categorizes the protein as unstable (II of > 40 indicates instability), the experimental result of the recombinant PspA1-5c+p expression and purification showed that this protein was stable [51]. The molecular weight of the PspA1-5c+p construct was 67.93 kDa. The SDS-PAGE and western bot results of recombinant PspA1-5c+p expression confirmed the estimated molecular weight of PspA1-5c+p construct. It has been reported that proteins with a molecular weight of less than 100 kDa are suitable for vaccine design due to their easy expression and purification steps [46]. Therefore, this designed protein was an acceptable vaccine candidate. Also, the results of the codon adaptation index (CAI) and GC content of 0.84 and 42.97%, respectively, showed a good efficiency of the final vaccine transcription and translation in the E. coli host. So that, the Gibbs free energy after sequence optimization for PspA1-5c+p construct mRNA was -445.5 kcal.mol−1, showing the lowest free energy and stable structure, and there was no observed unsuitable pseudo-knot or loop at 5’of mRNA. These computational results were confirmed with the expression of PspA1-5c+p in E. coli BL21 using 1 mM IPTG. The presence of a 67 kDa recombinant PspA1-5c+p sharp band by 12% SDS-PAGE showed acceptable expression and codon optimization. The computationally predicted overexpression and soluble feature of PspA1-5c+p using the SOLpro prediction were validated by purification of recombinant PspA1-5c+p under the native condition in Ni–NTA affinity chromatography with a high concentration of 0.8 mg/ml. The expression of the PspA1-5c+p construct was verified by the Western blot on PspA1-5c+p using an anti-His tag antibody.

Understanding the secondary and tertiary structures of the target protein is critical to vaccine design. The secondary structure of PspA1-5c+p contained 83.22% alpha-helix, 0.49% extended strand, and 16.28% random coil using the GOR V prediction server. It has been reported that the important shapes of “structural antigens” are natively unfolded protein regions and alpha-helical coiled-coil peptides. Both structural forms can be retreated into their native structure and therefore be identified by antibodies naturally induced in response to infection [55]. The PspA1-5c+p 3D structure was modeled using the I-TASSER server. This server is one of the best and most widely used servers for designing three-dimensional protein structures. I-TASSER server uses the multiple threading alignments from PDB to identify structural templates and designs the 3D structures using repetitive fragment assembly simulations [71]. According to many recent papers which have cited to I-TASSER server for protein 3D structure predictions [72,73,74,75], the I-TASSER server was ranked as the No 1 server for protein structure prediction in recent community-wide CASP7, CASP8, CASP9, CASP10, CASP11, CASP12, CASP13, and CASP14 experiments. It was also ranked as the best for function prediction in CASP9. The server is in active development with the goal to provide the most accurate protein structure and function predictions using state-of-the-art algorithms [36]. Using structural refinement servers, we could improve the overall quality factor of the initial PspA1-5c+p 3D model predicted by I-TASSER from 89.66% to 98.14%, and in the Ramachandran plot, disallowed region residues were reduced from 1.6% to 1.2% after the refinement process. Ramachandran plot also revealed that most of the residues are located in the favored and allowed regions (98.8%), demonstrating that the overall model quality is satisfactory. The structural refinement servers optimized the hydrogen-bonding network, minimized the atomic energy of the model, and improved the 3D structure by molecular dynamics simulation. In this study, the MD simulation was applied to verify the stability and flexibility of the structure of the designed PspA1-5c+p protein. Analysis of the MD simulation trajectory revealed that the designed structure of the PspA1-5c+p reaches a stable state with low deviations from 50 to 85 ns. This can indicate the stability of the 3D structure. In addition, using the RMSF plot, we found that the C-terminal of PspA1-5c+p protein is the fluctuating region of the protein. During the simulation, the fluctuation of this region occurred around 0.7 nm. Nonetheless, the rest of the protein had a fluctuating value of less than 0.35 nm. These residues (C-terminus region of the protein) have more freedom of action in the environment due to the coil structure. Furthermore, the ClusPro and DimPlot results of PspA1-5c+p and HLA-DRB1*01:01 (the most common binding allele in the Iran population [46]) docking complex showed the lowest energy binding of -744.3 kcal.mol−1 and 64 cluster members indicating good binding affinity and coupling of this protein with human MHCII via sixteen hydrogen bonds and six salt bridges. However, in order to improve and examine the precise interaction between the protein and HLA-DRB1*01:01, the docking between the T-cell epitope placed in the groove of HLA-DRB1*01:01 chains with the T lymphocyte receptor (TCR) [76] or docking of the human ternary complex of the T-cell receptor, peptide-MHCII molecule, and CD4 are recommended [77]. Since the PspA has a lactoferrin binding domain in the CDR region[59], to furthermore validation of the 3D structure of the modeled PspA1-5c+p, we docked the PspA1-5c+p protein with human lactoferrin N-lobe (HLF). We demonstrated the PspA1-5c+p protein can be attached to HLF molecules effectively via both regions representing PspA Families 1 and 2 in PspA1-5c+p protein with the lowest energy binding of -1128.9 and -987.2 kcal.mol−1 and maximum cluster members of 58 and 80, respectively. In coordination with the docked control molecule (PDB id: 2PMS) [59], in two models of the PspA1-5c+p-HLF docked complex, most residues of HLF that have been in contact with CDR residues include Arg4, Arg5, Arg25, Arg28, Arg31, Arg40, Gln14, Gln24, and lys39. It has been reported that the negatively charged surface of PspA helices can interact with the highly cationic lactoferricin moiety of lactoferrin and inhibit its bactericidal effect against pneumococci. Our results were in line with the study conducted by Senkovich et al. and could show a good 3D structure of the CDR region in the PspA1-5c+p vaccine formulation that could bind to HLF correctly. Senkovich et al. also suggested that inhibition of this interaction using small molecules or antibodies may permit lactoferrin’s natural bactericidal effects to preserve the host from pneumococcal colonization and infection and can be used for designing therapeutic strategies for the prevention and treatment of pneumococcal diseases [59]. Therefore, further studies can be performed to evaluate the binding of antibodies generated against PspA1-5c+p to PspA on the surface of the different pneumococcal strains in the presence of the labeled human lactoferrin.

