Designing and comparative analysis of anti-oxidant and heat shock proteins based multi-epitopic filarial vaccines

Abstract

Background

Lymphatic Filariasis (LF) is a neglected tropical disease affecting more than 882 million people in 44 countries of the world. A multi-epitope prophylactic/therapeutic vaccination targeting filarial defense proteins would be invaluable to achieve the current LF elimination goal.

Method

Two groups of proteins, namely Anti-oxidant (AO) and Heat shock proteins (HSPs), have been implicated in the effective survival of the filarial parasites in their hosts. Several B-cell, CTL, and T-helper epitopes were predicted from the three anti-oxidant proteins GST, GPx, and SOD. Likewise, epitopes were also predicted for HSP110, HSP90, and HSP70. Among the predicted epitopes, screening was applied to include only non-allergenic, non-toxic epitopes to construct two MEVs, PV_AO and PV_HSP. The epitopes for each group of proteins were connected to each other by the inclusion of suitable linkers and an adjuvant. The 3D models for PV_AO and PV_HSP were predicted, and validated, followed by prediction of physicochemical properties using bioinformatics tools. The binding free energy of PV_AO and PV_HSP with Toll like Receptors (TLR) TLR1/2, TLR4, TLR5, TLR6, and TLR9 was calculated with HawkDock. The immunogenicity of both the MEVs were assessed by Immune simulation after which codon adaptation and in-silico cloning were carried out.

Results

Conservation of the selected AOs and HSPs in other parasitic nematode species suggested that both the generated chimera could be helpful in cross-protection too. The 3D models of both MEVs contained more than 97% residues in allowed regions, as predicted by PROCHECK server. High MMGBSA and docking scores were obtained between MEVs and TLR4, TLR1/2, TLR6, and TLR9. Molecular dynamics simulation confirmed the stability of candidate vaccines in dynamic conditions present in the biological systems. The in-silico immune simulation indicated significantly high levels of IgG₁, T-helper, T-cytotoxic cells, INF-γ, and IL-2 responses following immunization with PV_AO and PV_HSP.

Conclusion

The immunoinformatics approaches used in this study confirmed that, the designed vaccines are capable of eliciting sustained immunity against LF, however, additional in-vivo studies would be required to confirm their efficacy. Furthermore, by employing multi-epitope structures and constructing two different cocktail vaccines for LF, this study can form an important milestone in the development of future LF vaccine/s.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12879-024-10272-9.

Background

Lymphatic filariasis (LF) is a neglected tropical disease and is a major barrier to socioeconomic development in endemic nations [1]. Filarial parasites Wuchereria bancrofti, Brugia malayi, and Brugia timori are the causative agents of Lymphatic filariasis and are transmitted by mosquito vectors of the genus Anopheles, Culex, and Aedes [2]. W. bancrofti is responsible for around 90% of filarial parasite infections, followed by B. malayi and B. timori (https://www.who.int/news-room/fact-sheets/detail/lymphatic-filariasis). Lymphatic filariasis is common in ~ 44 countries and affects around 882 million individuals [3, 4]. According to World Health Organization (WHO), LF alone accounts for approximately 2.8 million disability-adjusted life-years, making it the second most frequent cause of long-term disability after mental illness [5]. WHO has approved a combination of Ivermectin (IVM), Diethylcarbamazine (DEC), and Albendazole for Mass drug administration (MDA) for LF treatment and control. This drug combination has the ability to kill microfilariae and late embryonic stages but has little effect on adult filarial worms, because of which the adult worms remain alive for several years with in the human host [6].

In chronic patients, LF infections culminate into hydrocele, lymphedema, and ultimately elephantiasis [7]. The WHO launched the Global Program to Eliminate Lymphatic Filariasis (GPELF) in 2000, which recommended MDA to the entire population at risk of LF. Though the MDA program has considerably decreased global LF transmission, there are still questions about its long-term benefits and prevention of subsequent infections. In addition, the development of resistance to the MDA drugs is an upcoming challenge for LF elimination [8]. As a result, a more justifiable modus operandi would be, the development of an LF vaccine to prevent the occurrence or relapse of the disease [9]. No licensed LF vaccines are available till date, although several studies on vaccine candidates like BmHSP and BmALT-1 etc., have been conducted in the past [10, 11]. Recently, multi-epitope vaccines (MEVs) have emerged as a novel approach to vaccine development since they can circumvent the impact of damaging epitopes of the intact antigen [12]. Epitope-based vaccines offer additional advantages of safety, stability, and can effectively address the issues associated with conventional vaccines such as poor safety record, disease outbreak due to immunization etc [13]. The approach has already been used to design vaccines against SARS CoV-2, Mycobacterium tuberculosis [14], Schistosoma mansoni [15], and Onchocerca volvulus [16]. Earlier studies are indicative of protective immunity in LF, which is evidenced by the presence of endemic normal people living in endemic regions [17]. Cross-reacting antibodies and high levels of pro-inflammatory cytokines are both associated with potential immunity against the filarial parasites in people living in endemic areas. Therefore, developing a vaccine might be an effective way to combat the pathophysiology and to prevent new LF infections [18].

The filarial worms have highly developed enzymatic and non-enzymatic antioxidants for reducing the harmful effects of pro-oxidants, which are reactive oxygen and nitrogen species produced by the host immune cells. The filarial nematodes antioxidant system comprises of enzymes like Glutathione-S-Transferase (GST), Superoxide Dismutase (SOD), and Glutathione Peroxidase (GPx) which neutralize the hydrogen peroxide, superoxide radicals, hydroxyl ions, and other pro-oxidants generated by the hosts’ immune cells [19]. Thus, current antifilarial studies employ targeted chemo- and immunotherapeutic strategies to circumvent the filarial redox regulation system’s defenses [20]. Notably, an earlier vaccine targeting the filarial Glutathione redox system has been shown to reduce worm burden in test animals [21]. Since, the parasites’ antioxidant machinery is crucial for the survival therefore the anti-oxidant enzymes may be considered as promising vaccine candidates. Another, highly conserved class of proteins known as heat shock proteins (HSP) are well-known for their functions as molecular chaperones [22, 23]. The expression of HSPs, for either refolding damaged proteins or selecting them for degradation, is often upregulated in response to a rise in temperature in the majority of cells. HSPs expression increases quickly in response to environmental changes in temperature, pH, and pressure [24]. The filarial larvae (microfilariae L1 stage), in particular, must adapt to a substantial decrease in temperature during transmission from the homeoiotherm mammalian host to the poikilotherm arthropod vector and vice versa as different classes of HSPs are expressed when L3 is transferred from the vector to the mammalian host [25]. Therefore, in this work, the epitopic region of the crucial anti-oxidant enzymes and antigenic Heat shock proteins were predicted and two MEVs PV_AO and PV_HSP were constructed using in-silico approaches [26]. The constituent proteins were assessed for amino acid conservation with homologous proteins in other related parasites and humans to address the cross-protection potential of the developed vaccine candidates.

Methods

Retrieval of protein sequences and preliminary characterization

The protein sequences for W. bancrofti GST (A0A0H5S7P0, 208 aa) [27] SOD (A0A0K0IYY8, 223 aa), GPx (J9E3A6, 220 aa), HSP70 (EJW86287.1, 645 aa) [28], HSP 90 (EJW88125.1, 717 aa), and HSP110 (A0A0K0IYH1, 817 aa) were downloaded in FASTA format from the UniProt (https://www.uniprot.org/) and GenBank database (https://www.ncbi.nlm.nih.gov/genbank/). To discriminate secretory proteins from non-secretory, SignalP6 server (https://services.healthtech.dtu.dk/services/SignalP-6.0/) was used [29]. The DeepLoc 1.0 (https://services.healthtech.dtu.dk/services/DeepLoc-1.0/) server was used to predict the subcellular and cellular membrane localization of the chosen PV_AO and PV_HSP proteins. The DeepLoc 1.0 server uses neural networks trained on UniProt proteins combined with experimental evidences of subcellular localization [30].

Protein conservation analysis in related nematodes

A BLASTp search was conducted on the UniProt database to obtain the percentage identities of homologous relatives of the selected protein. The potential for selected proteins to provide cross-protection against additional filarial infections was examined. After performing consecutive sequence alignment of homologous proteins in related filarial worms, the protein conservation was predicted. O. volvulus, L. loa, O. ochengi, B. malayi, L. sigmodontis and W. bancrofti filarial parasites was examined and epitopes were chosen based on their significance for human and animal health.

