In silico design and assessment of a multiepitope and multiantigen potential vaccine candidate against Leishmania donovani

Abstract

Visceral Leishmaniasis (VL), primarily caused by Leishmania donovani, presents a formidable public health challenge, with existing therapeutic options—such as pentamidine, amphotericin B, miltefosine, and paromomycin—being limited by adverse effects, high costs, restricted availability, and the emergence of drug-resistant strains. Despite the use of combination therapies, which have achieved an approximately 85% cure rate, the lack of a safe, effective, and durable vaccine remains a critical barrier to VL eradication. This study employed a comprehensive immuno-bioinformatics approach to design and analyse a multi-epitope vaccine candidate targeting essential immunogenic proteins in Leishmania spp: stress-induced proteins, Prohibitin (PHB), and GP63. These proteins play crucial roles in parasite survival, virulence, and host immune modulation. Our computational analysis identified epitopes from these proteins that were predicted to elicit strong humoral and cellular immune responses while being non-allergenic. The designed vaccine construct demonstrates substantial immunogenic potential, highlighting its promise as a candidate for further preclinical and clinical evaluation in pursuing an effective VL vaccine.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40203-025-00433-2.

Introduction

Visceral Leishmaniasis (VL) is one of the lethal manifestations of leishmaniasis is being primarily cause by Leishmania donovani (Mann et al. 2021). Pentamidine, amphotericin B, miltefosine, and paromomycin are among the few medications used to treat this illness (Singh et al. 2023). Nevertheless, these medications frequently have disadvantages, such as severe side effects, exorbitant prices, restricted availability, and the development of drug resistance (Singh et al. 2023). According to Pacheco-Fernandez et al. (2023), a combination of multiple drugs now has an 85% cure rate; yet, the continuous appearance of new patients is a problem. VL cases in non-endemic areas and the emergence of novel Leishmania strains triggered by genetic alterations in response to medication exposure are two factors accelerating the spread of leishmaniasis (Singh et al. 2023). Additionally, attempts to eradicate VL have been limited by the lack of a viable vaccine (Parkash et al. 2021).

Strategies for developing a vaccine against VL include the use of live-attenuated or dead parasites, cellular extracts, and individual or transgenic parasite immunogens (Lage et al. 2020). Various previous development projects of vaccines achieved clinical trials till Phase I and III (Carvalho et al. 2023). In the Phase I trial, second-generation bioengineered leishmanial antigens such as monopeptide and protein chain, LEISH-F3 adjunct with GLA-SE and ChAd63-KH, a third-generation DNA vaccine showed promising outcomes against VL and PKDL. In the phase III trial, only one first-generation vaccine could perform a suboptimal efficacy (Saha et al. 2022). However, developing a safe, efficacious, durable, and cost-effective preventive VL vaccine remains a formidable challenge. Epitope-based chimeric or subunit vaccines offer compelling advantages over traditional vaccines by avoiding the inclusion of entire pathogens; whether live or inactivated, they provide improved specificity and stability. Hence, identifying the antigenic regions of immune-stimulating proteins becomes crucial in the rational design and development of peptide vaccines.

Proteomic analyses have identified numerous crucial immunogenic candidates in Leishmania parasites, closely associated with infection and pathogenicity (Jesus-Oliveira et al. 2023). Stress-induced proteins (STI1), Prohibitin (PHB) and GP63 are paramount in immunogenicity. Stress-induced proteins help the antigen-processing pathway by acting as immunodominant antigens and chaperones (Requena et al. 2015). In particular, heat shock proteins (HSPs) like HSP70 and HSP83 play a crucial part in this process; there is data that they help infected individuals produce Th-1 cytokine responses (Binder 2014). The survival and virulence of the parasite, especially under stressful situations, depend on these proteins. In order to create a multi-epitope subunit vaccination against VL, we used a thorough immunoinformatics strategy in this work. Three immunogenic proteins—Stress-Induced Protein 1 (STI1), Prohibitin (PHB), and GP63—were chosen because of their vital functions in antigenicity, host immune regulation, and parasite survival. To find antigenic, non-allergic, and non-toxic B-cell, CTL, and HTL epitopes, computational techniques were employed.

These were assembled into a single vaccine construct with appropriate linkers and an adjuvant to enhance immunogenicity. The designed vaccine (Fig. 1A) was further evaluated for population coverage, structural integrity, docking with immune receptors (e.g., TLR4), and immune simulation, indicating strong potential for initiating robust cellular and humoral immune responses. The vaccine shows substantial immunogenic potential, warranting further in-vivo and in-vitro validation in future research phases.

Fig. 1.

A Schematic illustration of the finalized multi-epitope vaccine construct. B Visualization of the refined 3D vaccine model highlighting helical, sheet, and loop regions. C Ramachandran plot analysis performed using the PROCHECK server to assess stereochemical quality D Structural validation of the vaccine model through ERRAT, yielding a quality score of 97.13. E ProSA-web analysis showing the Z score of the vaccine model

Materials and methods

Retrieval of protein sequences and screening for antigenic and allergic Properties

The amino acid sequences of Prohibitin (LDBPK_161710), stress-induced proteins (LDBPK_081020, LDBPK_360080), and GP63 glycoproteins (LDBPK_280600, LDBPK_280610) were retrieved from the KEGG genome database (Kanehisa et al. 2023) in FASTA format for immunoinformatic analysis. LDBPK_081020 and LDBPK_360080 represent distinct homologs within the stress-response protein family, while LDBPK_280600 and LDBPK_280610 are isoforms of the GP63 virulence factor. All were evaluated independently to capture sequence-level diversity and maximize epitope coverage for robust vaccine design. Antigenicity, allergenicity, and toxicity of the selected proteins were assessed using VaxiJen v2.0 (threshold ≥ 0.5) https://www.ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen.html (Doytchinova and Flower 2007), AllerTOP v2.0 http://www.ddg-pharmfac.net/AllerTOP/ (Dimitrov et al. 2014), and ToxinPred https://webs.iiitd.edu.in/raghava/toxinpred2 (Gupta et al. 2013a, b), respectively.