The results of predicting the conformational B-cell epitopes of the PspA1-5c+p construct showed that after designing the structure, the conformational B-cell epitopes of each clade with a score of > 0.5 could be identified by the ElliPro server. These results can be indicated by the high potential of the PspA1-5c+p to stimulate humoral immunity with the help of antibodies. One of the first steps in confirming a vaccine candidate is immunoreactivity detection using the serological test. According to the antigenicity score of 0.77 for the final PspA1-5c+p construct from the Vaxijen server, this protein was considered a good antigen to stimulate the immune system. The experimental results confirmed and validated the computational antigenicity analysis of this protein. This protein was able to raise anti-PspA1-5c+p IgG titers in immunized mice with PspA1-5c+p construct compared to the control group (p < 0.0001) at different times of administration (Fig. 12). In addition, using immunoinformatics predictions, PspA1-5c+p was considered a non-toxic, and non-allergen. So that, in the experimental results, this protein provided a very good and effective immunological response without causing any allergenicity or toxicity in the animal model. So that, after injection of the PspA1-5c+p construct, we did not observe any increase in body temperature, weight loss, allergic reaction, sensitivity, or restlessness in the animal model. As in past studies conducted on the PspA protein, there were no reports of any deleterious nature of PspA. Sanofi Pasteur has also studied phase 1 of the clinical trial of PspA [9, 74, 75]. In this study, we demonstrated that anti-PspA1-5c+p IgG reacted strongly with no significant difference (p-value = 0.2) against the surface of all three pneumococcal strains representing both PspA families. These results can indicate the high coverage of the cross-reactivity and binding ability of the anti-PspA1-5c+p IgG among different used PspA clades, and cover the limitation of different cross-reaction levels in the PspA-based construct designed so far. In this context, Akbari et al. demonstrated that an antibody against the PspAB1-5 antigen containing the single B region from all clades compared to PspA4ABC could increase the cross-reactivity against pneumococcus strains representing Clades 1, 2, and 5. However, the strong binding ability of the anti-PspAB1-5 antibody was against strain ATCC 6305 (Clade 2) with an optical density of ~ 2.1. Although, they suggested that for the construction of a PspA-based vaccine, the B region from all clades should be included [7] but is not sufficient due to the significant difference observed between optical densities of the cross-reactivity ability of the anti-PspAB1-5 antibody against all three pneumococcus strains [7]. In this study, no different cross-reactivity ability of the anti-PspA1-5c+p antibody was seen against two PspA families. In contrast to Akbari et al., our whole-cell ELISA results showed the optical density of the cross-reactivity ability of the anti-PspA1-5c+p antibody was the same between three stains (Clades 1, 2, and 5) and increased to 3. This increase in the tendency of anti-PspA1-5c+p antibody to bind to the bacterial surface may be due to two factors: the use of all cross-reactive truncated domain of CDRs together with highly conserved NPB region and using repetitive proline-rich motifs that cover the diversity of each clade. This study was also able to solve problems related to cross-reactivity differences in the studies of other research that used the various recombinant PspA proteins consisting of N-terminal and proline-rich regions from two PspA families or each region alone [23, 25, 78,79,80].

We also applied the Opsonophagocytosis test to assess the in vitro potential protective effects of PspA-based vaccines against pneumococcus strains representing both PspA families. The gold standard in-vitro test for assessing the polysaccharide-base pneumococcal vaccine effectiveness is the Opsonophagocytosis assay [16, 81]. Opsonophagocytosis is thought to be considered an important function in the host defense for the elimination of pneumococci. This process is started by complement activation in the presence of antibodies that are attached to the surface of pneumococci. Then, using phagocytic cells, pneumococci are swallowed and killed [16, 81]. The results showed that the anti-PspA1-5c+p antibodies act as a good opsonin for killing pneumococcal strains and can attach to the native protein from each PspA clade on the surface of pneumococcal strains representing both PspA families.

We also analyzed the complement-mediated killing activity of anti-PspA1-5c+p antibody as a serum bactericidal assay against three strains of pneumococcus, expressing two PspA families. The highest bactericidal activity was detected at a 1:4 dilution in order to kill more than 50% of pneumococci compared to the control group. No significant difference was seen between the antibody’s activities against three strains of pneumococcus. These results suggest that this antibody not only has a high titer with strong and uniform cross-reactivity coverage against three pneumococcal strains but also has high bioactivity for pneumococcal clearance using complement or phagocytic cells. Goulart et al. reported that the level of complement-mediated antibody-dependent phagocytosis depends on the similarity between anti-PspA antibodies and PspA that are expressed on the pneumococcal surface [80].

