In silico design and analysis of a multiepitope vaccine against Chlamydia.

Chlamydia trachomatis ( Ct ) is the most common sexually transmitted bacterial pathogen globally, with an approximate prevalence of 4% in women and 2.5% in men (Newman et al. 2015 , World Health Organization 2023 ). Recently, an estimated 129 million new infections occurred worldwide, primarily affecting individuals aged 15–49 years. However, many cases go unreported due to the commonality of asymptomatic infections, with rates as high as 70% in women and 50% in men (WHO 2016 ). In women, infection primarily occurs within the columnar epithelial cells of the cervix. Acute inflammation is characterized by cervicitis and urethritis when limited to the lower genital tract (Centers for Disease Control and Prevention 2023a ). However, ascension to the upper genital tract can lead to pelvic inflammatory disease (PID), affecting the uterus, fallopian tubes, and ovaries. Complications include ectopic pregnancy and fallopian tube infertility (Price et al. 2016 ). About 10%–15% of Chlamydia -infected women will develop PID (Centers for Disease Control and Prevention 2023b ). Current strategies to combat infections primarily rely on antimicrobial therapy and screening programs. Despite their temporary effectiveness, these strategies have not been sufficient in reducing Chlamydia prevalence (van Bergen et al. 2021 , Hou et al. 2022 ). Many countries have implemented screening programs for Chlamydia , yet success often hinges on resources and funding, posing barriers to both high- and low-income countries (Igietseme et al. 2011 ). Additionally, treating individuals with antibiotics early-on hinders the development of immunity against future infections (Brunham and Rekart 2008 ). The diagnosis and treatment of Chlamydia contribute to an annual estimated $1 billion in medical costs to the USA (Centers for Disease Control and Prevention 2022 ). A vaccination program would be a more cost-effective and impactful approach to controlling infections. Currently, no licensed human vaccine is available; however, research into the design of a suitable vaccine has supported its feasibility. Several approaches to designing vaccines against Ct have been explored, including live attenuated vaccines, inactivated vaccines, and subunit vaccines. Early trials of whole organism vaccines in humans and nonhuman primates demonstrated the potential for protection against infection (Peterson et al. 1999 , Lu et al. 2002 ). However, this protection was often short-lived and sometimes associated with a risk of reversion to virulence. Consequently, many researchers shifted their efforts toward the development of subunit vaccines. Most experimental vaccines were based on the structurally and immunologically dominant major outer membrane protein (MOMP). Purified native MOMP (nMOMP), recombinant MOMP (rMOMP), and MOMP peptide preparations showed variable results in providing partial protection (Pal et al. 2001 , Cheng et al. 2009 , Hickey et al. 2009 , Cunningham et al. 2011 ). Other chlamydial proteins have been targeted as antigenic candidates, including polymorphic membrane proteins (Pmps), three of which are included in the multiepitope vaccine studied here. Pmps constitute a group of nine (PmpA-I) surface-exposed proteins with highly conserved regions, functioning as autotransporter adhesins during the initial phase of infection (Taylor et al. 2011 , Vasilevsky et al. 2016 ). Fragments of Pmps (E–H), derived from C. muridarum , have been shown to accelerate infection clearance following genital challenge in various mouse strains (Yu et al. 2014 ). Moreover, a vaccine combining four Pmps with MOMP significantly reduced vaginal shedding and inflammation in C57BL/6 mice after transcervical infection (Karunakaran et al. 2015 ). Outer membrane complex B (OmcB), another adhesin, has also been explored as a target for vaccine development (Follmann et al. 2008 ). Studies have indicated its potential role in both B- and T-cell responses during infection. Fusion proteins containing OmcB and rl16 have demonstrated reduced bacterial shedding and elicited a partially CD4+ T-cell-dependent protective response (Penttilä et al. 2004 , Olsen et al. 2010 ). Previous research has also investigated multisubunit vaccines incorporating MOMP, Pmps, OmcB, and another chlamydial protein, porin B (PorB) (Eko et al. 2004 , Ifere et al. 2007 , Pais et al. 2017 , 2019 ), with observed partial protection in mouse models of infection. Rational vaccine design is based on a comprehensive understanding of immune mechanisms. Current evidence suggests that immunity against Ct primarily relies on the cellular immune response, with CD4+ T cells and Th1 cytokines, such as IFN-γ and IL-12, playing pivotal roles in infection clearance (Morrison et al. 2000 , Brunham and Rey-Ladino 2005 ). Experimental studies in mice lacking CD4+ T cells, IFN-γ, or IL-12 have demonstrated compromised control of Ct infection (Morrison et al. 1995 , Lu and Zhong 1999 , Mercado et al. 2021 ). Additionally, adoptive transfer of CD4+ T cell clones in nude mice has conferred protection against infection (Su and Caldwell 1995 ). While historically, the emphasis was on cellular immunity, recent insights have highlighted the importance of humoral responses. In humans, reduced bacterial burden has been associated with mucosal IgG and IgA (Brunham et al. 1983 , Cotter et al. 1995 ). Therefore, the focus of our study is on the in silico generation of a multiepitope vaccine capable of eliciting robust cell-mediated and humoral responses. We aim to design and validate the vaccine using immunoinformatic analyses of five Chlamydia proteins (PmpC, PmpD, PmpG, OmcB, and PorB), selected based on their immunogenic properties observed in previous studies. Furthermore, the immunogenicity of the vaccine will be evaluated through in vivo analysis in mice.