Prediction and assessment of CTL epitopes

CTL epitopes are crucial for eliciting MHC-I-regulated cellular immune responses (Saha and Raghava 2006). The IEDB MHC-I binding tool, which may be found at (http://tools.iedb.org/mhci/) (Fleri et al. 2017), was used to predict 9-mer cytotoxic T lymphocyte (CTL) epitopes. The SMM algorithm was chosen as the primary prediction method within the IEDB MHC-I binding tool, and the generated peptides were organized according to their projected Ic50 values.

The projected IC50 value, derived from the Stabilized Matrix Method (SMM), represents the concentration (in nM) required to inhibit 50% of peptide-MHC binding. Lower IC50 values correlate with higher binding affinities; epitopes with IC50 < 50 nM are considered high-affinity binders, while those < 500 nM are generally considered biologically significant.

The output format was set up to produce an XHTML table, and the MHC source species was carefully chosen to be human. The immunogenicity for the target epitopes was evaluated using the class I immunogenicity tool from the IEDB Analysis Resource (http://tools.iedb.org/immunogenicity/). Only the epitopes with a positive immunogenicity value were kept for the following analysis phases. A positive immunogenicity value, as computed by the IEDB Class I Immunogenicity tool, indicates a predicted likelihood of an epitope to effectively elicit a CD8⁺ T-cell-mediated immune response. Only epitopes with immunogenicity scores greater than zero were selected for further analysis.

To ensure translational relevance and broad immunogenic coverage, epitope predictions were made against a panel of highly prevalent HLA class I alleles, including HLA-A01:01, A02:01, A03:01, A24:02, A26:01, B07:02, B08:01, B27:05, B39:01, B40:01, B*58:01, and B*15:01. These alleles were selected based on their global and South Asian population frequency data as reported by the IEDB population coverage resource. The selected CTL epitopes were assessed based on their antigenicity using the VaxiJen v2.0 server.

Prediction and assessment of HTL epitopes

Helper T lymphocytes (HTLs) play a central role in orchestrating adaptive immune responses by activating B cells, cytotoxic T cells, and macrophages. In this study, 15-mer HTL epitopes were predicted using the IEDB MHC-II binding tool (http://tools.iedb.org/mhcii/), employing the SMM-align algorithm (Paul et al. 2013), with Homo sapiens selected as the host species.

For MHC class II-restricted epitopes, adjusted percentile rank values were used to normalize binding predictions across different alleles. A lower adjusted rank indicates stronger predicted binding affinity; epitopes with ranks < 2% are typically considered high-confidence MHC-II binders (Lim et al. 2021).

A panel of particularly prevalent HLA class II alleles were selected for the binding predictions in order ensure coverage and applicability across the entire population. These included, corresponding to their reported global and South Asian allele frequencies in the IEDB database, HLA-DRB103:01, DRB107:01, DRB1*11:01, and DRB1*15:01. The purpose of this selection technique was to improve the anticipated HTL epitopes' wide immunogenic potential and relevance.

The epitopes selected for this study underwent a thorough screening process. Initially, the VaxiJen v2.0 and AllergenFP v1.0 programs were used to assess their antigenicity and allergenicity. The IFNepitope tool http://crdd.osdd.net/raghava/ifnepitope/ was then used to thoroughly evaluate the remaining epitopes' ability to trigger certain cytokines Dhanda et al. (2013). A hybrid approach that combines a support vector machine (SVM) and motif-based analysis was used to investigate the induction of interferon (IFN) (Zhu et al. 2024). The inducibility of interleukin-10 (IL-10) was evaluated using the IL10pred software, which uses an SVM prediction model with a threshold of -0.3 (Sarvmeili et al. 2024).

B-cell epitope prediction& assessment

B cells, a subset of antigen-presenting cells, possess specific receptors identifying B-cell epitopes on parasites. Upon interaction, B cells become activated, initiating a humoral immune response. Memory B cells are also generated and can mount a high-affinity humoral response upon subsequent exposure to the same antigen. Thus, analysing B-cell epitopes is crucial for the development of effective vaccines. Using a 0.5 cutoff value, the BePipred linear epitope prediction 2.0 program (http://tools.iedb.org/bcell/) was used to predict linear B-cell lymphocyte (LBL) epitopes for this study.

Epitope toxicity prediction

The toxicity of selected peptides is assessed using the ToxinPred (Gupta et al. 2013a, b). In addition to being beneficial for predicting peptide toxicity, computational techniques can also be utilized to create more usable peptides with lower toxicity levels (Robles-Loaiza et al. 2022). These rigorous criteria ultimately resulted in the inclusion of only epitopes with enhanced antigenic, non-allergenic, and non-toxic characteristics as well as the ability to trigger the specified cytokines in the ultimate vaccine formulation.

Designing a multi-epitope vaccine construct

The vaccine candidate has been developed by joining B and T cell epitopes employing linkers. To increase the vaccine's immunogenicity, beta defensin 3 (Q5U7J2) (Naz et al. 2020) was fused to the N-terminal region applying the EAAAK linker as an adjuvant. The AAY linker was used to link the CTL epitopes, the GPGPG linker was used to link the HTL epitopes, and the KK linker was used to link the B-cell epitopes (Naz et al. 2020).

Population coverage

A key component of computational vaccine design is population coverage prediction. The general efficacy of the produced vaccine is assessed by examining the percentage of human leukocyte antigen (HLA) alleles associated with the relevant epitopes. As study input data, the total coverage of the selected MHC class I and II alleles and T cell epitopes is assessed.The IEDB population coverage tool (http://tools.iedb.org/population/) was used for this analysis (http://tools.iedb.org/population/). These included highly prevalent alleles across global populations, particularly from South Asia and India, ensuring optimal coverage and translational relevance.

Allergenicity, antigenicity, solubility and physicochemical analysis of candidate vaccine

Using tools called AllerTop and VaxiJen, respectively, the allergenicity and antigenicity of the vaccine design were predicted. The physicochemical characteristics of the vaccine were described via the ProtParam (https://web.expasy.org/protparam/)tool on the Expasy online site (Kalita et al. 2020).