In addition, according to immune simulation servers, PspA1-5c+p was predicted to compose antigenic regions that have the potency to induce IL-4 and IL-10 cytokines. Furthermore, the PspA1-5c+p construct was predicted to have many IFN-γ inducing MHC class II binding peptides throughout its sequence. These bioinformatics results showed that PspA1-5c+p might induce both humoral and cellular immune pathways. Overall, these results show the success of the bioinformatics tool in designing a PspA-based vaccine candidate to cover the cross-reactivity of the vaccine candidate against all used PspA clades. As the efficacy and reliability of the immunoinformatics approach have been proven in a lot of pioneering work regarding the design and development of epitope-based vaccines [26, 27, 47, 82, 83]. Our results are in accordance with these studies. In this context, Ahmadi et al. designed a novel Hla-MntC-SACOL0723 fusion protein using immunoinformatics tools. They then showed that this fusion protein could elicit high specific IgG titer with high opsonophagosytosis’s killing activity against S. aureus resulting in a decrease in the bacterial burden in the spleen and kidneys [47]. Hasanzadeh et al. also demonstrated that the computational design of their epitope‑based vaccine candidate could induce immune responses and provide high potency in the protection of the urinary tract against uropathogenic Escherichia coli (UTEC) [83].

The limitation of this study was the lack of access to standard pneumococcal strains expressing other clades of PspA for assessing the full cross-reactive feature of the anti-PspA1-5c+p IgG. In the future, we will resolve the mentioned limitation and also analyze the profile of subclasses of specific IgG1 and IgG2a against PspA1-5c+p construct immunization, levels of the IL-4 and IFN- γ cytokines, and the protection ability of this construct in immunized groups against pneumococcal infections to confirm our computational immune simulation results.

Conclusion

Our experimental data revealed that immunoinformatics helps us to design protective serotype-independent vaccine candidates. Experimental assessments on three clades of PspA showed promising results with a strong cross-reactivity feature that should be further investigated in vitro and in vivo experiments with other pneumococcal clades to confirm the full cross-reactivity and cross-protection.

Availability of data and materials

Not applicable.

Abbreviations

PspA:

Pneumococcal surface protein A

IPD:

Invasive pneumococcal disease

PPV:

Pneumococcal protein-based vaccine

CDR:

Clade-defining region

PRD:

Proline-rich domain

NPB:

Non-proline block

NCBI:

National Center for Biotechnology Information

SVM:

Support vector machine

PDB:

Protein data bank

C-score:

Confidence score

RMSD:

Root Mean Square Deviation score

TM-score:

Template modeling score

HTL:

Helper T-cell

HLF:

Human lactoferrin

PSSM:

Position-specific score matrix

IPTG:

Isopropyl-β-D Thiogalactopyranoside

LB:

Luria–Bertani broth

LAL:

Limulus amebocyte lysate

TMB:

Tetramethylbenzidine

ELISA:

Enzyme-linked immunosorbent assay

SBA:

Serum Bactericidal Assay

OPK:

Opsonophagocytic killing activity

ANOVA:

Analysis of variances

GRAVY:

Grand average of hydropathicity

MW:

Molecular weight

pI:

Isoelectric point

II:

Instability index

CAI:

Codon adaptation index

MD:

Molecular dynamic

References

  1. Weiser JN, Ferreira DM, Paton JC. Streptococcus pneumoniae: transmission, colonization and invasion. Nat Rev Microbiol. 2018;16(6):355–67. https://doi.org/10.1038/s41579-018-0001-8.

    Article  CAS  Google Scholar 

  2. Engholm DH, Kilian M, Goodsell DS, Andersen ES, Kjærgaard RS. A visual review of the human pathogen Streptococcus pneumoniae. FEMS Microbiol Rev. 2017;41(6):854–79. https://doi.org/10.1093/femsre/fux037.

    Article  CAS  Google Scholar 

  3. Lane JR, Tata M, Briles DE, Orihuela CJ. A jack of all trades: the role of pneumococcal surface protein A in the pathogenesis of Streptococcus pneumoniae. Front Cell Infect Microbiol. 2022. https://doi.org/10.3389/fcimb.2022.826264.

    Article  Google Scholar 

  4. GBD 2016 Lower Respiratory Infections Collaborators. Estimates of the global, regional, and national morbidity, mortality, and aetiologies of lower respiratory infections in 195 countries, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet Infect Dis. 2018;18(11):1191.

    Article  Google Scholar 

  5. Cherazard R, Epstein M, Doan T-L, Salim T, Bharti S, Smith MA. Antimicrobial resistant Streptococcus pneumoniae: prevalence, mechanisms, and clinical implications. Am J Ther. 2017;24(3):361–9. https://doi.org/10.1097/MJT.0000000000000551.

    Article  Google Scholar 

  6. Aceil J, Avci FY. Pneumococcal surface proteins as virulence factors, immunogens, and conserved vaccine targets. Front Cell Infect Microbiol. 2022. https://doi.org/10.3389/fcimb.2022.832254.

    Article  Google Scholar 

  7. Akbari E, Negahdari B, Faraji F, Behdani M, Kazemi-Lomedasht F, Habibi-Anbouhi M. Protective responses of an engineered PspA recombinant antigen against Streptococcus pneumoniae. Biotechnol Rep. 2019;24:1–7. https://doi.org/10.1016/j.btre.2019.e00385.

    Article  Google Scholar 

  8. Feldman C, Anderson R. Current and new generation pneumococcal vaccines. J Infect. 2014;69(4):309–25. https://doi.org/10.1016/j.jinf.2014.06.006.