This study strictly adhered to the recommendations outlined in the National Institutes of Health’s Guide for the Care and Use of Laboratory Animals. Approval for the study protocol was granted by the Institutional Animal Care and Use Committee (IACUC) of Morehouse School of Medicine. MSM-IACUC adheres to federal guidelines including the National Institute of Health (NIH), the Public Health Service (PHS), and the Animal Welfare Act. Female C57BL/6 J mice aged 6–8 weeks ( n  = 24) were obtained from the Jackson Laboratory (Bar Harbor, ME) and allowed to acclimate for 3 days in the animal facility at Morehouse School of Medicine prior to experimentation. The complete amino acid sequences of the five C. trachomatis (serovar D/UW-3/CX) proteins PmpD, PmpC, PmpG, PorB, and OmcB, and the adjuvant cholera toxin A1 (CTA1) were retrieved from NCBI ( https://www.ncbi.nlm.nih.gov/genbank/ ) in the FASTA format and used for computational prediction. Highly expressed in the chlamydial elementary or reticulate bodies, the Chlamydia antigens were selected based on previous reports of immunogenicity or ability to protect against genital chlamydial infection. CTA1 was chosen for its characterization as a potent mucosal adjuvant (Cunningham et al. 2009 , Farris and Morrison 2011 ). BepiPred-2.0 server ( https://services.healthtech.dtu.dk/services/BepiPred-2.0/ ) was used to predict linear B-cell epitopes from the protein sequences. The method applies a random forest algorithm to distinguish between epitope and nonepitope amino acids. Each protein residue is assigned a score based on a specificity and sensitivity-dependent threshold. Scores above the threshold are predicted to be part of an epitope. For this analysis, the default threshold of 0.5 was applied. Peptide-binding affinity to human alleles was evaluated through the NetMHCII 2.2 server ( http://www.cbs.dtu.dk/services/NetMHCII/ ). The server predicts peptide binding via neural networks trained on binding affinity data from the Immune Epitope Database (IEDB). The output is IC50 values correlating to binding affinity with <50 nM indicating high affinity, <500 nM intermediate affinity, and <5000 nM low affinity. Percentile ranks are also generated comparing the peptides’ scores against scores derived from a set of random natural peptides. A smaller percentile rank denotes high affinity. Predicted B- and T-cell epitopes were submitted to the IEDB’s Population Coverage tool ( http://tools.iedb.org/population/ ) to determine the coverage rate of the vaccine. The tool calculates the fraction of individuals predicted to respond to an epitope based on HLA and MHC binding parameters. Calculation options for MHC-I, MHC-II, and combined MHC were selected for global and regional coverage. Selected high-scoring B- and T-cell epitopes were integrated into the design of the final vaccine construct. Epitopes were joined by GPGPG linkers. The CTA1 adjuvant was added at the N-terminal using an EAAAK linker and a FLAG tag at the C-terminal for protein purification and detection. The adjuvant and the main vaccine sequence were assessed for IFN-γ-inducing epitopes through the IFNepitope server ( http://crdd.osdd.net/raghava/ifnepitope/predict.php ). Predictions are derived from a dataset of IFN-γ-inducing and noninducing MHC II binders. Motif and support vector machine (SVM) algorithms are applied to model IFN-γ versus non-IFN-γ epitopes at a threshold of 0. Separate analyses for the adjuvant and main sequence (MECA) were executed due to residue number restrictions. From sequence input, the tool generates overlapping peptides and their IFN-γ-inducing ability. The antigenicity of CTA1-MECA was determined by ANTIGENpro ( http://scratch.proteomics.ics.uci.edu/explanation.html ) and VaxiJen v2.0 ( http://www.ddgpharmfac.net/vaxijen/VaxiJen/VaxiJen.html ). ANTIGENpro creates an antigenicity index from protein microarray data. The output is a percentage of predicted probability of antigenicity. The accuracy of the server was estimated to be 76% following cross-validation experiments (Magnan et al. 2010 ). VaxiJen uses auto cross covariance (ACC) to transform sequences into vectors of principal amino acid properties. This allows for antigen classification based on physiochemical properties. The results from the analysis include the probability of antigenicity and a statement of antigen or nonantigen according to a predefined cutoff of 0.5. In addition, the average antigenic propensity was determined using the Predicted Antigenic Peptides tool from the Immunomedicine Group of the University of Complutense, Madrid, Spain ( http://imed.med.ucm.es/Tools/antigenic.pl ). The output is a list of antigenic determinants using the method of