Prediction of linear B-cell epitopes

The protein sequences in FASTA format were used for prediction of B-cell epitopes. ABCpred (https://webs.iiitd.edu.in/raghava/abcpred/ABC_submission.html) and B-cell tools of the Immune Epitope Database (IEDB) (http://tools.iedb.org/bcell/) servers were used [31]. ABCpred can predict linear B-cell epitopes using artificial neural network and the server has been trained using BciPep database and has better performed for B-cell epitope prediction. The IEDB server predicts regions that are most likely to be recognized as epitopes in a typical B-cell response. The antigen’s sequence properties and Hidden Markov Models (HMMs) were used to predict the linear B cell epitopes. Since the use of multiple servers to confirm the predicted epitopes is recommended, hence the epitopes that were predicted both by ABCpred and IEDB servers only were chosen for vaccine constructions.

Cytotoxic T lymphocytes (CTL) and helper T-cell epitope prediction

For the chosen proteins, cytotoxic-T-lymphocyte (CTL) epitopes were predicted using the freely accessible NetCTL 1.2 server (MHC supertype A1) (http://www.cbs.dtu.dk/services/NetCTL/). The technique combines the prediction of TAP transport efficiency, proteasomal C terminal cleavage, and peptide MHC class I binding [32]. Additionally, a weight matrix was used to predict TAP effectiveness. The NetMHCII 2.3 server (http://www.cbs.dtu.dk/services/NetMHCII/) has been used to predict HTL epitopes with a length of 9-mer for human alleles. The NetMHCII 2.3 server predicts peptide binding to the HLA-DR, HLA-DQ, and HLA-DP alleles [33]. MHC II epitopes were predicted based on receptor affinity and percentile scores assigned to each anticipated epitope [34]. The percentile rank is directly related to IC₅₀ and inversely related to epitope affinity. The predicted CTL and HTL epitopes were reassessed for antigenicity by VaxiJen server. For the estimation of CTL and HTL epitopes, the threshold value was set at 0.5 following earlier researches in the field [35].

Designing of vaccine candidates

The candidate vaccine sequence was predicted by using a combination of high scoring CTLs, high affinity HTLs, and B-cell epitopes predicted by BepiPred 3.0.(https://services.healthtech.dtu.dk/services/BepiPred-3.0/) Some predicted sequences had the capacity to function as both B-cell and T-cell epitopes [36]. Using linkers GPGPG and AAY, various epitopes were connected and to increase the vaccine’s immunogenicity, a 50 S ribosomal protein L7/L12 (Locus RL7 MYCTU) adjuvant, agonist of TLR4 (Accession number P9WHE3) was added with an EAAAK linker at the N-terminal of the constructed vaccines. UniProt database was used to obtain the adjuvant’s sequence.

IFN-γ inducing epitope scanning

Interferon gamma (IFN-γ) is a cytokine that effectively responds to MHC antigens, stimulates macrophages and natural killer cells, and participates in both innate and adaptive immune responses. IFN-γ epitopes were predicted for the selected antioxidant and heat shock proteins using the IFNepitope server (http://crdd.osdd.net/raghava/ifnepitope/scan.php). The predicted method was based on a hybrid approach using the Support Vector Machine (SVM) and a motif.

Evaluation of toxicity, allergenicity, Antigenicity, and physicochemical properties of the developed vaccine candidates

The toxicity of all the epitopes were analysed by ToxinPred server (https://webs.iiitd.edu.in/raghava/toxinpred/multi_submit.php). The VaxiJen v2.0 server (https://www.ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen.html) was used for the analysis of antigenicity of the whole vaccine and the ANTIGENpro server (https://scratch.proteomics.ics.uci.edu/) was used for the validation. The allergenicity of the vaccine candidates was evaluated with AllergenFP v.1.0 (http://ddg-pharmfac.net/AllergenFP/) [37] and AllerTOP v.2.0 (https://www.ddg-pharmfac.net/AllerTOP/) [38]. The solubility of the vaccine candidates were predicted against the average solubility of the Escherichia coli protein in the ProteinSol server (https://protein-sol.manchester.ac.uk/) [39]. The ExPASy-ProtParam (http://web.expasy.org/protparam/) was used for the evaluation of the physicochemical properties of the vaccine candidates [40].

Secondary and tertiary structure construction and refinement

Secondary structure of the vaccine candidates was predicted from PSIPRED 4.0 server (http://bioinf.cu.ucl.ac.uk.psipred/) [41]. For structural determination of candidate vaccines, the tertiary structures were constructed with I-TASSER server (https://zhanggroup.org/I-TASSER/) [42]. Out of the five models provided by the I-TASSER server, the model having best RMSD and TM scores was selected. The 3D model of the multi-epitope vaccine peptide was improved by a three-step method in the ModRefiner (https://zhanglab.ccmb.med.umich.edu/ModRefiner/) server. In the ModRefiner server firstly, any missing residues were added, secondly, mild relaxation was performed and lastly, a more intense relaxation was carried out for the construction and refinement of protein structure.

Tertiary structure validation

A critical step in the model-building process is detecting any probable errors in the predicted 3-D models. For a single input structure, ProSA (https://prosa.services.came.sbg.ac.at/prosa.php) determines an overall quality score; this score is then displayed alongside all previously identified protein structures [43]. The structure probably contains mistakes if the estimated score is outside the range that is typical of native proteins then the 3D molecular viewer will highlight and display such structural flaws. In addition, non-bonded atom-atom interactions were examined using the ERRAT site (http://services.mbi.ucla.edu/ERRAT/) which compares the predicted structures to trustworthy high-resolution crystallographic structures. A Ramachandran plot, which is based on the Van der Waal radius of the side chains of amino acids, was obtained using the PROCHECK server (https://saves.mbi.ucla.edu/results?job=17609&p=procheck). The Ramachandran plot, is a method for visualizing the energetically permissible and disallowed dihedral angles psi (ψ) and phi (ϕ) of an amino acid. The PROCHECK server was employed to verify protein structures using the Ramachandran plot separating the Glycine and Proline residues [44].

Prediction of discontinuous B-cell epitopes, promiscuous MHC class-I epitopes and IFN-γ inducing epitopes in candidate vaccines

Both the candidate vaccines PV_AO and PV_HSP were submitted in FASTA format to the IEDB server for prediction of discontinuous B-cell epitopes in the final vaccine constructs. Promiscuous MHC class-I epitopes were predicted by the ProPred1 server (http://crdd.osdd.net/raghava/propred1/) for both PV_AO and PV_HSP. A quantitative matrix for forecasting promiscuous MHC class-I binding peptides is used by the ProPred1 server. The ProPred1 achieves a high-level accuracy of 75% at the preset threshold of 4%. IFN-γ epitopes were predicted for the candidate vaccine peptide with the IFNepitope server (http://crdd.osdd.net/raghava/ifnepitope/scan.php).

Molecular docking analysis and interaction studies

Molecular docking was performed for the analysis of the binding affinity between the structurally refined vaccine constructs and the human TLRs as well as MHCI and MHCII. TLR1/2 (PDB ID-2Z7X) TLR4 (PDB ID-3FXI), TLR5 (PDB ID-3J0A), TLR6 (PDB ID-4OM7) TLR9 (PDB ID-5GMH) MHC-I (PDB ID-7RK7) and MHC-II (PDB ID-8JRK) were retrieved from the RCSB protein databank (https://www.rcsb.org/) and were used for docking. The molecular docking process was rendered by PatchDock (https://bioinfo3d.cs.tau.ac.il/PatchDock/php.php) and HawkDock (http://cadd.zju.edu.cn/hawkdock/) servers based on protein-protein rigid-body docking. Subsequently, lowest energy-weighted scores of the various docked complexes were obtained to identify the best ligand-receptor interaction from a group of ligand-receptor complexes. The final selected docked complex interactions were visualized by BIOVIA Discovery Studio 3.0 software and PDBsum (https://www.ebi.ac.uk/thornton-srv/databases/pdbsum/).

Calculation of free binding energies (MM/GBSA)

Binding free energy was calculated to confirm the docked data’s reliability and the bound molecules’ stability. This study utilized the HawkDock server to assess binding free energy via the Molecular Mechanics/Generalized Born Surface Area (MM/GBSA) method (http://cadd.zju.edu.cn/hawkdock/#MMGBSA_submit) [45] for the docked complexes of PV_AO and PV_HSP with MHC-I, MHC-II, TLR1/2, TLR4, TLR5, TLR6, and TLR9.