Tertiary structure prediction, validation and model refinement

The tertiary structure of the vaccine construct was initially predicted using the Robetta server (https://robetta.bakerlab.org/) The quality of the predicted structure was then assessed with the SAVES v6.0 server (https://saves.mbi.ucla.edu/), focusing on ERRAT and PROCHECK parameters. The predicted 3D structures were refined using the GalaxyRefine tool (https://galaxy.seoklab.org/cgi-bin/submit.cgi?type=REFINE), which applies CASP10-based methodologies for repacking and molecular dynamics modelling to enhance global and local structural features. The refined vaccine structures were subsequently re-validated using ERRAT, PROCHECK, and ProSA-web servers (https://prosa.services.came.sbg.ac.at/prosa.php). ProSA-web calculates the Z-score to assess overall quality, while PROCHECK evaluates stereochemical quality through residue geometry analysis and Ramachandran plots.

Molecular docking with host immune receptor

Molecular docking studies elucidated the intricate binding interactions between the engineered vaccine constructs and immune receptor proteins. The refined tertiary structure of the vaccine construct underwent docking simulations against the TLR4 receptor (PDB: 3FXI) utilizing the HADDOCK 2.0 web server (https://wenmr.science.uu.nl/), a sophisticated tool for macromolecular interactions (de Vries et al. 2007). PDBsum (Laskowski et al. 2018) was used to do a thorough study and visualization of the molecular interactions within the vaccine-TLR4 complex, providing thorough insights into residue-level interactions and binding affinities.

Normal mode analysis

The accomplished vaccination-TLR4 complex was subjected to Normal Mode Analysis (NMA) using the iMODS (https://imods.iqf.csic.es/) server, which extensively evaluated the internal dihedral coordinates and defined the cooperative functional dynamics of the peptide vaccination (López-Blanco et al. 2014) energy minimization, molecular stability, and in-depth atomic mobility characterization of the selected vaccine design inside the docked complex were guaranteed using the iMODS essential dynamics simulation framework. Using sophisticated criteria like B-factors, eigenvalues, root-mean-square deviation (RMSD), complex deformability metrics, covariance matrices, and elastic network models, this advanced analysis gave insight on the likely motions of the V1-TLR4 complex and provided a thorough understanding of the system's dynamic behavior.

Immune response simulation

In silico investigation using the C-ImmSim online dynamic immune simulation server (https://kraken.iac.rm.cnr.it/C-IMMSIM) which models the mammalian immune system through agent-based methods. examined the vaccine construct's immunological responses (Elshafei et al. 2024). C-ImmSim simulates mammalian immunological responses after vaccination including T-cell activation, B-cell proliferation, cytokine secretion, and memory cell formation. C-ImmSim uses a matrix-based epitope prediction and machine-learning immune interaction prediction on an agent-based model. The bone marrow, lymphatic organs, and thymus, which select naive T-cells to prevent autoimmune reactions, are included in this simulation model. For the simulation, three injections were administered at day 0, day 28, and day 56, mimicking a typical prime-boost-boost vaccination schedule. The simulation time steps were set to cover up to 1050 time steps (approximately 350 days), allowing analysis of both short- and long-term immune responses. The default set of human HLA alleles was used, which includes common Class I and Class II alleles to reflect global population diversity (Elshafei et al. 2024). This simulation setup enabled a detailed prediction of the designed vaccine construct’s potential to trigger innate and adaptive immune responses.

Codon optimization and in-silico cloning

Codon optimization is crucial to increase or promote the expression and production of recombinant vaccine constructs. The JCat tool (https://pubmed.ncbi.nlm.nih.gov/15980527/) was used to ensure optimal expression within the chosen system (Bhowmik et al. 2023). This tool requires two inputs: the protein sequence and the organism for vaccine production. It outputs sequence assessment, Codon Adaptation Index (CAI), GC content percentage, and the optimized sequence. A CAI above 0.8 and 30–70% GC content indicate effective protein translation. Integrated with the PRODORIC database, JCat provides species-specific information. The recombinant plasmid was then created using SnapGene (https://www.snapgene.com/), adding EcoRI and XhoI restriction sites to the vaccine construct's 5' and 3' ends, respectively. With restriction sites, this refined sequence was cloned into the pET8a vector.

Results

The amino acid sequences of various L. donovani proteins were obtained from the KEGG database to construct a multi-subunit vaccine (Supplementary Table S1). Protein selection was based on antigenic scores exceeding 0.5 (Table 1). Additionally, all proteins were verified to be non-allergenic and non-toxic (Table 1).

Table 1.

Predicted Antigenicity, allergenicity and toxicity of selected proteins

Sl No	Protein id	Vaxijen score	Allergenicity	Toxicity
1	LDBPK_280600	0.6031	Non- allergen	Non-toxic
2	LDBPK_280610	0.6215	Non- allergen	Non-toxic
3	LDBPK_161710	0.5002	Non- allergen	Non-toxic
4	LDBPK_081020	0.5164	Non- allergen	Non-toxic
5	LDBPK_360080	0.6295	Non- allergen	Non-toxic

Identification of CTL epitope (MHC-I) and assessment of immunogenicity

Seven cytotoxic T lymphocyte (CTL) epitopes were identified based on their high binding affinity to MHC class I alleles, low predicted IC50 values (ranging from 3.72 to 52.47 nM), and favorable immunological characteristics (Table 2). The majority of epitopes exhibited strong binding, with SMM percentile ranks ≤ 0.5, including FQANAYVRV and SPRVTTVAW, which showed the lowest IC50 values (3.72 and 8.79 nM, respectively) and high antigenicity scores, supporting their potential as dominant CD8 + T cell targets. All selected peptides were predicted to be non-allergenic and non-toxic, and VaxiJen scores ranged from 0.6724 to 2.0432, indicating strong antigenic potential. Although epitopes such as FGLDYDERI and SISPGVAGA presented slightly higher IC50 values and ranks, they remained within the accepted thresholds for MHC-I binding affinity. Collectively, these epitopes demonstrate robust immunogenicity, broad HLA allele coverage, and safety, validating their inclusion in the final multi-epitope vaccine construct.

Table 2.