Kolaskar and Tongaonkar (Kolaskar and Tongaonkar 1990 ). AllerTOP v2.0 ( http://www.ddg-pharmfac.net/AllerTOP ) and AllergenFP ( http://ddg-pharmfac.net/AllergenFP/ ) were used to predict allergenicity. AllerTOP v2.0 also relies on physiochemical properties for its predictions. The alignment-free tool utilizes amino acid E-descriptors, ACC, and k-nearest neighbors (kNN) algorithms to classify allergens. The method has been reported to be 85.3% accurate in a 5-fold cross-validation (Dimitrov et al. 2014a ). AllergenFP is an alignment-free, descriptor-based fingerprint approach used to distinguish allergens and nonallergens. The application depends on a four-step algorithm. First, properties of the protein sequences are described including hydrophobicity, size, relative abundance, and helix and β-strand forming propensities. Generated strings of varying lengths are then converted into vectors of equal length by ACC. Subsequently, vectors are transformed into binary fingerprints and compared by the Tanimoto coefficient. In a comparison between allergens and nonallergens, 88% were identified correctly with a Matthews correlation coefficient of 0.759101 (Dimitrov et al. 2014b ). Physiochemical properties of the vaccine candidate were determined using the ProtParam server ( http://web.expasy.org/protparam/ ). This included its amino acid composition, theoretical pI, instability index, in vitro and in vivo half-life, aliphatic index, molecular weight, and the grand average of hydropathicity (GRAVY) value. Solubility was verified by the PROSO II server ( http://mbiljj45.bio.med.uni-muenchen.de:8888/prosoII/prosoII .seam). PROSO II exploits differences between soluble and insoluble proteins derived from TargetDB, a repository for public information on targets. At a default threshold of 0.6, PROSO II was 71% accurate in a 10-fold cross-validation test (Smialowski et al. 2012 ). Prediction of the secondary structure was performed by the PSIPRED 4.0. To generate the structure, PSIPRED ( http://bioinf.cs.ucl.ac.uk/psipred/ ) first designs a sequence profile constructed by position-specific iterated BLAST (PSI-BLAST). Based on this output, neural networks deliver a secondary structure. PSIPRED 3.2 accomplished a normal Q3 score of 81.6% when a cross-approval strategy was utilized to assess its performance (Khatoon et al. 2017 ). The properties of the secondary structure were further described by DISOPRED3 ( http://bioinf.cs.ucl.ac.uk/disopred/ ) and RaptorX ( http://raptorx.uchicago.edu/StructurePropertyPred/predict/ ). Also implementing PSI-BLAST and neural networks, DISOPRED3 detected intrinsically disordered regions within the structure at a cutoff of 0.5 RaptorX assessed solvent accessibility using the Deep Convolutional Neural Fields (DeepCNF) machine learning model for its predictions. The server accomplished 84% Q3 accuracy for 3-state SS, 72% Q8 accuracy for 8-state SS, and 66% Q3 accuracy for 3-state solvent accessibility (Yang et al. 2018 ). The tertiary structure of the vaccine candidate was modeled by the Iterative Threading Assembly Refinement (I-TASSER) server ( https://zhanglab.ccmb.med.umich.edu/I-TASSER/ ). I-TASSER identifies structural templates from the Protein Data Bank (PDB) using a multiple threading approach. 3D atomic models are then constructed through iterative template-based fragment assembly simulations. From thousands of potential tertiary models, five top models are reported with their associated C-scores, a confidence score estimating the quality of the models. C-scores usually range (−5, 2) with a higher value signifying high confidence. The 3D model chosen from the I-TASSER output was refined through ModRefiner ( https://zhanglab.ccmb.med.umich.edu/ModRefiner/ ) and GalaxyRefine ( http://galaxy.seoklab.org/cgi-bin/submit.cgi?type=REFINE ). ModRefiner applies an algorithm to refine C-α trace, main-chain models, or full atomic models. The aim of the application is to refine models closer to their native state in hydrogen bonds, backbone topology, and side-chain positioning. GalaxyRefine adopts an iterative optimization approach involving the rebuilding of side-chain conformations and repeatedly relaxing the structure after side-chain packing. Five refined models are presented with GDT-HA, RMSD, and MolProbity scoring describing the structural changes following refinement. GDT-HA is the global distance score with a higher value indicating better model accuracy. RMSD measures the root mean square deviation of Cα atom positioning in comparison to the native structure. A lower RMSD is favorable. MolProbity which accounts for clash, poor rotamers, and Ramachandran favored scores, assesses the quality of all atoms of the model with a