Disulfide engineering

To ensure the vaccination model’s stability, disulfide engineering was performed on the Design 2.0 webserver (http://cptweb.cpt.wayne.edu/DbD2/). The parameters χ³ angle were set for − 87° and + 97° and the Cα-Cβ-Sγ angle to 114.6° [46]. The introduction of novel disulfide bonds into proteins has been extensively utilized to improve protein stability, modify functional characteristics, and enable the study of protein dynamics.

Normal mode analysis

The conformational stability of the selected docked complex was analyzed through iMODS (https://imods.iqfr.csic.es/) web server [47]. NMA mobility is represented with arrows, which indicates motion dynamics. Deformability plots show the parts of the structures that are not rigid, whereas eigenvalues show the level of rigidity of these structures. The B-factor, variance, covariance map, and linking matrix of the desired structure are also provided by this server.

Molecular Dynamics Simulation of MEVs

The online server SiBioLead for Macromolecular Simulations (https://sibiolead.com/MDSIM) was used to perform molecular dynamics (MD) simulations of the protein-water complexes. The SiBioLead server force field was chosen to be the OPLS/AA. The other input parameters included the water model SPC, the box type Triclinic, the salt type NaCl, and the options to neutralize and add 0.15 M NaCl salt. The 5,000-step integrator of the steepest descent was chosen using energy minimization parameters. Equilibration and MD run parameters were chosen as NVT/NPT equilibration form, 300 K temperature, 1 (one) bar pressure, leap-frog MD integrator, 100 ns simulation period, and roughly 5,000 frames per simulation. The simulation output has been evaluated based on the average number of H-bonds in each frame over time, the root mean square deviation (RMSD) of the given structure over time, the root mean square fluctuation (RMSF) of each residue in the given structure, the radius of gyration (Rg) or structural compactness, the solvent-accessible surface area (SASA), and Gromacs energy analysis [48].

Java Codon Adaptation and In-silico cloning

The multi-epitope vaccine constructs were expressed using reverse translation and codon optimization using the Java Codon Adaptation Tool (JCat) server (http://www.prodoric.de/JCat). Codon optimization was performed to express the MEV vaccine constructs in the E. coli (strain K12) host, since the codon usage of E. coli differs from that of the native host filarial parasites [49]. Three additional parameters were introduced to escape the prokaryotic ribosome binding site, restriction enzyme cleavage sites, and rho-independent transcription termination [50]. To estimate the expression levels of proteins, the Codon Adaptation Index (CAI) and percentage GC content are included in the JCat output. CAI offers insights into codon use biases. A score of 1.0 is considered optimal, but a value of > 0.8 is considered good [51]. The GC content of a sequence should fall within a range of 30–70%, as values outside this range may negatively impact the transcriptional and translational efficiency [52]. In order to clone the optimized gene sequence of the final vaccine constructs in the E. coli pET-30a (+) vector, NdeI, and XhoI restriction sites were introduced to the N and C-terminals of the candidate vaccine sequences, respectively. Finally, this sequence was inserted into the pET-30a (+) vector with SnapGene 6.1.2 software for in-silico cloning.

Immune simulation

The C-ImmSim server (http://150.146.2.1/C-IMMSIM/index.php) was used to simulate an immunological response in a computational environment [53]. The agent-based model C-ImmSim predicts immunological interactions by using machine learning approaches and a position-specific scoring matrix (PSSM) for immune epitope prediction. Simultaneous simulation of three compartments defines three distinct anatomical regions in mammals: (i) the bone marrow, where new lymphoid and myeloid cells are produced by hematopoietic stem cells; (ii) the thymus, where naive T-cells are selected to prevent autoimmunity; and (iii) a tertiary lymphatic organ, such as a lymph node. The time intervals were set at 1, 84, and 168, where each interval represents 8 h. The first interval corresponds to the injection at time = 0. The remaining simulation settings were set to their default values [54] and three shots were administered, with a time gap of 28 days. To evaluate clonal selection, 12 doses of the injections were administered four weeks apart to mimic frequent antigen exposure typical of an endemic area. The plot also allows for the deduction of Simpson index, D, which is a measure of diversity.

Results

Selection of proteins and phylogenetic analysis

For construction of MEVs, the adult filarial antigens were selected either from W. bancrofti and B. malayi databases (Table 1). The antigens GST and GPx belonging to the family of anti-oxidant proteins are expressed in all life stages of the filarial parasites thus are crucial for the survival [55, 56]. The antigens SOD [57], HSP70 [58], HSP90 [59] and HSP110 were chosen based on their reactivities in immunological studies [60]. The homology analysis of AO and HSP proteins was confirmed by performing a BLASTp for proteins against the UniProt and GenBank database (Table 1). The cellular localization of the proteins was predicted by DeepLoc to be located either in the Cytoplasm/soluble (A0A0H5S7P0, J9E3A6, A0A0K0IYY8, EJW86287.1, EJW88125.1, A0A0K0IYH1), Lysosomes (EJW86287.1, EJW88125.1, A0A0K0IYH1), Cell membrane (A0A0K0IYY8, J9E3A6), or Nucleus (A0A0H5S7P0, EJW88125.1, A0A0K0IYH1) (Supplementary Table 1). The signal peptide was checked in all the protein sequences using SignalP 6.0 server and none of the sequences contained signal peptides. Antigenic properties of antioxidant and heat shock proteins chosen for construction of multi-epitopic vaccines were predicted by the ANTIGENpro server. The highest antigenicity was shown by HSP110 (0.915) and the lowest antigenicity by GST (0.517). The BLASTp was performed against Homo sapiens proteins to assess the safety profile of the proposed vaccine candidates. The vaccine constructs exhibited no similarity with any human protein. This shows that vaccine constructs may be entirely safe for usage, with no risk of an autoimmune response post-vaccination (Supplementary Table 2).

Table 1.

Similarity of filarial proteins with Homo sapiens

Protein	Percent identity
	W. bancrofti	B. malayi	Loa-loa	O. volvulus	O. ochengi	Homo sapiens
Superoxide dismutase	95.19	100	91.40	86.10	86.55	6.20
Glutathione peroxidase	100	95.68	94.37	NA	88.12	11.2
Glutathione-S-transferase	99.04	100	84.26	79.81	80.29	8.0
HSP110 (VDM08680.1)	97.31	100	88.39	NA	88.74	9.0
HSP90 (EJW88125.1)	100	99.58	NA	NA	NA	4.5
HSP70 (EJW86287.1)	100	98.08	85.28	81.25	81.73	11.5

B-cell, CTL and HTL epitope prediction

In the next step, linear B-cell and T-cell epitope prediction was applied to the functional sequences of the proteins. Linear B-cell epitopes of PV_AO having 7 to 20 amino acids residues were predicted with two different servers i.e. ABCpred and IDEB server. After prediction, only those epitopes were chosen which were jointly predicted by both servers (Table 2). The linear B-cell epitopes, with antigen scores higher than 0.75, in ABCpred analysis were included in the vaccine construction. Out of the three antioxidant proteins, a total of 15 CTLs (MHC supertype A1) were predicted using NetCTL1.2 server, out of which 9 were selected based on high antigenicity scores (> 0.75) as given by ANTIGENpro server. Amongst the three HSPs sequences, a total of 15 CTLs were predicted and all the 15 CTLs were included for PV_HSP construction (Table 3). MHC-II epitopes having high binding affinity with HLA DR, DQ and DP (IC₅₀ scores) were selected for construction of MEVs. All the highly antigenic (VaxiJen score > 0.5) epitopes as predicted by NetMHCII 2.3 server were included in the construction of PV_AO, and PV_HSP vaccine constructs (Table 4).

Table 2.