Selected CTL epitopes for vaccine construct

Sl no	Protein ID	Epitopes	Alleles	SMM_IC50	SMM_rank	Vaxijen score	Immunogenicity	Allergenicity	Toxicity
1	LDBPK_280600	IAAMEDTGY	HLA-B*35:01	23.15154614	0.5	0.6724	0.00694	Non- allergen	Non_toxic
2	LDBPK_280610	FQANAYVRV	HLA-A*02:06	3.720060475	0.2	2.0432	0.10589	Non- allergen	Non_toxic
3		SPRVTTVAW	HLA-B*07:02	8.787999024	0.1	0.9849	0.19038	Non- allergen	Non_toxic
4	LDBPK_161710	RVRPRTLHT	HLA-A*30:01	21.92653467	0.4	0.9878	0.10212	Non- allergen	Non_toxic
5		FGLDYDERI	HLA-A*02:06	41.73976504	1.4	1.6533	0.15074	Non- allergen	Non_toxic
6	LDBPK_081020	HAYFWTKQY	HLA-B*35:01	21.21192095	0.4	1.2564	0.11916	Non- allergen	Non_toxic
7	LDBPK_360080	SISPGVAGA	HLA-A*02:03	52.4747043	1.8	0.9722	0.05982	Non- allergen	Non_toxic

Identification of HTL epitope (MHC-II) and assessment of immunogenicity

Five helper T lymphocyte (HTL) epitopes were identified based on strong binding affinity to HLA class II alleles, low percentile ranks, and favourable immunogenic profiles. The selected epitopes exhibited SMM-aligned IC50 values ranging from 6 to 26 nM and percentile ranks ≤ 3.9, indicating efficient MHC-II binding (Table 3). Notably, QEPSGAGGAVPTTAV and PTGELAVAAAATGAW showed the strongest binding affinities with IC50 values of 6 and 9 nM, respectively, and were restricted by the HLA-DQA105:01/DQB103:01 allele. All epitopes demonstrated antigenic potential with VaxiJen scores ranging from 0.6342 to 1.2683. IFN-γ induction scores were positive for all candidates, supporting their potential to elicit Th1-type immune responses, although none were predicted to induce IL-10. Collectively, these epitopes demonstrated strong binding, high antigenicity, and favorable cytokine profiles, supporting their inclusion in the final vaccine construct to promote robust CD4 + T cell-mediated immunity.

Table 3.

Selected HTL epitopes for vaccine construct

SlNo	Protein id	Selected epitopes	Biding alleles	SMM align ic50	Percen -tile rank	vaxijen score	IFN-pred score	IL-10 pred score	IL-10 induction
1	LDBPK_280600	HLA-DQA1*05:01/DQB1*03:01	PTnGELAVAAAATGAW	9	0.39	0.6342	0.48221471	− 0.92475308	Non-inducer
2	LDBPK_280610	HLA-DQA1*05:01/DQB1*03:01	QEPSGAGGAVPTTAV	6	0.21	1.2683	0.49693853	− 0.67345334	Non-inducer
3	LDBPK_161710	HLA-DRB1*11:01	NVRLRVLFRPMADRL	26	1.1	0.8827	0.58096866	0.096077933	Non-inducer
4	LDBPK_081020	HLA-DRB1*01:01	SGISAKINKLISAGI	24.00	3.9	0.6658	0.45955353	− 0.33786956	Non-inducer
5	LDBPK_360080	HLA-DQA1*05:01/DQB1*03:01	KAQSISPGVAGASQG	13.00	0.73	1.0766	0.55187607	− 0.33786956	Non-inducer

Prediction and assessment of linner B-cell epitopes

Three linear B-cell epitopes were selected based on high AbcPred scores (ranging from 0.69 to 0.97), strong antigenicity, and favourable safety profiles. Among them, GGTGGRPLRMDSNSHS exhibited the highest prediction score (0.97), while KISGLKDSVYGEGLQG demonstrated the highest VaxiJen score (1.941), indicating strong potential for B-cell activation and antibody generation. All selected epitopes were predicted to be non-allergenic and non-toxic, supporting their inclusion in the final construct as safe and immunogenic B-cell component. (Table 4).

Table 4.

Selected LBL epitopes for vaccine construct

SL NO	Protein id	B-cell epitopes	Abcpred score	Vaxijen score	Allergenicity	Toxicity
1	LDBPK_280610	GGTGGRPLRMDSNSHS	0.97	1.3888	Non-allergen	Non-toxic
2	LDBPK_280610	DSVYGEGLQGRIIGLD	0.76	1.4462	Non-allergen	Non-toxic
3		KISGLKDSVYGEGLQG	0.69	1.941	Non-allergen	Non-toxic

Designing multi-epitope vaccine construct

A multi-epitope subunit vaccine comprising 5 HTL, 7 CTL, and 3 B-cell epitopes was constructed, resulting in a 317-amino-acid sequence. Epitope segments were joined using specific linkers—AAY for CTL, GPGPG for HTL, and KK for B-cell epitopes—while a β-defensin 3 adjuvant was fused at the N-terminus via an EAAAK linker to enhance immunogenicity (Fig. 1 A). The construct’s antigenicity, allergenicity, and toxicity were evaluated using VaxiJen v2.0, AllerTOP v2.0, and ToxinPred, respectively. The VaxiJen score of 0.7801 classified the construct as a probable antigen, while predictions confirmed its non-allergenic and non-toxic nature. To assess cross-reactivity, all selected peptides were subjected to BLASTp analysis against the human proteome, revealing no significant sequence homology (identity < 70%, limited alignment), thus minimizing the risk of autoimmune responses.

Population coverage

Predicting population coverage is crucial for computational vaccine design. The prevalence of HLA alleles linked to the epitopes is examined to assess the overall effectiveness of the vaccination. The IEDB population coverage tool evaluated the coverage of epitopes. As a result, all epitopes had average Class I and Class 2 coverage scores for the world, South Asia, and India of 92.01, 87.56, and 86.48%, respectively. Population-wide MHC allele frequencies (A01:01, A02:01, A03:01, A24:02, A26:01, B07:02, B08:01, B27:05, B39:01, B40:01, B58:01, B15:01, HLA-DRB103:01, HLA-DRB107:01, HLA-DRB1*15:01) were employed to analyze the entire global population (Supplementary Fig. 1 and Table S2).