    Article  Google Scholar 

  9. Lagousi T, Basdeki P, Routsias J, Spoulou V. Novel protein-based pneumococcal vaccines: assessing the use of distinct protein fragments instead of full-length proteins as vaccine antigens. Vaccines. 2019;7(1):9–27. https://doi.org/10.3390/vaccines7010009.

    Article  CAS  Google Scholar 

  10. Yuki Y, Uchida Y, Sawada S-I, Nakahashi-Ouchida R, Sugiura K, Mori H, et al. Characterization and specification of a trivalent protein-based pneumococcal vaccine formulation using an adjuvant-free nanogel nasal delivery system. Mol Pharmaceutics. 2021;18(4):1582–92. https://doi.org/10.1021/acs.molpharmaceut.0c01003.

    Article  CAS  Google Scholar 

  11. Datta A, Kapre K, Andi-Lolo I, Kapre S. Multi-valent pneumococcal conjugate vaccine for global health: From problem to platform to production. Hum Vaccin Immunother. 2022. https://doi.org/10.1080/21645515.2022.2117949.

    Article  Google Scholar 

  12. Converso T, Assoni L, André G, Darrieux M, Leite LCDC. The long search for a serotype independent pneumococcal vaccine. Expert Rev Vaccin. 2020;19(1):57–70. https://doi.org/10.1080/14760584.2020.1711055.

    Article  CAS  Google Scholar 

  13. Oliveira GS, Oliveira MLS, Miyaji EN, Rodrigues TC. Pneumococcal vaccines: past findings, present work, and future strategies. Vaccines. 2021;9(11):1338–54. https://doi.org/10.3390/vaccines9111338.

    Article  CAS  Google Scholar 

  14. Zhang Y, Guo X, Guo M, Chen X, Li B, Yu J, et al. Combined prime-boost immunization with systemic and mucosal pneumococcal vaccines based on Pneumococcal surface protein A to enhance protection against lethal pneumococcal infections. Immunol Res. 2019;67(4):398–407. https://doi.org/10.1007/s12026-019-09107-6.

    Article  CAS  Google Scholar 

  15. Silva PH, Vazquez Y, Campusano C, Retamal-Díaz A, Lay MK, Muñoz CA, et al. Non-capsular based immunization approaches to prevent Streptococcus pneumoniae infection. Front Cell Infect Microbiol. 2022. https://doi.org/10.3389/2Ffcimb.2022.949469.

    Article  Google Scholar 

  16. Converso TR, Goulart C, Rodriguez D, Darrieux M, Leite L. Rational selection of broadly cross-reactive family 2 PspA molecules for inclusion in chimeric pneumococcal vaccines. Microb Pathog. 2017;109:233–8. https://doi.org/10.1016/j.micpath.2017.06.004.

    Article  CAS  Google Scholar 

  17. Sempere J, Llamosí M, del Río MI, López Ruiz B, Domenech M, González-Camacho F. Pneumococcal choline-binding proteins involved in virulence as vaccine candidates. Vaccines. 2021;9(2):181–200. https://doi.org/10.3390/vaccines9020181.

    Article  CAS  Google Scholar 

  18. Roberts S, Williams CM, Salmon SL, Bonin JL, Metzger DW, Furuya Y. Evaluation of pneumococcal surface protein A as a vaccine antigen against secondary Streptococcus pneumoniae challenge during Influenza A infection. Vaccines. 2019;7(4):146–55. https://doi.org/10.3390/vaccines7040146.

    Article  CAS  Google Scholar 

  19. Jedrzejas MJ, Hollingshead SK, Lebowitz J, Chantalat L, Briles DE, Lamani E. Production and characterization of the functional fragment of pneumococcal surface protein A. Arch Biochem Biophys. 2000;373(1):116–25. https://doi.org/10.1006/abbi.1999.1544.

    Article  CAS  Google Scholar 

  20. Chang B, Kinjo Y, Morita M, Tamura K, Watanabe H, Tanabe Y, et al. Distribution and variation of serotypes and pneumococcal surface protein A clades of Streptococcus pneumoniae Strains isolated from adult patients with invasive pneumococcal disease in Japan. Front Cell Infect Microbiol. 2021;11:192–200. https://doi.org/10.3389/fcimb.2021.617573.

    Article  CAS  Google Scholar 

  21. Mukerji R, Hendrickson C, Genschmer KR, Park S-S, Bouchet V, Goldstein R, et al. The diversity of the proline-rich domain of pneumococcal surface protein A (PspA): potential relevance to a broad-spectrum vaccine. Vaccine. 2018;36(45):6834–43. https://doi.org/10.1016/j.vaccine.2018.08.045.

    Article  CAS  Google Scholar 

  22. Darrieux M, Miyaji EN, Ferreira D, Lopes L, Lopes APY, Ren B, et al. Fusion proteins containing family 1 and family 2 PspA fragments elicit protection against Streptococcus pneumoniae that correlates with antibody-mediated enhancement of complement deposition. Infect Immun. 2007;75(12):5930–8. https://doi.org/10.1128/IAI.00940-07.

    Article  CAS  Google Scholar 

  23. Melin M, Coan P, Hollingshead S. Development of cross-reactive antibodies to the proline-rich region of pneumococcal surface protein A in children. Vaccine. 2012;30(50):7157–60. https://doi.org/10.1016/j.vaccine.2012.10.004.

    Article  CAS  Google Scholar 

  24. Scott NR, Mann B, Tuomanen EI, Orihuela CJ. Multi-valent protein hybrid pneumococcal vaccines: a strategy for the next generation of vaccines. Vaccines. 2021;9(3):209–25. https://doi.org/10.3390/vaccines9030209.