lower score being ideal. ProSA-web ( https://prosa.services.came.sbg.ac.at/prosa.php ) was used to validate the refined tertiary structure. An overall quality score in the form of a Z-score is provided in the context of all known protein structures. A Ramachandran plot was generated through PDBsum ( http://www.ebi.ac.uk/thornton-srv/databases/pdbsum/ ). The server deploys PROCHECK, a suite of programs, to assess the geometry of the model and predict its stereochemical quality. The Ramachandran plot is a visual representation of the energetically feasible torsional angles, phi (ϕ) and psi (ψ), of the residues in a peptide. The quality of the model is defined by percentages of allowed and disallowed regions. Toll-like receptor 4 (TLR4) plays a role in the recognition of Chlamydia and the induction of inflammatory responses (Nosratababadi et al. 2017 ). Molecular docking of the vaccine with TLR4 (PDB ID: 3FXI) was performed through PatchDock ( http://bioinfo3d.cs.tau.ac.il/PatchDock/ ). Protein–protein docking is based on shape complementarity principles. Submitted molecules are divided into concave, convex, and flat patches. Transformations are generated by matching complementary patches. Candidate transformations are then further evaluated for geometric fit and atomic desolvation energy. A final application of RMSD clustering discards redundant solutions. Potential models are distinguished by a geometric score, interface area size, and desolvation energy (Schneidman-Duhovny et al. 2005 ). Reverse translation and codon optimization was executed in the Java Codon Adaptation Tool (JCat) ( http://www.prodoric.de/JCat ) server. Adaptation is based on codon adaptation index (CAI) values. Codon usage was adapted to Escherichia coli (strain K12). All other parameters were kept at default. From the sequence, JCat measures the CAI and GC-content percentage. CAI measures synonymous codon usage bias with values closer to 1.0 as ideal. The GC-content percentage relates to protein expression levels with a favorable percentage ranging between 30% and 70%. The application then creates an optimized sequence and outputs its CAI index and GC-content. SnapGene ( https://www.snapgene.com/ ) was used to clone the optimized sequence into the pET-32a vector using the restriction enzymes EcoRV and BstBI. To investigate potential CTA1-MECA-induced immune responses, the C-ImmSim server was implemented. C-ImmSim is a model that employs position-specific scoring matrices and machine learning algorithms to predict immune reactions at the cellular and molecular levels. A vaccine regimen of two injections, 3 weeks apart were set through time steps of 1 and 63 (each time step is 8 hours with time step 1 representing injection at time = 0). The remaining simulation parameters were kept as default inputs. Genes synthesis, cloning, protein expression, and purification of the vaccine was performed by Biomatik (Wilmington, DE). The vaccine vector was constructed such that the codon optimized 1980 bp fragment of the N-terminal CTA1-MECA coding sequence was inserted in frame with the C-terminal FLAG contained in plasmid pUC57. Protein expression was detected by SDS-PAGE and immunoblotting analysis using Anti-His antibody. Lyophilized preparations were stored at −80°C. Groups of mice (6/group) were intranasally (IN) immunized with 2, 10, or 20 µg lyophilized CTA1-MECA in 20 µl deionized water or 20 µl Phosphate-Buffered Saline (PBS) alone and boosted at a 3-week interval. The concentration of antigen-specific antibodies (IgG2c and IgA) in pooled sera and vaginal washes collected 2 weeks postimmunization was measured by a standard ELISA procedure described previously (Macmillan et al. 2007 ). Briefly, 96-well plates (Corning) were coated overnight with 10 µg/ml CTA1-MECA and IgA or IgG2c standards. Plates were blocked with 1% bovine serum albumin containing 5% goat serum in PBS and then incubated with sera or vaginal washes. Following incubation with horseradish peroxidase-conjugated goat-antimouse IgG2c or IgA isotype (Southern Biotechnology Associates, Birmingham, AL), plates were developed with 2,2’-azino-bis (3-ethylbenzthiazoline-6-sulfonic acid) substrate (MilliporeSigma, Milwaukee, WI). The absorbance was read on a microtiter plate reader at 492 nm. The results were generated simultaneously with a standard curve and display data sets corresponding to absorbance values as mean concentrations (ng/ml) +/− SD of triplicate wells for each experiment. Statistical analyses were performed with the GraphPad Prism 9 package (GraphPad Software, Inc. La Jolla, CA, USA). Statistical differences between more than two groups by one-way ANOVA. Differences were considered significant at P  < .05.