Analyses of physiochemical properties and solubility prediction of vaccine construct

Gasteiger's 2005 comprehensive study evaluated the physicochemical properties of a subunit vaccination protein by examining its theoretical isoelectric point (pI), amino acid composition, in vitro and in vivo half-lives, instability index, aliphatic index, molecular weight (MW), and grand average of hydropathicity (GRAVY). The complete physicochemical characteristics of the subunit vaccination protein were determined using the ExPASy ProtParam and are shown in Supplementary Table S3. The vaccine's essential characteristics is demonstrated by its theoretical pI of 9.28 and molecular weight of 32.54 kDa. It indicated stability with anticipated instability indices < 40 (Singh et al. 2020). A GRAVY score of -0.245 indicated that the protein was hydrophilic, whereas the chimeric construct's aliphatic index of 70.00 indicated thermal stability. Overall, it is anticipated that the developed vaccine will be hydrophilic, thermostable, and very basic (Singh et al. 2020; Rabienia et al. 2024). The half-life of the vaccine in vitro is more than 20 h in yeast and 30 h in human reticulocytes. We also used the Protein-Sol service at https://protein-sol.manchester.ac.uk/ to evaluate the solubility of the vaccine construct (Supplementary Table S3). This website uses prediction models based on isoelectric pH and amino acid compositions to estimate vaccine protein solubility. The physiological behavior of the vaccine protein is understood by this study, assisting in optimizing vaccination delivery and dispersion. The suggested vaccine architecture is extremely soluble in water, as evidenced by its 9.95 pI score and solubility score of 0.446.

Secondary structure prediction

The PSIPRED 4.0 web service predicted the multi-epitope subunit vaccine's secondary structure. This prediction helps assess the vaccine construct's stability and functionality by revealing its structural properties (Singh et al. 2020). Alpha-helix, beta-strand, and coil creation required 74, 53 & 190 amino acids. Thus, the construct has 23.34% alpha helix, 16.71% beta-strand, and 59.93% coil (Supplementary Fig. 2A).

Tertiary structure prediction and structure validation

The Rosetta server was used to construct the tertiary structure of the subunit vaccine protein, which was used for functional characterization (Supplementary Fig. 2B). The optimal model was chosen based on the overall quality of the multi-epitope vaccine's protein structure. According to the Ramachandran plot generated by the SAVES v6.0 server, the model exhibited 91.9% of residues in the most favoured regions, 5.3% in additional allowed regions, 1.8% in generously allowed regions, and 0.9% in disallowed regions (Supplementary Fig. 2D), respectively. The effectiveness of the modelled structure was also examined using the ERRAT. The three-dimensional model's overall quality factor was 92.67 (Supplementary Fig. 2C). More residues may have been found in acceptable side chain areas, as shown by the Verify 3D score of 83.09 (Supplementary Fig. 2E), which was also higher than average. The modelled structure was further validated and found to be of good quality and have a high stability level; it then underwent model refinement.

Energy minimization of the model and its validation

Energy minimization was performed on the specified model using GalaxyRefine (Fig. 1B). The anticipated tertiary structures were validated with the RAMPAGE server, which performed a Ramachandran plot analysis of the modelled vaccine construct. The results revealed that 94.1, 3.7, and 3.7% of the protein residues were in favoured, allowed, and outlier regions (Fig. 1C). The preferred region and GalaxyRefine-selected scores for refined models are practically identical. The ProSA-web and ERRAT servers determined the quality and probable faults. The selected refined model had an overall quality factor of 97.13 on the ERRAT server (Fig. 1D), The achieved results satisfy the goal of validation because the generated Z score of -4.06 is within the range of frequently occurring native proteins of comparable size (Fig. 1E).

Molecular docking of vaccine construct with TLR4 receptor

Understanding the molecular interactions between a vaccine construct and human immune receptors is critical for assessing its immunogenicity. TLR4, an essential immune receptor, recognizes foreign proteins and triggers inflammatory cytokine production, ultimately initiating the innate immune response against various pathogens (Kim et al. 2023; Faria et al. 2012). Molecular docking simulations were performed to investigate these interactions for the designed vaccine construct.

TLR-4 monomers and the improved vaccination design were also submitted to HADDOCK2.2. 8 structures were found in a total of 4 clusters as a result of the HADDOCK2.2 Vaccine and TLR 4 interaction, which accounted for 4% of the water-refined models. Model 1 in cluster 1 with the lowest Z-score was chosen (Fig. 2A). Hydrogen bond interactions between the vaccine and various TLRs were identified using PyMOL (Fig. 2B), while PDBsum provided detailed analysis of interface residues involved in hydrogen bonds, salt bridges, disulfide bonds, and non-bonded contacts (Fig. 3C). PDBsum data shows that TLR-4 produces 5 hydrogen bonds and 2 salt bridges among Asp 308, Glu 313, and Gly 312. Gln 316 & Lys 21, Gly 44, Ser 42, Arg 41 from Chain A (vaccine construct) & Chain B (TLR-4) respectively. Gibbs free energy, or G value, was obtained from the PRODIGY website to further support the vaccine's affinity for the TLR receptor (Table 5).

Fig. 2.

A Docking findings for the vaccine-TLR4 complex are shown. TLR4 is illustrated in blue, while the vaccine design is indicated in purple. B The vaccine-TLR4 complex interaction in three dimensions. C Total 2D interaction residues and links between TLR4 protein chains and the vaccination

Fig. 3.

A Vaccine-TLR4 complex (Black) only vaccine (Blue) & just TLR 4 (Brown) RMSD plots; B Radius of gyration analysis of Vaccine-TLR4 complex (Black) vaccine (Blue) & only TLR 4 (Brown); C Hydrogen bonds created inside the TLR-vaccine complexes. Using the MM-GBSA method, the binding free energy of Vaccine-TLR4 has been evaluated

Table 5.