    Article  CAS  Google Scholar 

  25. Piao Z, Akeda Y, Takeuchi D, Ishii KJ, Ubukata K, Briles DE, et al. Protective properties of a fusion pneumococcal surface protein A (PspA) vaccine against pneumococcal challenge by five different PspA clades in mice. Vaccine. 2014;32(43):5607–13. https://doi.org/10.1016/j.vaccine.2014.07.108.

    Article  CAS  Google Scholar 

  26. Khan MT, Islam MJ, Parihar A, Islam R, Jerin TJ, Dhote R, et al. Immunoinformatics and molecular modeling approach to design universal multi-epitope vaccine for SARS-CoV-2. Inform Med Unlocked. 2021;24:100578. https://doi.org/10.1016/j.imu.2021.100578.

    Article  Google Scholar 

  27. Adar Y, Singer Y, Levi R, Tzehoval E, Perk S, Banet-Noach C, et al. A universal epitope-based influenza vaccine and its efficacy against H5N1. Vaccine. 2009;27(15):2099–107. https://doi.org/10.1016/j.vaccine.2009.02.011.

    Article  CAS  Google Scholar 

  28. Khan M, Khan S, Ali A, Akbar H, Sayaf AM, Khan A, et al. Immunoinformatics approaches to explore Helicobacter pylori proteome (Virulence Factors) to design B and T cell multi-epitope subunit vaccine. Sci Rep. 2019;9(1):1–13. https://doi.org/10.1038/s41598-019-49354-z.

    Article  CAS  Google Scholar 

  29. Rahman M, Puspo JA, Adib AA, Hossain ME, Alam MM, Sultana S, et al. An immunoinformatics prediction of novel multi-epitope vaccines candidate against surface antigens of Nipah Virus. Int J Pept Res Ther. 2022;28(4):1–25. https://doi.org/10.1007/s10989-022-10431-z.

    Article  CAS  Google Scholar 

  30. Rodrigues RR, Ferreira MRA, Kremer FS, Donassolo RA, Júnior CM, Alves MLF, et al. Recombinant vaccine design against Clostridium spp. toxins using immunoinformatics tools. In: Thomas S, editor., et al., Vaccine Design. New York: Springer; 2022. p. 457–70.

    Chapter  Google Scholar 

  31. Yu CS, Chen YC, Lu CH, Hwang JK. Prediction of protein subcellular localization. Proteins. 2006;64(3):643–51. https://doi.org/10.1002/prot.21018.

    Article  CAS  Google Scholar 

  32. Möller S, Croning MD, Apweiler R. Evaluation of methods for the prediction of membrane spanning regions. Bioinformatics. 2001;17(7):646–53. https://doi.org/10.1093/bioinformatics/17.7.646.

    Article  Google Scholar 

  33. Nielsen H, Tsirigos KD, Brunak S, von Heijne G. A brief history of protein sorting prediction. Protein J. 2019;38(3):200–16. https://doi.org/10.1007/s10930-019-09838-3.

    Article  CAS  Google Scholar 

  34. Doytchinova IA, Flower DR. VaxiJen: a server for prediction of protective antigens, tumour antigens and subunit vaccines. BMC Bioinformatics. 2007;8(1):1–7. https://doi.org/10.1186/1471-2105-8-4.

    Article  CAS  Google Scholar 

  35. Chen J, Liu H, Yang J, Chou K-C. Prediction of linear B-cell epitopes using amino acid pair antigenicity scale. Amino Acids. 2007;33(3):423–8. https://doi.org/10.1007/s00726-006-0485-9.

    Article  CAS  Google Scholar 

  36. Yang J, Yan R, Roy A, Xu D, Poisson J, Zhang Y. The I-TASSER Suite: protein structure and function prediction. Nat Methods. 2015;12(1):7–8. https://doi.org/10.1038/nmeth.3213.

    Article  CAS  Google Scholar 

  37. Lee GR, Heo L, Seok C. Effective protein model structure refinement by loop modeling and overall relaxation. Proteins. 2016;84:293–301. https://doi.org/10.1002/prot.24858.

    Article  CAS  Google Scholar 

  38. Wiederstein M, Sippl MJ. ProSA-web: interactive web service for the recognition of errors in three-dimensional structures of proteins. Nucleic Acids Res. 2007;35(suppl 2):1–4. https://doi.org/10.1093/nar/gkm290.

    Article  Google Scholar 

  39. Laskowski RA, Rullmann JAC, MacArthur MW, Kaptein R, Thornton JM. AQUA and PROCHECK-NMR: programs for checking the quality of protein structures solved by NMR. J Biomol NMR. 1996;8(4):477–86. https://doi.org/10.1007/BF00228148.

    Article  CAS  Google Scholar 

  40. Ponomarenko J, Bui H-H, Li W, Fusseder N, Bourne PE, Sette A, et al. ElliPro: a new structure-based tool for the prediction of antibody epitopes. BMC Bioinformatics. 2008;9(1):1–8. https://doi.org/10.1186/1471-2105-9-514.

    Article  CAS  Google Scholar 

  41. Wang P, Sidney J, Kim Y, Sette A, Lund O, Nielsen M, et al. Peptide binding predictions for HLA DR, DP and DQ molecules. BMC Bioinformatics. 2010;11(1):1–12. https://doi.org/10.1186/1471-2105-11-568.