Selection of B-cell epitopes and MHC-I epitopes for PV_AO construction (VaxiJen score > 0.50)

S.N.	B-cell Epitope Sequence	Start point	Length	MHC-I Epitope Sequence	VaxiJen Score	MHC-I binding Allele
1	LRGDAGVSG	8	9	VAEVYINSY (2.2877)	1.2984	HLA-A1
						HLA-B62
2	IIYFQQGSGGSITTI	17	15	DAGVSGIIY (0.9587)	0.7657	HLA-A1
3	SGSVSGLT	32	8
4	PGLHGFH	40	7
5	VHQYGDQT	47	8
6	NGCTSAGDHY	55	10
7	NPFGKTHGG	65	9
8	PNDRIKHIGDLGNIVAGANG	74	20
9	DIKLRGPLSVIGHSLVVHA	103	19
10	ANTDDLGQG	121	20
11	TGNMREESLKTGNAGSRLAC	130	8
12	YEFTVKDI	6		MTAIKADDY (3.2387)	1.1358	HLA-A1
13	NGVDVSLEKYRGHVCLIVNV	14	20	LTNSNYTEL (0.8885)	1.0250	HLA-A1
14	CKSGATDKNYRQLQEMHTRL	35	20	GKDVSLEKY (0.8132)	0.5930	HLA-A*0201
15	NQFGGQEPWAEAEI	67	14
16	KKFVTEKYGVQFDMFSKIK	81	19
17	VNGSDADDLYK	100	11
18	FLKSRQHGTLTNNIKWNFSK	111	20
19	QPVKRYSPTTAPYDIEGDIM	138	20
20	PIRLVLVDQGIKFTD	15	15	YTKMIYQAY(2.7894)	1.5349	HLA-A1
						HLA-A26
						HLA-B8
						HLA-62
21	DRINASDWPSMKSHFHFGQ	30	19	HTKYTKMIY (2.2179)	1.5031	HLA-A1
						HLA-A26
						HLA-B62
22	LPCLYDGDHQIVQSG	49	15	FTDDRINAS (1.2726)	1.3981	HLA-A1
23	AILRHLA	64	7	AYDTEKDSY (0.9831)	1.0288	HLA-A1
24	RKHNLNGGNELETTHID	71	17
25	MFCEGVRDLHTKYTKMIYQA	88	20
26	ILPVELAKFE	119	10
27	KLLATRD	129	20
28	DGKNFILGEKISYVDFVLFE	136	7
29	CLDKFPLLKAYHQRMEDRPG	167	20

Table 3.

Selection of B-cell epitopes and MHC-I epitopes for PV_HSP constructions (VaxiJen score > 0.50)

S.N.	B-cell Epitope sequence	Start point	Length	MHC-I Epitope sequence	VaxiJen Score	MHC-I Binding Allele
1	AFTTRGRSMGVAARQQLNTN	40	20	FSKLFYDTY (2.3138)	1.6016	HLA-A1
						HLA-A26
2	RVRILLTNLLRESGKKAEEV	318	20	KLTVLGATY (0.8951)	1.4512	HLA-A1
						HLA-A3
						HLA-B58
						HLA-B62
3	AAMKCAILSPAFKVRDFSVK	378	20	NTNIKNTII (0.7662)	1.4508	HLA-A1
4	ETDDTKG	520		MSEEMNGET(0.9951)	2.1441	HLA-A1
5	NQEKEENTNAPSNDVVLP	527	7	QLEFRALLF (0.9500)	2.0230	HLA-A1
						HLA-B62
6	PPKAEPAKEPEKGSDGQKGE	775	18	MSLIINTFY(0.9468)	1.9252	HKA-A1
						HLA-A3
						HLA-B58
						HLA-B62
7	DDEAEEE	222	7	LMSLIINTF (0.7788)	1.6891	HLA-A1
						HLA-A24
						HLA-B8
						HLA-B58
						HLA-B62
8	KKDEDKEKKE	229	10	ETALLSSGF (0.7567)	1.5783	HLA-A26
9	GEIEDVGE	239	8	VKDLVVLLF (0.8568)	1.2065	HLA-A1
						HLA-A24
10	DEEEDKKEKDKK	247	12	MKENQKQIY (0.9639)	0.9724	HLA-A1
11	LAVKHFSVEGQLEFRALLFV	306	20	TRGTKITLY (0.7913)	0.8785	HLA-A1
						HLA-B27
12	SAGGMPGG	613	8	YMTDPIDEY (1.7255)	0.8258	HLA-A1
						HLA-A26
						HLA-B62
13	MPGGMPSGAPGAGSTGGGPT	621	20	YSDNQPGVL (1.7835)	1.2493	HLA-A1
						HLA-B39
14	KSTAGDTHLGGEDFDNRMVN	220	20	YTNITRARF (0.9825)	0.8258	HLA-A1
						HLA-A26
						HLA-B58
						HLA-B62
15	AVITVPAYFNDSQRQATKDS	142	20	SSSQASIEI (0.7980)	0.5904	HLA-A1
						HLA-B58

Table 4.

Selection of MHC-II epitopes for constructions of PV_AO and PV_HSP (VaxiJen score > 0.50)

Name of vaccine constructs	MHC-II Epitope sequence	VaxiJen Score	MHC-II Binding Allele	No. of MHC-II binding allele
PVAO	KNFILGEKI	0.6758	DRB1_0101	2
			HLA-DPA10201-DPB1010
	DIKLRGPLS	2.9048	DRB1_0101	2
			DRB1_1602
	PLSVIGHSL	1.1870	DRB1_0101	2
			HLA-DQA10501-DQB10303
	LSVIGHSLV	0.6973	DRB1_0101	1
PVHSP	YIKITPNKA	0.7987	DRB1_0101	11
			DRB1_0401
			DRB1_0405
			BRB1_1602
			DRB3_0202
			DRB5_0101
			DRB1_0701
			DRB1_0802
			DRB1_1001
			DRB1_1101
			DRB1_1301
	ERIMKAQAL	0.5397	DRB1_0101	1
	AILSPAFKV	1.3767	DRB1_0101	1
			DRB1_0801	4
			DRB1_1001
			DRB1_1302
	VRILLTNLL	1.1471	DRB1_0101	10
			DRB1_0103
			DRB1_0404
			DRB1_405
			DRB1_1501
			DRB4_0101
			DRB1_0701
			DRB1_1302
			HLA-DPA10301-DPB10402
			HLA-DQA10102-DQB10501

Construction of multi-epitopic vaccine candidates

Two separate potential multi-epitope vaccines PV_AO consisting of GST, GPx and SOD epitopes and PV_HSP consisting of HSP70, HSP90 and HSP110 epitopes were constructed. The PV_HSP and PV_AO were constructed by combining highly antigenic CTLs (score > 0.5), high affinity B-cell epitopes (Threshold value = 0.50), and HTL epitopes (score > 0.75). A linker GPGPG connected HTL and B-cell epitopes and another linker AAY was used for adding CTL (MHC supertype A1) epitopes in the vaccine constructs. A 50 S ribosomal protein L7/L12, chosen as an adjuvant was added to the N-terminal of vaccine constructs through an EAAAK linker which increased the vaccine`s stability and immunogenicity (Fig. 1A).

Fig. 1.

Secondary structure of the candidate vaccines (A) Schematic diagram of polypeptide multi-epitope vaccine with adjuvant, epitopes, and linkers. The secondary structure features strand, helix, and coiled regions in (B) PVAO and (C) PVHSP, respectively

Prediction of discontinuous B-cell epitopes, promiscuous MHC class-I epitopes and IFN-γ inducing epitopes in candidate vaccine constructs

The IFN-γ epitopes of the PV_AO and PV_HSP vaccine constructs were predicted by IFNepitope server. Twenty three (23) IFN-γ epitopes were predicted for PV_AO and eighteen (18) IFN- epitopes were predicted for PV_HSP vaccine constructs (Table 5). The candidate vaccine PV_AO was predicted to contain 8 and PV_HSP to contain 3 discontinuous B-cell epitopes by the IEDB server (Supplementary Table 5). Promiscuous MHC class-I epitopes of HLA-A1, HLA-A2, HLA-A*0201 and HLA-A*0205 were predicted for both PV_AO and PV_HSP (Supplementary Table 6).

Table 5.