Results for binding energetics. Predicted results of binding affinity (Gibbs free energy), dissociation constant, number of interfacial contacts per property, non-interacting surface per property by PRODIGY web server

(With 25degree)
Protein–protein complex	ΔG (kcal mol−1)	Kd (M) at ℃	ICs charged-charged	ICs charged-polar	ICs charged-apolar	ICs polar-polar	ICs polar-apolar	ICs apolar-apolar	NIS charged	NIS apolar
Cluster_5_1	− 7.6	2.7e-06	4	4	13	0	3	4	23.25	40.2

(With 37 degree)
Protein–protein complex	ΔG (kcal mol−1)	Kd (M) at ℃	ICs charged-charged	ICs charged-polar	ICs charged-apolar	ICs polar-polar	ICs polar-apolar	ICs apolar-apolar	NIS charged	NIS apolar
Cluster_5_1	− 7.6	4.5e-06	4	4	13	0	3	4	23.25	40.2

Normal mode analysis

Normal mode analysis (NMA) was performed to assess the molecular stability and functional movements of the construct-TLR4 complex (Supplementary Fig. 3). Deformability peaks indicate primary chain residue deformations, revealing "hinges/linkers" in high deformability areas (Supplementary Fig. 3A). The experimental B-factor plot (Supplementary Fig. 3D) shows the relationship between NMA mobility and the V1-TLR4 complex (Supplementary Fig. 3B), with average RMSD values representing the docked complex. The eigenvalue of 6.748725e-05 reflects motion stiffness (Supplementary Fig. 3C), and a negative correlation between variance and eigenvalue is observed. The covariance map indicates correlated (red), uncorrelated (white), and anti-correlated (blue) atomic motions in the complex (supplementary Fig. 3E). The elastic network map shows atom pairs connected by springs, with darker greys indicating rigid springs (Supplementary Fig. 3F). NMA analysis predicts a stable interaction between TLR4 and the vaccine construct.

Molecular dynamic simulation

The root mean square deviation (RMSD) of the vaccine construct and TLR4 receptor, in both bound and unbound states. The dynamic behavior and conformational stability from the initial to final structure were assessed through calculations and graphs based on the Cα atoms. Stable complex formation is indicated by minor deviations in the RMSD curve, whereas more significant deviations suggest instability. For the bound state, the average RMSD of > 1 nm indicates minor fluctuations throughout the simulation, with significant drifts observed towards the end between 70 to 100 ns, after which the system fully converged with a standard deviation of 0.59. Individual chains of the TLR4 receptor and vaccine construct exhibited mean RMSD values of < 1 nm and ~ 1 nm, respectively, with standard deviations of 0.43 and 0.65 (Fig. 3A). Molecular dynamics simulations revealed minimal conformational rearrangements within the vaccine construct upon TLR4 binding. This suggests that the majority of the vaccine structure remained stable throughout the simulation, potentially due to favourable interactions with the receptor that minimized strain-induced fluctuations.

The stability of the vaccine-TLR4 complex was further assessed using the radius of gyration (Rg) metric. Molecular dynamics (MD) simulations showed initial stability, which improved over time despite minor fluctuations towards the end (Fig. 3B). A detailed local analysis revealed a substantial number of hydrogen bonds and consistent total contact numbers between the vaccine and TLR4 throughout the simulation trajectory (Fig. 3C).

Binding energy calculations between the vaccine and TLR4 receptor were performed using MMPBSA and MMGBSA methods, considering electrostatic, van der Waals, solvent-accessible surface area, and polar solvation energies. The polar solvation energy was computed using the Poisson-Boltzmann equation for MMPBSA and the Generalized Born equation for MMGBSA. Free energy analysis revealed a strong binding affinity, with a binding energy of − 4650.569 kcal/mol (MMGBSA) (Fig. 3D and Table 6). Electrostatic and polar solvation energies were vital contributors to this interaction.

Table 6.

The evaluation of various MM/PBSA terms calculated using GMMBSA module

TLR4-Vaccine interaction (kcal/mol)
van der Waal energy	− 198.032 ± 198.025
Electrostatic energy	− 5345.184 ± 3285.448
Polar solvation energy	935.054 ± 915.248
SASA energy	− 42.407 ± 29.272
SAV energy	0
WCA energy	0
Binding energy	− 4650.569 ± 2597.497

In conclusion, MD simulations and free energy analyses confirm the stability of the vaccine-TLR4 complex and demonstrate the vaccine's high binding affinity for TLR4, highlighting its potential efficacy.

Immune response simulation

The immune simulation results significantly enhanced primary and secondary immune responses to the designed vaccine construct, and the results are shown in Supplementary Fig. 4A–F. The simulation, which involved three successive immunizations, showed a notable increase inIgG1 + , IgG2, IgM, and IgM + IgG antibodies, indicating a robust immune response. B-cell populations expanded with repeated exposure, establishing humoral immune memory. Concurrently, cytotoxic and helper T cell populations rose significantly, with a marked reduction in antigen levels during secondary and tertiary responses, indicating enhanced adaptive immunity (Dan et al. 2021). After each vaccination, the anticipated development of macrophages, dendritic cells, and natural killer cells was noted. Additionally, the suggested vaccine caused a notable release of cytokines that are essential for boosting immune responses against infections, including as IFN-gamma, IL-23, IL-10, and IFN-beta (Novelli and Casanova 2004). Over the course of the three vaccinations, repeated exposure to the antigen resulted in increased levels of TGF-B, IFN-gamma, and other interleukins, along with varied quantities of other cytokines. The vaccine's wide analytical influence was shown by the average measurement of the Simpson's Index (D) (Porter et al. 2022). According to the results of immunological simulation, the prioritized vaccine may stimulate T and B cells, resulting in the production of antibodies and the development of long-lasting memory cells after repeated exposure to the antigen. The ability of the vaccination to elicit strong innate and adaptive immune responses against human leishmaniasis is highlighted by this.

Codon adaptation & in-silico cloning

The final phase was cloning the vaccine nucleic acid sequence into an expression vector. The amino acid sequence was reverse-translated and codon-optimized for the E. coli K12 strain using the Java Codon Adaptation Tool (JCAT) service. The resulting cDNA sequence achieved a codon adaptation index (CAI) value of 1 and a GC percentage of 50.73%, within the ideal range of 0.8 to 1 for CAI and GC content (Elshafei et al. 2024). The vaccination sequence had HinDIII and BamHI sites added at the 5' and 3' ends to produce sticky ends after restriction digestion. The vaccine construct was cloned into the pET28a vector using Snap gene software. The circular plasmid pET28a's acquired length, including the insert, was 6206 base pairs (Fig. 4).