    Article  CAS  Google Scholar 

  42. Reche PA, Glutting J-P, Zhang H, Reinherz EL. Enhancement to the RANKPEP resource for the prediction of peptide binding to MHC molecules using profiles. Immunogenetics. 2004;56(6):405–19. https://doi.org/10.1007/s00251-004-0709-7.

    Article  CAS  Google Scholar 

  43. Hattotuwagama CK, Guan P, Doytchinova IA, Zygouri C, Flower DR. Quantitative online prediction of peptide binding to the major histocompatibility complex. J Mol Graph Model. 2004;22(3):195–207. https://doi.org/10.1016/S1093-3263(03)00160-8.

    Article  CAS  Google Scholar 

  44. van de Garde MD, van Westen E, Poelen MC, Rots NY, van Els CA. Prediction and validation of immunogenic domains of pneumococcal proteins recognized by human CD4+ T cells. Infect Immun. 2019;87(6):1–18. https://doi.org/10.1128/IAI.00098-19.

    Article  Google Scholar 

  45. Dorosti H, Eslami M, Negahdaripour M, Ghoshoon MB, Gholami A, Heidari R, et al. Vaccinomics approach for developing multi-epitope peptide pneumococcal vaccine. J Biomol Struct Dyn. 2019;37(13):3524–35. https://doi.org/10.1080/07391102.2018.1519460.

    Article  CAS  Google Scholar 

  46. Sanami S, Zandi M, Pourhossein B, Mobini G-R, Safaei M, Abed A, et al. Design of a multi-epitope vaccine against SARS-CoV-2 using immunoinformatics approach. Int J Biol Macromol. 2020;164:871–83. https://doi.org/10.1016/j.ijbiomac.2020.07.117.

    Article  CAS  Google Scholar 

  47. Ahmadi K, Pouladfar G, Kalani M, Faezi S, Pourmand MR, Hasanzadeh S, et al. Epitope-based immunoinformatics study of a novel Hla-MntC-SACOL0723 fusion protein from Staphylococcus aureus: Induction of multi-pattern immune responses. Mol Immunol. 2019;114:88–99. https://doi.org/10.1016/j.molimm.2019.05.016.

    Article  CAS  Google Scholar 

  48. Chen C, Li Z, Huang H, Suzek BE, Wu CH, Consortium U. A fast peptide match service for UniProt knowledgebase. Bioinformatics. 2013;29(21):2808–9. https://doi.org/10.1093/bioinformatics/btt484.

    Article  CAS  Google Scholar 

  49. Chen X, Zaro JL, Shen W-C. Fusion protein linkers: property, design and functionality. Adv Drug Deliv Rev. 2013;65(10):1357–69. https://doi.org/10.1016/j.addr.2012.09.039.

    Article  CAS  Google Scholar 

  50. Gupta S, Kapoor P, Chaudhary K, Gautam A, Kumar R, Consortium OSDD, et al. In silico approach for predicting toxicity of peptides and proteins. PLoS ONE. 2013;8(9):e73957. https://doi.org/10.1371/journal.pone.0073957.

    Article  CAS  Google Scholar 

  51. Gasteiger E, Hoogland C, Gattiker A, Wilkins MR, Appel RD, Bairoch A. Protein identification and analysis tools on the ExPASy server. In: The proteomics protocols handbook. 2005, pp. 571–607. https://doi.org/10.1385/1-59259-890-0:571.

  52. Yu M, Zhu Y, Li Y, Chen Z, Sha T, Li Z, et al. Design of a novel multi-epitope vaccine against Echinococcus granulosus in immunoinformatics. Front Immunol. 2021;12:1–17. https://doi.org/10.3389/2Ffimmu.2021.668492.

    Article  Google Scholar 

  53. Yang Z, Bogdan P, Nazarian S. An in silico deep learning approach to multi-epitope vaccine design: a SARS-CoV-2 case study. Sci Rep. 2021;11(1):1–21. https://doi.org/10.1038/s41598-021-81749-9.

    Article  CAS  Google Scholar 

  54. Bhattacharya D, Nowotny J, Cao R, Cheng J. 3Drefine: an interactive web server for efficient protein structure refinement. Nucleic Acids Res. 2016;44(W1):406–9. https://doi.org/10.1016/S1093-3263(03)00160-8.

    Article  CAS  Google Scholar 

  55. Shey RA, Ghogomu SM, Esoh KK, Nebangwa ND, Shintouo CM, Nongley NF, et al. In-silico design of a multi-epitope vaccine candidate against onchocerciasis and related filarial diseases. Sci Rep. 2019;9(1):1–18. https://doi.org/10.1038/s41598-019-40833-x.

    Article  CAS  Google Scholar 

  56. Abraham MJ, Murtola T, Schulz R, Páll S, Smith JC, Hess B, et al. GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX. 2015;1:19–25. https://doi.org/10.1016/j.softx.2015.06.001.

    Article  Google Scholar 

  57. Mafakher L, Rismani E, Rahimi H, Enayatkhani M, Azadmanesh K, Teimoori-Toolabi L. Computational design of antagonist peptides based on the structure of secreted frizzled-related protein-1 (SFRP1) aiming to inhibit Wnt signaling pathway. J Biomol Struct Dyn. 2022;40(5):2169–88. https://doi.org/10.1080/07391102.2020.1835718.

    Article  CAS  Google Scholar 

  58. Murthy VL, Stern LJ. The class II MHC protein HLA-DR1 in complex with an endogenous peptide: implications for the structural basis of the specificity of peptide binding. Structure. 1997;5(10):1385–96. https://doi.org/10.1016/S0969-2126(97)00288-8.