IFN-γ epitope of chimeric vaccine candidates of antioxidant and heat shock proteins of filarial parasite

Peptide	Sequence	Method	Result	Score
PVAO
	EGDIMGPGPGPIRLVL	SVM	NEGATIVE	-0.27343023
	YAAYYAAYAYDTEKDS	MERCI	POSITIVE	1
	VLSYKVEKAGRCDIH	SVM	NEGATIVE	-0.56939587
	LGEKISYVDFVLFEGP	SVM	POSITIVE	0.319
	LQEMHTRLGPGPGNQF	SVM	NEGATIVE	-0.4640616
	PGPGILPVELAKFEGP	SVM	NEGATIVE	-0.17574
	PVELAKFEGPGPGCLD	SVM	NEGATIVE	-0.406873
	GDRINASDWPSMKSHF	SVM	NEGATIVE	-0.13892
	PGPGMFCEGVRDLHTK	SVM	NEGATIVE	-0.2640
	MFSKIKGPGPGFLKSR	SVM	NEGATIVE	-0.3202906
	GPGPGAILRHLAGPGP	SVM	POSITIVE	0.32084744
	QIVQSGGPGPGAILRH	SVM	NEGATIVE	-0.17945
	PGCLDKFPLLKAYHQR	SVM	NEGATIVE	-0.01395
	GVNGSDADDLYKGPGP	MERCI	NEGATIVE	1
	PGIIYAAYVAEVYAAY	MERCI	NEGATIVE	4
	GKLLATRDGPGPGDGK	SVM	NEGATIVE	-0.47270355
	KDLVDGAPKPLLEKVA	MERCI	NEGATIVE	1
	PCLYDGDHQIVQSGGP	MERCI	POSITIVE	1
	LKTGNAGSRLACGPGP	MERCI	NEGATIVE	2
	PGRKHNLNGGNELETT	SVM	NEGATIVE	-0.575877
	PGPGDGKNFILGEKIS	SVM	POSITIVE	0.07909736
	VVREIVSGLGLKEAKD	MERCI	NEGATIVE	2
	YDTEKDSYAAYHHHHH	SVM	POSITIVE	0.82982
PV HSP
	QATKDSGPGPGAILSP	MERCI	NEGATIVE	1
	GSTGGGPTGPGPGKST	MERCI	NEGATIVE	3
	GETDDTKGGPGPGNQE	SVM	NEGATIVE	-0.59720259
	TDPIDEYAAYYSDNQP	MERCI	POSITIVE	1
	GMPGGGPGPGMPGGMP	SVM	POSITIVE	1.0220518
	QKQIYAAYTRGTKITL	SVM	NEGATIVE	-0.32601724
	GDTHLGGEDFDNRMVN	MERCI	POSITIVE	2
	MPGGMPSGAPGAGSTG	MERCI	NEGATIVE	1
	GDTHLGGEDFDNRMVN	MERCI	POSITIVE	2
	VEAAEEQSEFDVILEA	SVM	POSITIVE	0.31781947
	YAAYYSDNQPGVLAAY	SVM	POSITIVE	0.11582581
	GATYAAYNTNIKNTII	SVM	NEGATIVE	-0.061137575
	LGLKEAKDLVDGAPKP	MERCI	NEGATIVE	1
	YMSEEMNGETAAYQLE	SVM	NEGATIVE	-0.6340472
	DVILEAAGDKKIGVIK	SVM	POSITIVE	0.5297006
	RVRILLTNLLRESGKK	MERCI	POSITIVE	1
	PGPGLAVKHFSVEGQL	SVM	POSITIVE	0.18118755
	EVTAAAPVAVAAAGAA	MERCI	NEGATIVE	1

Prediction of toxicity, allergenicity, antigenicity, and physiochemical properties of the vaccine candidates

All the epitopes used in this study were found to be non-toxic (Supplementary Table 7). The antigenicity of PV_AO as predicted by VaxiJen v2.0 was 0.530 and by ANTIGENpro was 0.652, which showed the antigenic nature of this construct. The score of PV_HSP obtained by VaxiJen was 0.879 and by ANTIGENpro the score was 0.9234, which indicated that PV_HSP could be more antigenic than PV_AO. Both the PV_HSP and PV_AO were non-allergenic as predicted by AllerTOPv2. The predicted molecular masses from ExPASy-ProtParam serve were 82.24 kDa for PV_AO and 70.24 kDa for PV_HSP. The PV_AO and PV_HSP constructs were slightly acidic in nature as per the PI values. The GRAVY scores for PV_AO -0.438 and − 0.343 for PV_HSP suggesting the hydrophilic nature of these molecules (Table 6). The physicochemical properties such as a number of amino acids, molecular weight, formula, total no of atoms, estimated half-life, instability index, and aliphatic index, for both the constructs were satisfactory for their further development as vaccine candidates [40].

Table 6.

VaxiJen score, ANTIGENpro score, and Physiochemical properties of the vaccine candidates PV_AO and PV_HSP

Name of Vaccines	VaxiJen score	Antigen pro score	Solubility	No. of amino acids	Molecular weight (Daltons)	Theoretical isoelectric Point (PI)	Formula	Total no. of atoms	Estimate half life	Instability index	Aliphatic index	(GRAVY)
PPVAO	0.5304	0.6527	0.9018	807	82241.26	6.30	C 3663 3663 H 5635 5635 N 1025 1025 O 1104 1104 S 17 17	11,444	30 h (mammalian reticulocytes, in vitro). > 20 h (yeast, in vivo). > 10 h (E. coli, in vivo).	21.31	65.07	-0.438
PV HSP	0.8749	0.9234	0.9018	677	70247.17	5.17	C 3126 3663 H 4892 5635 N 830 1025 O 977 1104 S 16 17	9841	30 h (mammalian reticulocytes, in vitro). >20 h (yeast, in vivo). >10 h (E. coli, in vivo).	33.77	70.64	-0.343

Prediction of secondary and tertiary structures

The secondary structure of PV_AO and PV_HSP and the helix, strand and coil region of both constructs are compared in Fig. 1B and C. The secondary structure of the peptide vaccines was predicted by PSIPRED that works on the probable sequence produced by PSI-BLAST. PSIPRED works on the secondary structure with the highest score for generating a protein prediction in terms of helix, strands, and coil. PV_AO consisted of 56% helix, 0% beta-sheet, and 43% random coil, whereas, PV_HSP consisted of 37% helix, 14% beta-sheet and, 48% random coil. The candidate vaccine sequences were subjected to 3D structure prediction with I-TASSER and 5 structures were predicted for each peptide vaccines (Fig. 2). The PV_AO model with the highest C-score of 0.57 and PV_HSP model with the highest C-score of -1.94 was chosen for further studies. Additionally, the model topology was also validated by TM score and RMSD value (Supplementary Fig. 4). A negative “Z” score was calculated by the ProSA web server (PV_AO -2.61 and PV_HSP -5.37) which evidenced that the predicted vaccines scored within the range typically found for native proteins (Supplementary Fig. 5). Since the estimated ERRAT values were greater than 50 (PV_AO 78.907 and PV_HSP 76.097) (Supplementary Fig. 5), the improved 3D structures were considered acceptable to serve as trustworthy models for additional evaluations [61]. The Ramachandran plot analysis showed 84.4%, 11.8%, 2.4%, and 1.6% residues of PV_AO and 64.1%, 26.6%, 7.0%, and 2.3% residues of PV_HSP in most favored, additionally allowed, generously allowed, and disallowed regions respectively (Supplementary Fig. 6). ModRefiner server was utilized to validate 3D structure in terms of highest values for parameters like RMSD and TM score. The model 1 provided by ModRefiner server for PV_AO had RMSD and TM scores of 0.558 and 0.9967 respectively. Similarly for PV_HSP the model 1 had RMSD and TM scores as 0.438 and 0.9977 respectively (Fig. 2B and E). In both constructs, model 1 showed the best scores, hence these were chosen for further studies. Disulfide engineering is a crucial technique for incorporating disulfide bonds into a target protein to improve its stability, increase functional properties, and facilitate the comprehension of its dynamics. The PV_AO vaccine construct had 78 potential pairs for forming disulfide bonds while 60 potential disulfide sites were predicted for PV_HSP vaccine constructs (Supplementary Table 8).

Fig. 2.

Protein modeling, refinement, and validation of PV_AO and PV_HSP vaccine constructs. (A) Tertiary structure of PV_AO. (B) Refined structure of PV_AO. (C) Ramachandran plot illustrating the quality of PV_AO. (D) Tertiary structure of PV_HSP. (E) Refined structure of PV_HSP. (F) Ramachandran plot illustrating the quality of PV_HSP

Molecular docking of PVAO and PVHSP and calculation of binding energy

The involvement of TLR4s in lymphatic filariasis infection is well studied and TLR4 has been implicated for conferring protective immunity [62]. Hence the in-silico interactions of PV_AO and PV_HSP with TLR4 were examined by PatchDock server and visualized by Biovia Discovery Studio (Figs. 3 and 4). The top-scoring PV_AO-TLR4 complex had a GSC score of 19,308, Area 3202.80, and ACE 413.30, whereas the PV_HSP-TLR4 complex had a GSC score of 19,562, Area 3485.60, and ACE 484.59. The amino acids residues K119, D792, Q39, N58, D95, Q99, G120, Q129, L19, V42, E121, G150, G200, G734, D792, P28, V30, G96, S102, S123, K153, H179, L600 in PV_AO, and K74, K487, K153, G158, T319, T162, G373, K497, R264, N265, T319, H529, N531, S552, K153, G352, S353, G369, K487, S317, H458, Q507, H555, in PV_HSP formed hydrogen bonds with the TLR4 molecule (Supplementary Table 9). In summary, each vaccine candidates formed more than 20 hydrogen bonds with TLR4 molecule in the docked complex.