Fig. 4.

In silico restriction cloning of the vaccine design into the expression vector pET-28a ( +)

Discussions

Currently, there are no licensed vaccines for preventing leishmaniasis in humans, posing a significant global health challenge. Existing chemotherapy and treatments face issues such as toxicity and resistance (Dinc 2022). Vaccination is recognized as the most effective and economical approach to disease control. Several vaccine strategies, including DNA vaccines, subunit vaccines, and killed parasites, have demonstrated efficacy against L. donovani. However, developing standard vaccine formulations typically spans up to a decade (Soltan et Melby et al. 2001). Immunoinformatics-based vaccine design can accelerate this process by predicting protein epitopes and vaccine characteristics. These insights guide the development of multi-epitope vaccines, which offer advantages in enhancing innate, humoral, and cellular immune responses compared to single-epitope vaccines (García-Machorro et al. 2022).

This study used in-silico immunoinformatics to develop a multi-epitope vaccine against Leishmania. HTL, CTL, and B-cell epitopes from four selected L. donovani antigenic protein sequences were mapped. HTLs are crucial for inducing pro-inflammatory cytokines, stimulating antibody secretion by B cells, facilitating cell type transitions, and activating macrophages in various infections while CTLs play a pivotal role in generating secondary immune responses following low-dose parasite infections (Costa-da-Silva et al. 2022). Adoptive transfer of antigen-specific CTLs during chronic L. donovani infection resulted in a 90% reduction in splenic parasite load (Polley et al. 2006).

This study used in-silico immunoinformatics to develop a multi-epitope vaccine against Leishmania. HTL, CTL, and B-cell epitopes from four selected L. donovani antigenic protein sequences were mapped. HTLs are crucial for inducing pro-inflammatory cytokines, stimulating antibody secretion by B cells, facilitating cell type transitions, and activating macrophages in various infections while CTLs play a pivotal role in generating secondary immune responses following low-dose parasite infections (Costa-da-Silva et al. 2022). Adoptive transfer of antigen-specific CTLs during chronic L. donovani infection resulted in a 90% reduction in splenic parasite load (Polley et al. 2006).

Regarding B cells' role in intracellular parasite infections like Leishmania, there have been conflicting reports. However, studies indicate that IgG antibodies in L. major-infected C57BL/6 mice enhance antigen uptake by dendritic cells (Woelbing et al. 2006). Mitchell and Handman (1983) demonstrated that serum transfer from cured animals does not confer protection against Leishmania. In this study, continuous (linear) B-cell epitopes were selected on the besis of a threshold value of 0.5 identifying antigenic linear epitopes. These candidates were further evaluated for antigenicity using VaxiJen, allergenicity using AllerTOP, and toxicity using ToxinPred to ensure immunological relevance and safety.

Because of their high epitope density, potent binding affinity to host MHC molecules, capacity to trigger interferon production, and wide population coverage, prohibitin, STI1, and GP63 were chosen as antigenic components for the vaccine. These proteins play key roles in the biology of Leishmania and are well-characterized immunogens. For parasite life and cellular homeostasis, STI1, a co-chaperone of HSP90, is essential (Hombach et al. 2013). Prohibitin retains mitochondrial integrity and controls the growth of parasites; in animal models, its recombinant version (rPHB) has been demonstrated to elicit potent Th1 responses and significantly lower parasite burden (Dias et al. 2018). A prominent surface glycoprotein called GP63 impacts host immunological signaling and promotes macrophage invasion. In preclinical animals, vaccination trials employing GP63 in DNA and recombinant formats have shown complete protection against VL (Kaur et al. 2016; Gomez et al. 2009). The inclusion of these proteins in the rational construct of a multiepitope vaccine is supported by their shown immunogenicity and functional significance.

A 317-amino acid vaccine construct was designed by linking 5 HTL, 7 CTL, and 2 B-cell epitopes from target proteins using appropriate linkers and an adjuvant. BLASTP analysis revealed no significant homology with human proteins, indicating a low risk of autoimmune cross-reactivity and supporting the safety and specificity of the vaccine peptides. Physicochemical analysis using the ProtParam server indicated a low instability index (Kaur et al. 2020), confirming the vaccine's stability. A higher GRAVY score also suggested hydrophobicity, complemented by a high aliphatic index, highlighting the vaccine's resilience to high temperatures (Droppa-Almeidet al. 2018). AllerTOP analysis predicted the vaccine to be non-allergenic.The selected T-cell epitopes covered the entire global population, indicating broad potential for protective immunity against leishmaniasis worldwide. While efficacy against Leishmania major and Leishmania braziliensis may vary due to genetic diversity among species, this is less critical given the vaccine's primary focus on combating visceral leishmaniasis caused by L. donovani. Importantly, several of the selected epitopes particularly from GP63 and Prohibitin exhibit high sequence conservation across Leishmania species, including L. major, L. infantum, and L. braziliensis his cross-species conservation is critical for developing a broadly protective vaccine. Prior studies have demonstrated that GP63 orthologs maintain over 80–90% identity across various Leishmania strains, while Prohibitin is a highly conserved mitochondrial protein involved in vital cellular processes such conservation increases the likelihood of immunological cross-reactivity and potential vaccine efficacy in regions where multiple species co-circulate. Although this study was focused on L. donovani, the inclusion of conserved epitopes suggests the potential for broader applicability pending further cross-species validation.