B-cell epitopes of varying lengths and distribution throughout the protein sequence were predicted for each Chlamydia antigen. High-scoring epitopes were selected and incorporated in the vaccine design (Table  1 ). B- and T-cell epitope predictions. Predicted linear B- and T-cell epitopes from each Ct antigen selected for the vaccine design. High-binding MHC-II epitopes for human alleles were predicted for each antigen. An overlap between predicted B- and T-cell epitopes was observed. T-cell epitopes were also selected and included in the final construct (Table  1 ). According to the population coverage analysis, the predicted epitopes exhibited a 60.29% and 82.35% worldwide coverage for MHC-I and MHC-II epitopes, respectively (Fig.  1 ). For combined MHC-I and II epitopes, a 92.99% worldwide coverage rate was noted. The highest population coverage rate for combined MHC-I and II was observed in Europe (96.21%), followed by North America (96.09%), West Africa (95.22%), South Asia (94.37%), North Africa (93.18%), East Africa (91.94%), Southwest Asia (91.49%), Central Africa (90.9%), Northeast Asia (90.38%), Oceania (89.11%), Southeast Asia (89.01%), West Indies (82.57%), East Asia (82.42%), South America (82.29%), Central America (59.98%), and South Africa (26.35%). Global population coverage. Population coverage rate of predicted T-cell epitopes (MHC-I, MHC-II, and combined MHC). The final construct comprised 20 B- and T-cell epitopes: PmpD (10 epitopes), PmpC (five epitopes), PmpG (one epitope), PorB (three epitopes), and OmcB (one epitope). Epitopes were joined by GPGPG linkers, constituting the main vaccine sequence. The adjuvant CTA1 was linked to main sequence at the N-terminal by EAAAK linkers. In addition, a FLAG tag was added at the C-terminal for protein purification and detection. The final vaccine sequence consisted of 660 amino acids (Fig.  2 ). Schematic representation of CTA1-MECA. The vaccine construct is 660 amino acids long with a CTA1 adjuvant at the N-terminal linked to the multiepitope sequence through an EAAAK linker. Epitopes are linked using GPGPG linkers. A FLAG tag is added at the C-terminal for protein purification and detection. A total of 90 potential IFN-γ-inducing epitopes (15-mer) were predicted for the adjuvant in which 31 epitopes scored above the threshold of 0. For the main vaccine sequence, 12 potential epitopes were predicted with five epitopes scoring positively. All predictions were based on the software’s SVM method. The results indicate that CTA1-MECA induces IFN-γ production. The vaccine candidate was determined to be both antigenic and nonallergenic with and without the adjuvant. In a bacterial model, VaxiJen scored the antigenicity of CTA1-MECA as 0.7335 at a threshold of 0.5. In comparison, an antigenicity index of 0.634732 was assigned by ANTIGENpro. Without the adjuvant, a score of 0.8452 was assigned through VaxiJen and 0.757844 in ANTIGENpro. Furthermore, the Predicted Antigenic Peptides tool predicted 24 antigenic determinants in CTA1-MECA (Fig.  3 ). AllerTOP and AllergenFP predicted the vaccine to be nonallergenic for both with and without the adjuvant. Predicted antigenic propensity (A) Antigenic plot for the vaccine sequence. (B) Server output displaying the sequence and start and end positions of 24 antigenic determinants. The physiochemical properties of the final construct were obtained from ProtParam. The vaccine candidate was predicted to be 67.3 kDa in molecular weight with a theoretical isoelectric point (pI) of 4.94. The half-life of the peptide was estimated to be 1.4 hours in mammalian reticulocytes in vitro and 3 minutes in yeast and >10 hours in E. coli in vitro . Thermostability was confirmed through an instability index of 32.49 and a 67.94 aliphatic index. Ideally, an instability index <40 and higher aliphatic index correlates with a stable structure. Solubility was also confirmed through a negative GRAVY score of −0.3115. The secondary structure of the vaccine was determined to contain 12% alpha helices, 19% beta strands, and 69% coils (Fig.  4A ). Of the 660 amino acids, 6% were predicted to be in disordered domains (Fig.  4B ). Regarding solvent accessibility, 45% of residues were predicted to be exposed, 29% medium exposed, and 26% buried. Secondary structure prediction. (A) Cartoon displaying the structural features of CTA1-MECA. The peptide is predicted to contain 12% alpha helices, 19% beta strands, and 69% coils. (B) Disorder profile of the sequence. Disorder was predicted to be 6%. I-TASSER predicted five tertiary structures of which 1tiiA, 7e2cI, 1lt4, 2nbiA, and ltii threading templates were ideal. The C-scores for the five structures ranged from −3.52 to −1.16, signifying high confidence in modeling. The model with the highest C-score was selected for further refinement (Fig.  5A ). The TM score for the chosen model was estimated to be 0.57 ± 0.15 and RMSD 10.7 ± 4.6, indicating correct topology. Tertiary structure modeling, refinement, and validation. (A) Chosen 3D model of CTA1-MECA from the I-TASSER server and visualized in PyMOL. (B) Refined model by ModRefiner and GalaxyRefiner. The refined structure is superimposed (colored) on the “crude model” (gray). (C) Pro-SA overall model quality depiction with a Z score of −3.94. (D) Ramachandran plot showing the distribution of phi–psi torsion angles (blue) relative