Fig. 3.

PV_AO vaccine constructs 3D structural analysis and molecular docking (A) Diagram of docking mode of the vaccine construct (green) and TLR4 (blue) complex. (B) Docked conformation and hydrogen bond interaction maps of the vaccine construct and TLR4. (C) The interacting residues between the docked vaccine construct (Chain A) and TLR4 (Chain B)

Fig. 4.

PV_HSP vaccine constructs 3D structural analysis and molecular docking (A) Diagram of docking mode of the vaccine construct (green) and TLR4 (blue) complex. (B) Docked conformation and hydrogen bond interaction maps of the vaccine construct and TLR4. (C) The interacting residues between the docked vaccine construct (Chain A) and TLR4 (Chain B)

Calculation of binding energies (MM/GBSA)

Molecular docking was also performed with HawkDock server for the analysis of the binding energy between the vaccine constructs and other TLRs such as TLR1/2, TLR4, TLR5, TLR6, and TLR9, as well as with MHC I and MHC II (Table 7). A higher docking score generally suggests a better binding affinity between the target receptors and the compounds or epitopes being tested. The MMGBSA scores obtained for TLR1/2, TLR6, TLR9, and MHC-II were in the range of -19.78 kcal/mol to -49.47 kcal/mol, which indicated that the candidate vaccines could also engage many TLRs effectively thus, leading to a stronger immune response against the filarial parasites.

Table 7.

Molecular Docking performed by HawkDock server, showing docking score and MM/GBSA energy

Vaccine		TLRs					MHCs
		TLR1/2	TLR4	TLR5	TLR6	TLR9	I	II
PVAO	PDB ID	2Z7X	3FXI	3J0A	4OM7	5GMH	7RK7	8JRK
	Docking Score	-5336.90	-6367.14	-5524.64	-3466.35	-5781.04	-6237.58	-5771.49
	MM/GBSA(kcal/mol)	-27.86	-20.48	-20.19	-19.78	-29.03	-39.65	-23.3
PVHSP	Docking Score	-9213.25	-4359.40	-7321.85	-3579.20	-6528.98	-5401.40	-4907.02
	MM/GBSA(kcal/mol)	-49.47	-19.28	-10.65	-30.06	-43.79	-8.26	-24.6

Normal mode analysis

The normal mode analysis of the docked MEV-TLR4 complexes has been carried out by the iMODS server. The NMA results gave information about domain mobility towards each other as shown with the arrow (Figs. 5A and 6A). The deformability plot indicated the flexible region of the complexes in the form of peaks (Figs. 5B and 6B). The B-factor represented atomic positional changes in the flexible region of the docked complexes (Figs. 5C and 6C). The eigenvalue of the PV_AO-TLR4 complex was 2.090283e-05 and eigenvalue of PV_HSP-TLR4 complex was 4.79102e-05 reflecting formation of stable structures (Figs. 5D and 6D). Eigenvalue is inversely related to variance as shown in Figs. 5E and 6E where purple color represents individual and green color represents cumulative variance. Connection between residues was deducted by covariance map (Figs. 5F and 6F), where red signified correlated, white signified un-correlated and blue anti-correlated residues.

Fig. 5.

A normal mode analysis of the PV_AO-TLR4 complex showing NMA mobility (affine-model arrows), deformability, B-factor, Eigenvalue, variance (purple color shows individual variance and green color shows cumulative variances), and co-variance map (red correlated, white uncorrelated, and blue anti-correlated)

Fig. 6.

Normal mode analysis of PV_HSP-TLR4 complex (A) NMA mobility (affine-model arrows), (B) deformability, (C) B-factor, (D) Eigenvalue, (E) variance (purple color represents individual variance and green color indicates cumulative variances), and (F) co-variance map (red correlate, white uncorrelated, and blue anti-correlated)

Molecular dynamics simulation analysis

In a simulated environment that is closely analogous to real systems, MD simulation is a well-accepted technique for effectively analyzing the stability of protein complexes. Molecular dynamics (MD) simulations were carried out for 100 ns using the Linux-based molecular dynamics program of SiBioLead server to further support the stability of the vaccine constructs. The SPC water model was applied to solvate the protein in addition to the OPLS/AA force field. It is crucial to reduce the global energy of molecules, hence to stabilize the conformations, an energy minimization step was added for the first 5000 steps. In order to learn more about the construct’s Radius of Gyration (Rg), Root Mean Square Deviation (RMSD), and Root Mean Square Fluctuation (RMSF), the simulation trajectories were plotted against time. During the MD simulation, it was observed that initially, both PV_AO and PV_HSP systems fluctuated slightly from 2 ns to 15 ns, after which the trajectories were stable over the entire run of 100 ns. The PV_AO system had a minimum RMSD value of 1.30 nm and a maximum of 1.75 nm with a total fluctuation of 0.45 nm (Fig. 7A). The RMSD value for the PV_HSP vaccine over a time period of 100 ns varied between 0.80 nm and 1.09 nm (Fig. 8A). The compactness and stability was assessed by Rg during the simulation run. Both the MD trajectories showed higher values for Rg at the beginning of the run however as the simulation progressed, the vaccines attained a much steady plateaued form that continued till the completion as shown in Figs. 7B and 8B.

Fig. 7.

Molecular dynamics (MD) simulation of PV_AO (A) Root mean square deviation (RMSD) plot. (B) Root mean square fluctuation (RMSF) plot (C) Radius of gyration plot. (D) Alteration in protein exposure to the solvent is shown by the complex’s Solvent Accessible Surface Area (SASA)

Fig. 8.

Molecular dynamics (MD) simulation of PV_HSP (A) Root mean square deviation (RMSD) plot. (B) Root mean square fluctuation (RMSF) plot. (C) Radius of gyration plot. (D) Alteration in protein exposure to the solvent is shown by the complex’s Solvent Accessible Surface Area (SASA)

The RMSF trajectory of the two candidate vaccines were plotted against time, and was observed that the amino acid residues of PV_AO from 80 to 110 (1.49 nm) was the highest and for 410 to 425 (0.71 nm) was the lowest in the entire run. In case of PV_HSP, the RMSF was maximal for residues 130 to 190 (1 nm) while residues 308 to 335 (0.89 nm) showed lowest fluctuations (Figs. 7C and 8C). Similar observations were also seen in the Solvent Accessible Surface Area (SASA) analysis of both PV_AO and PV_HSP (Figs. 7D and 8D).

Optimization of codon for peptide vaccine and in-silico cloning

To achieve maximal expression in E-coli (strain k12), the codon optimization was performed with Java Codon Adaptation Tool (JCat). The length of the optimized PV_AO codon was 2421 nucleotides and for PV_HSP 2031 nucleotides. The Codon Adaptation index (CAI) for PV_AO was 0.9497 and for PV_HSP was 0.9553, which was much above the threshold CAI of 0.5. In the next step, recombinant plasmid pET30a (+) was designed separately for PV_AO and PV_HSP (Fig. 9). The in-silico integration of PV_HSP in pET30a (+) expression vector between NdeI and XhoI restriction sites resulted in a product size of 7260 base-pair. Similarly, PV_AO was also inserted into the pET30a (+) expression vector between the NdeI and XhoI restriction sites resulting in a product of 7650 base-pairs.

Fig. 9.

In silico cloning of the PV_AO and PV_HSP (A) Cloning vector pET-30a (+) has multiple restriction sites. (B) Insert of PV_HSP vaccine constructs. (C) Insert of PV_AO vaccine constructs (D) Final product of PV_HSP (7260 bp) (E) Final product of PV_AO (7650 bp)

Immune simulation

Immunological simulations were done by the C-ImmSim server, which produced results that were in line with the real immunological reactions as evidence by the generation of secondary responses too. The primary response was associated with an elevation in the IgM levels. The secondary and tertiary responses showed significant increases in the B-cell population of IgG₁, IgG₂ IgM, and IgG⁺ IgM producing cells in the host. The simulation showed a marked decrease in antigen concentration after secondary immune response, which confirms generation of immunological memory (Figs. 10 and 11). The simulation programmed 12 injections of MEVs repeated for a period of 28 days to sufficiently increase IgG₁ levels and population of T-helper cells during the entire period of vaccination. Increase in the population of helper T-cell and cytotoxic T-cells are usually associated with the production of memory cells which is important for the effective clearance of antigens upon further exposure.

Fig. 10.