    Article  CAS  Google Scholar 

  59. Senkovich O, Cook WJ, Mirza S, Hollingshead SK, Protasevich II, Briles DE, et al. Structure of a complex of human lactoferrin N-lobe with pneumococcal surface protein a provides insight into microbial defense mechanism. J Mol Biol. 2007;370(4):701–13. https://doi.org/10.1016/2Fj.jmb.2007.04.075.

    Article  CAS  Google Scholar 

  60. Umar A, Haque A, Alghamdi YS, Mashraqi MM, Rehman A, Shahid F, et al. Development of a candidate multi-epitope subunit vaccine against Klebsiella aerogenes: subtractive proteomics and immuno-informatics approach. Vaccines. 2021;9(11):1373–92. https://doi.org/10.3390/vaccines9111373.

    Article  CAS  Google Scholar 

  61. Villalobos A, Ness JE, Gustafsson C, Minshull J, Govindarajan S. Gene designer: a synthetic biology tool for constructing artificial DNA segments. BMC Bioinformatics. 2006;7(1):1–8.

    Article  Google Scholar 

  62. Zuker M. Prediction of RNA secondary structure by energy minimization. In: Annette M. Griffin, Hugh G. Griffin eds., Computer analysis of sequence data. Springer; 1994. p. 267–94. https://doi.org/10.1385/0-89603-276-0:267

  63. Jung S-K, McDonald K. Visual gene developer: a fully programmable bioinformatics software for synthetic gene optimization. BMC Bioinformatics. 2011;12(1):1–13. https://doi.org/10.1186/1471-2105-12-340.

    Article  CAS  Google Scholar 

  64. Malekan M, Siadat SD, Aghasadeghi M, Shahrokhi N, Afrough P, Behrouzi A, et al. Evaluation of protective immunity responses against pneumococcal PhtD and its C-terminal in combination with outer-membrane vesicles as adjuvants. J Med Microbiol. 2020;69(3):465–77. https://doi.org/10.1099/jmm.0.001103.

    Article  CAS  Google Scholar 

  65. Afrough P, Bouzari S, Mousavi SF, Karam MRA, Vaziri F, Fateh A, et al. Evaluation of immunological responses to recombinant Porin A protein (rPoA) from native strains of Neisseria meningitidis serogroups A and B using OMV as an adjuvant in BALB/c mice. Microb Pathog. 2017;112:209–14. https://doi.org/10.1016/j.micpath.2017.09.038.

    Article  CAS  Google Scholar 

  66. Colichio GB, Oliveira GS, Rodrigues TC, Oliveira MLS, Miyaji EN. Efficacy of a protein vaccine and a conjugate vaccine against co-colonization with vaccine-type and non-vaccine type pneumococci in mice. Pathogens. 2020;9(4):278–91. https://doi.org/10.3390/pathogens9040278.

    Article  CAS  Google Scholar 

  67. Khan N, Jan AT. Towards identifying protective B-cell epitopes: the PspA story. Front Microbiol. 2017;8:742–50. https://doi.org/10.3389/fmicb.2017.00742.

    Article  Google Scholar 

  68. McDaniel LS, Ralph BA, McDaniel DO, Briles DE. Localization of protection-eliciting epitopes on PspA of Streptococcus pneumoniae between amino acid residues 192 and 260. Microb Pathog. 1994;17(5):323–37.

    Article  CAS  Google Scholar 

  69. Singh R, Singh S, Sharma PK, Singh UP, Briles DE, Hollingshead SK, et al. Helper T cell epitope-mapping reveals MHC-peptide binding affinities that correlate with T helper cell responses to pneumococcal surface protein A. PLoS ONE. 2010;5(2):1–16. https://doi.org/10.1371/journal.pone.0009432.

    Article  CAS  Google Scholar 

  70. Daniels CC, Coan P, King J, Hale J, Benton KA, Briles DE, et al. The proline-rich region of pneumococcal surface proteins A and C contains surface-accessible epitopes common to all pneumococci and elicits antibody-mediated protection against sepsis. Infect Immun. 2010;78(5):2163–72. https://doi.org/10.1128/IAI.01199-09.

    Article  CAS  Google Scholar 

  71. Yang J, Zhang Y. I-TASSER server: new development for protein structure and function predictions. Nucleic Acids Res. 2015;43(W1):W174–81. https://doi.org/10.1093/nar/2Fgkv342.

    Article  CAS  Google Scholar 

  72. Gorai S, Das NC, Gupta PSS, Panda SK, Rana MK, Mukherjee S. Designing efficient multi-epitope peptide-based vaccine by targeting the antioxidant thioredoxin of bancroftian filarial parasite. Infect Genet Evol. 2022;98:105237. https://doi.org/10.1016/j.meegid.2022.105237.

    Article  CAS  Google Scholar 

  73. Kar T, Narsaria U, Basak S, Deb D, Castiglione F, Mueller DM, et al. A candidate multi-epitope vaccine against SARS-CoV-2. Sci Rep. 2020;10(1):1–24. https://doi.org/10.1038/s41598-020-67749-1.

    Article  CAS  Google Scholar 

  74. Damas MSF, Mazur FG, de Melo Freire CC, da Cunha AF, da Silva Pranchevicius M-C. A systematic immuno-informatic approach to design a multiepitope-based vaccine against emerging multiple drug resistant. research, development and clinical trials for peptides-based vaccines. Front Immunol. 2022. 10.3389%2Ffimmu.2022.768569.