to 78.6% most favored (red), 17.1% additional allowed (brown), 1.1% generously allowed (dark yellow), and 3.2% disallowed (pale yellow) regions. The chosen I-TASSER model was refined through ModRefiner and GalaxyRefine. The latter tool presented five refined models. Based on the GDT-HA (0.9288), RMSD (0.494), and MolProbity scores (2.322), model 2 was identified as the best model (Fig.  5B ). The clash score was 13.9, poor rotamers 0.2, and Ramachandran favored 84.2%. The refined model was further validated by Pro-SA-web. The results obtained for the vaccine construct indicated that the refined tertiary structure had a Z-score value of −3.94 (Fig.  5C ). It has been reported that a negative score is indicative of a reliable, good-quality model (Gupta et al. 2013 ). The quality of CTA1-MECA was further validated through a Ramachandran plot analysis, revealing that 78.6% of residues are in the most favored regions (Fig.  5D ). Additionally, 17.1% of residues were in additional allowed regions and 1.1% in generously allowed regions. Only 3.2% of residues were in disallowed regions. To study interactions between TLR4 and CTA1-MECA, molecular docking was conducted via PatchDock. The best-docked complex among 100 models was selected by the highest docking score of 17 992 kcal/mol, indicating the best geometrically complementary relationship between the vaccine and TLR4 (Fig.  6 ). Molecular docking of vaccine candidate with TLR4. Docked conformation of the vaccine construct (blue) with TLR4 (red) obtained from PatchDock. Codon optimization in E. coli (strain K12) was completed in JCat. The codon optimized sequence was 1980 amino acids in length with a CAI of 1 and GC-content of 61.77%, indicating favorable expression in the E. coli host. Vaccine cloning was conducted in the pET32a vector (Fig.  7 ). In silico restriction cloning of the vaccine construct in pET32a expression vector. The optimized sequence was inserted into the vector backbone. In silico vaccine-induced immune responses were evaluated in the C-ImmSim server. There were notable increases in all measures of immunity. An increase in antibody levels (IgM + IgG, IgM, IgG1 + IgG2, IgG1, and IgG2) coincided with a decrease in antigen levels (Fig.  8A ). Consistent with the IFNepitope prediction of CTA1-MECA as IFN-γ-inducing, high titers of IFN-γ were produced (Fig.  8B ). The development of immune memory is also suggested by the increase of T-helper and B-cells following a boost injection of the vaccine candidate (Fig.  8C and  D ). C-ImmSim server prediction results of immune response after administering vaccine construct. (A) Antigen and immunoglobulins. (B) Induced levels of the cytokine and Simpson index. (C) CD4 helper T-cell population subdivided per entity state (active, resting, anergic, and duplicating). (D) B-lymphocytes cell population. Vaccine cloning was conducted in the pUC57 vector (Fig.  9A ). The purity of the purified protein was determined by Coomassie staining (Fig.  9B ). Following transformation of E. coli BL21 competent cells with pUC57 plasmid, expression of the CTA1-MECA protein was confirmed by western immunoblotting analysis using Anti-His antibody (Fig.  9C ). Construction of vaccine vector and expression of recombinant protein. (A) The vaccine vector was constructed by genetically inserting the amplified 1980 bp fragment of CTA1-MECA in plasmid vector pUC57. (B) Coomassie stained SDS-PAGE gel. Lane M1, protein marker; Lane 1, CTA1-MECA protein. (C) Protein expression was detected by immunoblotting analysis using Anti-His antibody. Lane M2, protein marker; Lane 2, CTA1-MECA protein. CTA1-MECA-specific antibody responses elicited 2 weeks after the last immunization were measured by titrating serum and vaginal secretions of vaccinated and control mice against CTA1-MECA. Significantly elevated levels of CTA1-MECA-specific IgG2c antibodies were observed in both serum (Fig.  10A ) and vaginal secretions (Fig.  10B ) of mice immunized with CTA1-MECA compared to the PBS control group. Similarly, higher levels of CTA1-MECA-specific IgA antibodies were detected in both serum (Fig.  10C ) and vaginal secretions (Fig.  10D ) of immunized mice compared to the PBS control group. In both serum and genital secretions, IgG2c levels were significantly higher compared to IgA levels. The antibody responses exhibited a dose-dependent trend in serum, with the 20-µg dose of CTA1-MECA demonstrating an immunogenic advantage. However, in the vaginal secretions, a nonlinear response was observed across different doses. The results collectively indicate that immunization with the vaccine candidate induces robust antichlamydial antibodies in systemic and mucosal tissues. Chlamydia -specific systemic and mucosal IgG2c and IgA antibody responses. Mice were immunized IN twice, 3 weeks apart. Serum and vaginal wash samples were collected from each mouse 2 weeks after the last immunization. A standard ELISA procedure was used to assess the concentration of IgG2c and IgA elicited in serum (A–C) and vaginal wash (B–D). The results were generated simultaneously with a standard curve and display data sets corresponding to absorbance values as mean concentrations (ng/ml) +/− SD of triplicate wells for each experiment. Significant differences between groups were evaluated by one-way ANOVA with Tukey’s post multiple comparison test at P  < .05.