In-silico immune simulation of PV_AO in the C-ImmSim server (A) The primary B cell antibody produced after antigen stimulation is IgM + IgG (yellow) and changes over time. (B) Active B cell (purple) secretion after antigen stimulation. (C) The number and proportion of helper T cell subtypes. (D) Secretion from active epithelial cells (purple) after antigen stimulation (E, F) Changes in the secretion level of helper T lymphocytes and the secretion levels of helper T lymphocytes of different memory types. (G, H) Changes in the level of CD8 + T secretion after antigen stimulation and the secretion of different types of CD8 + T cells. (I) Expression of NK cells after antigen stimulation. (J) Expression of DC cells (black) after antigen stimulation. (K) Expression in Macrophages (black) after antigen stimulation. (L) Changes in secretion levels of cytokines, mainly IFN-γ (purple) and IL-2 (yellow)

Fig. 11.

In-silico immune simulation of PV_HSP in the C-ImmSim server. (A) The primary B cell antibody produced after antigen stimulation is IgM + IgG (yellow) and changes over time. (B) Active B cell (purple) secretion after antigen stimulation. (C) The number and proportion of helper T cell subtypes. (D) Secretion from active epithelial cells (purple) after antigen stimulation (E, F) Changes in the secretion level of helper T lymphocytes and the secretion levels of helper T lymphocytes of different memory types. (G) Changes in the number of regulatory T cell subtypes. (H, I) Changes in the level of CD8 + T secretion after antigen stimulation and the secretion of different types of CD8 + T cells. (J) NK cell expression after antigen stimulation. (K) Expression in Macrophages (black) after antigen stimulation. (L) Expression of DC cells (black) after antigen stimulation. (M) Changes in secretion levels of cytokines, mainly IFN-γ (purple) and IL-2 (yellow)

The administration of booster doses led to an improved immune response in terms of combined levels of IgG₁ and IgG₂, as well as levels of IgM and IgG₁ (Supplementary Table 10). The total IgG and IgM antibody titer for both vaccines was between 780,000 per ml and 700,000 per ml following the third booster dose. The IgM titer was around 650,000 per ml for PV_AO and construct vaccines, compared to 330,000 for PV_HSP vaccines, respectively. Additionally, with the PV_AO and PV_HSP construct vaccines, the combined IgG1 and IgG2 levels were 180,000 per ml and 400,000 per ml, respectively. In addition, a significant amount of IgG1 was produced in response to all vaccine booster shots (Figs. 10 and 11). The result for Simpson Index D analysis shows that the total B-cell response induced by PV_AO was higher than PV_HSP although both vaccines induced sufficient humoral memory and effector responses (Fig. 10 (B) and 11 (B)). The immunological response upon vaccinations is significantly influenced by interleukins and cytokines. After vaccination, the antigen-presenting cells (APCs) process the antigen and present them to the T-cells. This mechanism activates T-cells which subsequently release cytokines including IFN-γ and IL-2, leading to B-cell and T-cell proliferation and differentiation. Subsequently, the antibodies produced by B-cells can then neutralize the pathogen against which the vaccine was created. The levels of interleukins (IL) and cytokines significantly increased after the vaccinations were given. The IFN-γ levels increased gradually and peaked at about 1,000,000 ng/ml (PV_AO) and 425,000 ng/ml (PV_HSP), further increased levels of cytokines such IFN-γ, IL-2, IL-10, IL-12, and TGF-β were also seen in response to the booster doses.

Discussion

Till date, LF treatment and control strategies across the globe are completely dependent upon the success of MDA programs. Neither the MDA programs are cost-effective nor the drugs are adulticidal; hence, the development of an LF vaccine proves to be the best alternative to control filarial infection. Earlier vaccine candidates like HSP70 [63] and GST [21], though successful in the laboratory, could not be translated into commercial LF vaccines. Therefore the approach has now shifted to developing epitope based vaccines. Multi-epitope vaccines comprise of overlapping epitopes that are recognized by several TCR clones and can elicit humoral, cytotoxic, and helper T-cell immune responses simultaneously. MEVs also possess the capability to produce long term immunogenicity [45, 64]. Epitope-based vaccines offer several potential benefits, such as enhanced safety and the flexibility to strategically modify epitopes to enhance their effectiveness and range [12].

The filarial parasites are long-lived organisms surviving for several years inside the human body. Owing to their robust antioxidant enzymatic system, filarial worms easily defend themselves against the reactive oxygen species generated during the oxidative burst of the hosts’ macrophages, neutrophils, and eosinophils. The anti-oxidant enzymes such as, SOD, GST, and GPx are crucial for the survival of the parasites [21, 57, 65]. In previous studies, mice vaccinated with naked DNA constructs expressing GP_X, Cu/Zn cytosolic superoxide dismutase, demonstrated a significant level of protection in comparison to a control group against S. mansoni infections [66, 67]. Moreover, filarial GSTs have already been investigated as prospective LF vaccine candidates [68]. Another key player in the survival and development of filarial parasites are the HSPs. The filarial HSP’s are prominent immunogens and have been explored as potential vaccine candidates [28]. Filarial HSP70 from W. bancrofti and Brugia spp. have been investigated as an immunogenic proteins leading to the generation of high IgG₄ levels in humans [58]. Therefore, in this study an attempt was made to design two MEVs comprising of filarial AOs GST, GPx, and SOD (PV_AO) and HSPs 70, 90, and 110 (PV_HSP) and to compare their in-silico efficacies using immuno-informatics approaches.

Additionally, it seemed imperative to create an MEV that could offer cross-protection against other filarial parasites. The minimal identity percentage that might preclude self-peptide and epitope cross-reactivity has not been documented by any investigation. Therefore, non-homology of the filarial AO and HSP sequences with the human proteome was a key consideration in selecting the antigens initially [69]. The primary sequence of candidate vaccines PV_AO and PV_HSP had multiple CD8⁺, CD4⁺, and B-cell epitopes that could trigger a potent immune response which is very important for vaccine development since B-cells produce antibodies and also help CD8⁺ and CD4⁺ T-cells in generating long-lasting adaptive immunity [70]. The chosen epitope sequences demonstrated their strong immunogenic potential, as evidenced by the results of the ANTIGENpro server. The induction of allergenicity rather than an immunological response by vaccination is one of the main barriers to the development of vaccines; therefore, the list of epitopes was narrowed down to those that wouldn’t be expected to trigger an allergic reaction.

For designing the multi-epitope-based vaccination, each selected B-cell, CD8⁺ and CD4⁺ epitope was connected with its neighbors using GPGPG and AAY linkers. The sequences GPGPG and AAY linkers have been seen to produce very less immunogenicity [7]. Adjuvants are crucial parts of MEVs as boosting innate immunity is also critical to improving the therapeutic efficacy of a vaccine. Additionally, a 50 S ribosomal protein L7/L12 adjuvant was combined with the first MHC-I epitopes using the EAAAK linker. The presence of the EAAAK linker, imparts a distinctive ability to auto-cleave, hence allowing for the efficient isolation of the desired protein from the fusion protein without protease treatment. Also, the linker has proven cost-effective in protein purification [71]. In the next step, the preliminary 3D structures of the candidate vaccines PV_AO and PV_HSP were obtained with I-TASSER server. Further improvement to obtain a high-quality 3D model, was attained following loop refinement and energy minimization of the initial model with ModRefiner. The resulting Z-score of the improved modelled structure of PV_AO and PV_HSP were comparable to previous studies, indicating that the predicted structure corresponded well with the X-ray crystallography values of comparable-sized proteins [72].

The refined model of candidate vaccines was next subjected to prediction for IFN-γ epitopes, discontinuous and promiscuous B-cell epitopes. PV_AO and PV_HSP vaccine constructs were predicted to contain IFN-γ epitopes that can elicit IFN-γ secretion in the host. Earlier IFN-γ secretions have also proven useful in controlling filarial infections in clinical cases [73]. The discontinuous B-cell epitopes play an important role in humoral immune responses by eliciting the production of antibodies. PV_AO and PV_HSP had several discontinuous B-cell epitopes, showing that both the candidate vaccines could induce ample quantities of antibody production. Identifying promiscuous epitopes with robust binding affinity to various HLA (HLA-A1, HLA-A2, HLA-A*0201, and HLA-A*0205 supertype) alleles is essential for eliciting protective immunity, and both candidate vaccines had several promiscuous B-cell epitopes, hence confirming for extensive population coverage. Further, to confirm the antigenicity of the designed PV_AO and PV_HSP vaccines, VaxiJen, and ANTIGENpro servers were employed. When compared to other LF MEVs constructed using Aspartic protease and Thioredoxin proteins, the PV_HSP had the highest ANTIGENpro score (0.9234). Upon comparing the VaxiJen scores, Thioredoxin MEV demonstrated a slightly greater score of 0.9513 than PV_HSP (0.8749) which confirms that the MEVs designed in this study had sufficient immunogenicity too [19, 62].