  75. Farhani I, Nezafat N, Mahmoodi S. Designing a novel multi-epitope peptide vaccine against pathogenic Shigella spp. based immunoinformatics approaches. Int J Pept Res Ther. 2019;25(2):541–53. https://doi.org/10.1007/s10989-018-9698-5.

    Article  CAS  Google Scholar 

  76. Wang L, Zhao Y, Li Z, Guo Y, Jones LL, Kranz DM, et al. Crystal structure of a complete ternary complex of TCR, superantigen and peptide-MHC. Nat Struct Mol Biol. 2007;14(2):169–71. https://doi.org/10.1038/nsmb1193.

    Article  CAS  Google Scholar 

  77. Basu A. In silico epitope-based vaccine prediction against fungal infection aspergillosis. Challenges. 2022;13(2):29. https://doi.org/10.3390/challe13020029.

    Article  Google Scholar 

  78. Darrieux M, Moreno AT, Ferreira DM, Pimenta FC, de Andrade ALS, Lopes AP, et al. Recognition of pneumococcal isolates by antisera raised against PspA fragments from different clades. J Med Microbiol. 2008;57(3):273–8. https://doi.org/10.1099/jmm.0.47661-0.

    Article  CAS  Google Scholar 

  79. Moreno AT, Oliveira MLS, Ferreira DM, Ho PL, Darrieux M, Leite LC, et al. Immunization of mice with single PspA fragments induces antibodies capable of mediating complement deposition on different pneumococcal strains and cross-protection. Clin Vaccine Immunol. 2010;17(3):439–46. https://doi.org/10.1128/CVI.00430-09.

    Article  CAS  Google Scholar 

  80. Goulart C, Darrieux M, Rodriguez D, Pimenta FC, Brandileone MCC, de Andrade ALS, et al. Selection of family 1 PspA molecules capable of inducing broad-ranging cross-reactivity by complement deposition and opsonophagocytosis by murine peritoneal cells. Vaccine. 2011;29(8):1634–42. https://doi.org/10.1016/j.vaccine.2010.12.074.

    Article  CAS  Google Scholar 

  81. Toh ZQ, Higgins RA, Mazarakis N, Abbott E, Nathanielsz J, Balloch A, et al. Evaluating functional immunity following encapsulated bacterial infection and vaccination. Vaccines. 2021;9(6):677–85. https://doi.org/10.3390/vaccines9060677.

    Article  CAS  Google Scholar 

  82. Gottlieb T, Ben-Yedidia T. Epitope-based approaches to a universal influenza vaccine. J Autoimmun. 2014;54:15–20. https://doi.org/10.1016/j.jaut.2014.07.005.

    Article  CAS  Google Scholar 

  83. Hasanzadeh S, Habibi M, Shokrgozar MA, Ahangari Cohan R, Ahmadi K, Asadi Karam MR, et al. In silico analysis and in vivo assessment of a novel epitope-based vaccine candidate against uropathogenic Escherichia coli. Sci Rep. 2020;10(1):1–16. https://doi.org/10.1038/s41598-020-73179-w.

    Article  CAS  Google Scholar 

Download references

Acknowledgements

The authors thank the Pasteur Institute of Iran, especially the personnel of the department of microbiology and nanobiotechnology at the Pasteur Institute of Iran for their assistance and cooperation with this project.

Funding

This research did not receive any specific grants from funding agencies in the public, commercial, or not-for-profit sectors.

Author information

Authors and Affiliations

Authors

Contributions

Conceptualization: SFM., RAC, FS, and EA; methodology: RAC and EA; validation: SFM, RAC, FS; formal analysis: SFM, RAC, FS, and EA; investigation: EA; writing-original draft preparation: EA; writing-review and editing: SFM, RAC, EA; visualization: EA; project administration: SFM and RAC; Resources: SFM and EA; Supervision: SFM. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Seyed Fazlollah Mousavi.

Ethics declarations

Competing interests

The authors declare that they have no conflict of interest, known competing for financial interests, or personal relationships that could have appeared to influence the work reported in this paper.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1: Figure S1.

The schematic results of IEDB server for antigenicity, surface accessibility, flexibility, hydrophilicity, beta turn, linear and continuous predicted epitope analysis in PspA proteins. Figure S2. 3D modeling and validation of PspA clades. Figure S3. Graphical representation of features of secondary structure of the final pspA1-5C+P construct sequence using PSIPRED server. Table S1. PspA candidates accession numbers and characteristics. Table S2. Predicted linear B-cell epitopes for PspA proteins using BCPred, IEDB, and Ellipro servers. Table S3. Assessment of refined and validated scores for 3D modelling of PspA clades structures. Table S4. Predicted conformational B-cell epitopes for PspA clades using Ellipro server. Table S5. Predicted helper T-cell epitopes for PspA Proteins using IEDB server (Percentile Rank ≤ 20). Table S6. Predicted helper T-cell epitopes for PspA proteins using RANKPEP server. Table S7. Predicted helper T-cell epitopes for PspA Proteins with IC50 value ≤ 100 (nM) using MHCPred server. Table S8. The best IL4 inducing analog/peptide from PspA1-5c+p construct.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Afshari, E., Cohan, R.A., Sotoodehnejadnematalahi, F. et al. In-silico design and evaluation of an epitope-based serotype-independent promising vaccine candidate for highly cross-reactive regions of pneumococcal surface protein A. J Transl Med 21, 13 (2023). https://doi.org/10.1186/s12967-022-03864-z

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12967-022-03864-z

Keywords