The absence of an approved vaccine for genital chlamydial infections in humans poses a significant public health challenge. The repercussions of untreated infections, particularly in women, are profound, encompassing complications such as PID, ectopic pregnancy, and tubal infertility (Paavonen and Eggert-Kruse 1999 ). Despite the availability of effective antimicrobial treatments, the persistent rise in the incidence of Ct infections remains a pressing concern (ref). In 2021 alone, the CDC reported a staggering 2 million new cases of infection in the USA, emphasizing the urgency for proactive intervention (Centers for Disease Control and Prevention 2023c ). The imperative for a prophylactic vaccine to combat asymptomatic infections and curtail transmission is firmly grounded in evidence (Brunham and Rappuoli 2013 , de la Maza et al. 2021 ). Mathematical models have meticulously evaluated the potential impact of such a vaccine across various efficacy and coverage rates, revealing promising outcomes (Gray et al. 2009 , Owusu-Edusei et al. 2015 ). These findings not only highlight the cost-effectiveness of vaccination but also underscore its practicality in stemming infection rates. However, amidst these promising prospects, the importance of developing a viable vaccine to control and prevent chlamydial infections looms large, necessitating a multifaceted approach to address this persistent public health challenge. Traditional vaccine development approaches are facing logistical challenges and safety concerns, prompting a transition towards exploring alternative vaccination methods (Madewell et al. 2021 ). Epitope-based vaccines stand out for their ability to trigger targeted immune responses and streamline vaccine development through advancements in immunoinformatics (Shawan et al. 2023 ). Contrasting conventional approaches, epitope-based vaccines offer advantages such as cost-effectiveness, safety, and customization, rendering them particularly appealing in the fight against infectious diseases (Excler et al. 2021 ). Previous research has delved into in silico vaccine design targeting a range of organisms, establishing a strong foundation for future exploration and innovation in this domain (Michel-Todó et al. 2019 , Shey et al. 2019 , Khalid et al. 2022 , Evangelista et al. 2023 ). Our prior research has focused on the development of Vibrio cholerae ghost (VCG)-based subunit vaccines expressing various Chlamydia antigens. VCGs are bacterial cell envelopes that have been emptied of their cytoplasmic contents and cholera toxin through genetic inactivation, leading to the expulsion of cellular material (Eko et al. 2000 ). They are appealing as nonliving vaccine carriers because they are nontoxic, preserve the structural and functional integrity of expressed antigens, and excel at delivering vaccine antigens to primary antigen-presenting cells (Szostak et al. 1996 ). A VCG-based chlamydial vaccine (rVCG-PmpD/PorB) was evaluated for its potential to confer cross-protection against heterologous genital Ct infection and associated upper genital tract pathologies in mice (Pais et al. 2017 ). Immunization prompted robust CD4+ T-cell and humoral responses, reducing chlamydial shedding and inflammation while protecting against infertility. Furthermore, we compared single (rVCG-MOMP) and multisubunit (rVCG-MOMP/OMP2) vaccines, with the latter demonstrating superior immunogenicity, characterized by an increased frequency of Th1 cells and enhanced capacity to provide protective immunity (Eko et al. 2004 ). Building upon these in vivo findings, we explored an in silico vaccine design integrating these antigens alongside other immunogenic antigens. Recognizing the importance of adjuvants in augmenting vaccine immunogenicity, we investigated the incorporation of CTA1, a potent mucosal adjuvant derived from cholera toxin, into our multiepitope vaccine design. CTA1 has been shown to elicit a robust and well-balanced CD4+ T-cell response while significantly enhancing specific antibody production (Berry et al. 2004 ). Integrating insights from both our studies and animal or clinical research, we meticulously selected five Chlamydia antigens (PmpC, PmpD, PmpG, OmcB, and PorB). The Chlamydia genome encodes ~900 proteins (Stephens et al. 1998 ). Prior studies have utilized immunoinformatics to identify immunogenic antigens. Recently, Shiragannavar et al. ( 2020 ) employed a reverse vaccinology approach to identify immunogens from the entire proteome of Ct . Out of the 895 proteins analyzed, 477 were identified as probable antigens. However, further analysis focused on five proteins derived from PmpD, PmpF, PknD, and Yop translocation protein, selected based on epitope conservancy and HLA MHC allele binding. In another study, in the assessment of B- and T-cell epitopes within PmpD, epitopes were found to be distributed across the sequence (Russi et al. 2018 ). Specifically, 70 T-cell epitopes exhibited strong affinity to MHC I molecules, 742 to MHC II molecules, and 24 linear B-cell epitopes. In this study, 20 high-scoring epitopes were chosen from the five antigens and employed in designing the novel vaccine candidate, CTA1-MECA. Widely used to enhance conformational stability in vaccine constructs, the construct comprises epitopes linked by GPGPG linkers and the EAAAK-adjoined adjuvant, CTA1. Effective vaccine design relies on the comprehensive evaluation of multiple factors governing immune recognition and response. These factors include MHC class I/II binding, antigenicity, allergenicity, and the ability to induce IFN-γ production. The selected epitopes underwent rigorous analysis to assess these properties. Our findings revealed that CTA1-MECA demonstrated promising