The initiation and establishment of Lymphatic filariasis disease are significantly influenced by TLR-mediated responses [74, 75]. TLRs are a subclass of pattern recognition receptors (PRRs) that are expressed on the surface of many different cell types, including immune cells (such as DCs and macrophages) and non-immune cells (like fibroblasts and epithelial cells). A wide range of molecular patterns present on the pathogen also known as Pathogen Associated Molecular Patterns (PAMPS), are recognized by TLRs [76]. Immediate response to pathogens by the host immune system via TLR recognition involves the production of pro-inflammatory cytokines and amplified co-stimulatory signal. Furthermore, the development of new vaccines is largely dependent on TLR-mediated immune activation because TLRs trigger signaling cascades that produce type 3-interferon (INF-γ), which in turn is necessary for macrophage activation and elevated expression of MHC-I and MHC-II [77]. Subsequently, the candidate vaccines were studied for immune cell receptor interactions by the HawkDock server. Both PV_AO and PV_HSP showed stable binding, evidenced by high docking scores and binding energies with MHC-II, TLR1/2, TLR4, TLR6, and TLR9, which strongly indicate towards the generation of long term immunity against LF through multiple TLR mediated pathways as well as cytotoxic responses.

The most important and primary innate immune response against LF infection is known to be generated by TLR4 [78]. Therefore, the molecular docking of TLR4 with PV_AO and PV_HSP was accomplished by the PatchDock server to obtain detailed molecular interactions. The results clearly highlighted the importance of electrostatic and Van der Waal interactions in the binding of MEVs with TLR4 [62]. The MEVs and TLR4 complex was additionally stabilized because of the formation of several H-bonds (Figs. 3 and 4, and Supplementary Table 9). These findings show that PV_AO and PV_HSP could stably bind to the immune receptors, suggesting that MEVs might elicit a potent immune response in-vivo too. Moreover, the normal mode analysis from iMODS showed that the PV_AO and PV_HSP with TLR4 complex had sufficiently high eigenvalues, which establishes the lower level of deformability. Consequently, the deformability graph also showed a potential increase in complex rigidity for the designed vaccines. Non-uniform stiffness/immobility profiles, as well as high B-factor, deformability, and mobility indices, gave much flexibility to the PV_AO and PV_HSP vaccines.

The PV_AO and PV_HSP could be expressed in the E. coli system, as supported by the estimated half-life of PV_AO and PV_HSP (> 10 h for both PV_AO and PV_HSP), the index of instability (PV_AO 21.31 and PV_HSP 33.77), and the aliphatic index. The candidate vaccines’ predicted solubility score also added support towards their successful production. In-silico serological analysis for immune reactivity is the initial step in verifying a multi-epitope candidate vaccines; therefore, the recombinant protein must be expressed in an appropriate host, such as E. coli expression systems [50, 79]. Codon optimization for PV_AO and PV_HSP were done so that the recombinant vaccine protein is abundantly expressed in high levels in E. coli (strain K12). Both the codon adaptability index and the GC content for PV_AO and PV_HSP was 0.9497, 0.9553, and 57.001%, 53.126%, respectively, which is suitable for expression of MEVs in bacteria. Therefore, a well-characterized expression vector pET30a (+) was used for in-silico cloning of the MEV.

The stability of the PV_AO and PV_HSP constructs in a dynamic environment was assessed by molecular dynamics simulation using SiBioLead services. The proposed vaccines displayed constant conventional thermodynamic behavior, with RMSDs leveling up for less than 20 ns only. The RMSD estimates molecular deviation from the specified original/reference structure, which may be used to ensure ligand-target stability/confinement as well as the validity of the MD procedure [80]. Obtaining low RMSD values that are swiftly equilibrated has been linked to the effective convergence of the simulated models, requiring no additional molecular manipulation. In general, the C-RMSF flexibility analysis tool estimates the deviations of protein residues from their reference positions, providing an evaluation of protein residues in terms of their respective dynamic behavior as represented by flexibility and fluctuation. Within the vaccine’s planned structure, the addition of lengthy helices with flexible loop connections would rationalize the early relaxation and considerable convergence into more stable compacted conformations. Broadly, Rg is responsible for the global stability of protein ternary complexes, where the mass-weighted RMSD for atom groups in regard to their respective common center of mass is specified as a stability parameter. Therefore, the observed dynamic behavior of the simulated candidate vaccines to exhibit low Rgs’ demonstrates substantial stability and compactness of PV_AO and PV_HSP. The results of the SASA analysis for candidate vaccines were consistent with the above preferential complex stability as the simulated both PV_AO and PV_HSP showed a steadier SASA tone along the mean equilibration plateau and up to the end of the simulation. As SASA is a quantitative measurement of the interaction between protein and solvent, correlating for the molecular surface area (MSA) accessible to solvent, low SASA tones indicate relative structural reduction under the stress of solvent surface charges resulting in more compact, stable conformations.

The vaccine candidates designed must be capable of eliciting a potent immunological response from the host in order to be effective. The dynamics of the human immune system in response to the candidate vaccines was monitored by in-silico immunological simulation using the C-ImmSim server. It was observed that after exposure to PV_AO and PV_HSP, the immune responses were strongly stimulated and the levels of antibodies (IgG, IgM) were sufficiently high. As a result, the secondary immune response related to the injections of the built vaccine was also activated after the initial immune reaction was generated by the vaccine injections. Post-vaccinations, there was a significant rise in the concentration of several immune cells, including memory B-cells, plasma B-cells, helper T-cells, cytotoxic T-cells, and various types of antibodies. The results also indicated that the population of cytotoxic T-cells and helper T-cells was increased along with an increase in the population of memory cells [19]. Additionally, low Simpson indexes were predicted for PV_AO and PV_HSP by the C-ImmSim server, which is indicative of multiple forms of immune responses.

In this study, multiple approaches and servers were used to assess the immunogenicity of the predicted vaccines; still, in general multi-epitope vaccines suffer from certain drawbacks. It has been seen that the epitope prediction tools fail to appropriately take into account the necessity of identifying the appropriate antigen processing sites. For instance, the composition of antigen processing pathways differs according to pro-inflammatory signals and changes among distinct cell types; thus, present prediction algorithms may be inadequate for assessing the processing efficacy of parasite antigens in an infected target cell. Further in-vitro and in-vivo experiments like expressing the candidate vaccines in bacterial systems and validating their immunogenicity by diverse immunological assays are needed to ascertain the safety and efficacy before human administration.

Conclusion

The global burden of lymphatic filarial infections warrants better chemotherapeutic and diagnostic strategies as well as, effective vaccines too. In this work, two vaccines (PV_AO and PV_HSP) comprising several B-cell and T-cell (HTL and CTL) epitopes were designed using immuno-informatics methods. The peptide vaccines may have preventive and therapeutic advantages since the proteins bearing these epitopes are expressed in the microfilariae and infective stages of the parasite. The in-silico findings showed that the multi-epitope-based peptide vaccines were stable in 3D architecture and may have the capacity to elicit powerful immune responses against the filarial parasites. Moreover, the vaccine model substantially supports the capabilities to induce both humoral and cell-mediated immune responses. Based on antigenicity, physiochemical properties, and molecular dynamics simulation, PV_HSP turned out to be a better candidate vaccine than PV_AO. Though our in-silico analysis revealed a robust immunogenic response, in-vitro and in-vivo validation for immunogenicity and host safety is still required before human administration. Also, the modified vaccine candidates were anticipated to provide cross-protection, which will allow additional Neglected tropical disease control projects focused on related filarial diseases such as onchocerciasis and loiasis to reap benefits from the initiative.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

SK, AS, VK: Conceptualization, Designing, Data Curation, Study implementation and Analysis. SK, AS, AKS: Formal analysis, Writing, and editing of the manuscript. SK, AM, TS: Investigating, Methodology analysis. AS: Supervision, Project administration, and Validation. All authors read and approved the final manuscript.

Funding

No specific funding was received for this work.

Data availability

All data supporting the findings of this study are available within the paper and its Supplementary Information.

Declarations

Ethical approval

This study does not involve any human or animal subject.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.