The vaccine construct was modelled using Rosetta and refined via GalaxyRefine, followed by rigorous validation using PROCHECK Ramachandran plot analysis and ProSA Z-Score assessment. Analysis of the Ramachandran Plot revealed almost 94% of residues in favoured regions and only 0.3% in disallowed regions, indicating a high-quality model structure. A Z-Score of -4.06, falling within the typical range for experimentally validated proteins of similar size, further validated the model's reliability. To investigate the interactions and stability of vaccine-TLR complexes, we performed docking between the vaccine model and TLR construct. Detailed analysis of the docked complex revealed that TLR-4 forms 5 hydrogen bonds and 2 salt bridge. The PRODIGY server calculated a negative G value of -10.2 kcal mol-1 for the docked complex, confirming the feasibility of this interaction. The vaccination-TLR-4 complex's stability and binding strength were evaluated using molecular dynamics (MD) simulations. The simulations showed that stability was quickly attained while retaining constant hydrogen bonds and total contacts.

Although the primary function is to recognize lipopolysaccharide (LPS) from Gram-negative bacteria, TLR-4 may also interact with a variety of non-LPS protein ligands, including viral envelope glycoproteins, HMGB1, and heat shock proteins (HSP60, HSP70). This demonstrates its wider function in innate immune recognition. Controlling Leishmania donovani infection requires Th1-biased immune responses, which are triggered by TLR-4. Again, study has demonstrated that animals lacking TLR-4 had increased parasite loads and reduced cytokine production, highlighting the protective function of the receptor. Targeting TLR-4 was therefore an intentional choice because of its shown function in triggering Th1-skewed immune responses, which are essential for the removal of Leishmania. Additionally, our molecular docking, normal mode analysis, and molecular dynamics simulations indicate the ability of the proposed vaccine to successfully engage and activate TLR-4 by confirming a stable and energetically advantageous contact between the two receptors.

These results were verified by free energy calculations, which revealed that the vaccination and TLR-4 had favorable binding energies. Overall, our docking results were validated by MD simulations, demonstrating the vaccine's strong and consistent affinity for the TLR-4 receptor. Strong humoral and cell-mediated immunity against infections is the ultimate objective of a successful vaccine (Atapour et al. 2021; Ikeogu et al. 2020). Our vaccine design demonstrated the ability to elicit robust T-cell and B-cell responses in our in-silico experiments.

Both humoral and cellular immune responses must work in tandem to protect against visceral leishmaniasis (VL). Effective parasite removal depends on the cellular immune response, which is mostly mediated by CD4 + Th1 cells and CD8 + cytotoxic T lymphocytes. Interleukin-12 (IL-12) and interferon-gamma (IFN-γ) are two cytokines secreted by Th1 cells that stimulate macrophages to eliminate intracellular Leishmania parasites. By directly lysing infected host cells and generating more IFN-γ, CD8 + T cells help. On the other hand, the humoral response, which is defined by B cells producing certain antibodies, helps neutralize extracellular parasites and opsonize Leishmania for improved phagocytosis. Robust and sustained cellular immunity is thought to be necessary for long-term protection and disease resolution in VL, even if antibody responses are crucial for preventing the spread of parasites. Thus, our vaccine construct's capacity to elicit robust humoral and cellular responses is a major benefit, raising the possibility of thorough and long-lasting protection against L. donovani infection.

Significant increases in CTL and activated HTL proliferation were seen following first dosages, suggesting successful activation of cell-mediated immunity. Following first injection, B-cell responses displayed comparable patterns and increased IgM and IgG levels. B-cell responses showed similar trends and elevated IgM and IgG levels post-initial administration. The vaccine also induced inflammation, evidenced by increased IFN-γ levels and minimal IL-10 production. Analysis revealed a diverse B and T-cell response with negligible Simpson index. Increased memory B and T cells suggest readiness to combat future infections. Lastly, the vaccine construct was optimized for high-level expression in E. coli, achieving a CAI of 1 and a GC% of 52, ideal for efficient protein production in the host system.

Additionally, secondary structure analysis using RNAfold confirmed the absence of inhibitory structures near the 5′ untranslated region and start codon, which supports efficient ribosome binding and translation initiation (Supplementary Fig. 5A and BS). Although minor stem-loop formations were present in internal regions, they were not predicted to interfere with transcript stability or elongation. These results validate the translational feasibility of the codon-optimized construct and further support its utility for recombinant expression in E. coli.

While this study employed a robust computational pipeline to design and evaluate a multi-epitope vaccine candidate against Leishmania donovani, experimental validation is critical to translating these in silico findings into practical immunoprophylactic solutions. In vitro experiments, such as enzyme-linked immunosorbent assay (ELISA), can be employed to confirm the antigenicity and immunoreactivity of individual epitopes or the full construct using sera from infected or vaccinated animals. Synthetic peptides representing predicted epitopes can be tested for their ability to stimulate cytokine release (e.g., IFN-γ, IL-12) in human peripheral blood mononuclear cells (PBMCs) or splenocyte cultures from mice. Additionally, in vivo studies using BALB/c or C57BL/6 mice infected with L. donovani can be used to evaluate the protective efficacy of the vaccine. Parameters such as parasite load in spleen and liver, antibody titers, cytokine profiling, and T-cell proliferation can help assess immune protection. Recombinant protein expression followed by immunization in animal models would allow validation of predicted immunogenicity and the potential to confer protection, ultimately bridging the gap between computational design and clinical application.

Conclusion

The suggested multi-epitope vaccine model shows promise for future success. It is antigenic and can induce an immunological response, remarkably increasing IFN-γ production. Structural examination confirms the validity and stability of its 3D model. The vaccine model demonstrates a strong affinity for the TLR-4 receptor, suggesting activation of an inflammation-promoting signalling pathway. T-cell epitopes cover 100% of the global population, indicating potential efficacy worldwide. Epitopes are relatively conserved in L. infantum and moderately conserved in selected strains of L. donovani, highlighting broad applicability. Simulation tests affirm the vaccine's strong binding to TLR-4, supporting its efficacy in combating leishmaniasis. These findings pave the way for further in vitro and in vivo studies to validate its antigenicity and protective efficacy as a potential chimeric vaccine candidate against leishmaniasis.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

SS, SL, SG, GK and DK carried out the experiment. SS and DK wrote the manuscript. SS, SL, SG, GK and DK contributed to the analysis of the results. DK supervised the project and conceived the original idea.

Data availability

All data generated or analysed during this study are included in this article [and its supplementary information files].

Declarations

Conflict of interest

The authors declare that no conflict of interest exists.

Ethical approval

This declaration is not applicable.