compatibility with both MHC class I and II binding, covering an estimated 92.99% of the global population. Notably, Aslam et. al. ( 2021 ) demonstrated a Ct vaccine candidate with 99.9% population coverage, highlighting the varying degrees of coverage achievable with different vaccine designs. The distribution of HLA alleles varies significantly across different ethnicities and geographical regions, underscoring the need for vaccine designs that account for genetic diversity to ensure equitable distribution and effectiveness across populations (Adhikari et al. 2018 ). In addition, we also observed that the vaccine induced IFN-γ, IgG + IgM antibodies, and activated T- and B-cell populations in immune simulations. By triggering both proinflammatory and adaptive immune pathways, the vaccine demonstrates its ability to engage various components of the immune system. Furthermore, the vaccine was predicted to be antigenic and nonallergenic, supporting its potential as a viable vaccine candidate. The interplay among antigen structure, stability, and immunogenicity is intricate. In line with recommendations for structure-based vaccine design, we employed deep-learning methods to analyze the structural properties of CTA1-MECA (Byrne and McLellan 2022 ). The vaccine candidate, characterized by a molecular weight of 67.3 kDa, demonstrated thermostability, solubility, and an acidic nature, with a theoretical pI of 4.94. Secondary structure analyses revealed that the vaccine primarily consists of 69% coils and 6% disordered regions. These coils, known for their stability, along with the flexibility provided by disordered regions, may significantly influence vaccine efficacy (van der Lee et al. 2014 ). Furthermore, the refined tertiary structure exhibited favorable characteristics, including 96.8% of residues located in favored and allowed regions and only 3.2% in disallowed regions. Also essential to our vaccine’s structural design is the assessment of TLR4 binding and its integration into a suitable vector. The best understood and most prominent pathogen recognition receptors in chlamydial infections are toll-like receptors (TLRs). TLR4 is known as the primary signal transducer in the recognition of lipopolysaccharide, a component of the outer membrane of Ct (Joyee and Yang 2008 ). Upon recognition, TLR4 triggers a signaling cascade that leads to the activation of innate immune responses (Nosratababadi et al. 2017 ). A notable interaction was observed between CTA1-MECA and TLR4, evidenced by the docking score of the complex. To confirm proper expression in a host, the codon-optimized vaccine sequence exhibited optimal expression levels in E. coli (strain K12). The vaccine’s structural attributes highlight its optimal design as a stable formulation, indicating encouraging stability, safety, and efficacy potential. With this confirmation, CTA1-MECA was expressed in pUC57 and investigated in mice. Our exploration into the immunological effects of intranasal immunization with CTA1-MECA unveiled the initiation of humoral responses in vivo . Significantly, both systemic and mucosal IgG2c and IgA antibodies were elevated in serum and vaginal wash secretions compared to the PBS control group, suggesting the effectiveness of vaccination. IgG2c is primarily associated with systemic immunity and Th1-type responses, whereas IgA is linked to mucosal immunity, often displaying a shorter duration and greater variability in comparison to IgG responses (Piura et al. 1985 , Geisler et al. 2012 ). Consistent with our earlier investigations, we observed increased levels of IgG2c relative to secretory IgA in vaginal secretions, underscoring the importance of the Th1-associated IgG2c antibody isotype in conferring immunity against Chlamydia (Moore et al. 2003 , Agrawal et al. 2009 ). Moreover, previous studies have emphasized the role of humoral responses in safeguarding against Ct (Su et al. 1997 , Morrison and Morrison 2005 ), reinforcing the significance of our antibody responses. The antibody response patterns introduce a layer of complexity to our findings. While the highest dosage elicited a more pronounced systemic immune response, a nonlinear pattern emerged in vaginal secretions, suggesting intricate interactions within the local immune environment. While our study sheds light on the development and immunogenicity of the CTA1-MECA vaccine, it has limitations. Depending solely on computational predictions for epitope selection and vaccine design might introduce biases and overlook aspects of immune responses best studied empirically. Furthermore, the in silico approach may not fully capture the intricate complexity of host–pathogen interactions or immune response variability. Although we conducted preliminary in vivo experiments in mice, further experimental validation is necessary. Additionally, the chosen immunization route might not fully represent the most effective vaccination strategy against Ct . Exploring alternative routes and adjuvants could enhance vaccine delivery and immune response. Lastly, while our study presents promising outcomes, translating findings into clinical applications may encounter regulatory and logistical hurdles, necessitating further research efforts.

Current strategies to combat rising Ct infections are limited in efficacy, calling for the development of a proper vaccine. In this study, an immunoinformatics approach was utilized to construct and validate a multiepitope Ct vaccine antigen (CTA1-MECA). With the selection of previously tested Chlamydia antigens, in silico predictions, and findings in mice, the vaccine candidate has the potential to be prophylactic. Future work involves additional in vivo studies to confirm its functionality.